MXPA06005387A - Ignition timing controlling device and method - Google Patents

Abstract

A method of controlling ignition timing of an engine. In the control method, final ignition timing for performing ignition is calculated by adding a variation component to a set ignition timing. According to the final ignition timing, an indicated average effective pressure of an in-cylinder pressure detected when ignition is performed is calculated. An ignition timing characteristic curve indicating the correlation between the indicated average effective pressure and the variation component is estimated and optimal ignition timing is calculated from the characteristic curve. Feedback control for converging the set ignition timing to the optimal ignition timing is then performed. Consequently, the ignition timing is controlled to an optimal ignition timing corresponding to a current operational state of the engine.

Description

APPARATUS AND METHOD TO CONTROL THE IGNITION TIME Field of the Invention The present invention relates to an apparatus and method for controlling the ignition timing of an internal combustion engine.

BACKGROUND OF THE INVENTION A method for detecting a pressure within a combustion chamber (referred to hereinafter as a pressure inside the cylinder) of the internal combustion engine (referred to hereinafter as an engine) to control the ignition time that has been proposed. According to the method shown in the Unexamined Publication of Japanese Patent Application (Kokai) No. 2003-262177, a difference of? P between the pressure Ptdc inside the cylinder in the upper dead center (TDC) and the maximum Pmax of the pressure inside the cylinder is compared to a threshold value. If the difference? P is less than the threshold value, the ignition time is advanced. It is preferred that the ignition be performed at an optimum ignition time. The optimal ignition time is usually called MBT (Minimum Advance for the Best Torque). The ignition at the optimum ignition time improves the efficiency of the combustion and the operation of purification of the exhaust gas. In general, the ignition times corresponding to various operating conditions of the engine are stored in a memory in the form of a map. We refer to the map based on the detected operating condition of the engine to determine the ignition time corresponding to the detected operating condition. In a vehicle comprising mechanisms, such as the timing mechanism of the valve and a variable compression ratio mechanism, the number of possible operating conditions of the engine is large, and hence the number of ignition times to be stored on a map be great. It can be difficult to define such a large number of ignition times on a map. Because recent vehicles comprise several parts associated with the engine, variations in the combustion condition can occur and deterioration over time can vary from part to part. Therefore, it can be difficult to set ignition times adapted to different parts. . If the optimal MBT ignition time corresponding to the current engine operating condition can not be determined, the ignition time needs to be restarted as to avoid the drift. The control to delay the ignition time can reduce the combustion efficiency. According to the prior art, the ignition time is moved gradually to the optimum ignition time while periodically comparing the pressure inside the cylinder and a threshold value. Because this scheme takes time to cause the ignition timing to converge at the optimum ignition timing, combustion efficiency can be reduced. Therefore, there is a need for an apparatus and method for estimating the optimal ignition time MBT corresponding to the current operating condition of the engine and then causing the ignition timing to rapidly converge to the estimated optimum ignition timing.

SUMMARY OF THE INVENTION In accordance with an aspect of the present invention, an apparatus for controlling the ignition timing of an engine is provided. The apparatus comprises an ignition timing calculator for adding a fluctuation component to a set of ignition times to calculate a final ignition time to ignite the engine, an effective calculating means of pressure to calculate an average effective pressure indicated for a pressure within the cylinder detected when the engine has been turned on according to the final ignition time, a MBT calculator to estimate the ignition timing characteristic curve representing a correlation between the indicated average effective pressure and the fluctuation and to determine the optimal ignition time of the characteristic curve and a controller to control the ignition timing adjustment to converge on the optimal ignition timing. In accordance with the present invention, by adding the fluctuation component to the ignition time, the optimum ignition time corresponding to the current operating condition of the engine can be determined. The convergence of the ignition time to the optimum ignition time increases the pressure inside the cylinder and prevents the combustion efficiency from being reduced. Because the optimal ignition timing corresponding to the current operating condition can be determined, it is not required to store a large number of ignition times in a memory in advance.

According to one embodiment of the present invention, the ignition timing characteristic curve is represented by a function. An input of the function is the fluctuation component and an output of the function is the indicated average effective pressure. The MBT calculator further includes an identifier to identify coefficients associated with the fluctuation component in the function based on the indicated average effective pressure calculated by the average effective pressure calculator. The characteristic curve is estimated based on the identification of the coefficients. Therefore, the coefficients included in the function are identified in a more correct way, thus improving the accuracy of the estimation of the characteristic curve. In accordance with one embodiment of the present invention, the apparatus further comprises a generator for generating the jitter component. The generator generates the fluctuation component to cover self-excitation conditions to identify the coefficients of the function. In one example, the number of self-excitation conditions is equal to or greater than a value obtained by adding one to the number of coefficients to be identified. Therefore, a signal is generated in an appropriate manner to estimate the characteristic curve. According to one embodiment of the present invention, the identifier is further configured to determine update components for the coefficients, such that an error between the indicated average effective pressure calculated by the average effective pressure calculator and an estimated average effective pressure estimated from the function converge to zero, add the update component to the previously determined reference values to determine the coefficients. The coefficients converge on the reference values as the error converges to zero. The reference values are previously determined, so that the control to control the ignition timing adjustment that converges to converge at the optimum ignition time is stopped when the coefficients have converged with the reference value. According to the present invention, when a real average effective pressure is equal to the estimated average effective pressure of the characteristic curve (ie, when the error identification methods approach zero), the coefficients converge to the values of reference, avoiding in this way, the deviation of the coefficients. In addition, because the reference values are set so that the feedback control for the on-time is stopped when the coefficients have converged on the reference values, an erroneous identification is prevented. According to one embodiment of the present invention, a limiting process is applied to at least one of the coefficients, so that the characteristic curve is prevented from being estimated as a convex curve downward. When the ignition timing setting has converged near the optimal ignition time, the curvature of the estimated characteristic curve is flat. According to the present invention, it is avoided that in said condition the curvature of the characteristic curve is estimated in a wrong way. According to one embodiment of the present invention, the average effective pressure calculator is configured to extract an alternating component of the pressure within the detected cylinder and to calculate the indicated average effective pressure based on the alternating component. Therefore, even if the influence caused by the pyroelectric effect and the thermal deviation appear at the outlet of the pressure sensor inside the cylinder, this influence is eliminated from the determination of the average effective pressure indicated. Therefore, a ceramic-type piezoelectric element can be used for the pressure sensor inside the cylinder. further, the pressure sensor inside the cylinder can be placed near the wall of the motor cylinder. In accordance with one embodiment of the present invention, the controller uses a response allocation control to control the ignition timing adjustment. The response allocation control has the ability to specify the characteristic response of the ignition timing setting to the optimal ignition timing. Therefore, the ignition timing adjustment converges on the optimum ignition timing without excessive blasting. It prevents the combustion efficiency from being reduced, because the ignition time is not required to be retarded or advanced excessively.

Brief Description of the Figures Figure 1 is a block diagram showing an engine and its control unit according to an embodiment of the present invention. Figure 2 is a diagram for explaining a general principle of ignition timing control according to an embodiment of the present invention. Figure 3 is a block diagram showing the ignition timing control apparatus according to an embodiment of the present invention. Figure 4 shows a map for specifying a reference value of the ignition timing according to an embodiment of the present invention. The fi re 5 shows a relation between a volume and a pressure inside the cylinder of a combustion chamber. Figure 6 shows characteristics of a first order filter and a second order filter according to an embodiment of the present invention. Figure 7 is a diagram for explaining a method for extracting a first order component and a second order component of the pressure inside the cylinder according to an embodiment of the present invention. Fig. 8 is a diagram for explaining the effect of calculating an indicated average effective pressure based on an alternating component of the pressure within the cylinder according to an embodiment of the present invention. Figure 9 shows a relationship between the ignition time and an indicated average effective pressure. Fig. 10 shows a waveform of a jitter signal according to an embodiment of the present invention. Figure 11 is a diagram for explaining the fluctuation of the ignition timing caused by the jitter signal according to an embodiment of the present invention. Figure 12 shows an estimated turn-on time curve and an optimal turn-on time calculated according to an embodiment of the present invention. Figure 13 is a diagram to explain a reason for performing a process of limits on one of the coefficients to be identified according to an embodiment of the present invention. Figure 14 shows a change function in a response allocation control according to an embodiment of the present invention. Figure 15 shows a rate of convergence of a controlled variable specified by a response allocation parameter in a response allocation control according to an embodiment of the present invention. Figure 16 shows an average effective pressure indicated when a feedback control is not performed for an optimal ignition time. Figure 17 shows the behavior of several parameters when the feedback control for the optimal ignition time is performed in accordance with an embodiment of the present invention. Fig. 18 is a flow chart of a main routine of an ignition timing control according to an embodiment of the present invention. Fig. 19 is a flow chart of a feedback control for an optimal turn-on time according to an embodiment of the present invention. Figure 20 is a flow chart of a sample process for the pressure inside the cylinder according to an embodiment of the present invention. Detailed Description of the Invention. Structure of an engine and a control unit With reference to the drawings, the specific embodiments of the present invention will be described.

Fig. 1 is a block diagram showing an engine and a control unit for the engine according to an embodiment of the present invention. An electronic control unit (hereinafter referred to as an ECU) 1 comprises an input interface for receiving data sent from each part of the vehicle, a CPU Ib for carrying out operations to control different parts of the vehicle, a memory lc including a read-only memory (ROM) and a random access memory (RAM) and an output interface Id for sending a control signal to different parts of the vehicle. The programs and the different data to control each part of the vehicle are stored in the ROM memory. A program and the data to implement the ignition timing control according to the present The invention is stored in the ROM memory. The ROM memory can be a rewritten memory such as an EPROM memory. The RAM provides the work areas for the CPU Ib operations, in which data sent from each part of the vehicle, as well as . as the control signals that are going to be sent to each part of the vehicle are stored temporarily. A motor 2 is, for example, a 4-cycle motor. The motor 2 is connected to an intake manifold 4 to through an intake valve 3 and connected to an exhaust manifold 6 through the exhaust valve 5. A fuel injection valve 7 is provided on the intake manifold 4 for each cylinder. The fuel injection valve 7 injects 5 fuel according to a control signal of the ECU 1. The engine 2 introduces into the combustion chamber 8 a mixture of intake air of the intake manifold 4 and the fuel injected from the fuel injection valve 7 A spark plug 9 is provided to generate a spark according to an ignition time signal from the ECU 1 to the combustion chamber 8. The spark of the spark plug 9 causes combustion of the mixture. The combustion increases the volume of the mixture, thereby driving the piston 10 downwards. The reciprocal movement of the piston 10 is converted into a rotational movement in the crankshaft 11. A pressure sensor inside the cylinder 15 is, for example, a sensor comprising a piezoelectric element. The pressure sensor inside the cylinder 15 is embedded in the connection portion between the cylinder and the spark plug. The pressure sensor inside the cylinder 15 produces a signal Pcyl inside the cylinder corresponding to the pressure inside the combustion chamber 8. The signal inside the cylinder is sent to the ECU 1. A crank angle sensor 17 is connected in the 2. The crank angle sensor 17 produces a CRK signal and a TDC signal to the ECU 1 in accordance with the rotation of the crankshaft 11. The CRK signal is a pulse signal that is produced at each previously determined crank angle (for example, example, 15 degrees). The ECU 1 calculates a rotation speed NE of the motor 2 according to the signal CRK. The signal TDC is also a pulse signal that is produced at a crank angle associated with a position of the TDC (upper dead center) of the piston 10. An acceleration valve 18 is located in the intake manifold 4 of the engine 2. A The opening degree of the acceleration valve 18 is controlled by a control signal of the ECU 1. The acceleration valve opening sensor (? TH) 19, which is connected to the acceleration valve 18, supplies the ECU 1 the electric signal corresponding to the opening angle of the acceleration valve 18. An intake manifold pressure sensor (Pb) 20 is placed below the acceleration valve 18. The pressure Pb of the intake manifold detected by the Pb sensor 20 is sent to the ECU 1. An air flow meter (AFM) 21 is positioned above the acceleration valve 18. The air flow meter 21 detects the amount of air passing through the valve d e acceleration 18 and sends it to the ECU 1. An accelerator opening sensor 25 is connected to the ECU 1. The accelerator opening sensor 25 detects a degree of opening of the accelerator pedal and sends it to the ECU 1. A mechanism (not shown) for variably operating a quantity of phase and / or elevation of the intake valve and / or the exhaust valve. A mechanism (not shown) can be provided to change the compression ratio of the combustion chamber. The signals sent to the ECU 1 are passed to the input interface la. The input interface 5a converts the analog signal values into digital signal values. The CPU Ib processes the resulting digital signals, performs operations according to one or more programs stored in the memory lc and creates control signals. The output interface Id sends these control signals to the actuators for the fuel injection valve 7, the spark plug 9, the acceleration valve 18 and other mechanical components. General Principle of the Invention For reasons of easier understanding of the present invention, the general principle of the invention will now be described first. Referring to Figure 2, the vertical axis indicates an average effective pressure indicated in the pressure inside the cylinder. The horizontal axis indicates the on time. The characteristic curve 31 indicates a correlation between the average effective pressure indicated and the ignition time. A method for calculating the indicated average effective pressure will be described later. As shown in the figure, the characteristic curve 31 has a maximum value 32. The ignition time corresponding to the maximum value 32 is called an optimal ignition time MBT. It is assumed that the currently set ignition time is IG1. In accordance with one embodiment of the present invention, a fluctuation component is added to the ignition time IG1. Said addition of the fluctuation component causes the ignition timing to fluctuate within a predetermined range. In one embodiment, as shown by arrow 33, the fluctuation component is determined, so that the ignition time fluctuates between more and less with respect to IG1 over continuous cycles. Therefore, the average effective pressure indicated is acquired when the ignition time is fluctuated. Portion 34 (solid line) of characteristic curve 31 corresponding to range 33 is acquired based on the fluctuation component and the indicated average effective pressure acquired. A shape of characteristic curve 31 is calculated from portion 3. The optimal ignition time MBT is determined from the estimated characteristic curve 31. The ignition time is controlled so that it converges on the optimum ignition time MBT. According to the present invention, because the optimal ignition time MBT is determined according to the current operating condition of the engine, the ignition timing can quickly converge to the optimum ignition time MBT. By causing the ignition timing to rapidly converge to the optimum ignition time MBT, the pressure inside the cylinder is maximized and combustion efficiency is reduced. Furthermore, in accordance with the present invention, it is not required that a large number of ignition times corresponding to different operating conditions of the engine and various parts associated therewith be stored in a memory in advance. Ignition time control apparatus Figure 3 shows a block diagram of an ignition timing control apparatus according to an embodiment of the present invention. The functions of each block are usually implemented by one or more programs stored in the memory lc. Alternatively, these functions can be implemented by any type of hardware. A jitter signal generator 41 generates a DIGID jitter signal. The jitter signal DIGID indicates the jitter component as described above with reference to Fig. 2. The DIGID jitter signal is passed to a start-up time signal generator 42. In one embodiment, the value of the signal of fluctuation DIGID can be stored in advance in the memory lc. The ignition timing signal generator 42 refers to a map based on the current detected operating condition of the engine to determine an IGBASE reference value for the ignition timing. In this mode, the IGBASE reference value is determined by referring to a map based on the detected intake air quantity Gcyl and the detected engine rotation speed NE. Figure 4 shows an example of said map, which can be stored in the memory lc of figure 1. Preferably, the map specifies the ignition times only for the typical operating conditions of the engine. Ignition times for a large number of operating conditions are not required on the map. It must be understood that said map is not necessarily required, as will be described later. However, said map is available, a convergence speed of the optimal ignition time MB can be improved. The amount of intake air Gcyl is calculated according to equation (1). In the equation, Gth represents a value detected by the air flow meter 21 (figure 1). Pb represents a value detected in the pressure sensor of the intake manifold 20 (figure 1). Vp represents a volume (m3) of the intake manifold. Tb represents a temperature (K) of the intake manifold. R represents the gas constant. "k" which is used to identify a control cycle, (k) indicates the current control cycle and (k-1) indicates the previous control cycle. In the following description, the processes are performed in a control cycle represented by "k" in synchronization with the combustion cycle (for example, the combustion cycle of a four-cycle engine is from 0 to 720 degrees of the crank angle ).

The ignition timing signal generator 42 calculates an IGLOG ignition time signal by adding the jitter signal DIGID received from the jitter signal generator 41 to a sum of the IGBASE reference value and a correction value DIGOP, as shown in equation (2). The spark plug 9 (figure 1) is operated according to IGLOG ignition time signal. IGLOG = IGBASE + DIGOP + DIGID (2) The DIGOP correction value is used to cause the ignition timing to converge on the optimal ignition time MBT. It should be noted that the DIGID fluctuation signal is included in the IGLOG ignition time signal. Incidentally including said DIGID fluctuation signal in the ignition time signal ISLOG, (IGBASE + DIGOP) that causes it to fluctuate within a predetermined range. In the following description, we can refer to (IGBASE + DIGOP) as an adjustment of the on time. The ignition timing adjustment is based on the current operating condition of the engine and is an object controlled by a control to cause the ignition timing to converge at the optimum ignition timing.

As described above with reference to Figure 2, the DIGID jitter signal is aggregated so that the ignition timing fluctuates within a predetermined range in relation to the ignition timing adjustment. It is preferred that the DIGID jitter signal be generated, so that the overall jitter signal does not thereby cause a large variation in the combustion condition. When an engine has been ignited in accordance with the IGLOG ignition timing signal including the DIGID jitter signal, the pressure inside the Pcycle cylinder is detected by the pressure sensor inside the cylinder 15. An average effective pressure calculator 43 calculates a Average effective pressure indicated Pmi based on the pressure inside the cylinder Pcyl. An MBT calculator 44 calculates a characteristic curve of an ignition time based on the indicated average effective pressure Pmi_act and the fluctuation signal DIGID corresponding to the indicated average effective pressure Pmi_act. The optimal ignition time MBT is calculated from the estimated characteristic curve. An ignition timing controller 45 calculates the DIGOP correction value described above, so that the ignition timing adjustment converges to the optimum ignition time MBT. In this mode, a sum of the IGBASE reference value and the DIGOP correction value is controlled to converge on MBT. The use of the reference value has the following advantage. The operating condition of the motor can change abruptly. If the reference value corresponding to the operating condition detected after said change is used, the controller 45 can cause the ignition timing to converge more rapidly with the optimal ignition time MBT. However, alternatively, the controller 45 may be configured to calculate the ignition timing in each control cycle so as to cause the ignition timing to converge to the optimum ignition time without using said reference value. Average Effective Pressure Calculator Referring to Figure 5, the indicated average effective pressure will be described. Figure 5 shows a relation between a volume of the combustion chamber of the engine and a pressure inside the cylinder of the combustion chamber of the engine. At point P, the intake valve is opened to initiate the intake stroke. The pressure inside the cylinder decreases to a point U where the pressure is minimal through a point N corresponding to the upper dead center TDC of the piston. Then, the pressure inside the • cylinder increases through a point K corresponding to the lower dead center BDC. A compression stroke starts at a point Q. The pressure inside the cylinder increases during the compression stroke. At point R, initiates a combustion path and the pressure inside a cylinder increases abruptly due to the combustion of the air-fuel mixture. At point S, the pressure inside the cylinder is maximum. The piston is driven downwards due to the combustion of the air-fuel mixture. The piston moves to the BDC shown by a point N. The pressure inside the cylinder decreases according to this movement. At a point T, the exhaust valve is opened to initiate an escape route. The pressure inside the cylinder also decreases during the exhaust stroke. A value obtained by dividing the area surrounded by the pressure curve inside the cylinder, as illustrated in Figure 5, by the volume of the piston stroke is what we refer to as an indicated average effective pressure. In an alternative embodiment, the average effective pressure from the point K corresponding to the BDC to the point M corresponding to the BDC through the point L corresponding to the TDC can be calculated as an indicated average effective pressure. In one embodiment of the present invention, the average effective pressure calculator 46 samples the detection value of the pressure sensor inside the cylinder at each previously determined crank angle (15 degrees in this mode). A pressure inside the marked cylinder represented by Pcyl (n). "n" indicates a sampling cycle. The average effective pressure calculator 46 calculates an indicated average effective pressure Pmi_act according to equation (3). The equation shows a method to calculate the indicated average effective pressure Pmi_act by extracting only the alternating components of the pressure inside the cylinder. The detail of this method is shown in the Japanese Patent Application Examined Publication (Kokoku) No. H08-20339. In equation (3), "h" is a coefficient according to the number of courses required for a combustion cycle, in the case of a four-cycle engine, h = l / 2 and in the case of an engine of two cycles, h = l. "?" is a ratio of a length? s "of the connecting rod to a radius" r "of the crankshaft. That is,? = S / r.

Cl represents an amplitude of a first-order component of the speed of rotation of the motor (i.e., an amplitude of a frequency component corresponding to the speed of rotation of the motor) in the pressure inside the cylinder Pcyl. fl represents a phase error with respect to the TDC of the first order component of the speed of rotation of the motor in the pressure inside the cylinder Pcyl. C2 represents an amplitude of a second-order component of the speed of rotation of the motor (i.e., an amplitude of a frequency component corresponding to twice the speed of rotation of the motor) in the pressure inside the cylinder Pcyl. f2 represents an error of phase with respect to the TDC of the second order component of the speed of rotation of the motor in the pressure inside the cylinder Pcyl. As described above, k represents a control cycle in synchronization with the combustion cycle. The calculation of the average effective pressure indicated Pmi_act is carried out in each combustion cycle. p (1 ^ Pmi _ act (k) = - ^ - Cí (k) cos (fl (k)) + - C2 (k) cos (f2 (k)) (3) 2h v In this way, the average effective pressure calculator 43 calculates the indicated average effective pressure Pmi_act based on the alternate components (the first order component and the second order component of this mode) of the pressure within the cylinder Pcyl. A method for extracting the first order component and the second order component of the pressure inside the cylinder Pcyl will be described. It should be noted that this method of extraction is different from that of the Examined Publication of Japanese Patent Application (Kokoku) No. H08-20339 to which reference was made above. The method shown in this publication extracts the component using analogous filters. In contrast, the method according to one embodiment of the present invention extracts the components using digital filters. The average effective pressure calculator 43 applies a first order filter and a second order filter to the cylinder center pressure Pcyl (n) sampled as shown in equations (4) and (5), respectively. As described above, "n" indicates a sampling cycle that is synchronized with the previously determined crank angle (e.g., 15 degrees).

Pcylod \ (n) = aodí 1 • Pcylodí (n - 1) + aodll • Pcy? Odí (n - 2) + aodYh • Pcylodl (n - 3) + aod? A • Pcylodí (n - 4) + bodl • Pcyl (?) + bodl 1 • Pcyl (n - 1) + bodí2 • _P? y / («- 2) + bodU • Pcyl (n - 3) + bodlA • Pcyl (n - 4) (4) filter coefficients: aod \ i (i = 1 4), bod \ j (j = 0 5) Pcylod2 (ri) = aod 21 • Pcylod \ (n - 1) + ao 22 • Pcylod \ (n - 2) + aoc 23 • Pcylod \ (n - 3) + ao¿ 24 • Pcylod \ (n - 4) + bod 20 • P? and / + bod 21 • Pcv («- 1) + boí 22 • cj / ^ - 2) + bo¿23 • cj; 7 (?? - 3) + bod24 • cy C» - 4) (5) filter coefficients: aod2i (i = 1 4), bod2j (j = 0 5) The characteristics of these digital sites are shown in Figure 6 (a) and 6 (b), respectively. The first-order filter (a) is a bandpass filter that has the characteristic of extracting the first-order component from the speed of rotation of the motor. The second-order filter (b) is a bandpass filter that has a characteristic of extracting the second-order component from the rotation speed of the motor. A horizontal axis represents a frequency that is normalized by a Nyquist frequency. The Nyquist frequency changes according to the rotation speed of the motor, because the pressure inside the cylinder Pcyl is sampled in a cycle that is synchronized with the speed of rotation of the motor. This normalization by the Nyquist frequency makes it possible for the first-order component and the second-order component of the motor rotation speed to be extracted from the cylinder pressure Pcyl without changing the coefficients of these filters, regardless of the current value of the speed of engine rotation. According to a method to apply the filters in a constant interval of time, the pass band can become an extremely low frequency when the rotation speed of the motor is low and hence the filter coefficients can become extremely small. This can make the filter outputs unstable. In the above method according to one embodiment of the present invention, said phenomenon can be avoided, because the filters are applied in synchronization with the rotation speed of the motor. An analogous waveform representing the Pcylodl (n) that is obtained by applying the first-order filter is expressed as Cl ° sin (? Ne + fl). An analogous waveform representing the Pcylod2 (n) that is obtained by applying the second order filter is expressed as C2 ° sin (2? Ne + f2). In this case,? Ne represents the angle of rotation of the motor that has a value of 0 to 2 p. When the piston is placed in TDC,? Ne = Orad. Cl ° cos (fl) and C2 ° cos (f2) in equation (3) can be expressed as shown in equations (6) and (7), respectively.

Clcos ^ l) = Clsin (- + fl) (6) C2cos ($ * 2) = C2sin (- + ^ 2) = C2sin (2 (-) + f2) (7) As observed by comparing the first order filter output Cl ° sin (? Ne + fl) and equation (6), the first-order component Cl ° cos (fl) of equation (3) can be obtained sampling the output of the first order filter when the rotation angle of the motor is p / 2 (ie, when? ne = p / 2). Similarly, as can be seen from the comparison between the second order filter output C2 ° sin (2? Ne + f2) and equation (7), the second order component C2 ° cos (f2) of Equation (3) can be obtained by sampling the output of the second order filter when the angle of rotation of the motor is / 4 (that is, when? ne = p / 4). In this case, referring to Figure 7, the pressure inside the cylinder Pcyl, the analogous waveform Cl ° sin (? Ne + fl) of the first order filter output and the analogous waveform C2 ° are shown sin (2? ne + f2) of the second order filter output. The horizontal axis indicates the crank angle. The analogous waveform Cl ° sin (? Ne + fl) and the analogous waveform Cl ° sin (? Ne + fl) are shown by bold lines. For comparison reasons, Cl ° without (? Ne) and C2 ° without (2? Ne) are shown by thin lines. As described above, a value 51 that is obtained by sampling the first order filter output Cl ° sin (? Ne + fl) when? Ne = p / 2 (ie, sampling 90 degrees after the TDC) represents the component of first order Cl ° cos (fl) of the equation (3) . A value 52 that is obtained by sampling the second order filter output C2 ° sin (2? Ne + f2) when? Ne = p / 4 (that is, sampled at 45 degrees after TDC) represents the second order component C2 ° cos (f2) of equation (3). Therefore, by sampling and maintaining the output of the first order filter and the second order filter output at the previously determined crank angles, the indicated average pressure Pmi_exact may be calculated according to equation (3). Alternatively, the internal cylinder pressure sampling cycle can be adjusted to an integral submultiple of 45 degrees crank angle (for example, sampling can be performed every 5 or every 3 crank angles) because it is sufficient if The outputs of the digital filter can be sampled at 45 and 90 degrees after the TDC. Referring to Figure 8, the effect of the above-described method for calculating the indicated average effective pressure by extracting only the alternating components of the pressure within the cylinder will be described. A waveform 55 shows the detection value of the pressure sensor inside the onboard cylinder 15 as shown in Figure 1 when the sensor is mounted on a vehicle. The waveform 56 shows the detection value of a sensor that is used for experimental purposes. The pressure sensor inside the cylinder for the experiment is provided to directly touch the air-fuel mixture inside the combustion chamber. A piezoelectric element provided in the pressure sensor inside the cylinder for the experiment is made of a single crystal, which is very expensive. On the other hand, the piezoelectric element used for the pressure sensor inside the cylinder on board is generally made of polycrystalline ceramics from the point of view of cost and durability. Because the on-board sensor is mounted on a vehicle, it can be difficult to maintain the temperature of the piezoelectric element at a constant level. Accordingly, as seen by comparison with waveform 56 representing the output of the sensor used for the experiment, waveform 55 for the on-board sensor has a "derivation" in the detection value Pcyl which is caused by the pyroelectric effect and the deviation of heat In order to avoid such deviation, it is required that the piezoelectric element be made of a single expensive crystal and that the sensor be placed far from the combustion chamber as to avoid the influence of the temperature inside the combustion chamber caused by variations in the operating conditions of the engine., this adaptation of the sensor is of high cost. In addition, the S / N ratio may decrease because the absolute value of the sensor output becomes small. The frequency components that may appear due to the pyroelectric effect and the heat deflection are slower than the first order component. According to one embodiment of the present invention, said unwanted frequency components can be eliminated, because the indicated average effective pressure Pmi_act is calculated based on the alternate components of the pressure inside the cylinder. As shown in Fig. 8 (b), the indicated average effective pressure Pmi_act (shown by waveform 57) which is calculated based on the detection value of the pressure sensor inside the onboard cylinder 15 exhibits almost the same value than the indicated average effective pressure (shown by waveform 58) that is calculated based on the sensing value of the sensor used for the experiment. MBT fluctuation signal generator and calculator FIG. 9 shows a diagram similar to FIG. 2. A characteristic curve 71 of the ignition timing has a maximum value 72. The ignition timing corresponding to the maximum value 72 is the optimum ignition time. MBT. Because the actual combustion conditions vary in each combustion cycle, the average effective pressure indicated Pmi_act is generally distributed within a range as shown by the shaded area 74 having a width 73. In an environment for testing the engine, a characteristic curve 71 can be obtained by measuring the average effective pressure indicated while changing the ignition time from "retarded" to "advanced". However, performing this operation when the vehicle is actually traveling can cause the degradation of the handling capacity. If the ignition time is set to a value (IG1, for example) extracted from a map as in conventional schemes, the average effective pressure indicated is distributed on a line 75. The shape (curvature and inclination) of characteristic curve 71 can not be estimated from said one-dimensional distribution of the average effective pressure indicated. In order to calculate the ignition characteristic curve 71 without reducing the handling capacity, according to one embodiment of the present invention, the jitter signal generator 41 is introduced as explained above with reference to Figure 3. The jitter signal generator 41 'generates a jitter signal that covers the auto-excitation conditions to calculate the characteristic curve 71. The number of auto-excitation conditions is equal to or greater than a value that is obtained by adding one to the number of coefficients contained in a function that expresses characteristic curve 71. The function will be described later. In this mode, because three coefficients are included in the function that expresses the characteristic curve 71, the number of auto-excitation conditions (PE) is set to four. Therefore, the jitter signal generator 41 generates a DIGID signal by combining three sine waves as shown in Fig. (8). dl, d2 and d3 represent the respective amplitudes,? l,? 2 and? 3 are adjusted to be equal to an integral submultiple of the control frequency (in this mode, a frequency corresponding to the combustion cycle). f and f 'represent respective phases. These parameters are previously determined. DIGID (k) = dhsm (? Lk) + d2 -sm (? 2 -k +?) + Dl-sm (? 3 -k +? ') (8) Alternatively, the DIGID fluctuation signal can be generated to cover five or more conditions of auto excitation. For example, a random wave containing a finite number of sine waves can be generated as the DIGID jitter signal. Or, the DIGID jitter signal can be generated as a series of impulse signals (eg, M-sequence). Figure 10 shows an example of the waveform of the DIGID jitter signal. The horizontal axis indicates the value of a Cdigid counter. The DIGID fluctuation signal is generated so that it has the cycle length of Cdigid__max. The fluctuation signal DIGID corresponding to each value of the Cdigid counter can be stored as a map in the memory lc (figure 1). The value of the counter is increased in each control cycle. The fluctuation signal DIGID corresponding to the value of the counter is extracted from the map. If the value of the counter reaches the Cdigid_max, the counter is reset to zero. The reference number 77 indicates a range of the value from which the DIGID fluctuation signal can be taken. The DIGID fluctuation signal is generated to fluctuate between positive and negative with respect to zero. Alternatively, the range where the DIGID fluctuation signal fluctuates can be tilted towards either the positive or negative. The fluctuation width 77 of the DIGID jitter signal is preferably adjusted so that it is within the fluctuation width 73 of the average effective pressure indicated as shown in Figure 9, which is observed in a regular operating condition. the motor. In said adjustment of the fluctuation width 77, the DIGID fluctuation signal is prevented from influencing the combustion condition.

Referring to Figure 11, a method will be described, which is performed by the MBT calculator 44 to calculate the ignition timing characteristic curve 71 using the DIGID jitter signal. A range 81 shown in the figure corresponds to the width 77 within which fluctuates the DIGID jitter signal of Figure 10. As described above, the ignition timing adjustment is a sum of the JGBASE reference value and the DIGOP correction value. . Adding the DIGID jitter signal to the ignition timing setting, the signal of the resulting ignition time IGLOG fluctuates within the range 81. The shaded area 82 represents a range within which the indicated average effective pressure is distributed when the ignition time fluctuates within the range 81 due to the DIGID fluctuation signal. The characteristic curve 71 is estimated based on the indicated average effective pressure that is distributed within the area 82. As described above with reference to figure 9, the shape (the inclination and curvature) of the characteristic curve can not be calculated when the ignition time is set to a value extracted from a map, because the indicated average effective pressure is distributed on line 75. However, according to one embodiment of the present invention, the characteristic curve can be calculated by fluctuating the ignition time within the range 81 through the use of the DIGID fluctuation signal due to the indicated average effective pressure distributed within the range 82 which is not one dimensional as the line 81, but which has a two-dimensional degree, the which is acquired. The specific method for calculating characteristic curve 71 will now be described. First, characteristic curve 71 is defined by a function Fmbt of the second order of the DIGID jitter signal, as shown in equation (9). Fmbt (DIGID) = Aigop? DIGID2 + Bigop ::: DIGID + Cigop (9) Aigop, Bigop and Cigop are coefficients that are going to be identified. These coefficients are identified from the indicated average effective pressure that is distributed within the range 82 due to the DIGID fluctuation signal. The method of identification will be described later. Figure 12 shows an estimated curve 83 that is derived by the identified coefficients Aigop, Bigop and Cigop. It can be seen that the estimated curve 83 is almost identical to the actual characteristic curve 71. An EIGOP error between the setting of the on time and the optimum on time MBT is shown by an arrow 84. Because the maximum value 72 of the estimated curve 83 can be determined by differentiating equation (9), the EIGOP error is calculated as shown in equation (10).

EIG0P = - Bi8 ° P (10) 2 • Aigop The ignition time setting has the error EIGOP with respect to the optimal ignition time MBT. By conling the ignition timing setting so as to eliminate the error, the ignition can be implemented at the optimum ignition time MBT. This con scheme will be described later in the "ignition time conler" section. The characteristic curve 71 is not a second order function in the strict sense. Therefore, when the ignition timing setting is separate from the optimal ignition time MBT, the estimated curve may include an error. However, by causing the EIGOP error to be converted to zero by the ignition timing conler 45, the ignition timing adjustment can converge to the optimum ignition time MBT. Now, an identification method for the Aigop, Bigop and Cigop coefficients included in the Fmbt function described above will be described. These coefficients are identified, so that an estimated average effective pressure Pmi is determined by substituting the fluctuation signals DIGID determined in the previous con cycle within the function Fmbt of the estimated curve equal to Pmi_act that is calculated in the current cycle by means of the average effective pressure calculator 43 based on the. pressure within the cylinder detected as a result of using the DIGID fluctuation signal determined in the previous con cycle. The identification method can use a technique well known as the least square method and a maximum likelihood method. In one embodiment of the present invention, a delta (d) correction method is used, which is a more efficient technique. The detail of the delta correction method is described in Japanese Patent No. 3304845. A method for identifying these coefficients using the delta correction method will be briefly described. A recursive identification algorithm that uses the delta correction method is expressed as shown in equation (11). A vector of the coefficient? (k) is expressed by a sum of a reference value? _base (k) and its update component d? (k). d is a vector for forgetting the coefficient, which is expressed in equation (16). ? (k) =? _ base (k) + d? (k) (11) d? (k) = d • d? (k - 1) + KP (k) • E _ id (k) (12) where? t (k) = [Aigop (k), Bigop (k), Cigop (k)] (13) d? t (k) = [Aigop (k) - Aigop _ base, dBigop (k), ddgop (k) ] (14)? _ base t (k) = [Aigop _ base (k), 0, Cigop _ base (k)] (15) In the coefficient forgetting vector d, an element corresponding to the Aigop is set to a value of one of the elements corresponding to Bigop and Cigop are adjusted to a value greater than zero but less than one. This adjustment has the effect that only the Aigop remains and the Bigop and Cigop coefficients are forgotten when the identification error E__id converges to zero. The identification error E_id (k) of equation (12) is expressed by equation (17). That is, the identification error E_id is an error between the indicated average effective pressure Pmi_act calculated by the average effective pressure calculator 43 based on the pressure inside the cylinder that is detected as a result of including the ignition time signal of the DIGID fluctuation signal determined in a previous con cycle, and the estimated effective average effective pressure Pmi_act that is calculated based on the Fmbt function using as input the DIGID fluctuation signal determined in the previous con cycle.

E_ id (k) = Pmi_ act (k) - Pmi_hat (k) (17) where Pmi_ hat (k) =? T (k) -? (K) = Aigop (k) • DIGID (k - 1) 2 + Bigop • DIGID (k - 1) + Cigop (k) (18)? T (k) = [DIGID (k - \, DIGID (k - 1), 1] (19) A gain KP (k) is expressed by the equation (twenty) . P is expressed by equation (21). Depending on the adjustment of the coefficients? L and? 2 of the equation (20), it is determined the type of identification algorithm in the following way:? l = 1 and? 2 = 0: fixed gain algorithm? l = ly? 2 = l: least squares algorithm? l = ly? 2 =?: gain algorithm that decreases gradually (? is a previously determined value that is not zero and 1) l =? and 2 = 1: the least squares weighted algorithm (? is a previously determined value that is not 1 and 1) P (k - l) -? (k) KP (k) = (20) l +? t (k ) - P (k - l) -? (K) (21) where I is a unit matrix of 3x3) When the ignition time converges completely with the optimal ignition time MBT, the variations in the average effective pressure indicated with respect to the fluctuation of ignition time become small. In said stable condition, according to other identification methods, the identified coefficients can be deviated. In contrast, according to the method described above in the present invention, the coefficient vector? (k) is expressed by the sum of the reference value? _base (k) and its update component d? (k) as shown in the equation (eleven) . When the identification error E_id approaches completely zero, the update component d? converges to (Aigop (k-1) - Aigop_base, 0, 0), as can be seen in equation (12). Therefore, the vector of the coefficient? converges in (Aigo (k-1), 0, Cigop_base) as can be seen in equation (11). Because the Aigop coefficient is identified so that the Aigop value does not become zero, the division between zero of the equation (10) can be avoided. Consequently, it is prevented that the feedback control diverges for the optimal power on time MBT. In addition, when the identification error E_id approaches completely zero, the coefficient converges to zero and hence the error EIGOP (shown in equation (10)) of the optimal ignition time MBT becomes zero. Because the EIGOP error converges to zero, the feedback control to cause the ignition timing to converge on the optimal start time MBT ends automatically. When variations in the average effective pressure indicated with respect to ignition timing fluctuation become large due to some abnormality in combustion, a correlation between the DIGID fluctuation signal and the average effective pressure Pmi_act can not be maintained. In this case, the error E_id appears as a white noise and the average error E__id becomes zero. As a result, the vector of the coefficient? converges on the reference value? _base, automatically ending the feedback control in this way. Therefore, according to the delta correction method, the identified coefficients are prevented from being diverted in a stable condition where the identification error E_id is very small. In one embodiment of the present invention, a Lim_a function is applied to the Aigop coefficient identified as shown in equation (22). The function Lim_a (x) is a function of restricting "x" to a value that is less than zero. The Lim_a (Aigop) function restricts the Aigop so that it has a negative value.

Aigop < = - Lima _ a (Aigop) (22) The reason for applying the Lim_a function that acts to restrict the Aigop coefficient to a negative value will be described with reference to Figure 13. Figure 13 shows a condition in which the ignition timing setting was completely converted to the ignition time optimal MBT and average effective pressure indicated Pmi_act almost does not exhibit fluctuation (ie, Pmi_act is almost flat). The actual characteristic curve is shown with the reference numeral 91. In such a condition, the estimated curve can be determined erroneously as a downward convex curve 94 (ie, Aigop = 0). Said erroneous calculation can cause an error in the calculation of the optimal ignition time MBT. In order to avoid such miscalculation, the Lim_a function is applied to calculate the estimated curve as a curve 93 having a convex upward shape (ie, Aigop < 0). Ignition time controller The ignition timing controller 45 controls the ignition timing to converge on the optimal ignition time which is calculated from the calculated curve. More specifically, the DIGOP correction value is calculated so as to cause the EIGOP error of the optimum start time MBT to converge to zero. Adding the DIGOP correction value to the IGBASE reference value compensates for the EIGOP error. The ignition timing controller 45 uses a response assignment control to calculate a control input (i.e., a DIGOP correction value). The calculation of the DIGOP is expressed in equation (23). k DIGOP (k) = -Krch • s (k) - Kadp? s (i) (23) s (k) = EIGOP (k) + POLE • EIGOP. { k-1) (24) where Krch, Kadp > 0 The response assignment control is a control that has the ability to specify a convergence capacity of a controlled variable (the EIGOP error, in this case) to a desired value (zero, in this case). According to the response allocation control, the EIGOP error can converge to zero without exceeding. A change function s is established in the response allocation control. POLE is a parameter of the response assignment of the change function s to define a convergence speed of the EIGOP error. POLE is preferably adjusted to meet -1 <; POLE < 0. The equation s (k) = 0 is called an equivalent input system, which specifies the convergence characteristic of the EIGOP error. Assuming that s (k) = 0, the change function s of equation (24) is expressed as shown in equation (25). EIGOP (k) = -POLE-EIGOP (k-?) (25) In this case, referring to Figure 14, the change function will be described. In a phase plane with EIGOP (k) on the vertical axis and EIGOP (kl) on the horizontal axis, the change function s in equation (25) is expressed as a line 95. This line 95 is called a line of change Assuming that an initial value of the quantity of the condition (EIGOP (kl), EIGOP (k)) that is a condition of (EIGOP (kl) and EIGOP (k)) is represented by a point 96. The control of assignment of response acts to place the amount of condition represented by point 96 of line 95 and then restricts it on line 95. According to the response allocation control, because the amount of the condition is held on the line of change 95, the condition quantity can converge stably with the zero origin without being influenced by the disturbances. In other words, by restricting the amount of the condition (EIGOP (kl), EIGOP (k)) in said stable system, which does not have an input shown in equation (25), the EIGOP error can robustly converge to zero against disturbances and design errors. Because the phase plane with respect to the shift function s has two dimensions in this mode, the line of change is represented by a straight line 95. When the phase plane has three dimensions, the line of change is represented by a plane. When the phase plane has four or more dimensions, the line of change is represented by a hyperplane. The POLE response assignment parameter can be adjusted variably. By adjusting the value of the POLE response assignment parameter, the convergence speed of the EIGOP error can be designated.

Referring to Figure 15, the reference numbers 97, 98 and 99 show the convergence speed of the EIGOP error in cases where the POLE response assignment parameter takes a value of -1, -0.8 and -0.5, respectively . The convergence speed of the EIGOP deviation increases as the absolute value of the POLE response allocation parameter decreases. Effect of the ignition timing control according to a modality With reference to figures 16 and 17, the effect of the ignition timing control according to an embodiment of the present invention will be described. Figure 16 shows the actual average effective pressure indicated Pmi_act when the ignition time IGLOG is calculated by adding the DIGID jitter signal to the IGBASE reference value obtained from a previously determined map based on the current operating condition of the engine. In the example, feedback control is not performed for the optimal power on time MBT using the correction value. At time ti, the operating condition of the motor changes. The IGBASE reference value based on the detected operation condition after the change is extracted from the map. It is assumed that the IGBASE reference value extracted in this way from the map has a delayed value with respect to the optimal ignition time. As a result, the ignition time is delayed. The actual average effective pressure indicated Pmi_act decreases as the ignition time is delayed. The level of the indicated average effective pressure corresponding to the optimal ignition time MBT is shown with the reference number 101. Because the ignition timing can not converge on the MBT, a "deviation" between the average effective pressure is not eliminated. indicated actual Pmi_act and level 101.

Therefore, it is not realized if the feedback control to cause the ignition timing to converge at the optimum time, then the indicated average effective pressure is maintained at a decreased level, which can reduce the combustion efficiency. Figure 17 shows a case where the feedback control for the ignition timing is performed according to an embodiment of the present invention. Because the DIGID jitter signal is added to the ignition time signal setting (IGBASE + IGLOG) the IGLOG ignition time signal fluctuates.

Over a period of time of tO to you, the ignition time signal IGLOG has converged on the optimal ignition time MBT and hence the actual indicated average effective pressure Pmi_act is maintained at a level corresponding to the MBT. Because the ignition time signal IGLOG has converged with the MBT, the value of the DIGOP correction value is almost zero. At time ti, the operating condition of the motor changes. Due to this change, the IGBASE reference value deviates from the MBT. And, hence IGLOG's ignition time signal deviates from the MBT. As a result, the indicated effective average pressure Pmi_act real decreases below the level 105 corresponding to the MBT. The MBT calculator 44 identifies the Aigop, Bigop and Cigop coefficients, so that the indicated average effective pressure Pmi_act that is calculated by the Fmbt function based on the DIGID fluctuation signal becomes equal to the actual average effective pressure indicated Pmi_act. As a result, the estimated effective average effective pressure Pmi_act follows the actual average effective pressure indicated Pmi_act. By identifying the Aigop, Bigop and Cigop coefficients, the optimal start time MBT is calculated. In addition, the EIGOP error of the ignition time setting (IGBASE + DIOP) with respect to the MBT is calculated. It can be seen that the EIGOP error is rising around time t2. The ignition timing controller 45 calculates the DIGOP correction value to compensate for the EIGOP error. It is observed that the DIGOP correction value rises to follow the elevation of the EIGOP error. By adding the DIGOP correction value, the ignition time signal IGLOG is corrected to advance. As a result, the ignition time signal IGLOG returns to the optimal ignition time MBT around time t3. Because the ignition time signal IGLOG has converged with the MBT, the actual average effective pressure indicated Pmi_act returns to the optimum level 105. In this simulation, the reference values are adjusted as follows: Aigop_base = 2, Bigop_base = 0 and Cigop__base = 300. As described above, when the identification error converges completely to zero and hence the ignition time signal IGLOG converges on the optimal ignition time MBT, the Bigop coefficient converges to its reference value (= zero). As a result, the EIGOP error is set to zero, automatically ending the feedback control (an arrow 106 indicates that the feedback control has reached automatic termination around time t4). Control flow Figure 18 shows a main route of a process for ignition time control. This routine is performed in synchronization with the combustion cycle. This flow chart shows an example process for a single-cylinder engine. In the case of the multi-cylinder engine, the following process is performed for each combustion cycle of each cylinder. For example, in the case of a four-cylinder engine, the process starts for one of the four cylinders at each 180 degree crank angle. In step SI, it is determined whether a fault is detected in a valve operating system (which includes a variable phase mechanism and a variable lift mechanism) or a variable compression ratio mechanism. Due to torsion control of part of the valve operating system that can not be performed if a fault is detected, safe fault control is performed to calculate an ignition time to maintain engine rotation speed at a level constant (S2). Fail safe control can be implemented, for example, through the response assignment control described above. The ignition time Ig_fs is calculated so that the rotation speed of the motor converges to a desired predetermined value, (for example, 2000 rpm). The calculation to implement the response assignment control for safe fault control is performed, for example, in accordance with the following equations: k Ig _ fs = Ig _ fs _ base - Krch '-s' (k) - Kadp '•] s' (i) (26)? = 0 sk) = Enfs (k) + POLE' -Enfs (k - 1) (27) Enfs (k) = NE (k) - NE _ fs (28) Krch ', Kadp': feedback gain POLE ': response allocation parameter (-1 <POLE' <0) NE_fs: desired value for the motor rotation speed (eg, 2000 rpm) Ig_fs_base: reference value for safe failure control (eg 0 degrees). In step S3, the calculated Ig_fs is set to IGLOG turn-on time. If no fault is detected in the SI step, it is determined if the engine is in the start mode (S4). If the engine is in start-up mode, the ignition IGLOG time is adjusted to a previously determined value (for example, +10 degrees) (S5).

If the engine is not in the start mode, it is determined in step S6 whether the accelerator pedal is completely closed. If the accelerator pedal is completely closed, this indicates that the engine is in an inactive condition. Then, in step S7, it is determined whether the previously determined time that is set to perform the rapid heating control of the catalyst has elapsed. If the previously determined time has not elapsed, this indicates that the rapid heating control of the catalyst is still operating. The rapid heating control of the catalyst is a control to increase the temperature of the catalyst so as to rapidly activate the catalyst. During the rapid heating control of the catalyst, the ignition timing is delayed, so that the speed of rotation of the engine converges to a desired value. This control is implemented through the control of response allocation in a manner similar to step S2. The following are equations for implementing response allocation control. k Ig _ ast = Ig_ ast _ base - Krch "-a" (k) - Kadp "? s" (i) (29) / = os "(k) = Enast { k) + POLE'-Enast ( k - 1) (30) Enastik) = NE (k) - NE _ast (31) Krch ", Kadp ': feedback gain POLE": response assignment parameter (- 1 <POLE "<0) NE_ast: desired value for the motor rotation speed (eg, 1800 rpm) Ig_ast_base: reference value for heating the catalyst temperature (eg +5 degrees). In step S9, the calculated Ig_ast is set to the ignition time IGLOG. If the rapid heating control of the catalyst is completed in step S7, the feedback control is performed (Fig. 19) for the optimal ignition time MBT according to the present invention (S10). Fig. 19 shows a flow chart of the optimization time feedback control MBT. In step S21, the values obtained from the output sampling of the first order filter and the output of the second order filter are received and the indicated average effective pressure Pmi_act is calculated according to equation (3). A flow chart for the sampling of the first order filter output and the second order filter output is shown in FIG. 20. In step S22, the Aigop, Bigop and Cigop are calculated according to the equations from (11) to (22) described above to determine the estimated curve expressed in the equation (9) In step S23, the EIGOP error is calculated based on equation (10). In step S24, the response assignment control is performed as expressed in equations (23) and (24) to calculate the correction value to cause the EIGOP error to converge to zero. In step S25, the reference value is determined IGBASE referring to a map as shown in Figure 4 based on the current engine rotation speed NE and 'the current admission air amount Gcyl. In step S26, the value of the Cdigid counter is incremented by one. As described above with reference to Fig. 10, the jitter signal depends on the value of the Cdigid counter. If the value of the Cdigid counter exceeds the Cdigid_max which indicates a cycle length of the DIGID jitter signal in step S27, the counter becomes adjusted (S28). If the value of the Cdigid counter is equal to or less than the Cdigid max, the process proceeds to step S29. In step S29, a table is consulted as shown in Figure 10 to determine the current value of the DIGID jitter signal corresponding to the value of the Cdigid counter. In step S30, the IGBASE reference value, the DIGOP correction value and the DIGID jitter signal are added until the IGLOG ignition time signal is determined. Steps from S21 to S24 can be performed in parallel with steps S25 through S29. Figure 20 is a flow chart of a sampling process for the pressure inside the cylinder. This routine is performed at each 15 degree crank angle. In step S31, the detection value is sampled Pcyl of the pressure sensor inside the cylinder. In step S32, the first order filter is applied to the sampled Pcyl detection value. In step S33, the second order filter is applied to the detection value Pcyl. In step S34, it is determined whether the current crank angle is 45 degrees after the TDC. If the response to the step is Yes, the output of the second order filter is sampled and stored in a memory (S35). In step S36, it is determined whether the current crank angle is 90 degrees after the TDC. If the response of the step is Yes, the output of the first order filter is sampled and stored in a memory (S37). The second order output C2 ° cos (f2) sampled in step S35 and the first order output Cl ° cos (fl) sampled in step S37 are passed to step S21 of figure 19. The present invention can be applied to a general purpose motor (for example, an outboard motor).

Claims (14)

NOVELTY OF THE INVENTION Having described the present invention, it is considered as a novelty and, therefore, the content of the following is claimed as property:
1. CLAIMS 1.- An apparatus to control the ignition time of an engine, which includes: a ignition time calculator to add a fluctuation component to an ignition timing adjustment to calculate a final ignition time to start the engine; an average effective pressure calculator to calculate the average effective pressure indicated for a cylinder pressure detected when the engine has been turned on according to the final ignition time; an MBT calculator for an ignition time characteristic curve representing a correlation between the indicated average effective pressure and the fluctuation component and for determining an optimal ignition time of the characteristic curve; and a controller to control the ignition timing adjustment so that it converges on the optimum ignition timing.
2. 2. - The apparatus according to claim 1, characterized in that the characteristic curve of the ignition time is represented by a function, an input of the function being the fluctuation component, the output of the function being the average effective pressure indicated; wherein the MBT calculator further includes an identifier to identify the coefficients associated with the fluctuation component in the function based on the indicated average effective pressure calculated by the average effective pressure calculator to calculate the characteristic curve based on the identification of the coefficients .
3. 3. The apparatus according to claim 2, which further comprises a generator for generating the fluctuation component, wherein the generator generates the fluctuation component to cover the self-excitation conditions to identify the coefficients of the function.
4. 4. The apparatus according to claim 2, characterized in that the identifier is further configured to: determine the update components for the coefficients, so that an error between the indicated average effective pressure calculated by the average effective pressure calculator and a calculated average effective pressure calculated from the function converges to zero; adding the update components to previously determined reference values to determine the coefficients, thereby causing the coefficients to converge on the reference values as the error converges to zero; wherein the reference values are previously determined so that the control to control the adjustment of the ignition timing converges on the optimum ignition time stops when the coefficients have converged on the reference values.
5. 5. The apparatus according to claim 2, characterized in that a limit process is applied to at least one of the coefficients so that the characteristic curve is prevented from being estimated as a convex curve downward.
6. 6. The apparatus according to claim 1, characterized in that the average effective pressure calculator is further configured to extract an alternating component of the pressure within the detected cylinder and to calculate the indicated average effective pressure based on the alternating component.
7. 7. The apparatus according to claim 1, characterized in that the controller uses a response allocation control to control the ignition timing adjustment, the capacity having the response assignment control to specify a response characteristic of the adjustment of the ignition time for the optimal ignition time.
8. 8. A method for controlling the ignition timing of an engine, comprising the steps of: (a) adding a jitter component to an ignition timing adjustment to calculate a final ignition time to start the engine; (b) calculating the average effective pressure indicated for a pressure within the detected cylinder when the engine has been turned on in accordance with the final ignition timing; (c) calculating a characteristic curve of ignition time that represents a correlation between the indicated average effective pressure and the fluctuation component; (d) determining an optimal ignition time of the characteristic curve; (e) control the ignition timing adjustment to converge on the optimal ignition timing.
9. 9. The method according to claim 8, characterized in that the characteristic curve of the ignition time is represented by a function, an input of the function being the fluctuation component and an output of the function being the indicated average effective pressure; wherein step (c) further comprises the step of: (cl) identifying the coefficients associated with the fluctuation component in the function based on the average effective pressure indicated to calculate the characteristic curve based on the identification of the coefficients.
10. 10.- The method according to the claim 9, which further comprises the step of generating the fluctuation component to cover the auto-excitation conditions to identify the coefficients of the function.
11. 11. The method according to claim 9, characterized in that step (cl) further comprises the steps of: determining the update component for the coefficients so that an error between the indicated average effective pressure calculated in step (b) ) and the indicated average effective pressure calculated from the function converge to zero; and adding the update components to the previously determined reference values to determine the coefficients, thereby causing the convergence of the coefficients with the reference values as the error converges to zero; wherein the reference values are previously determined so that the control for controlling the ignition timing adjustment converges on the optimum ignition time and stops when the coefficients have converged on the reference values.
12. 12. The method according to claim 9, which further comprises the step of: applying a limit process to at least one of the coefficients so that the characteristic curve is prevented from being estimated with a convex curve downward .
13. 13. The method according to claim 8, characterized in that step (b) further comprises the steps of: extracting an alternating component of the pressure detected within the cylinder; and calculate the average effective pressure indicated based on the alternate component.
14. 14. The method according to claim 8, characterized in that the step (e) further comprises the step of: using a response assignment control to control the ignition timing adjustment, the capacity having the response assignment control to specify a response characteristic of the setting the on time to the optimal on time.