JP4735620B2 - Injection amount learning device - Google Patents

Description

  The present invention relates to an injection amount learning device, and more particularly to an injection amount learning device that learns an injection amount of fuel injected from an injector in a common rail fuel injection device of a diesel engine.

  Conventionally, in a diesel engine, in order to reduce combustion noise and NOx emission, it is effective to perform so-called pilot injection in which a small amount of fuel is injected prior to main injection that is main fuel injection. However, in pilot injection with a small injection amount, it is important to improve the accuracy of the injection amount in order to fully exhibit the effect. For this reason, it is necessary to appropriately correct a deviation between a target injection amount in pilot injection (hereinafter referred to as “target injection amount”) and an actual injection amount (hereinafter referred to as “actual injection amount”).

Therefore, in Patent Document 1, for example, a single micro-injection is performed when there is no injection when the target injection amount from the injector becomes 0 or less, such as during deceleration of a vehicle in which the diesel engine is in no-load operation. As a result, the learning of the injection amount, particularly a small amount of injection amount, is carried out with high accuracy from the change in the rotation of the engine accompanying the fuel injection.
JP 2005-36788 A

  However, the fuel pressure inside the common rail is widely distributed. Therefore, in the case of Patent Document 1, it takes a long time to learn the injection amount at a plurality of pressures according to the pressure of the fuel inside the common rail, and the traveling distance of a vehicle equipped with a diesel engine increases. Further, in the case of Patent Document 1, since the injection amount is learned during the no-load operation of the diesel engine, there is a problem that the learning frequency decreases on the side where the fuel pressure is high inside the common rail.

  Therefore, the present invention has been made in view of the above problems, and an object thereof is to provide an injection amount learning device that learns an injection amount of fuel in a wide pressure range in a short period of time.

  According to the first aspect of the present invention, the fuel injection amount is learned as a detected pressure injection amount at an arbitrary detection pressure among a plurality of set detection pressures. Then, using the detected pressure injection amount acquired by learning, the fuel injection amount at an undetected detected pressure that is not learned among a plurality of set detected pressures is estimated. That is, when the fuel injection amount is learned at a certain detection pressure, the fuel injection amount at another detection pressure is calculated as an estimated value using the fuel injection amount acquired by learning. Therefore, it is possible to learn the change over time in the injection amount over a wide pressure range in a short period of time.

In the first aspect of the invention, when estimating the fuel injection amount at the undetected pressure, the undetected pressure injection amount estimating means searches for the detected pressure having the highest correlation. For example, when estimating the fuel injection amount at an undetected pressure on the high pressure side, the accuracy may be low even if the detected pressure injection amount on the low pressure side is used. Therefore, the fuel injection amount at the undetected pressure is estimated using the detected pressure whose injection amount characteristics are close to the undetected pressure on the high-pressure side, that is, a highly correlated pressure. Thereby, the accuracy of the estimated fuel injection amount can be increased.

Hereinafter, an embodiment of an injection amount learning device of the present invention will be described with reference to the drawings.
(Configuration of fuel injection system)
FIG. 2 is a schematic diagram showing a fuel injection system to which an injection amount learning device according to an embodiment of the present invention is applied. In the case of this embodiment, the injection amount learning device 10 is applied to a diesel engine 21 in which fuel injection is controlled by a common rail fuel injection system 20. The fuel injection system 20 includes a fuel tank 22, a suction amount control valve 23, a supply pump 24, a common rail 25, and an injector 40. The injection amount learning device 10 includes an engine control device (hereinafter referred to as “ECU: Engine Control Unit”) 11. The suction amount control valve 23 and the supply pump 24 constitute an integral pump unit 27.

  The fuel tank 22 stores normal pressure fuel. The fuel inside the fuel tank 22 is supplied to the intake amount control valve 23 via the intake pipe portion 31 by a low-pressure pump (not shown). The supply pump 24 pressurizes fuel sucked into a pressurizing chamber (not shown) by reciprocating a plunger (not shown). In the supply pump 24, the amount of fuel discharged changes according to the amount of fuel sucked into the pressurizing chamber. The plunger receives driving force from a crankshaft 28 of the diesel engine 21. The fuel pressurized by the supply pump 24 is discharged to the common rail 25. A fuel pipe portion 32 is connected to the discharge side of the supply pump 24. The fuel pipe portion 32 connects the supply pump 24 and the common rail 25.

  The common rail 25 is connected to the fuel pipe portion 32 and stores the fuel pressurized by the supply pump 24 in a pressure accumulation state. An injector 40 that injects fuel into each cylinder 29 of the diesel engine 21 is connected to the common rail 25. The injector 40 is provided for each cylinder 29. The fuel stored in the common rail 25 in an accumulated state is injected from the injector 40 into the combustion chamber formed in each cylinder 29. A reflux piping section 33 is connected to the supply pump 24, the common rail 25, and the injector 40. Surplus fuel in the supply pump 24, the common rail 25, and the injector 40 is returned to the fuel tank 22 via the return piping section 33.

  The ECU 11 is composed of, for example, a microcomputer having a CPU, a ROM, and a RAM. The CPU controls the entire fuel injection system 20 according to a computer program stored in the ROM. The ECU 11 functions as an injection amount learning unit and an undetected pressure injection amount estimating unit. In the ECU 11, a pressure sensor 13, an accelerator sensor 14, a rotation sensor 15 and the like are connected to a circuit on the input side. The pressure sensor 13 is provided on the common rail 25.

  The pressure sensor 13 detects the pressure of the fuel stored in the common rail 25. The pressure sensor 13 outputs the detected fuel pressure in the common rail 25 (hereinafter, the fuel pressure in the common rail is referred to as “rail pressure”) to the ECU 11 as an electrical signal. The accelerator sensor 14 outputs the depression amount of an accelerator pedal (not shown) to the ECU 11 as an electric signal. The rotation sensor 15 detects the rotation of the crankshaft 28 of the diesel engine 21. The rotation sensor 15 outputs the detected rotation of the crankshaft 28 to the ECU 11 as an electrical signal.

  The ECU 11 detects the operating state of the diesel engine 21 from, for example, an electrical signal related to the rotational state of the diesel engine 21 detected by the rotation sensor 15 and an accelerator pedal depression amount detected by the accelerator sensor 14. The ECU 11 sets the amount of fuel injected from the injector 40 according to the detected operating state of the diesel engine 21. The ECU 11 sets the rail pressure based on the set fuel injection amount.

  The ECU 11 is connected to a circuit on the output side by an intake amount control valve 23 and an electronic drive unit (hereinafter, the electronic drive unit is referred to as “EDU: Electronic Drive Unit”) 12. The intake amount control valve 23 controls the flow rate of fuel supplied to the supply pump 24 based on the control current output from the ECU 11. The EDU 12 is connected to the electromagnetic valve 41 of the injector 40. The EDU 12 outputs a pulsed drive signal to the electromagnetic valve 41 of the injector 40 based on the drive signal output from the ECU 11. In the injector 40, the solenoid valve 41 is driven based on the pulse-shaped drive signal output from the EDU 12, and fuel injection is intermittently performed. As a result, the injector 40 injects the fuel stored in the common rail 25 into the combustion chamber formed in each cylinder 29 of the diesel engine 21.

Next, the operation of the injection amount learning device 10 having the above configuration will be described.
Prior to main injection, which is the main fuel injection, when performing pilot injection that injects a small amount of fuel as compared to main injection, it is necessary to precisely adjust the fuel injection amount during pilot injection. That is, when the accuracy of the fuel injection amount in pilot injection is low, effects obtained by pilot injection, such as noise reduction and NOx reduction, are reduced. Therefore, it is necessary to appropriately correct the fuel injection amount in the pilot injection from each injector 40 and always maintain a predetermined injection amount.

  In the injector 40, as shown in FIG. 3, the relationship between the width of the pulse energized to the solenoid valve 41, that is, the energization time T and the fuel injection amount Q gradually changes over time. The cause of this change is mainly due to deterioration of the accuracy of the injector 40 over time, that is, deterioration due to an increase in the elapsed period during which the injector 40 is operating. The elapsed period during which the injector 40 is operating corresponds to the travel distance of the vehicle on which the diesel engine 21 is mounted. As the elapsed time during which the injector 40 is operating becomes longer, the fuel injection amount Q decreases due to deterioration even if the energization time T of the injector 40 is constant. Thus, the fuel injection amount Q from the injector changes with time even when the energization time T is constant. Therefore, the ECU 11 needs to periodically learn the relationship between the fuel injection amount characteristic from the injector 40, that is, the relationship between the energization time T and the fuel injection amount Q, and correct the injection amount.

  In this case, when the injector 40 is in a certain elapsed period, that is, even if the elapsed period in which the injector 40 is operating is the same, the fuel injection amount Q from the injector 40 varies depending on the rail pressure Cp in the common rail 25. . For example, even when the energization time T is constant, the fuel injection amount Q increases as the rail pressure Cp of the common rail 25 increases. On the other hand, a predetermined correlation exists between the rail pressure Cp of the common rail 25 and the fuel injection amount Q. Therefore, in the present embodiment, even if the rail pressure has a low learning opportunity, such as a rail pressure on the high pressure side such as 180 MPa, a lower rail is obtained by using the correlation between the rail pressure Cp and the fuel injection amount Q. Learning is performed while estimating the fuel injection amount Q at the rail pressure on the high pressure side based on the learning result on the pressure.

(Rail pressure index)
The injection amount learning device 10 sets the detected pressure according to the rail pressure as shown in FIG. That is, the injection amount learning device 10 does not learn the injection amount in the entire range of the rail pressure set in the common rail 25, but performs the injection amount learning for the representative rail pressure in the predetermined range. To do. The rail pressure shown in FIG. 4 is an exemplification, and an arbitrary rail pressure can be set as the detected pressure in accordance with the pressure set on the common rail 25. Also, the rail pressure classification can be arbitrarily set.

  In the case of this embodiment, six rail pressures “32 MPa”, “60 MPa”, “80 MPa”, “100 MPa”, “140 MPa”, and “180 MPa” are set as the rail pressures to be detected pressures. The injection amount learning device 10 stores that the rail pressure is “32 MPa” as an index number “# 0”, and similarly, “60 MPa” is “# 1”, “80 MPa” is “# 2”, “100 MPa”. Are stored as “# 3”, “140 MPa” as “# 4”, and “180 MPa” as “# 5”. The injection amount learning device 10 learns the injection amount of fuel injected from the injector 40 when the rail pressure in the common rail 25 is a pressure corresponding to each index. The ECU 11 of the injection amount learning device 10 stores a rail pressure index corresponding to the rail pressure in a ROM (not shown) or the like.

(Detailed map)
The injection amount learning device 10 stores a detailed map corresponding to the rail pressure index. As shown in FIG. 5, the detailed map includes a rail pressure index for each rail pressure corresponding to the detected pressure, and data related to a learning implementation state in each cylinder 29 of the diesel engine 21. In the case of the four-cylinder diesel engine 21, the cylinder index is set as "# 0" for the first cylinder, "# 1" for the second cylinder, "# 2" for the third cylinder, and "# 3" for the fourth cylinder. ing. That is, the detailed map is a map that shows the state of learning of the injection amount at each rail pressure at which the injector 40 provided in each cylinder 29 becomes the detected pressure. At this time, the detailed map indicates “0” if the injection amount learning is not performed, “1” if the injection amount learning is being performed, and “2” if the injection amount learning is completed. Written. When learning of the injection amount is not performed, the injection amount learning device 10 estimates the injection amount that has not been learned from the injection amount at the rail pressure at which learning of the injection amount has been completed. The ECU 11 of the injection amount learning device 10 stores a detailed map as electronic data in a RAM or ROM (not shown).

(Learning of injection amount)
The ECU 11 learns the fuel injection amount for each rail pressure indicated by the rail pressure index. At this time, the ECU 11 learns the fuel injection amount for each injector 40 provided in each cylinder 29 of the diesel engine 21. The ECU 11 cannot directly correct the injection amount of fuel injected from the injector 40. Therefore, the ECU 11 corrects the pulse width of the drive signal, that is, the energization time T (μsec) for each rail pressure index in each injector 40, and stores the correction value of the energization time T as a learning value. The learning of the injection amount, that is, the learning of the energization time T is performed by a known method, for example, detecting a change in the rotational speed of the crankshaft 28 that changes due to fuel injection by the rotation sensor 15. Therefore, the description of the specific means for learning the injection amount is omitted.

  The rail pressure in the common rail 25 changes according to the operating state of the diesel engine 21. That is, the rail pressure is low when the load on the diesel engine 21 is small, and the rail pressure is high when the load on the diesel engine 21 is large. On the other hand, at the time of normal acceleration or cruise, the upper limit of the rail pressure is about “140 MPa” indicated by the rail pressure index “# 4”. For this reason, there is little opportunity to learn the injection amount at the high rail pressure “180 MPa” indicated by the rail pressure index “# 5”. Therefore, in the present embodiment, a learning value at a high rail pressure with a small learning opportunity, such as 180 MPa, is estimated using a result learned in a range of normally used rail pressure.

  As shown in FIG. 1, it is assumed that learning of the injection amount (energization time T) is completed at a certain rail pressure A and rail pressure B, and a learned value is stored. At this time, the ECU 11 learns the injection amount at the rail pressure C. In the case shown in FIG. 1, it is assumed that the rail pressure increases toward the right. That is, the relationship of rail pressure A> rail pressure B> rail pressure C is established. As described above, for the rail pressure A and the rail pressure B, learning of the injection amount is completed. Therefore, the ECU 11 stores the pulse width of the rail pressure A as TQG [A], and stores the pulse width of the rail pressure B as TQG [B]. The injection amount corresponding to the learned value TQG [A] at the rail pressure A and the learned value TQG [B] at the rail pressure B corresponds to the detected pressure injection amount in the claims. The rail pressure C for estimating the learned value of the fuel injection amount corresponds to the undetected pressure in the claims.

Further, learning of the injection amount from the injector 40 is performed at predetermined time intervals as the operation period of the injector 40 elapses. Therefore, the ECU 11 has also learned the pulse width of the rail pressure C in the past, and stores the pulse width at that time as the previous value TQG [C] _i−1 . This previous value TQG [C] _i-1 may be calculated based on an actual measurement value, or may be calculated as an estimated value based on another rail pressure.

  Thus, when the pulse width TQG [A] or TQG [B] at the rail pressure A or the rail pressure B is learned, the ECU calculates the estimated value of the pulse width TQG [C] at the rail pressure C. . The pulse width TQG [C] at the rail pressure C is calculated by the following equations when the pulse width TQG [A] of the rail pressure A is used and when the pulse width TQG [B] of the rail pressure B is used, respectively. The

TQG [C] = (TQG [A] × m−TQG [C] _i−1 ) × k + TQG [C] _i−1
TQG [C] = (TQG [B] × m−TQG [C] _i−1 ) × k + TQG [C] _i−1
In the above formula, m represents a correlation coefficient, and k represents a reflection coefficient. The correlation coefficient m is a value indicating a correlation between the detected injection amount at the rail pressure A or the rail pressure B that has already been learned and the injection amount at the rail pressure C that has not yet been learned. On the other hand, the reflection coefficient k is calculated when the injection amount at the rail pressure C that has not been learned is estimated based on the detected injection amount at the rail pressure A or rail pressure B that has already been learned. This is a value indicating how much the detected injection amount at the pressure B is reflected in the injection amount at the rail pressure C.

  As shown in FIG. 6, a predetermined relationship is established between the correlation coefficient m and the reflection coefficient k. That is, when the estimated rail pressure is the reference P0, the variation is large and the correlation is low regardless of whether the pressure is higher or lower than the rail pressure P0. Specifically, the reflection coefficient k is such that as the learned rail pressure used for estimating the fuel injection amount becomes closer to the rail pressure P0 for estimating the fuel injection amount, the correction of the variation becomes smaller and its value approximates to 1. To do. On the other hand, as the reflection coefficient k is farther from the rail pressure P0 at which the learned rail pressure used for estimating the fuel injection amount is farther from the rail pressure P0 where the fuel injection amount is estimated, the variation correction becomes larger, and the value thereof is farther from 1. The correlation coefficient m is set as a predetermined function according to the learned rail pressure used for estimating the fuel injection amount.

As shown in FIG. 1, the previous learned value TQG [C] _i-1 at the rail pressure C, the learned value TQG [A] at the rail pressure A or the learned value TQG [B] at the rail pressure B, and the rail pressure A Alternatively, the correlation value TQG [C] _m of the learning value at the current rail pressure C is calculated from the correlation coefficient m set according to the rail pressure B. Thus, the correlation coefficient m is set so that the same learning value TQG [C] _{m at} the rail pressure C is calculated regardless of the learning value of the rail pressure A or the rail pressure B. This learning value TQG [C] _m is the upper limit of the learning value to be estimated.

Here, the learning value TQG [C] at the rail pressure C is calculated by calculating according to the above equation using the reflection coefficient k set according to the rail pressure A or the rail B. At this time, since the reflection coefficient k is set for each rail pressure, the reflection coefficient k is close to 1 at a highly correlated rail pressure, and the reflection coefficient is less than 1 at a rail pressure with low correlation. . As a result, between the learning value TQG [C] _A at the rail pressure C calculated based on the rail pressure _A and the learning value TQG [C] _B at the rail pressure C calculated based on the rail pressure B, A difference depending on the amount of reflection occurs. Even in this case, since the maximum value of the reflection coefficient k is 1, the correlation value TQG [C] _m is not exceeded. That is, the reflection coefficient k has a maximum value of 1, and the learning value TQG [C] _A or TQG [C] _B estimated for the rail pressure C based on the rail pressure A or the rail pressure _B is the correlation value TQG [C] _m. Is limited to the upper limit. Thereby, the deterioration of the accuracy of the fuel injection amount due to unnecessary reflection of the learned value is suppressed. In the case shown in FIG. 1, the reflection coefficient k of the rail pressure B close to the rail pressure C is larger than the rail pressure A, and the estimated value TQG [C] _B in the rail pressure C calculated based on the rail pressure _B Is adopted as a new learning value TQG [C] at the rail pressure C.

  As described above, when the fuel injection amount at the rail pressure C that has not been learned is estimated, the detected injection amount at the rail pressure A or the rail pressure B that has been learned is used. On the other hand, when there are a plurality of rail pressures for which learning has been completed as in the example illustrated in FIG. 1, the ECU 11 searches for rail pressures with higher correlation among the rail pressures for which learning has been completed.

Next, the flow of estimating the rail pressure for which learning has not been completed will be described with reference to FIG.
When the ECU 11 shifts to learning of a certain rail pressure (hereinafter referred to as “learning rail pressure”), the ECU 11 determines whether learning with another rail pressure is completed (S101). If learning at another rail pressure has not been completed, learning of the fuel injection amount at another rail pressure is continued without performing learning of the fuel injection amount at the learning rail pressure. When ECU 11 determines in step S101 that learning of the fuel injection amount at another rail pressure has been completed, ECU 11 determines whether or not the processing loop has ended (S102). That is, the ECU 11 determines whether or not learning of the fuel injection amount has been completed for all set rail pressures. If the ECU 11 determines that learning of the fuel injection amount has been completed for all the set rail pressures, the ECU 11 ends the process. Here, all the set rail pressures are pressures corresponding to indexes # 0 to # 5 as shown in FIG. 4, for example.

  For example, when rail pressure index # 5 (180 MPa) is used as the learned rail pressure, ECU 11 determines whether or not learning of the fuel injection amount has been completed at one of the rail pressures indicated by rail pressure indexes # 0 to # 4 in step S101. Determine. If the ECU 11 determines in step S101 that learning of the fuel injection amount has been completed at any of rail pressure indexes # 0 to # 4, whether or not learning of the fuel injection amount has been completed in all rail pressure indexes in step S102. Determine whether or not.

  When the ECU 11 determines in step S102 that the processing loop has not been completed, the ECU 11 determines whether or not learning of the fuel injection amount at the learning rail pressure has been completed (S103). When the ECU 11 determines that the learning of the fuel injection amount at the learning rail pressure has been completed, the ECU 11 returns to step S102 and performs processing at another rail pressure to be learned.

  For example, when the rail pressure index # 5 is used as the learning rail pressure, if the ECU 11 determines in step S102 that the processing loop has not been completed, the ECU 11 determines in step S103 whether learning of the rail pressure index # 5 has been completed. When the ECU 11 determines that learning of the rail pressure index # 5 is completed, the ECU 11 returns to step S102 and learns the fuel injection amount at other rail pressures for which learning has not been completed.

  If the ECU 11 determines in step S103 that learning of the fuel injection amount at the learning rail pressure has not been completed, the ECU 11 extracts an optimal correlation rail pressure index (S104). When shifting to extraction of the optimum correlation rail pressure index in step S104, the ECU 11 extracts the optimum correlation rail pressure index along the flow shown in FIG.

  As described above, when the learning value at a certain rail pressure is estimated, the correlation is different from the rail pressure at which the learning value is estimated for each rail pressure. Therefore, the ECU 11 extracts the rail pressure having the highest correlation when estimating the learning value in the learning rail pressure. When the ECU 11 proceeds to extraction of the optimum correlation rail pressure index as shown in FIG. 8, the ECU 11 determines whether or not the rail pressure processing loop has ended (S201). The rail pressure processing loop is performed in parallel for a plurality of rail pressures. Therefore, the ECU 11 determines whether or not a rail pressure processing loop is executed in step S201.

  When the ECU 11 determines in step S201 that the rail pressure processing loop is being performed, the ECU 11 determines whether or not the processing loop for the priority rail pressure has ended (S202). As described above, the rail pressure processing loop is performed in parallel for a plurality of rail pressures. Therefore, the ECU 11 determines whether or not a processing loop having a high priority suitable for extracting the optimal correlation rail pressure index has been completed. The relationship between the learning rail pressure and the priority rail pressure is stored as a map in a ROM (not shown) of the ECU 11.

  For example, when the rail pressure index # 5 is used as the learning rail pressure, the most suitable rail pressure for learning the fuel injection amount is the rail pressure index # 4 closest to the rail pressure index # 5. Further, when the rail pressure index # 5 is used as a learning rail pressure, when learning of the fuel injection amount at the rail pressure index # 4 is not performed, the next priority is the fuel injection amount at the rail pressure index # 3. Learning value. In this way, the ECU 11 gives priority to the fuel injection amount used for estimating the learning value for each rail pressure index that becomes the learning rail pressure.

  In addition, when learning the fuel injection amount with respect to the learning rail pressure, it is desirable to use the learning value acquired in the rail pressure processing loop that is closest in time. For example, when the rail pressure index # 5 is used as the learning rail pressure, when the learning at the rail pressure index # 3 is finished and the learning at the rail pressure index # 4 is not finished yet, the rail pressure index # 3 that is closer in time is used. It is desirable to learn the rail pressure index # 5 using the learned value of the fuel injection amount at.

  Therefore, the ECU 11 determines in step S202 whether or not the processing loop for the priority rail pressure has ended. When determining that the processing loop has ended, the ECU 11 determines that the corresponding rail pressure index is “none” and returns to step S201. (S203). On the other hand, when the processing loop for the priority rail pressure has not ended in step S202, the ECU 11 extracts the rail pressure having a higher priority from the processing loop for the priority rail pressure (S204).

  When the ECU 11 extracts a rail having a high priority, it determines whether learning of the fuel injection amount at the rail pressure has been completed (S205). When the ECU 11 determines that learning of the fuel injection amount has been completed at the rail pressure with high priority extracted in step S205, the ECU 11 determines the rail pressure extracted here as the optimum correlation rail pressure (S206). On the other hand, when the learning of the fuel injection amount is not completed at the rail pressure with the high priority extracted in step S205, the ECU 11 returns to step S201 and continues the process.

When the optimum correlation rail pressure is determined according to the flow shown in FIG. 8, the optimum correlation rail pressure index is extracted in step S104. The ECU 11 extracts the correlation coefficient m and the reflection coefficient k based on the optimum correlation rail pressure index extracted in step S104 (S105).
When the correlation coefficient m and the reflection coefficient k are extracted, the ECU 11 calculates a learning value at the learning rail pressure using the extracted correlation coefficient m and the reflection coefficient (S106). The learning value calculated here is calculated based on the following general formula as described above.

_{TQG = (TQG f × m-} TQG i-1) × k + TQG i-1
In the above formula,
TQG: Learning value at learning rail pressure to be calculated TQG _f : Learning value at other rail pressure at which learning has been completed TQG _i-1 : Previous learning value at learning rail pressure m: Between other rail pressure and learning rail pressure Correlation coefficient k: A reflection coefficient between the other rail pressure and the learning rail pressure.

  The learning value of the fuel injection amount at an arbitrary rail pressure is calculated by the above procedure. The ECU 11 calculates the learning value of the fuel injection amount for all the set rail pressures for each injector 40 provided in each cylinder 29 of the diesel engine 21. Then, the ECU 11 continues the above processing until calculation of the learning value of the fuel injection amount at all rail pressures for all the injectors 40 is completed.

  It should be noted that the ECU 11 does not estimate the learning value of the fuel injection amount from the injector 40 for all rail pressures. That is, when the ECU 11 obtains the learning value of the injection amount by actual measurement at a certain rail pressure, the ECU 11 gives priority to the actual measurement value and stores the actual measurement value as the learning value. On the other hand, when there are few opportunities for actual measurement of the fuel injection amount from the injector 40, such as the rail pressure index # 5, the fuel injection amount at the rail pressure with less learning opportunity based on the learning value measured or estimated as described above. Estimate the learning value of.

(Single injection correction)
The ECU 11 learns the change in the injection amount of the injector 40 over time for each rail pressure as described above, and also corrects the fuel injection amount by single injection. The correction of the fuel injection amount by single injection is, for example, as shown in Patent Document 1, by detecting a change in the rotation of the diesel engine 21 by single injection of a small amount of fuel at a predetermined time, and changing the injection amount from the injector 40. Learning.

  When learning by single injection is performed, if the learning of the injection amount of the injector 40 for each rail pressure (referred to as “deterioration learning”) is performed as described above, the correction amount in the deterioration learning as shown in FIG. Changes. As a result, the basic fuel injection amount changes depending on the correction amount in the deterioration learning, and an error occurs in learning by single injection. Therefore, when applying the correction value by single injection, the ECU 11 adjusts the update amount when updating the injection amount τ that is the basis before the correction to reflect the learning value in the deterioration learning at other rail pressures. ing. Thereby, ECU11 prevents the change of the injection quantity used as the foundation by deterioration learning, and has applied the correction quantity by single injection appropriately. Therefore, even when learning by single injection is performed, deterioration learning can be performed without hindering learning by single injection.

  As described above, in the embodiment of the present invention described above, when there is an undetected rail pressure for which learning has not been completed, learning of the fuel injection amount is performed based on the rail pressure for which learning has already been completed. Thereby, for example, even when the amount of learning is small, such as when the rail pressure in the common rail 25 is high, it is possible to learn the fuel injection amount that accompanies the change in the characteristics of the injector 40 over time. As a result, even if the change in the characteristic of the injector 40 due to the actual measurement value cannot be learned, the change in the characteristic of the injector 40 becomes the learning value based on the actual measurement value as shown in FIG. Approximate. Therefore, even when the characteristic change of the injector 40 cannot be learned for a long time, the error between the actually measured value and the learned value can be minimized over a wide range of rail pressure, and the fuel injection amount can be precisely controlled. .

  The present invention described above is not limited to the above-described embodiment, and can be applied to various embodiments without departing from the gist thereof.

The figure for demonstrating estimation of the fuel injection quantity by the injection quantity learning apparatus by one Embodiment of this invention. Schematic which shows the fuel-injection system of the diesel engine to which the injection quantity learning apparatus by one Embodiment of this invention is applied. Schematic diagram showing the relationship between the elapsed time and the change in the injection amount from the injector and the change in the learning value Schematic which shows the correspondence table of the rail pressure index in the injection quantity learning apparatus by one Embodiment of this invention. Schematic which shows the correspondence table of the detailed map in the injection quantity learning apparatus by one Embodiment of this invention Schematic diagram showing the relationship between reference rail pressure and reflection coefficient Schematic which shows the flow which calculates the learning value of fuel injection quantity in the injection quantity learning apparatus by one Embodiment of this invention. Schematic which shows the flow which determines the optimal correlation rail in the injection quantity learning apparatus by one Embodiment of this invention. The figure for demonstrating the setting of the update amount at the time of implementing single correction in the injection amount learning apparatus by one Embodiment of this invention.

Explanation of symbols

  In the drawing, 10 is an injection amount learning device, 11 is an ECU (injection amount learning means, undetected pressure injection amount estimation means), 20 is a fuel injection system (fuel injection device), 21 is a diesel engine, 25 is a common rail, and 29 is A cylinder 40 is an injector.

Claims (1)

In a common rail type fuel injection device that injects fuel stored in a common rail into each cylinder of a diesel engine from an injector, when the pressure of the fuel stored in the common rail is at a detected pressure, the amount of fuel injected from the injector An injection amount learning device for learning
An injection amount learning means for learning, as a detected pressure injection amount, an injection amount of fuel corresponding to the specific detection pressure when the pressure of the fuel in the common rail is at any one of the detection pressures;
Undetected pressure injection amount estimation means for estimating an injection amount of fuel corresponding to an undetected pressure that is a detected pressure other than the specific detected pressure based on the detected pressure injection amount learned by the injection amount learning means; With
The undetected pressure injection amount estimating means includes
Search for the detected pressure closest to the estimated undetected pressure or the detected pressure closest to the estimated undetected pressure.
A correlation coefficient set in advance as a value indicating a correlation between the detected pressure and the estimated undetected pressure in the detected pressure injection amount at the detected detected pressure, and the detected pressure injection amount An injection amount learning apparatus that estimates a fuel injection amount at the undetected pressure by multiplying a reflection coefficient indicating how much the undetected pressure is to be reflected .

Images