JPH10323070A - Motor controller - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a motor controller capable of command value for realizing a higher speed in operation with a small amount of calculations without deviating from restricting conditions even though vibration easily occurs due to a low rigidity of a machine driven by a motor or external disturbances such as friction or gravity is applied. SOLUTION: A restricted variable measured by actually operating a motor and a comparing means 16, 17 for comparing the variable with its allowable value, and allowable value correcting means 18, 19 for correcting the allowable value of the restricted variables by the output of these comparing means 16, 17 are added, and if the measured value exceeds the allowable value, then corresponding allowable value is corrected smaller and the command value is calculated by command value creating means 10 using the corrected allowable value.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a motor control device used for an automatic machine such as an automatic transfer machine or an automatic assembling machine or an industrial robot.

[0002]

2. Description of the Related Art In general automatic machines used for industrial purposes,
To improve productivity, it is strongly desired to increase the speed of operation. However, there is an upper limit to the torque and speed that can be output by the motor, and the upper limit is limited by the specifications of the motor. In addition, in order to guarantee the quality of the work by the automatic machine, the motor movement is also required by the work performed by the automatic machine, such as the accuracy of following the command value must be kept below the specified allowable value. Be restricted. For this reason, it is necessary to generate a command value for operating as fast as possible within a range that satisfies the constraint conditions determined by the above-described motor specifications and work requirements.

A command value generation method for realizing such a high-speed operation has been studied in the field of robots.
For example, FIG. 9 is a schematic diagram for explaining the method described in “Reduction of Manipulator Operation Time in Which a Spatial Path is Specified” (Transactions of the Society of Instrument and Control Engineers, Vol. 22, No. 10, October 1986). is there.

In the example described here, the constraints
It is considered that the motor torque is suppressed to the allowable value or less and the error in following the command value is suppressed to the allowable value or less. If the variable subject to the constraint is called a constrained variable, the constrained variable here is an error following the motor torque and the command value.

In FIG. 9, 1 is a motor, 2 is a machine driven by the motor, and 3 is a motor control means for controlling the motor 1 so as to follow a command value. Reference numeral 4 denotes a reference command value storage unit for storing a reference command value designating a desired operation of the motor 1, and 5 denotes a command value to be given to the motor control unit 3 by expanding and contracting the reference command value in the time axis direction. Command value calculating means. Reference numeral 6 denotes a time scale function storage unit for storing a time scale function. The command value calculation unit 5 expands and contracts the time axis based on the time scale function. Reference numeral 7 denotes a motor control model that models the characteristics of the motor 1, the machine, and the motor control means 3, and 8 predicts a motor torque as a constrained variable and an error following a command value based on the motor control model 7. The prediction model 9 is an optimizing means for optimizing the time scale function based on the output of the prediction model 8. Reference numeral 10 denotes a command value generating means, which comprises a reference command value storing means 4, a command value calculating means 5, a time scale function storing means 6, and an optimizing means 9, as shown in FIG. In addition, 11
Is an allowable torque value that is the upper limit of the motor torque, and 12 is an error allowable value that is the upper limit of the error following the command value.

Next, the operation of the conventional motor control device shown in FIG. 9 will be described.

The reference command value r (τ) is, for example, as shown in FIG.
(B) As shown in the above figure, the motion of the motor position is given as a function of time τ, given as a function of time τ. With this reference command value r (τ), the motor load is not so large at the beginning and end of the operation because it is a gradual movement, but the movement is abrupt near the center and the motor load is also large. For such a reference command value r (τ), where the motor load is small,
The operation can be speeded up by reducing the reference command value r (τ) in the time axis direction. Conversely, where the motor load is large, it is necessary to reduce the motor load by extending the time axis so that the motor load does not exceed the allowable value.

As described above, by expanding and contracting the reference command value r (τ) in the time axis direction according to the load of the motor, the command value r () for realizing high-speed operation within a range in which the motor load does not exceed the allowable value. t) can be generated. FIG. 10 (b)
The figure below shows an example of the command value r (t) generated in this way. In the following description, the time τ before the expansion and contraction of the time axis is referred to as virtual time, and the time t after the expansion and contraction is referred to as real time.

A function indicating the degree of expansion and contraction of the reference command value r (τ) in the time axis direction as described above is referred to as a time scale function κ (τ). Time scale function κ
(Τ) is defined as a function of virtual time τ. FIG. 1 shows an example of the time scale function κ (τ) for the above reference command value.
0 (a). The virtual time τ and the real time t are related by the time scale function as dt / dτ = κ (τ). That is, the time scale function κ
Where (τ) is greater than 1, the time is extended, and where (τ) is less than 1, the time is reduced.

The reference command value storage means 4 and the time scale function storage means 6 store the reference command value r (τ) and the time scale function κ (τ), respectively, as described above. In the means 5, a command value r (t) to be given to the motor control means 3 is generated based on both. The time scale function κ (τ) is calculated in advance before operating the motor, and is stored in the time scale function storage means 6.

The time scale function κ (τ) is generated as follows.

The motor control model 7 models the characteristics of the motor 1, the machine 2 driven by the motor 1, and the characteristics of the motor control means 3. The prediction model 8 is generated by the command value calculation means 5. Based on the motor control model 7, the motor torque when the command value is given to the motor control means 3 and the accuracy of following the command value are predicted.

The optimizing means 9 sets a time period within which the operation specified by the reference command value is completed in the shortest possible time within a range in which the motor torque and the tracking error predicted by the prediction model 8 do not exceed the allowable values. Calculate the scale function κ (τ). The time required to complete the specified action is Therefore, the optimizing means 6 determines the time scale function κ (τ) so as to minimize this. However,
τ0 is the operation start time in virtual time, and τ1 is the operation end time. Under the condition that the motor torque, the tracking error, and the like do not exceed the allowable values, the above equation (1)
The time scale function κ (τ) that minimizes the evaluation function of
It can be obtained by solving a nonlinear optimization problem with constraints in mathematical programming.

In order to more specifically show the operation and problems of the above-mentioned conventional method, the results of a simulation are shown below. For the simulation, a simple 1 shown in FIG.
A shaft motor drive system was used. In FIG. 11, 1 is a motor, 45 is an inertial load corresponding to a machine driven by the motor, and the motor 1 and the inertial load 45 are connected by a rigid shaft 43. Reference numeral 3 denotes a motor control unit which performs feedback control of the motor speed and position detected by the speed detector 41 and the position detector 42 attached to the motor 1, and controls the motor 1 in accordance with a command value. It is configured to control. FIG. 12 shows the reference command value. As the reference command value, a value which is rotated by one radian in two seconds from the stop state and stopped is used.

FIG. 13 shows a simulation result when the reference command value is directly input to the motor control means 3 as a command value. The dashed line in FIG. 13A indicates the command value input to the motor control means 3, and the solid line indicates the position of the motor 1 operating following the command value. FIG. 13B shows the motor torque, and FIG. 13C shows the error following the command value. In FIGS. 13B and 13C, the motor torque and the tracking error are normalized by respective allowable values, and −1 to +1 correspond to the allowable range. FIG.
(D) is the time scale function κ (τ), where
Since the time axis is not expanded or contracted, the time scale function κ
(Τ) is 1. The maximum value of the motor torque in FIG. 13B is about half of the allowable value, and the maximum value of the tracking error in FIG. 13C is about 80% of the allowable value, which indicates that there is room for the allowable value. .

FIG. 14 shows a simulation result when a command value is generated using the time scale function optimized by the optimizing means 9. The movement time is 2 of the reference command value.
Seconds to 1.2 seconds. Further, at the beginning and end of the operation, the motor torque reaches the allowable value, the tracking error reaches the allowable value in the middle part, and the maximum speed of the operation is performed within a range satisfying the constraint condition.

The conventional operation speed-up method is configured as described above, and a time scale function for minimizing the evaluation function of the equation (1) while keeping the error following the motor torque and the command value below an allowable value. Is calculated by the optimization means 9.

However, in the above-described conventional method, the time scale function is calculated based on the motor control model 7. Therefore, if there is a modeling error between the motor control model 7 and the characteristics of the actual motor control system, the above-described prediction is performed. The value of the constrained variable predicted by the model 8 does not match the actual value of the constrained variable. Therefore, even if the predicted value of the constrained variable does not exceed the allowable value, the actual value may not satisfy the constraint condition. For example, when the rigidity of the machine 2 is low and the machine is likely to vibrate, by increasing the speed of the operation, machine vibration is likely to occur, and the constraint may not be satisfied due to the influence of the vibration. Also, for a system that is subject to disturbances that are difficult to model, such as friction and gravity, it is necessary to drive the machine against friction and gravity. Therefore, a torque larger than the motor torque predicted by the prediction model 8 is required. I do.

In order to specifically illustrate such a problem,
FIG. 15 shows a simulation result when the command value r (t) generated using the time scale function κ (τ) calculated as described above is applied to a machine having low rigidity. FIG. 16 shows a model of a mechanical system having low rigidity used in the simulation, in which the motor 1 and the inertial load 45 are connected to the shaft 4 having low rigidity.
4 is the same as FIG.

Looking at the simulation result of FIG.
It can be seen that the motor torque exceeds the allowable value due to the mechanical vibration and does not satisfy the constraint. FIG. 17 plots the predicted values and the actual values on the same graph.
17 (a) shows the motor torque, and FIG. 17 (b) shows the following error to the command value. In each of these figures, the solid line shows the predicted value predicted based on the motor control model 7 and the prediction model 8. The broken line indicates the value when the motor is actually operated. Since the motor control model 7 does not include vibration characteristics, the effect of mechanical vibration cannot be predicted, and the actual value of the constrained variable cannot be predicted accurately. In particular, the prediction error of the motor torque is large.

[0021]

As described above, in the above-described conventional method, a time scale function is calculated using a model of a motor control system, and a command value is generated based on the time scale function. In some cases, there is a problem that a command value that does not satisfy the constraint condition is obtained.

A method of obtaining a time scale function using a more accurate model by including a vibration model and a disturbance model in the motor control model 7 is also conceivable. For this purpose, the vibration characteristics and the disturbance characteristics of the machine must be determined. It needs to be accurately identified. In reality, it is not easy to accurately identify the vibration characteristics and disturbance characteristics of a machine, and these characteristics may fluctuate due to aging, so it is necessary to identify the characteristics each time. There is a problem that it takes time and effort.

Further, in order to generate a command value for speeding up the operation as far as possible within a range satisfying the constraint condition, it is necessary to solve a nonlinear optimization problem with the constraint condition. Calculation time was required. In particular, for robots and automatic machines used for industrial purposes,
Many require synchronous operation of multiple axes. To calculate the optimal command value for such a multi-axis synchronous system, the amount of calculation increases in proportion to the square of the number of axes. There has been a problem that an enormous amount of calculation time is required as the number increases.

The present invention has been made in order to solve such a problem. Even when the rigidity of a machine driven by a motor is low and the vibration is easy, or when a disturbance such as friction or gravity is applied, the present invention is not limited to this. It is an object of the present invention to provide a motor control device that generates a command value for realizing a high-speed operation with a smaller calculation amount without departing from the conditions.

[0025]

The motor control device according to the first aspect of the present invention includes a conventional motor control device,
A comparison means for comparing the constrained variable actually measured by moving the motor with its permissible value, and a permissible value correcting means for correcting the permissible value of the constrained variable by the output of the comparing means are added. Things.

In the motor control device according to the second aspect of the present invention, when the measured value exceeds the allowable value, the corresponding allowable value is corrected to be small, and the command value generating means is used by using the corrected allowable value. Thus, a command value is calculated by the following formula.

According to a third aspect of the present invention, there is provided a motor control system comprising a motor control system including the motor, a mechanical system driven by the motor, and motor control means for controlling the motor so as to follow the command value. The frequency domain whose characteristics are accurately known is the first frequency domain, and the frequency domain whose characteristics are not exactly known is the second frequency domain,
In the conventional motor control device shown in FIG. 9, the motor control model is replaced with an approximate model that approximates the characteristic of the motor control model in the first frequency domain, and the value of the constrained variable is calculated using the approximate model. In addition to the first prediction model for predicting, a second prediction model for predicting the frequency component of the motor torque in the second frequency domain based on the approximation model is added, and in the command value generation means, While suppressing the predicted value of the constrained variable output from one prediction model to a set allowable value or less, the length of operation time required to perform a desired operation and the magnitude of the output of the second prediction model are At the same time, it is configured to generate a command value that is reduced.

According to a fourth aspect of the present invention, in the motor control device according to the third aspect, the command value generating means stores a reference command value describing a desired operation. Reference command value storage means, command value calculation means for generating a command value to be given to the motor control means by expanding and contracting the reference command value in the time axis direction, and a time expressing the rate of expansion and contraction of the reference command value in the time axis direction A time scale function storing means for storing a scale function, and an optimizing means for generating the time scale function, wherein the optimizing means calculates a predicted value of the constrained variable output from the first prediction model. Generate a time scale function that simultaneously reduces the length of the operation time required to perform a desired operation and the size of the output of the second prediction model while keeping the value below a set allowable value. Are those that you have configured so that.

A motor control device according to a seventh aspect of the present invention is the motor control device according to the conventional motor control device shown in FIG. 9 in which a motor control model is replaced by a natural vibration of a mechanical system driven by a motor of the motor control model. Replace with an approximation model that approximates the following characteristics, and in addition to the first prediction model for predicting the value of the constrained variable using the approximation model, a motor torque specific to the mechanical system based on the approximation model. A second prediction model for predicting a frequency component equal to or higher than the frequency is added, and the command value generation unit suppresses the prediction value of the constrained variable output from the first prediction model to a set tolerance or less. Generating a command value that simultaneously reduces the length of operation time required to perform a desired operation and the magnitude of the high-frequency component of the motor torque output from the second prediction model. It is obtained by sea urchin configuration.

An eighth aspect of the present invention relates to a method of generating a time scale function according to the fifth aspect of the present invention, wherein the high frequency of the motor torque output from the second prediction model is provided. The components are j (τ),
The constant multiplied by the high frequency component of the motor torque is W0, the value of the i-th constrained variable predicted by the first prediction model is Si (τ), and the weight function corresponding to the i-th constrained variable is Wi. When (τ) is set, in the above-mentioned optimization means, Is used as an evaluation function, a time scale function is calculated so as to minimize the evaluation function, and at the same time, a weighting function is corrected so that the constrained variable is equal to or less than an allowable value.

According to a ninth aspect of the present invention, there is provided a motor control device for operating a plurality of motors in synchronization with each other, wherein each motor is controlled by the motor control device according to any one of the fourth to sixth aspects. A time scale function synthesizing unit that sets the time scale function stored in the time scale function storage unit of each motor control device based on the time scale function calculated by the optimization unit of each motor control device controlled by the device is added. In the time scale function synthesizing means, when the time scale function calculated by the optimizing means of the i-th motor control device is κi (τ), The time scale function of each motor control device is set by the time scale function κ (τ) obtained by the above.

In the motor control device according to the first or second aspect of the present invention, the value of the constrained variable measured by actually moving the motor is compared with the allowable value, and the allowable value is corrected. Even when disturbance such as friction which is difficult to predict acts on the actual motor, a command value satisfying the constraint condition can be obtained.

[0033] In the motor control device according to any one of claims 3 to 6, the approximate model is a motor control system in a first frequency domain in which the characteristics of the motor control system are accurately known. Characteristics are approximated. Therefore, if the operating frequency range of the motor is limited to the first frequency range, the above approximation model is a good approximation of the actual motor control system, and the predicted value output by the prediction model is the actual motor motion. Matches well. However,
In the optimizing means, not only the time scale function is reduced, but also a time scale function that reduces the frequency component of the motor frequency predicted by the second prediction model in the second frequency region is calculated. The command value generated using such a time scale function hardly includes the frequency component in the second frequency domain. Therefore, if the motor is driven using the command value generated as described above, the operating frequency range of the motor can be limited to the first frequency range, and the actual motion of the motor can be accurately determined by the prediction model. Because it can be predicted, a command value that does not deviate from the constraint conditions can be generated even if the characteristics of the motor control system are not completely known.

According to a seventh aspect of the present invention, there is provided a motor control device according to the second or third aspect, wherein the first frequency region is driven by a motor. The frequency region is equal to or lower than the natural frequency, and the second frequency region is equal to or higher than the approximate natural frequency of the mechanical system. As a result, it is possible to generate a command value that hardly contains high-frequency components higher than the natural frequency of the mechanical system, so that a command value that satisfies the constraints without generating vibration is generated even for a machine having low rigidity and easy to vibrate. can do.

In the motor control device according to the present invention, an optimal time scale function is generated by solving a quadratic evaluation function optimization problem for a linear system. This enables the generation of command values that speed up the operation as much as possible within a range that satisfies the constraints with a simple algorithm, without solving complicated, nonlinear optimization problems with constraints that require a lot of computation time. Can be.

In the motor control device according to the ninth aspect of the present invention, when a plurality of motors are operated synchronously, all motor control devices are controlled by the maximum value of the time scale function calculated by each motor control device. The time scale function has been modified. Thereby, a command value that can realize a synchronous operation while all the motors satisfy the constraint condition can be generated with a smaller calculation amount.

[0037]

Embodiments of the present invention will be described below with reference to the accompanying drawings. In the following description, the same parts as those of the conventional control device shown in FIG. 9 are denoted by the same reference numerals.

Embodiment 1 FIG. 1 is a block diagram showing a configuration of a motor control device according to Embodiment 1 of the present invention. In FIG. 1, reference numeral 11 denotes an allowable value umax of the motor torque determined by the specification of the motor 1, and reference numeral 12 denotes an allowable value emax of a follow-up error to a command value determined by a work requirement.
Is a constrained variable measuring means for actually operating the motor 1 to measure a motor torque as a constrained variable and a tracking error to a command value; and 14 is a first maximum for detecting a maximum absolute value of the measured motor torque. Value detecting means 15; second maximum value detecting means 15 for detecting the maximum value of the absolute value of the measured tracking error;
6 is a first comparing means for calculating a ratio Ku between the output of the first maximum value detecting means 14 and the allowable value umax of the motor torque;
7 is second comparing means for calculating the ratio Ke between the output of the second maximum value detecting means 15 and the allowable value emax of the tracking error, 18
Is a first allowable value correcting means for correcting the set value of the allowable value of the motor torque based on the output Ku of the first comparing means 16, and 19 is a ratio outputted from the second allowable value comparing means 17. A second allowable value correcting means for correcting the set value of the allowable value of the following error based on Ke, 20 is a correction allowable value of the motor torque corrected by the first allowable value correcting means 16, and 21 is a second allowable value. This is a correction allowable value of the tracking error corrected by the allowable value correction unit 17.

Next, the operation of this embodiment will be described. First, the motor torque correction allowable value 2
The original motor torque allowance 11 and the original follow-up error allowance 12 are set as 0 and the correction allowance 21 of the tracking error, respectively. The time scale function is calculated by the method shown in (1) and stored in the time scale function storage means 6.

Next, a command value is generated by the command value calculating means 5 based on the reference command value and the time scale function, and the motor is driven. At the same time, the error following the motor torque and the command value is measured by the constrained variable measuring means 13. The maximum value of the absolute value of the measured motor torque is detected by the first maximum value detecting means 14, and further input to the first comparing means 16, and the output Ku of the first comparing means 16 is calculated. You. Similarly, the measured following error is input to the second maximum value detecting means 15 and the second allowable value comparing means 17, and the ratio Ke is calculated.

When the output Ku of the first comparing means 16 is larger than 1, it means that the measured motor torque is larger than the allowable value 11. At this time, the first allowable value correcting means 18 corrects the correction allowable value 20 so as to decrease the value. For example,
(The previous correction allowable value / Ku) is corrected as a new correction allowable value. Similarly, when the ratio Ke is greater than 1, the second allowable value correcting means 19 corrects the value of the correction allowable value 21 of the following error to be small.

Further, the time scale function is calculated by the optimizing means 9 using the corrected allowable values 20 and 21 thus corrected. In other words, when the constrained variable actually measured by moving the motor 1 does not satisfy the constraint, the constraint is strictly reset to generate the command value.

Thus, even when the motor control model 7 has a modeling error, it is possible to generate a command value such that the constrained variable obtained by actual measurement becomes equal to or smaller than an allowable value.

Embodiment 2 FIG. 2 is a block diagram showing a configuration of a motor control device according to Embodiment 2 of the present invention. In FIG. 2, reference numeral 22 denotes an approximation model that approximates characteristics of a motor, a motor control system including a mechanical system driven by the motor and a motor control unit in a frequency range equal to or lower than the natural frequency of the mechanical system.

Although the characteristics of the motor control system can be known relatively accurately in the low frequency range, it is difficult to accurately identify the characteristics in the high frequency range. In particular, in a frequency region equal to or higher than the natural frequency of the mechanical system, there is an influence of a higher-order vibration mode or the like, which makes it difficult to identify characteristics. Further, the vibration characteristics may fluctuate due to a change with time. Therefore, in the second embodiment, it is assumed that the characteristics of the mechanical system in the frequency range equal to or lower than the natural frequency are accurately known. Hereinafter, a frequency region below the natural frequency of the mechanical system is referred to as a first frequency region, and a frequency region above the natural frequency is referred to as a second frequency region.

In FIG. 2, reference numeral 8 denotes the approximate model 2
2 is a first prediction model for predicting a motor torque and a tracking error, which are constrained variables, based on the above-described approximation model 22, and a frequency component in the second frequency domain of the motor torque, 9 is a second prediction model that outputs a frequency component in a frequency region higher than the natural frequency of.

An optimizing means 26 generates a time scale function based on the first prediction model 8 and the second prediction model 23. The optimizing means 26 sets an evaluation function within a range in which the value of the constraint variable predicted by the first prediction model is equal to or less than an allowable value. Generate a time scale function κ (τ) in which is minimized. Here, τ0 is the operation start time in virtual time, τ1 is the operation end time, j (τ) is the frequency component of the motor torque output from the second prediction model in the second frequency domain,
W0 is a constant. The first term of the above equation (4) is a first evaluation function 25 for evaluating the length of operation time required to perform a desired operation given by the reference command value, and the second term is the second evaluation function. This is a second evaluation function 24 for evaluating the magnitude of the output of the prediction model, and the optimizing means 26 calculates a time scale function so that the sum of the two becomes minimum.

Next, the operation of the second embodiment will be described. FIG. 16 shows an example of the motor 1, a mechanical system driven by the motor 1, and a motor control system including the motor control means. That is, the machine 2 in FIG. 2 corresponds to the low rigid shaft 44 and the inertial load 45 connected thereto in FIG.

The approximate model 22 for the motor control system shown in FIG. 16 is as shown in FIG. FIG.
6 is obtained by replacing the low rigidity shaft 44 with the rigid high shaft 43. FIG. 3 shows the frequency characteristics of FIG. 16 and FIG. In FIG. 3, the dashed line is the frequency characteristic of the actual motor control system of FIG. 16, and the solid line is the frequency characteristic of the approximate model of FIG. The natural frequency of the mechanical system used here is about 2
4 rad / s, and the frequency characteristics of the two are different in a higher frequency region higher than that, but they are well matched in a low frequency region of 20 rad / s or less.

The virtual time τ and the real time t are related by using a time scale function κ (τ) as follows: dt / dτ = κ (τ) (5)

The state equation of the approximation model 22 in FIG. 2 is expressed as dx (t) / dt = A (x (t)) + B (x (t)) · r (t) (6) Here, x (t) is a state variable of the approximate model 22,
r (t) is a command value given to the motor control means, A (x
(T)) and B (x (t)) are matrices representing the characteristics of the approximate model 22.

At this time, the motor torque u (t) and the tracking error e (t) output from the first prediction model 8 are respectively u (t) = Cu · x (t) + Du · r (t) (7) e (t) = Ce × (t) + Der (t) (8) Here, Cu, Du, Ce, and De are matrices determined from the characteristics of the approximate model 22.

The high-frequency component of the motor torque, which is the output of the second prediction model 23, is given by: dxf (t) /dt=Af.xf (t) + Bfu.u (t) (9) j (t) = Cf Xf (t) + Dfu (t) (10) Here, equation (9) is a high-pass filter for removing a frequency component of 20 Hz or less lower than the natural frequency of the mechanical system, xf (t) is a state variable of the high-pass filter, and u (t) is a first prediction model 8. J (t) is the high-frequency component of the motor torque which is the output of the second predicting means 23, Af, Bf, Cf, D
f is a constant matrix representing the characteristics of the high-pass filter.

As a constraint, the motor torque u (t)
And the following error e (t) to the command value are the respective allowable values um.
do not exceed ax and emax. This constraint condition is defined by the motor torque u (t) and the error e following the command value.
When expressed by an expression using (t), -umax ≦ u (t) ≦ umax (11) −emax ≦ e (t) ≦ emax (12)

In the optimizing means 26, the evaluation function, Calculate a time scale function that minimizes. In other words, the optimizing unit applies the constraint conditions (11) and (1) to the system described by equations (5) to (10).
A time scale function κ (τ) that minimizes the evaluation function equation (13) within a range that satisfies 2) will be calculated. Such a time scale function κ (τ) is generally used. It can be obtained by using a nonlinear optimization method with constraints.

The time scale function calculated in this way is stored in the time scale function storage means 6. When the motor 1 is actually operated, the reference command value r (τ) stored in the reference command value storage means 4 and the time scale function κ (τ) stored in the time scale function storage means 6 are used.
The command value calculation means 5 generates the command value r (t) using The generated command value r (t) is input to the motor control means 3, and the motor 1 is controlled to follow the command value.

The command value r (t) thus generated
FIG. 4 shows a result obtained by simulating the movement when is given to the real model shown in FIG. The conditions of the simulation are exactly the same as those of the simulation by the conventional motor control device shown in FIG. From FIG. 4, the movement time is reduced from 2 seconds of the reference command value to 1.5 seconds. Although longer than 1.2 seconds according to the conventional method, both the motor torque and the tracking error are within the allowable values.

FIG. 5 shows a comparison between the predicted value output from the prediction model 8 and the measured value in the actual system. FIG. 5A shows the motor torque, FIG. 5B shows the error following the command value, the broken line shows the measured value, and the solid line shows the predicted value. Since the command value does not include a high-frequency component and the operating frequency region of the motor 1 is limited to the low-frequency region, the two values are in good agreement.

In the second embodiment, the square integral of the high frequency component of the motor torque is calculated as the second evaluation function 24. However, the second evaluation function 24 determines the magnitude of the high frequency component of the motor torque. The evaluation is not limited to the above. For example, the same applies to the integration of the absolute value of the high frequency component of the motor torque, the integration of the fourth power, or the maximum value of the absolute value.

Embodiment 3 FIG. 6 is a block diagram showing a configuration of a motor control device according to Embodiment 3 of the present invention. 6, the same parts as those in FIG. 2 are denoted by the same reference numerals. Reference numeral 31 denotes first weight function storage means for storing a weight function corresponding to a motor torque which is a first constrained variable, and 32 denotes a weight function corresponding to a tracking error which is a second constrained variable. Is a first comparing means for comparing the predicted value of the motor torque predicted by the first prediction model 8 with the allowable value 11 of the motor torque, and 28 is a first weighting function storing means for storing the first value. Prediction model 8
The second comparing means 29 for comparing the predicted value of the following error predicted by the above with the allowable value 12 of the following error, the first weighting function storing means 29 based on the comparison result of the first comparing means 27 A first weighting function correcting means for correcting the weighting function stored in 31, 30 corrects the weighting function stored in the second weighting function storing means 32 based on the comparison result of the second comparing means 28. This is the second weighting function correcting means.

As in the second embodiment, the state equation of the approximate model shown in FIG. 11 is expressed as follows: dx (t) / dt = A (x (t)) + B (x (t)) · r (t) (14) ) Motor torque u output from the first prediction model
(T) and the following error e (t) are expressed as follows: u (t) = CuＣｕx (t) + Duｒr (t) (15) e (t) = C ・ x (t) + De ・ r (t) ( 16) Further, the state equation of the high-pass filter included in the second prediction model is expressed as follows: dxf (t) / dt = Af × f (t) + Bfuu (t) (17) j (t) = Cf × f ( t) + Df · u (t) (18) The constraint condition is set as −umax ≦ u (t) ≦ umax (19) −emax ≦ e (t) ≦ emax (20)

When these equations are converted into equations in the virtual time τ domain using the above equation (5), dx (τ) / dτ = [A (x (τ)) + B (x (τ)) · r ( τ)] · κ (τ) (21) dxf (τ) / dτ = [Af · xf (τ) + Bf · u (τ)] · κ (τ) (22) u (τ) = Cu · x (τ ) + Du · r (τ) (23) e (τ) = Ce · x (τ) + De · r (τ) (24) j (τ) = Cf · xf (τ) + Dfu · (τ) (25) -Umax ≤ u (τ) ≤ umax (26)-emax ≤ e (τ) ≤ emax (27) is obtained.

In the optimizing means, the above equations (21) to (25)
Evaluation function for the system represented by A time scale function κ (τ) that minimizes is obtained. Here, W1 (τ) and W2 (τ) are the weight functions stored in the first and second weight function storage means 31 and 32, respectively.

However, the state equations of equations (21) and (22) are nonlinear with respect to the state variables x (τ) and xf (τ) and the time scale function κ (τ). To determine τ), it is necessary to solve a non-linear optimization problem, which requires complicated calculations. Therefore, equations (21) and (22) are linearly approximated in the vicinity of an appropriate time scale function κ0 (τ), and the time scale function κ that minimizes the evaluation function of equation (28) for the approximated linear system (Τ) is obtained.
For a linear system, κ (τ) that minimizes the quadratic form evaluation function of Equation (28) can be easily obtained by solving a matrix Riccati equation based on optimal control theory.

However, this alone only minimizes the evaluation function of equation (28), and does not guarantee that the constrained variable will be less than the allowable value. Therefore, the first and second weighting function correcting means 29 and 30 are set so that the constrained variable is equal to or less than the allowable value.
To correct the values of the weighting functions W1 (τ) and W2 (τ). At a time when the constrained variable is larger than the allowable value, the value of the corresponding constrained variable can be reduced by increasing the value of the weighting function.

By the above method, the time scale function κ
The algorithm for calculating (τ) is shown in the flowchart of FIG.

In FIG. 7, first, in step (a), the allowable values umax and emax of the motor torque u (τ), which is a constrained variable, and the error e (τ) following the command value are determined based on the motor specifications and work. Set based on request.

In step (b), the state equation of the approximate model is converted into a state equation in the virtual time domain. This conversion is as shown in Expressions (21) to (27).

In step (c), the time scale function κ
As an initial value of (τ), κ0 (τ) = 1 is substituted.

In step (d), the state equation in the virtual time domain obtained in step (b) is linearly approximated near κ0 (τ).

In step (e), an evaluation function is applied to the linear system obtained in step (b). Δκ (τ) is calculated such that is minimized. Here, κ (τ) = κ0 (τ) + Δκ (τ).

Such Δκ (τ) can be easily obtained by solving the matrix Riccati equation using the optimal control theory.

In step (f), the end of the repetitive calculation is determined. That is, as the determination conditions, (1) all the constraint conditions are satisfied, and Δ
The norm of κ (τ) is sufficiently small. (2) The number of repetition calculations is greater than a certain value. If either of the above (1) and (2) is satisfied using
0 (τ) is set to the optimal κ (τ), and the calculation is repeated. Otherwise, go to step (g).

In step (g), the weighting function W1 is set as follows so that the constrained variable becomes equal to or less than the allowable value.
(Τ) and W2 (τ) are corrected.・ W1 (τ) ・ u (τ) / umax is converted to a new W1
(Τ).・ W2 (τ) · e (τ) / emax is converted to a new W2
(Τ). That is, at the time when the constrained variable becomes larger than its allowable value, the size of the constrained variable is suppressed by increasing the weighting function. Conversely, at a time when the value of the constrained variable is smaller than the allowable value, the weighting function is modified to be small.

In step (h), κ0 (τ) + Δκ
(Τ) is set as a new κ0 (τ), the process returns to step (2), and the calculation is repeated.

In the above algorithm, step (g)
The weighting function is corrected at the time when the constrained variable exceeds the allowable value, and the weighting function is corrected small at the time when the variable does not exceed the allowable value. As a result, when the time exceeds the allowable value, the weight increases, and the value of the corresponding constrained variable decreases. On the other hand, at a time when the value does not exceed the allowable value, the value of the weight function is close to zero, so that the square sum of the time scale function and the torque high-frequency component is substantially obtained, that is, κ (τ) · κ (τ) + W0 · j A time scale function that reduces (τ) · j (τ) can be obtained.

That is, by the above-described iterative calculation algorithm, it is possible to find a time scale function that makes each constrained variable equal to or less than its allowable value and at the same time minimizes both the time required for the desired operation and the high frequency component of the motor torque as much as possible. it can.

In the above-mentioned algorithm, a constrained optimization problem for a nonlinear system is solved by repeatedly solving an optimization problem for minimizing a quadratic evaluation function for a linear system. In general, a constrained optimization problem for a nonlinear system is computationally complicated and requires a long calculation time, but a quadratic evaluation function minimization problem for a linear system can be solved relatively easily. The above algorithm solves the optimization problem in step (e) for the system linearly approximated in step (d), and can solve a complicated optimization problem by repeating relatively simple calculations.

Embodiment 4 The amount of calculation for finding the optimal time scale function for synchronously operating a plurality of motors increases in proportion to the square of the number of motor shafts. For example,
The order of the matrix Riccati equation solved in step (e) of the flowchart shown in FIG. 7 of the third embodiment is (order of all motor control systems) × (order of all motor control systems). The order of all motor control systems is (the order of one motor control system) x
(The number of motor axes), the amount of calculation for solving the matrix Riccati equation increases in proportion to the square of the number of motor axes.

FIG. 8 is a block diagram showing a configuration of a motor control device according to a fourth embodiment of the present invention, and shows an example in which the present invention is applied to a mechanical system that requires synchronous control of two motors.

In FIG. 8, reference numeral 1-1 denotes a first motor, 1-2 denotes a second motor, and the same parts as those in FIG. 2 are denoted by the same reference numerals. However, "-" attached to each code
"1" relates to the first motor, and "-2" relates to the second motor. Reference numeral 34 denotes a prediction model, which is a combination of the first prediction model 8 and the second prediction model 23 in FIG. 2, and reference numeral 35 denotes an evaluation function.
Is a combination of the first evaluation function 25 and the second evaluation function 24. Further, 36 is a time scale function synthesizing means.

Optimizing means 2 for first motor 1-1
In 6-1, the time scale function κ1 (τ) is calculated by the same method as in the second embodiment, and the time scale function κ2 (τ) is similarly calculated in the optimizing means 26-2 for the second motor 1-2. Is calculated. The time scale function calculated in this manner is input to the time scale function synthesizing means 36, and a new time scale function κ (τ) is obtained from the input κ1 (τ) and κ2 (τ) as shown in the following equation. ) Is generated.

Κ (τ) = max {κ1 (τ), κ2 (τ)} The time scale function thus generated is firstly
Time scale function storage means 6 for the second motor
1, 6-2 and stored.

As a result, the time axis is expanded and contracted for each reference command value using the same time scale function, so that the synchronous operation of the two motors can be realized, and the constraint conditions for both motors are satisfied. High-speed operation can be realized within the range. Further, the amount of calculation required for this only increases in proportion to the number of motor shafts, and the time scale function can be obtained with a small amount of calculation.

[0085]

As described above, the motor control device according to the present invention has the following excellent effects.

According to the motor control device of the first or second aspect, as described in the first embodiment, the allowable value is corrected by using the constrained variable measured by actually moving the motor. Therefore, even if there is a modeling error, there is an effect that a command value for accelerating the operation while satisfying the constraint condition can be generated.

According to the motor control device of any one of claims 3 to 7, as described in the second embodiment, the characteristics of the motor control system include an inaccurate frequency domain frequency component. This makes it possible to generate a high-speed command value that satisfies the constraint conditions even for a machine whose characteristics are not precisely known or for a machine with low rigidity and easy to vibrate.

According to the motor control device according to the fifth and eighth aspects, as apparent from the third embodiment, it is possible to solve a complicated optimization problem by repeating relatively simple calculations. There is an effect that an optimum speed-up command value can be obtained in a short calculation time.

According to the motor control device of the ninth aspect,
When controlling a machine that requires synchronous operation of multiple axes,
Since the time scale function is calculated independently for each motor and they are combined to obtain one time scale function, the amount of calculation for obtaining the time scale function increases in proportion to the square of the number of axes. Therefore, there is an effect that a high-speed command value capable of realizing a synchronous operation can be generated with a small amount of calculation without performing.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a motor control device according to a first embodiment of the present invention.

FIG. 2 is a block diagram illustrating a configuration of a motor control device according to a second embodiment of the present invention.

FIG. 3 is a diagram comparing frequency characteristics of an actual system and an approximate model according to the present invention.

FIG. 4 is a diagram showing a simulation result when the motor control device of FIG. 2 is used.

FIG. 5 is a diagram comparing a predicted value and a measured value by the motor control device of FIG. 2;

FIG. 6 is a block diagram illustrating a configuration of a motor control device according to a third embodiment of the present invention.

FIG. 7 is a flowchart illustrating an algorithm for calculating a time scale function κ (τ) according to the present invention.

FIG. 8 is a block diagram illustrating a configuration of a motor control device according to a fourth embodiment of the present invention.

FIG. 9 is a block diagram showing a conventional command value generation method.

FIG. 10 is a diagram showing a relationship between a virtual time τ and a time scale function κ (τ) according to the present invention.

FIG. 11 is a schematic view showing a one-axis rigid motor drive system.

FIG. 12 is a graph showing a waveform of a reference command value according to the present invention.

FIG. 13 is a diagram illustrating a simulation result when a reference command value according to the present invention is used.

FIG. 14 is a diagram illustrating a simulation result when a conventional method is applied to a rigid system.

FIG. 15 is a diagram illustrating a simulation result when a conventional method is applied to a vibration system.

FIG. 16 is a schematic view showing a motor drive system of one axis which easily vibrates.

FIG. 17 is a diagram comparing a predicted value and a measured value according to a conventional method.

[Explanation of symbols]

1 motor, 2 machines driven by motor, 3
Motor control means, 4 reference command value storage means, 5 command value calculation means, 6 time scale function storage means, 7 motor control model, 8 prediction model, 9 optimization means, 1
0 command value generating means, 11 torque allowable value, 12 error allowable value, 13 constrained variable measuring means, 14 first maximum value detecting means, 15 second maximum value detecting means, 16 first comparing means, 17th Second comparing means, 18 first allowable value correcting means, 19 second allowable value correcting means, 20 motor torque correcting allowable value, 21 tracking error correcting allowable value,
Reference Signs List 22 approximate model, 23 second prediction model, 24 second evaluation function, 25 first evaluation function, 26 optimization means, 27 first comparison means, 28 second comparison means, 2
9 first weight function correction means, 30 second weight function correction means, 31 first weight function storage means, 32 second weight function storage means, 33 optimization means, 34 prediction model, 35 evaluation function, 36 Time scale function synthesizing means, 43 axes, 44 axes with low rigidity, 45 inertia load.

Claims (9)

[Claims] 1. A constrained variable to which a constraint is imposed from a motor specification, a motor specification, and a work requirement using a motor, such as a motor torque, a motor speed, an acceleration, an effective torque, and a tracking error with a command value, is set in advance. A motor control device that generates a command value for performing a desired operation in the shortest time while controlling the motor to operate in accordance with the generated command value. A motor control model that models a driven machine and a motor control system including a motor control unit that controls the motor so as to follow the command value; and the constrained variable when the motor is actually operated. Constrained variable measuring means for measuring the value of the maximum value detecting means for detecting the maximum value of the measured absolute value of the constrained variable; Comparing means for comparing the allowable value with the allowable value; allowable value correcting means for correcting the allowable value of the constrained variable based on the comparison result of the comparing means; and a constrained variable predicted using the motor control model. Command value generating means for generating a command value for performing a desired operation in the shortest time and outputting the command value to the motor control means, while suppressing the predicted value to be equal to or less than the corrected correction allowable value. Motor control device. 2. The motor control device according to claim 1, wherein: (1) the motor is operated by a command value generated by the command value generation means; and (2) the maximum value detection means includes: Comparing the output with an original permissible value determined from the specification of the motor or a requirement in operation; (3) when the output of the maximum value detecting means is larger than the original permissible value, And (4) the command value generation means generates a command value again using the corrected correction value, and (5) the restricted variable is larger than the original allowable value. A motor control device characterized by repeating the above (1) to (4) until it becomes smaller. 3. A constrained variable to which a constraint is imposed from a motor specification such as a motor torque, a motor speed, an acceleration, an effective torque, an error in following a command value, and a work requirement using the motor is set in advance. A motor control device that generates a command value for performing a desired operation in the shortest time while controlling the motor so as to follow the generated command value, wherein the motor and the motor are driven by the motor. Mechanical systems,
The first frequency domain in which the characteristics of the motor control system including the motor control means for controlling the motor so as to follow the command value are accurately known, and the frequency domain in which the characteristics are not precisely known. The second frequency domain,
An approximate model that approximates the characteristic of the motor control system in the first frequency domain; a first prediction model that predicts the value of the constrained variable using the approximate model; and a motor torque based on the approximate model. A second prediction model for predicting a frequency component of the second frequency domain, and a command value generating means for generating a command value to be given to the motor control means, and an operation required to perform the desired operation. A first evaluation function that evaluates the length of time and a second evaluation function that evaluates the magnitude of the output of the second prediction model are set, and the command value generation unit sets the first prediction function A command that suppresses the predicted value of the constrained variable predicted by the model to be equal to or less than the set allowable value and minimizes the sum of the value of the first evaluation function and the value of the second evaluation function. Generating values Motor control device according to claim. 4. Preliminarily set a motor specification such as a motor torque, a motor speed, an acceleration, an effective torque, and a tracking error with a command value, and a constrained variable to which a constraint is imposed by a work requirement using the motor. Generate a command value that performs the desired operation in the shortest time at the same time as keeping it below the allowed value,
In a motor control device that controls a motor so as to follow a generated command value, a virtual time τ and an actual time t are represented by dt / dτ = κ.
Time scale function storage means for storing a time scale function κ (τ) related by (τ), a reference command value r given as a function of the virtual time τ
Reference command value storage means for storing (τ), the reference command value r (τ) and the time scale function κ
(Τ), a command value calculating means for calculating a command value r (t) at an actual time, the motor, a mechanical system driven by the motor, and controlling the motor so as to follow the command value. The frequency domain in which the characteristics of the motor control system comprising the motor control means are accurately known is the first frequency domain, and the frequency domain in which the characteristics are not exactly known is the second frequency domain,
An approximate model that approximates the characteristic in the first frequency domain of the motor control system; a first prediction model that predicts the value of the constrained variable based on the approximate model; and a motor based on the approximate model. A second prediction model for predicting a frequency component of the torque in the second frequency domain, and an optimization unit for calculating the time scale function κ (τ) based on the first and second prediction models. With, a first evaluation function to evaluate the length of operation time required to perform the desired operation, and a second evaluation function to evaluate the magnitude of the output of the second prediction model, The optimization means suppresses the predicted value of the constrained variable predicted by the first prediction model to a value equal to or less than the set allowable value, and at the same time, sets the value of the first evaluation function and the second evaluation function. Minimizes the sum of the values of Cormorant Do time scale function kappa (tau) motor control device and generates a. 5. Preliminarily set a motor specification such as a motor torque, a motor speed, an acceleration, an effective torque, an error following a command value, and a constrained variable to which a constraint is imposed by a work requirement using the motor. Generate a command value that performs the desired operation in the shortest time at the same time as keeping it below the allowed value,
In a motor control device that controls a motor so as to follow a generated command value, a virtual time τ and an actual time t are represented by dt / dτ = κ.
Time scale function storage means for storing a time scale function κ (τ) related by (τ), a reference command value r given as a function of the virtual time τ
Reference command value storage means for storing (τ), the reference command value r (τ) and the time scale function κ
(Τ), a command value calculating means for calculating a command value r (t) at an actual time, the motor, a mechanical system driven by the motor, and controlling the motor so as to follow the command value. The frequency domain in which the characteristics of the motor control system comprising the motor control means are accurately known is the first frequency domain, and the frequency domain in which the characteristics are not exactly known is the second frequency domain,
An approximate model that approximates the characteristic in the first frequency domain of the motor control system; a first prediction model that predicts the value of the constrained variable based on the approximate model; and a motor based on the approximate model. A second prediction model for predicting a frequency component of the torque in the second frequency domain; a weight function storage unit for storing a weight function corresponding to each constrained variable; Comparing means for comparing the value of the constraint variable with the allowable value of the constrained variable; weight function correcting means for correcting a weight function stored in the weight function storing means based on a result of the comparing means; Optimizing means for calculating the time scale function κ (τ) based on the outputs of the first and second prediction models and the value of the weight function stored in the weight function storage means. Characteristic motor control device. 6. The motor control device according to claim 5, wherein the optimizing unit calculates a square of each of the constrained variables output from the first prediction model and calculates a weighted function stored in the weighting function storage unit. The sum of the value obtained by multiplying the value, the constant times the square of the predicted value output from the second prediction model, and the square of the time scale function is calculated from the motor operation start time to the motor end time by the virtual time. Calculating the time scale function such that the value integrated in the area is minimized; and the weighting function correcting means, at a time when the constrained variable predicted by the prediction model is larger than its allowable value, Is increased, and conversely, if the predicted value of the constrained variable is smaller than the allowable value, the weight function is modified so as to decrease the corresponding weight function within a range not negative, Before The motor control device, wherein the optimization unit calculates the time scale function using the weight function corrected as described above. 7. The motor control device according to claim 3, wherein a frequency region equal to or lower than an approximate natural frequency of a mechanical system driven by the motor is set as the first frequency region, A motor control device, wherein a frequency region equal to or higher than the approximate natural frequency of the mechanical system is set as the second frequency region. 8. The motor control device according to claim 5, wherein said optimizing means calculates a time scale function κ (τ) in the following steps (a) to (h). . (B) Set the allowable value SiMax of the i-th constrained variable. (B) Convert the state equation of the approximate model into a state equation in the virtual time τ region. (C) The initial value of the time scale function κ (τ) is κ0
(Τ) = 1. (D) The state equation in the virtual time τ region obtained in (c) is linearly approximated near κ0 (τ). (E) Evaluation function for the linearly approximated state equation obtained in (d) Δκ (τ) for minimizing is obtained. Where κ (τ) = κ
0 (τ) + Δκ (τ), j (τ) is the predicted value output from the second prediction model, W0 is a constant that multiplies the j (τ), and Si (τ) is the i-th Constraint variables, Wi
(Τ) is a weight function corresponding to the i-th constrained variable. (F) All the constrained variables are less than or equal to their allowable values, and
If the norm of Δκ (τ) is sufficiently small, or if the number of repetitive calculations becomes equal to or more than a certain value, κ0 (τ) is terminated as the optimal κ (τ), otherwise, the process proceeds to step (g). . (G) In the weight function correcting means, Wi (τ) *
The weight function is corrected by setting Si (τ) / SiMax as a new Wi (τ). (H) Set κ0 (τ) + Δκ (τ) as a new κ0 (τ) and return to step (c). 9. A motor control device for operating a plurality of motors in synchronization with each other, wherein each motor is controlled by the motor control device according to any one of claims 4 to 6, and each motor control device calculates a value. A time scale function synthesizing unit that sets a time scale function stored in a time scale function storage unit of each motor control device based on the time scale function; When the time scale function calculated by the optimizing means is κi (τ), A time scale function of each control device is set according to a time scale function κ (τ) obtained by the motor control device.