JP7275389B2 - Servo controller - Google Patents

Description

The present disclosure relates to a servo control device that controls a servo motor that drives a control target axis of a machine tool or the like.

In general, in servo control devices that drive and control motors that drive controlled objects such as machine tools, robots, and industrial machines, the drive current to the motors is controlled so as to achieve the position and speed commanded by the command value generator. It is done. An example of a command value generating device is a numerical control (NC) device or motion controller. In particular, when the machining tool moves on the movement locus indicated by the machining program, the motor drive control is performed while precisely managing the position.

Patent Literature 1 discloses a control device that predicts a command value from a work to a production facility that produces a product using a prediction model, using input data of feature values of the work and attribute values of the environment in which the product is produced. . In the control device described in Patent Document 1, after starting the calculation processing of the prediction model, based on the remaining processing time until the calculation processing of the prediction model is completed, the command value is determined based on the output value obtained from the prediction model. is in time for the control timing for controlling the production operation by the production equipment. If the command value is determined in time for the control timing, the command value for the production device is determined based on the output value obtained by completing the arithmetic processing of the prediction model. On the other hand, if the command value cannot be determined in time for the control timing, the calculation process of the prediction model is stopped, and it is determined whether the stopped result is reliable as the prediction result. A command value to the production equipment is determined based on.

JP 2019-179468 A

By the way, in the motor drive control, there is known a technique of correcting the command value using a model for estimating the correction amount of the command value based on data such as the position and speed of the motor obtained by a sensor or the like. However, if the model does not fully consider the phenomenon that occurs in the motor drive control, the correction effect may not appear at all, or the correction may be overcorrected, and the result may be worse depending on the correction. rice field. In addition, there is a possibility that unexpected corrections may be made when the input to the model is disturbed by disturbance or the like or when there is an unknown input. Therefore, it is conceivable to judge how reliable the value estimated by the model is, that is, the uncertainty of the prediction result, and use it for control. However, the reliability in the technology described in Patent Document 1 is the reliability when compared with the result calculated to the end with the prediction model, and the reliability for determining the uncertainty of the prediction result obtained from the prediction model. isn't it. In other words, in the technique described in Patent Literature 1, the result of the computation to the end of the prediction model is used to generate the command value without considering the uncertainty of prediction. Therefore, when the command value is corrected using a model that estimates the correction amount based on the operating state of the motor, it is possible to determine how reliable the estimated correction amount is at the time of estimating the correction amount. A technology was needed.

The present disclosure has been made in view of the above, and provides a servo control device that can determine, at the time of estimating the correction amount, how reliable the estimated correction amount after the time of estimating the correction amount is. for the purpose.

In order to solve the above-described problems and achieve an object, a servo control device according to the present disclosure controls the operation of a servomotor based on a command value that is periodically input to instruct the operation of the servomotor. The servo control device includes a correction section, a determination section, and a servo amplifier. The correction unit includes a candidate command value that is a command value obtained by correcting the command value based on the command value and the actual measurement result of the operation of the servomotor, a reliability index that is an index for evaluating the reliability of the candidate command value, is determined, and a corrected command value, which is a command value after correction for controlling the servomotor, is output. The judging section judges permission or non-permission of application of the candidate command value to the control of the servomotor based on the reliability index, and outputs the judgment result to the correcting section. The servo amplifier controls the servomotor based on the corrected command value. The correction unit has a prediction unit and a correction selection unit. The prediction unit uses a prediction model that predicts the state quantity of the current or future servomotor operation, inputs the command value and the actual measurement result of the servomotor operation, and determines the candidate command value at the time after the time point A predicted state quantity obtained by predicting the state quantity of the operation of the servo motor is predicted, and the reliability of the predicted state quantity is evaluated. The correction selection section determines a candidate command value based on the predicted state quantity. The prediction unit uses the reliability of the predicted state quantity as a reliability index. A correction selection section of the correction section outputs the candidate command value as a post-correction command value to the servo amplifier based on the determination result.

The servo control device according to the present disclosure has the effect of being able to determine, at the time of estimating the correction amount, how reliable the estimated correction amount after the time of estimating the correction amount is.

1 is a block diagram schematically showing an example of a configuration of a servo control device according to a first embodiment; FIG. A diagram for conceptually explaining predicted values and variations in predicted values 1 is a flow chart showing an example of a procedure of a control method in a servo control device according to Embodiment 1; FIG. 2 is a block diagram schematically showing an example of the configuration of a servo control device according to Embodiment 2; FIG. 11 is a block diagram schematically showing an example of a configuration of a model information updating unit of a servo control device according to Embodiment 2; A block diagram schematically showing an example of the configuration of a servo control device according to a third embodiment. FIG. 2 is a diagram schematically showing an example of a hardware configuration for realizing servo control devices according to Embodiments 1, 2, and 3;

A servo control device according to an embodiment of the present disclosure will be described in detail below with reference to the drawings.

Embodiment 1.
In drive control of a servomotor, there is a problem that a trajectory error occurs due to the influence of frictional disturbance when the rotation direction of the servomotor is reversed. This trajectory error is also called quadrant protrusion or lost motion, and a technique for suppressing the trajectory error is desired. As an example of technology for suppressing trajectory errors, there is a method of correcting a command value using a model that estimates a correction amount for suppressing trajectory errors based on data such as the position and speed of a motor obtained by a sensor or the like. However, if the phenomenon that causes the trajectory error is not sufficiently considered, the correction effect may not appear at all, or the correction may be overcorrected, and depending on the correction, the result may rather deteriorate. In addition, there is a possibility that an unexpected correction may be made when the input to the model is disturbed by disturbance or the like or when there is an unknown input. Therefore, it is possible to suppress the trajectory error by judging how reliable the values estimated after the estimation of the correction amount using the model, that is, the uncertainty of the prediction result, and utilizing it for control. becomes possible. In the following embodiments, a servo control device will be described that can determine the uncertainty of the prediction of the result estimated using the model, and can suppress the occurrence of trajectory error in drive control of the servo motor.

FIG. 1 is a block diagram schematically showing an example of the configuration of a servo control device according to Embodiment 1. FIG. The servo control device 1 is electrically connected to the command value generator 2 and the servomotor 3 . The servo control device 1 performs control based on the command value periodically generated by the command value generator 2 so that the actual operating state of the servo motor 3 matches the corrected command value. An example of the state of operation is the position, velocity and acceleration of the servomotor 3 . The servo control device 1 generates current and voltage for driving the servomotor 3 based on the command value generated by the command value generator 2 and applies the current and voltage to the servomotor 3 .

The command value generator 2 is a device that outputs a command value to the servo control device 1 at a servo control cycle that is a predetermined time interval. The command value generator 2 is realized by, for example, a numerical control device or a motion controller, and generates command values using known techniques. Here, the command value is a target value for controlling the operating state or control amount of the servomotor 3 to a desired state, and is at least one of the position, speed, acceleration, torque, current, and model position of the servomotor 3. including one. The model position is an approximate position of the servomotor 3 obtained by calculation, and is an estimated actual position of the servomotor 3 in the current servo control cycle. By using a servo model that simulates the structure of the servo motor 3, it is possible to estimate the model position. This servo model is simply defined as a first-order lag filter with a cutoff frequency, and can generally be treated as a filter with low-pass characteristics.

The servomotor 3 is a driving device for performing drive control of a controlled object via a power transmission mechanism such as a ball screw (not shown), and receives voltage from a servo amplifier 11 in the servo control device 1 to rotate. Examples of controlled objects are machine tools, robots, or industrial machines. The servomotor 3 includes a position detector such as an encoder that detects the position of the servomotor 3 . The position detector outputs the detected position to the servo control device 1 . The position detected by the position detector is input to the servo amplifier 11 and the correction section 12 in the servo control device 1 .

The servo control device 1 includes a servo amplifier 11 , a correction section 12 and a determination section 13 . The servo amplifier 11 is a device that controls the position, speed, and acceleration of the servomotor 3 at predetermined time intervals, ie, servo control cycles. The servo amplifier 11 actually performs control so that the actual operating state of the servo motor 3 coincides with the command value corrected by the correction unit 12, which will be described later. Hereinafter, the command value used when controlling the operation of the servomotor 3 is referred to as a corrected command value.

Based on the command value generated by the command value generating unit 2 and state quantities such as the actual position and speed of the servomotor 3, the correction unit 12 generates a candidate command value, which is a candidate for the current or future post-correction command value. and a reliability index that is an index for evaluating the reliability of this candidate command value. Further, the correction unit 12 outputs a corrected command value, which is a corrected command value for controlling the servomotor 3 . Specifically, the correction unit 12 inputs the command value generated by the command value generation unit 2 and state quantities such as the actual position and speed of the servomotor 3, and calculates the candidate command value and the reliability by a method described later. Determine and output the degree index. The actual state quantity of the servomotor 3 is the actual measurement result of the operation of the servomotor 3 .

Based on the reliability index output from the correcting unit 12 , the determining unit 13 determines whether to permit or disallow application of the candidate command value to the control of the servomotor 3 , and outputs the determination result to the correcting unit 12 . Specifically, if the determination unit 13 determines that the application of the candidate command value is reliable based on the reliability index output from the correction unit 12, the determination unit 13 provides the correction unit 12 with the candidate command value. A determination result that permits output to the servo amplifier 11 is output. If the determination unit 13 determines that application of the candidate command value is unreliable based on the reliability index output from the correction unit 12, the determination unit 13 instructs the correction unit 12 to transfer the candidate command value to the servo amplifier 11. Output the judgment result that the output is not permitted. The reliability index is an index for determining the uncertainty of the predicted result, and is an index for determining whether or not the predicted result can be used to correct the command value. Since the correction unit 12 uses the prediction model to predict the future state of the servo motor 3, the uncertainty of the future prediction may increase if the accuracy of the prediction model is low. In addition, when the input to the prediction model is disturbed by disturbance or the like, or when there is an unknown input, the uncertainty of the result predicted by the prediction model may increase. The determination unit 13 determines the uncertainty of such prediction in the correction unit 12 . As a method for judging the reliability from the reliability index by the correction unit 12, there is a method such as judging from the magnitude relation with a threshold value which is a preset reference value, but the method is not limited to this method.

Further, the correction unit 12 selects a command value for controlling the servomotor 3 according to the determination result from the determination unit 13, and outputs the selected command value to the servo amplifier 11 as a corrected command value. Specifically, the correction unit 12 outputs the candidate command value to the servo amplifier 11 as a corrected command value when obtaining a permission determination result, and outputs the candidate command value to the servo amplifier 11 when obtaining a non-permission determination result. A command different from the candidate command value is output to the servo amplifier 11 as a command value after correction, depending on the method used.

Here, a more detailed configuration of the correction unit 12 will be described. The correction unit 12 has a prediction unit 121 and a correction selection unit 122 . The prediction unit 121 has a prediction model for predicting the state quantity of the operation such as the current or future position of the servomotor 3 . The prediction unit 121 inputs the command value obtained from the command value generation unit 2 and the state quantity of the actual operation of the servo motor 3 obtained from the servo motor 3, and predicts the present or future using a prediction model. A predicted state quantity, which is the operating state of the servomotor 3, is predicted. The prediction unit 121 also calculates a reliability index for evaluating the reliability of the predicted state quantity using the prediction model. The reliability index is used to evaluate the reliability of the predicted state quantity. It also evaluates degree.

As an example of calculating a prediction model, a predicted state quantity predicted from the prediction model, and a reliability index corresponding to this predicted state quantity, an example of a probability model assuming that the predicted state quantity is a random variable following a specific distribution A method using Gaussian process regression is mentioned. When calculating the predicted state quantity and the reliability index using Gaussian process regression, for example, the following calculations are performed. With N being a natural number, sampling is performed at N points while the servomotor 3 is being operated by processing or the like, input data is x, output data is y, and Gram matrix is C _N . At this time, one of the sampling input data is x _i (i is a natural number), one of the sampling output data is y _i (i is a natural number), and the values of the sampling input data are x ₁ , . x _N , the prediction value m(x N _{+1 ) of the output y N+1} for the new input x _{N+1 and} the variance σ ² (x _N+1 ), which is the basis of the reliability index, are given by the following equation (1 ) and the following equation (2).

m (x _N+1 )=k ^T (C _N ^-1 ) y (1)
σ ² (x _N+1 )=c−k ^T ·(C _N ⁻¹ )·k (2)

Here, as shown in the following equation (3), k is the kernel function when the sampled input data x ₁ , . . . , x _N and the new input x _N+1 are arguments. A vector of values. Also, c is a scalar value obtained by adding the accuracy parameter of the prediction model to the value of the kernel function between the new inputs x _N+1 . Although the variance σ ² (x _N+1 ) is obtained in equation (2), the standard deviation σ(x _N+1 ) can be obtained by calculating the square root of the variance.

Next, predicted values and variations in predicted values will be described. FIG. 2 is a diagram for conceptually explaining predicted values and variations in predicted values. FIG. 2 shows an example in which the predicted value and the range of variation of the predicted value are calculated using Gaussian process regression. The horizontal axis of FIG. 2 indicates the input data x, and the vertical axis indicates the output data y. The points indicated by the black circles in FIG. 2 indicate the data points obtained in advance. In prediction using Gaussian process regression, a predicted value of output data y is predicted assuming that output data y follows a Gaussian distribution. For this reason, if the predicted value is the average m(x) of the Gaussian distribution and the index indicating the uncertainty of the prediction is the standard deviation σ(x) of the Gaussian distribution, the actual output data y has a probability of about 95% , m(x)−2σ(x) or more and m(x)+2σ(x) or less. In FIG. 2, the solid curve indicates m(x), which is the predicted value of the output data y, and the dashed curves indicate m(x)−2σ(x) and m(x)+2σ. (x) curves are shown. As shown in FIG. 2, there is a tendency for the variation in predicted values to be small at locations close to the acquired data, and to be greater at locations distant from the acquired data.

From this statistical point of view, a reliability index is defined based on the standard deviation σ(x). For example, when the standard deviation σ(x) is used as the reliability index, the smaller the reliability index, the smaller the variation, so the predicted value becomes more reliable. In addition, when the actual output data y requires an allowable error δ (>0), if 2σ(x)≦δ is satisfied, there is a possibility that the error will be within the allowable error after correction. In another example, if the reliability index is 1/(1+α·σ(x)) and α=2/δ, the reliability index becomes 1 when the variation is 0, and the reliability index becomes 1 when the variation is infinite. The degree index becomes 0, and the larger the value of the reliability index, the smaller the variation and the more certain the predicted value. In addition, when the reliability index is 0.5 or more, 2σ(x)≦δ is satisfied, so there is a possibility that the error is within the allowable error when corrected. Thus, the prediction unit 121 can determine the reliability index based on the output data y predicted using the prediction model, that is, the variations in the state quantity of the operation of the servomotor 3 . Note that the reliability index is not limited to this.

Here, an example of using Gaussian process regression to calculate a prediction and a confidence index for the prediction has been described. However, the prediction method is not limited to this, and may be, for example, a method using machine learning such as decision tree, linear regression, boosting, and neural network. Also, the method of calculating the reliability index is not limited to this, and methods such as density estimation, mixture density network, etc. may be used, for example.

Returning to FIG. 1, the correction selection unit 122 calculates a correction amount so that the predicted state quantity output from the prediction unit 121 matches the command value, corrects the command value, and generates a candidate command value. Then, the correction selection section 122 controls whether or not to output the candidate command value to the servo amplifier 11 based on the determination result output from the determination section 13 . The correction selection unit 122 outputs the candidate command value to the servo amplifier 11 as a corrected command value when the determination result is permitted, and outputs a different command value from the candidate command value when the determination result is not permitted. The value is output to the servo amplifier 11 as a command value after correction. An example of another value is the command value after correction in the previous cycle, the command value output from the command value generator 2, that is, the command value not corrected, or a value specified in advance. However, other values are not limited to these. It should be noted that, in the case of disapproval, it is set in advance which of the above different values is to be output.

Next, the operation of the servo control device 1 according to Embodiment 1 will be described. FIG. 3 is a flow chart showing an example of a control method procedure in the servo control device according to the first embodiment. First, the command value generator 2 generates a command value in a servo control cycle and outputs it to the servo control device 1 . This command value is input to the prediction section 121 of the servo control device 1 . The servo motor 3 also outputs the state quantity of the actual operation of the servo motor 3 to the servo control device 1 . The state quantity of the operation of the servomotor 3 is input to the prediction section 121 .

When the prediction unit 121 acquires the command value and the state quantity of the operation of the servomotor 3 (step S11), the prediction unit 121 uses the prediction model to estimate the predicted state of the operation of the servomotor 3 after estimating the correction amount of the servomotor 3. Quantity and reliability indices are calculated (step S12). The prediction unit 121 outputs the calculated predicted state quantity of the operation of the servomotor 3 to the correction selection unit 122 and outputs the reliability index to the determination unit 13 .

The correction selection unit 122 generates a candidate command value by correcting the command value so that the predicted state quantity of the motion matches the command value (step S13). Further, the determination unit 13 determines whether or not to permit the output of the candidate command value to the servo amplifier 11 based on the reliability index (step S14). If the judgment unit 13 judges that the application of the candidate command value is reliable based on the reliability index, the judgment unit 13 notifies the correction selection unit 122 of the judgment result that the output of the candidate command value to the servo amplifier 11 is permitted. Output for Further, when determining that the application of the candidate command value is unreliable based on the reliability index, the determination unit 13 corrects and selects the result of determination that the output of the candidate command value to the servo amplifier 11 is not permitted. Output to the unit 122 .

The correction selection unit 122 determines whether the determination result is permission (step S15). When the determination result is permitted (Yes in step S15), the correction selection unit 122 outputs the candidate command value to the servo amplifier 11 as a corrected command value (step S16). On the other hand, if the determination result is not permitted, that is, if it is not permitted (No in step S15), the correction selection unit 122 sets another value other than the candidate command value as the corrected command value to the servo amplifier 11. (step S17). After step S16 or step S17, the servo amplifier 11 controls the servo motor 3 based on the corrected command value (step S18). Then, the process returns to step S11. The above processing is repeated each time a command value is input from the command value generator 2 in a servo control cycle.

In the first embodiment, the correction unit 12 inputs the command value generated by the command value generation unit 2 and the actual measurement result of the state quantity of the operation of the servomotor 3, the candidate command value obtained by correcting the command value, and the candidate command value and a confidence index indicating the uncertainty of the prediction of the command value. Further, the determination unit 13 determines whether or not to apply the candidate command value as the corrected command value to the control of the servomotor 3 based on the reliability index, and outputs this determination result to the correction unit 12 . Then, the correction unit 12 determines a corrected command value to be output to the servo amplifier 11 based on the determination result. As a result, it is possible to suppress deterioration in machining accuracy caused by using a candidate command value with high prediction uncertainty as a post-correction command value. That is, with the servo control device 1 according to the first embodiment, at the time of estimating the correction amount for correcting the command value, it is possible to judge how reliable the estimated correction amount after the time of estimating the correction amount is. have

Further, the correction unit 12 uses the prediction model to obtain a predicted state quantity that is a state quantity of the operation of the servomotor 3 at a time after the candidate command value is determined, and a reliability index corresponding to the predicted state quantity. and predict. Further, the correction unit 12 determines a candidate command value based on the predicted state quantity. In this way, since a prediction model capable of simultaneously outputting the current or future correction amount and the reliability index corresponding to this correction amount is used, it is possible to easily evaluate the reliability of the correction amount.

Furthermore, when the application of the candidate command value is permitted, the correction unit 12 outputs the candidate command value to the servo amplifier 11 as a corrected command value, and when the application of the candidate command value is not permitted, another The value is output to the servo amplifier 11 as a command value after correction. In this way, it is possible to automatically switch between correction using the candidate command value calculated from the prediction model using the reliability index and correction using other values. In addition, it is possible to prevent an unexpected correction amount from occurring when the input to the prediction model is disturbed due to disturbance or the like, or when there is an unknown input. Furthermore, since the reliability index is calculated based on the variation of the predicted values, the accuracy of the reliability can be improved.

Embodiment 2.
FIG. 4 is a block diagram schematically showing an example of the configuration of the servo control device according to the second embodiment. In the following, the same reference numerals are given to the same constituent elements as in the first embodiment, the description thereof is omitted, and the portions different from the first embodiment will be described. The servo control device 1A according to the second embodiment further includes an accumulation section 14 and a model information update section 15. FIG.

The accumulation unit 14 associates the command value from the command value generation unit 2 with the state quantity of the actual operation from the servomotor 3 and stores them as accumulation information. An example of the state quantity of operation is the position or velocity of the servomotor 3 or the current flowing through the servomotor 3 . In the second embodiment, the operation state quantity stored in the storage unit 14 is referred to as feedback information. That is, the accumulated information is information in which command values and feedback information are associated with each other. The feedback information is determined by the correction unit 12 based on the results of actual measurement of the operation of the servomotor 3. FIG. In one example, the state quantity of actual operation from the servo motors 3 used for constructing the prediction model is determined in the data collected as feedback information.

The accumulated information includes a combination of the command value and the feedback information when the judgment result of the judging section 13 is disapproval. Therefore, the storage unit 14 records the storage information at the timing when the determination result output from the determination unit 13 to the storage unit 14 becomes non-permissible. Note that the timing of recording is an example, and the timing may be earlier or later than the time set in advance based on the judgment result being disapproval. If the time is earlier than the predetermined time based on the judgment result being disapproval, in one example, the accumulated information may be recorded at a timing earlier than the timing at which the judgment result is issued by a predetermined time. Just do it. In this case, if the judgment result is not permitted, the recorded accumulated information is left as it is, and if the judgment result is permitted, the recorded accumulated information may be left as it is or deleted. Moreover, the accumulated information may be recorded at predetermined time intervals from the start timing instead of the time during which recording is not permitted.

The model information updating unit 15 uses the accumulated information stored in the accumulating unit 14 to update model information about the prediction model. The model information includes model parameters or hyperparameters of a prediction model that outputs a prediction state quantity and a reliability index corresponding to this prediction state quantity, or a prediction model. That is, the model information updating unit 15 updates the model parameters or hyperparameters of the prediction model as model information according to the accumulated information, or updates the prediction model as model information according to the accumulated information. Here, the case where the model parameters of the prediction model are updated using machine learning will be described as an example.

FIG. 5 is a block diagram schematically showing an example of the configuration of a model information updating unit of the servo control device according to the second embodiment. The model information updating unit 15 includes a state observing unit 151 , a learning unit 152 , a trained model storage unit 153 and an output unit 154 .

The state observation unit 151 observes accumulated information stored in the accumulation unit 14 as state variables. Here, the state observation unit 151 observes the accumulated information as a state variable, but it may at least observe the state variables related to the servo motor 3 or the servo control device 1A including the accumulated information. Examples of state variables relating to the servomotor 3 or the servo control device 1A are state quantities such as the position and speed of the servomotor 3 or current flowing through the servomotor 3, and command values.

The learning unit 152 learns the model parameters of the prediction model according to the training data set created based on the state variables. Any learning algorithm may be used by the learning unit 152 . As an example, a case of applying Gaussian process regression, which is one of supervised learning algorithms, will be described. When Gaussian process regression is used for the prediction model, the parameters used in the kernel function and the like are estimated using the accumulated information stored in the accumulation unit 14, and the prediction model is updated. For example, the time-series data of the command value output from the command value generation unit 2 is used as input data, and the actual state of the servo motor 3 is used as output data to create teacher data, and learning is performed based on these data, Estimate parameters. By using maximum likelihood estimation or the like as a parameter estimation method, a more probable prediction model can be constructed, but the parameter estimation method is not limited to this method. When the learning converges, the learning unit 152 sets the learned prediction model or the learned model parameters or hyperparameters applied to the prediction model as the learning result, which is the learned model information. A known determination method can be used to determine that learning has converged.

The learned model storage unit 153 stores learning results. As noted above, the learning result is a trained model, which is the model to which learning has converged, or updated model parameters or hyperparameters.

The output unit 154 acquires the learning result from the trained model storage unit 153 and applies the learning result to the prediction model of the prediction unit 121 at appropriate timing. That is, when the prediction model is learned, the output unit 154 reflects the updated prediction model in the prediction model of the prediction unit 121 . In addition, when the model parameters or hyperparameters of the prediction model are learned, the output unit 154 reflects the updated model parameters or hyperparameters in the prediction model of the prediction unit 121 . In this way, by sequentially updating the prediction model using the input/output data when the reliability is insufficient in the determination result of the determination unit 13, a prediction model that enables highly reliable correction can be constructed. becomes possible. That is, it is possible to reduce the cases where the application of the candidate command value to the servomotor 3 is not permitted.

Although FIG. 4 shows a case where the storage unit 14 and the model information updating unit 15 are provided within the servo control device 1A, the storage unit 14 and the model information updating unit 15 are separate from the servo control device 1A. It may be a device. In one example, the storage unit 14 and the model information update unit 15 may be built in an information processing device such as an external personal computer. In this case, the stored information is recorded and the prediction model is constructed on an external information processing device, and the results are reflected in the prediction section 121 in the servo control device 1A.

In the second embodiment, the accumulation unit 14 associates the command value from the command value generation unit 2 with the feedback information from the servomotor 3 and stores them as accumulated information. The accumulated information includes the command value and the feedback information for which the judgment result of the judging section 13 is disallowed. The model information update unit 15 updates result information, which is a prediction model or model parameters or hyperparameters, using accumulated information through machine learning, and reflects the updated result information in the prediction unit 121 . As a result, the prediction model can be sequentially updated using the input/output data when the determination result of the determining unit 13 indicates that the reliability is insufficient. That is, the prediction accuracy of the prediction model can be improved by updating the prediction model or the model parameters or hyperparameters again using the data when the reliability is low. In addition, it is possible to construct a more probable prediction model by machine learning.

Embodiment 3.
FIG. 6 is a block diagram schematically showing an example of the configuration of a servo control device according to Embodiment 3. As shown in FIG. In the following, the same reference numerals are given to the same constituent elements as in the first embodiment, the description thereof is omitted, and the portions different from the first embodiment will be described. In the servo control device 1B according to Embodiment 3, the correction section 12 has two or more prediction sections. In the example of FIG. 6, the correction unit 12 has a first prediction unit 121A and a second prediction unit 121B. Note that the correction unit 12 may have three or more prediction units.

The prediction models to be used are different between the first prediction unit 121A and the second prediction unit 121B. That is, the first prediction unit 121A and the second prediction unit 121B have different prediction model structures or prediction methods, or have the same prediction model structure or prediction method, but use model parameters or hyperparameters is different. For this reason, a difference usually occurs between the prediction results of the first prediction section 121A and the second prediction section 121B.

The first prediction unit 121A and the second prediction unit 121B acquire the command value output from the command value generation unit 2 and the state quantity of the actual operation of the servomotor 3 output from the servomotor 3, Each prediction model is used to predict a predicted state quantity and a confidence index corresponding to this predicted state quantity. The first prediction unit 121A and the second prediction unit 121B output the prediction state quantity to the correction selection unit 122 and output the reliability index to the determination unit 13 . Here, in order to distinguish the respective outputs of the first prediction unit 121A and the second prediction unit 121B, the prediction state quantity and the reliability index output from the first prediction unit 121A are the first prediction state quantity and the first reliability index, respectively. The prediction state quantity and reliability index output from the second prediction unit 121B are referred to as a second prediction state quantity and a second reliability index, respectively.

The determination unit 13 acquires the first reliability index output from the first prediction unit 121A and the second reliability index output from the second prediction unit 121B. The judging unit 13 judges reliability using the obtained first reliability index and second reliability index and a threshold that is a preset reference value. In this reliability determination, the first reliability indicator and the second reliability indicator both satisfy the threshold and the reliability is satisfied. The second case in which only one of the thresholds is satisfied and the reliability is satisfied, and the first reliability index and the second reliability index both do not satisfy the thresholds and the reliability It becomes either the 3rd case judged.

In the first case, the determination unit 13 selects the candidate command value corrected based on the predicted state quantity output from the prediction unit with the higher reliability among the first prediction unit 121A and the second prediction unit 121B. A determination result permitting output to the servo amplifier 11 is output to the correction selection unit 122 .

In the second case, the determination unit 13 permits the output of the candidate command value corrected based on the predicted state quantity output from one of the prediction units that satisfies the reliability to the servo amplifier 11. A determination result is output to the correction selection unit 122 .

In the third case, it outputs to correction selecting section 122 a determination result that the first predicted state quantity and the second predicted state quantity are not permitted to be used for correction.

For the first predicted state quantity output from the first prediction unit 121A, the correction selection unit 122 calculates a correction amount so that the first predicted state quantity matches the command value, corrects the command value, and selects a candidate command. generate a value. Further, the correction selection unit 122 calculates a correction amount for the second predicted state quantity output from the second prediction unit 121B so that the second predicted state quantity matches the command value, and corrects the command value. Generate candidate command values.

Then, the correction selection section 122 controls whether or not to output the candidate command value to the servo amplifier 11 based on the determination result output from the determination section 13 . In the first case, the correction selecting unit 122 corrects the candidate command value based on the predicted state quantity output from the predicting unit with higher reliability among the first predicting unit 121A and the second predicting unit 121B. is output to the servo amplifier 11 as a corrected command value. In the second case, the correction selection unit 122 sends the candidate command value corrected based on the predicted state quantity output from one of the prediction units that satisfies the reliability to the servo amplifier 11 as the corrected command value. Output. In the third case, the correction selector 122 outputs another value as the corrected command value. An example of another value is the command value after correction in the previous cycle, the command value output from the command value generator 2, that is, the command value not corrected, or a value specified in advance. However, other values are not limited to these.

In other words, when at least one of the determination results based on the reliability indices determined by the plurality of prediction units 121A and 121B is permitted, the correction selection unit 122 selects the candidate command value by the prediction unit for which the determination result is permitted. Among them, the candidate command value with the best reliability index is determined as the corrected command value. Further, when the judgment results based on the reliability indexes determined by the plurality of prediction units 121A and 121B are all disallowed, the correction selection unit 122 selects the command value after correction immediately before determining the candidate command value, A command value at the time of determining the candidate command value or a preset value is determined as the post-correction command value.

In Embodiment 3, the servo control device 1B includes a plurality of prediction units 121A and 121B, and the prediction models used by the prediction units 121A and 121B are different. As a result, different prediction state quantities and reliability indices are calculated from the respective prediction units 121A and 121B. The determination unit 13 selects candidate command values to be output to the servo amplifier 11 based on reliability indices output from the plurality of prediction units 121A and 121B. That is, among the reliability indices satisfying the threshold, the candidate command value corrected based on the predicted state quantity output from the prediction unit with the higher reliability is selected as the command value after correction, and all the reliability indices are If the threshold is not met, another value is selected as the corrected command value. In this manner, the accuracy of correction can be increased by comparing the reliability indices from among a plurality of prediction models and selecting the most probable correction amount. In addition, even for changes in the controlled object or inputs to the plurality of prediction units 121A and 121B, by using a more suitable prediction model according to the controlled object or data, more reliable correction is performed based on the reliability index. becomes possible. As a result, it is possible to prevent deterioration of machining accuracy due to inappropriate correction.

Note that the configuration of the second embodiment may be applied to the configuration of the third embodiment. Thereby, the prediction models in the plurality of prediction units 121A and 121B can be sequentially updated using the input/output data when the determination result of the determination unit 13 indicates that the reliability is insufficient. That is, the prediction accuracy of the prediction model can be improved by updating the prediction model or the model parameters or hyperparameters in the plurality of prediction units 121A and 121B again using the data when the reliability is low.

Next, a hardware configuration for realizing the correcting unit 12, the determining unit 13, the accumulating unit 14, and the model information updating unit 15 of the servo control devices 1, 1A, and 1B described in each embodiment will be described. FIG. 7 is a diagram schematically showing an example of a hardware configuration that implements the servo control devices according to the first, second and third embodiments. Correction unit 12, determination unit 13, storage unit 14, and model information update unit 15 described in the first, second, and third embodiments can be realized by processing circuit 100 shown in FIG.

Processing circuitry 100 includes processor 101 , memory 102 , input circuitry 103 and output circuitry 104 . The processor 101 is a CPU (Central Processing Unit, processing unit, arithmetic unit, microprocessor, microcomputer, processor, also called DSP), system LSI (Large Scale Integration), or the like. The memory 102 is a non-volatile or Volatile semiconductor memory, magnetic disc, flexible disc, optical disc, compact disc, mini disc, DVD (Digital Versatile Disc), and the like.

The correcting unit 12, the determining unit 13, the accumulating unit 14, and the model information updating unit 15 can be realized by reading corresponding programs from the memory 102 and executing them by the processor 101. FIG. The input circuit 103 is used to externally receive information processed by the processor 101, information stored in the memory 102, and the like. Used when outputting to

The servo amplifier 11 is implemented by a dedicated circuit including a conversion circuit that converts an externally supplied voltage to generate a voltage to be applied to the servo motor 3, a control circuit that controls the conversion circuit, and the like.

The configurations shown in the above embodiments are only examples, and can be combined with other known techniques, or can be combined with other embodiments, without departing from the scope of the invention. It is also possible to omit or change part of the configuration.

Reference Signs List 1, 1A, 1B servo control device, 2 command value generation unit, 3 servo motor, 11 servo amplifier, 12 correction unit, 13 determination unit, 14 accumulation unit, 15 model information update unit, 121 prediction unit, 121A first prediction unit , 121B second prediction unit, 122 correction selection unit, 151 state observation unit, 152 learning unit, 153 trained model storage unit, and 154 output unit.

Claims (6)

A servo control device that controls the operation of a servomotor based on a command value that is periodically input to instruct the operation of the servomotor,
a candidate command value that is a command value obtained by correcting the command value based on the command value and the actual measurement result of the operation of the servomotor; a reliability index that is an index for evaluating the reliability of the candidate command value; and a correction unit that outputs a corrected command value that is a corrected command value for controlling the servomotor;
a determination unit that determines permission or non-permission of application of the candidate command value to control of the servomotor based on the reliability index, and outputs a determination result to the correction unit;
a servo amplifier that controls the servo motor based on the corrected command value;
with
The correction unit is
Using a predictive model for predicting the state quantity of the current or future operation of the servomotor, inputting the command value and the actual measurement result of the operation of the servomotor, at the time after the time point of determining the candidate command value a prediction unit that predicts a predicted state quantity obtained by predicting the state quantity of the operation of the servomotor and evaluates the reliability of the predicted state quantity;
a correction selection unit that determines the candidate command value based on the predicted state quantity;
has
The prediction unit uses the reliability of the predicted state quantity as the reliability index,
The correction selection section of the correction section outputs the candidate command value to the servo amplifier as the post-correction command value based on the determination result. an accumulation unit that associates the command value with an actual measurement result of the operation of the servomotor and stores it as accumulated information;
If the determination result is disapproval, using the accumulated information, update the prediction model or model information including model parameters or hyperparameters of the prediction model, and update the prediction model of the prediction unit. a model information updating unit that reflects model information;
2. The servo controller of claim 1 , further comprising: The model information updating unit
a state observation unit that observes state variables relating to the servo motor or the servo control device that includes at least the accumulated information;
a learning unit that learns the model information according to a training data set created based on the state variables;
a learned model storage unit that stores the learned model information learned by the learning unit;
an output unit that reflects the learned model information in the prediction model of the prediction unit;
3. The servo control device according to claim 2 , comprising: The prediction unit determines the reliability index based on variations in the state quantity of the servo motor predicted using the prediction model,
2. The determination unit compares the reliability index output from the prediction model with a predetermined reference value, and determines the determination result based on the comparison result . 4. The servo control device according to any one of 3 . The correction selection unit outputs the candidate command value as the post-correction command value when the determination result is permitted, and outputs the candidate command value immediately before determining the candidate command value when the determination result is not permitted. 5. The command value after correction, the command value at the time of determining the candidate command value, or a value specified in advance is output as the command value after correction. servo controller. The correction unit has a plurality of prediction units,
When at least one of the determination results based on the reliability index determined by the plurality of prediction units is permission, the correction selection unit is configured to select the candidate command value by the prediction unit for which the determination result is permission. Among them, the candidate command value with the best reliability index is determined as the corrected command value, and when all the judgment results based on the reliability indices determined by the plurality of prediction units are disapproved, The command value after correction immediately before determining the candidate command value, the command value at the time of determining the candidate command value, or a preset value is determined as the command value after correction. 5. The servo control device according to any one of items 1 to 4 .

Images