JP2020025462A - Motor control system - Google Patents

Abstract

To allow appropriate processing to cope with mechanical abnormalities in an entire mechanical system.SOLUTION: A motor control system 100 carries out: a control procedure, performed as steps S15, S25, S40, S115, and S120, for determining a data abnormality on the basis of a comparison between a predetermined data abnormality determination threshold ath which is preset by normally driving a motor 12 and a Mahalanobis distance a(x') which is calculated based on time-series detection data detected by observationally driving the motor 12; and a control procedure, performed as a step S135, for determining the aging deterioration of a motor drive mechanism 1 on the basis of an occurrence frequency of the data abnormality. The motor control system 100 is configured to perform the same drive operation for the normal drive and the observational drive, with the time-series detection data being at least one of a torque command, a motor speed, and a motor position. Alternatively, the motor control system 100 is configured to perform different drive operations for the normal drive and the observational drive, with the time-series detection data being at least one of a disturbance observer estimation value and a speed deviation.SELECTED DRAWING: Figure 1

Description

  The disclosed embodiments relate to a motor control system.

  Patent Literature 1 and Patent Literature 2 disclose techniques for predicting diagnosis of the state of mechanical equipment by analyzing sensor data based on a statistical method.

Japanese Patent No. 5827425 Japanese Patent No. 5827426

  However, when the entire mechanical system is considered, it is not appropriate to determine the abnormality of the entire mechanical system by simply using a statistical method.

  The present invention has been made in view of such a problem, and an object of the present invention is to provide a motor control system capable of performing appropriate processing regarding a machine abnormality in the entire machine system.

  According to one aspect of the present invention, there is provided a motor control system for driving and controlling a motor for driving a motor driving mechanism, comprising: a predetermined data abnormality determination threshold; A data abnormality determining unit that determines data abnormality based on a comparison between the Mahalanobis distance calculated based on the series detection data, and a machine normality determining unit that determines machine normality of the motor drive mechanism based on the data abnormality determination And a motor control system having the following.

  According to another aspect of the present invention, there is provided a motor control system for driving and controlling a motor that drives a motor drive mechanism, the motor control system comprising: a predetermined data abnormality determination threshold; and a time series detection data at the time of driving the motor. A data abnormality determination unit that determines data abnormality based on a comparison with the Mahalanobis distance calculated in the above manner; a machine deterioration determination unit that determines aging of the motor drive mechanism based on the frequency of occurrence of the data abnormality; When the deterioration determination unit detects the occurrence of the aging, a motor control system including a motor stop unit for notifying the occurrence of the aging deterioration and stopping the drive control of the motor is applied.

  According to another aspect of the present invention, there is provided a motor control system for driving and controlling a motor that drives a motor driving mechanism, wherein the motor is driven normally to set a predetermined data abnormality determination threshold value, A data abnormality determining unit that determines a data abnormality based on a comparison with a Mahalanobis distance calculated based on time-series detection data detected by driving the motor for observation, and the motor driving based on a frequency of occurrence of the data abnormality. A mechanical deterioration determining unit that determines the aging of the mechanism, wherein the normal drive and the observation drive are the same drive operation, and the time series detection data is at least a torque command, a motor speed, and a motor position. One motor control system is applied.

  According to another aspect of the present invention, there is provided a motor control system for driving and controlling a motor that drives a motor driving mechanism, wherein the motor is driven normally to set a predetermined data abnormality determination threshold value, A data abnormality determination unit that determines a data abnormality based on a comparison with a Mahalanobis distance calculated based on time-series detection data detected by driving the motor by observing and driving the motor based on a frequency of occurrence of the data abnormality. A mechanical deterioration determining unit that determines aging of the mechanism, wherein the normal drive and the observation drive are different drive operations, and the time-series detection data is at least one of a disturbance observer estimated value and a speed deviation. Is applied.

  ADVANTAGE OF THE INVENTION According to this invention, appropriate processing regarding the machine abnormality in the whole machine system is attained.

FIG. 2 is a diagram illustrating a schematic block configuration of a motor control system. It is a figure showing the control block of the servo amplifier which acquires a torque command and an output speed as time series data in transfer function form. FIG. 9 is a diagram illustrating an example of time-series data when the degree of freedom is four. It is a figure explaining the relation of the chi-square distribution, the data abnormality judgment threshold value, and the Mahalanobis distance. 9 is a flowchart illustrating a control procedure of a preparation process executed by a CPU of a setup PC. It is a flowchart showing the control procedure of the aging deterioration determination processing which CPU of a high-order control apparatus performs. 5 is a graph in which time-series sample data of output speed, time-series observation data, and a determination result of a data abnormality are plotted when the degree of freedom is M = 1 and the movable slide position is different. 6 is a graph in which time series sample data of torque command, time series observation data, and a determination result of data abnormality are plotted when the degree of freedom is M = 1 and the movable slide position is different. 8 is a graph showing a normal distribution of time-series sample data of output speed and time-series observation data at time A1 in FIG. 7. 8 is a graph showing normal distribution and time-series observation data of time-series sample data of output speed at time A2 in FIG. 7. 9 is a graph showing normal distribution and time-series observation data of time-series sample data of a torque command at time B1 in FIG. 8. 9 is a graph showing normal distribution and time-series observation data of time-series sample data of a torque command at time B2 in FIG. 8. It is a graph which shows Mahalanobis distance of output speed and a data abnormality judgment threshold value in comparison. 6 is a graph showing a comparison between a Mahalanobis distance of a torque command and a data abnormality determination threshold. It is the graph which plotted the time-series sample data of the output speed at the time of oscillating, the time-series observation data, and the determination result of the data abnormality. It is the graph which plotted the time series sample data of the torque command at the time of oscillating, the time series observation data, and the determination result of the data abnormality. It is the graph which compared the Mahalanobis distance and the data abnormality judgment threshold value when the degree of freedom M = 2. 9 is a graph in which time-series sample data and time-series observation data in a case where the degree of freedom is M = 1 and the movable slide position is different are plotted by gain characteristics. 9 is a graph in which time-series sample data and time-series observation data in the case where the degree of freedom is M = 1 and the movable slide position is different are plotted by phase characteristics. 6 is a graph in which time-series sample data and time-series observation data in a case where the degree of freedom is M = 1 and the movable slide is at the same position are plotted by gain characteristics. It is the graph which plotted the time series sample data and the time series observation data in the case of the degree of freedom M = 1 and the same position of a movable slide at a phase characteristic. It is a figure showing the kind of machine abnormalities by the combination of the torque command and the data normal of an output position, and data abnormalities. It is a figure showing the control block of the servo amplifier which acquires the output of a disturbance observer as time series data in transfer function form. It is a figure showing the control block of the servo amplifier which acquires a speed deviation as time series data in transfer function form.

  Hereinafter, an embodiment will be described with reference to the drawings.

<1: Overall configuration of motor control system>
An example of the overall configuration of the motor control system according to the present embodiment will be described with reference to FIG.

  FIG. 1 shows a schematic block configuration of a motor control system. As shown in FIG. 1, the motor control system 100 includes a motor drive mechanism 1, a servo amplifier 2, a higher-level control device 3, an operator 4, and a setup PC 5.

  The motor drive mechanism 1 is a mechanical system whose drive is controlled by the motor control system 100 and in which various abnormalities relating to the drive are determined. The motor drive mechanism 1 includes a motor 12 having an encoder 11 and a drive machine 13 driven by the motor 12. In the example of the present embodiment, the motor 12 is a rotary electric motor, and the encoder 11 is a sensor that optically detects and outputs the rotational position of the motor.

  The servo amplifier 2 has a function (motor drive control function) of supplying a drive current to the motor 12 and controlling the drive so as to make the output position of the motor 12 follow a position command input from the higher-level control device 3 described later. I have. In the present embodiment, the servo amplifier 2 chronologically combines two data, a torque command generated in the process of supplying a drive current and an output speed generated based on the output position of the motor 12 output from the encoder 11. It also has a function of sequentially acquiring data as data and outputting it to the host control device 3 (see FIG. 2 described later).

  The host control device 3 has a function (motion control function) of sequentially outputting a position command of the motor 12 to cause the drive machine 13 to perform a desired temporal drive operation. Further, in the present embodiment, the higher-level control device 3 stores the time-series data input from the servo amplifier 2 and performs a data abnormality determination based on the time-series data and a data abnormality determination mode during observation driving described later. It also has a function of making a mechanical abnormality determination based on the information. Further, at the time of normal driving described later, the time series data input from the servo amplifier 2 is output to the setup PC 5 described later as it is.

  The operator 4 includes a display unit and an operation unit (not shown), and has a function as a user interface for displaying various information transmitted to and received from the host control device 3 and inputting various commands and parameters. At the time of the observation drive described later, the result of the machine abnormality determination input from the host control device 3 is displayed.

  The setup PC 5 is composed of, for example, a notebook-type general-purpose personal computer, and has a function of performing various initial settings for the host control device 3 before normal operation. Further, in the present embodiment, the setup PC 5 performs one of the above-described setting functions when the higher-level control device 3 performs the data abnormality determination based on time-series data input from the higher-level control device 3 during normal driving described later. A preparation process for calculating a sample average, a sample covariance matrix, and a data abnormality determination threshold required for the above and outputting the calculated data to the host controller 3 is performed (see FIG. 5 described later). During normal operation (that is, during observation driving described later), the setup PC 5 does not perform any processing and becomes unnecessary, so that the connection can be removed from the higher-level control device 3.

  In the above description, the servo amplifier 2, the host controller 3, the operator 4, and the setup PC 5 correspond to the motor control system described in each claim.

<2: Servo amplifier control block>
FIG. 2 shows a control block of the servo amplifier 2 in the present embodiment in the form of a transfer function. In the example of the present embodiment, the control block shown in FIG. 2 is realized by software executed by a CPU (not shown) provided in the servo amplifier 2.

  2, the servo amplifier 2 includes a subtractor 21, a position control unit 22, a subtractor 23, a speed control unit 24, a current control unit 25, and a speed conversion unit 26. The subtracter 21 outputs a position deviation by subtracting an output position (feedback position) described later from the position command input from the host controller 3. The position control unit 22 outputs a speed command based on the position deviation by so-called PID control or the like. The subtracter 23 subtracts an output speed (feedback speed) described later from the speed command to output a speed deviation. The speed control unit 24 outputs a torque command based on the speed deviation by so-called PID control or the like. The current control unit 25 outputs a drive current by power conversion based on the torque command, and supplies power to the motor 12. Then, the encoder 11 detects the output position when the motor 12 drives the drive machine 13 and feeds back the output position to the servo amplifier 2. This output position is subtracted from the position command in the subtractor 21 and is input to the speed conversion unit 26. The speed converter 26 outputs an output speed that is the driving speed of the motor 12 based on the output position. The speed conversion unit 26 may be constituted by a differentiator for differentiating the output position with time.

（ｋ：ばね係数、μ：摩擦係数、ｍ：可動部分の慣性モーメント）
の運動方程式に基づくフィードバック制御が行われているとみなせる。 The subtractor 21, the position control unit 22, the subtractor 23, the speed control unit 24, the current control unit 25, and the speed conversion unit 26 described above, together with the external motor 12 and the encoder 11, provide a so-called position control feedback loop and speed control feedback. The loop forms a double feedback loop. Note that a current control feedback loop is also provided inside the current control unit 25, but is omitted in the figure. In these feedback loops, the output of the position deviation by the subtractor 21 is equivalent to the time differential processing of the position command, and the output of the speed deviation by the subtracter 23 is equivalent to the time differential processing of the speed command. Therefore, in the above-described double feedback loop of the servo amplifier 2,

(K: spring coefficient, μ: friction coefficient, m: moment of inertia of movable part)
It can be considered that feedback control based on the equation of motion is performed.

  In this embodiment, the servo amplifier 2 sequentially detects the torque command and the output speed as time-series data in a short cycle such as a system cycle, and outputs the time-series data to the host controller 3.

<3: Features of the present embodiment>
In recent years, preventive maintenance has become a keyword as part of the added value enhancement to mechanical systems. Until now, a configuration has been adopted in which information that assists preventive maintenance is presented to the host control device 3 by means of a life monitor or an installation environment monitor. However, apart from these, information such as aging and oscillation of the motor drive mechanism 1 is provided. There is a need to detect mechanical abnormalities. The motor control system 100 of this embodiment is for detecting a mechanical abnormality of the motor drive mechanism 1 in response to this request.

  The state quantities that can be detected by the servo amplifier 2 are the torque input to the motor 12 and the speed and position output by the motor 12. In particular, in the case of the position / speed control, the influence of the reaction force on the driving machine 13 side is also reflected in the torque, so that it is considered that machine abnormality such as aging can be grasped by continuously observing. In the present embodiment, machine learning based on a statistical method is used as a method for detecting a change from an observed waveform.

  However, the abnormalities that can be detected by the above-described machine learning are only abnormal states that can be directly determined from data acquired instantaneously. On the other hand, in a mechanical system such as the motor drive mechanism 1, the position of the mechanism changes in a very short time, and depending on the conditions, a portion where the mechanism is abnormal and a portion where the mechanism is normal at continuous minute displacement occur. Therefore, it is necessary to determine a mechanical abnormality such as aging depending on every place. In addition, when the entire machine system is considered, it is not appropriate to determine the abnormality of the entire machine system using only a statistical method.

  Therefore, in the motor control system 100 of the present embodiment, an abnormal state directly determined from data by machine learning is regarded as a data abnormality, and an abnormal state corresponding to the aging or oscillation state of the motor drive mechanism 1 as described above. Are treated as machine abnormalities, and these data abnormalities and machine abnormalities are treated separately. Then, the motor control system 100 acquires time-series data relating to input and output of the motor 12 while the motor drive mechanism 1 is being driven, and determines a data abnormality from the time-series data. Then, a mechanical abnormality of the motor drive mechanism 1 is determined based on the acquisition mode (acquisition time, acquisition frequency, acquisition frequency, acquisition combination, and the like) of the time-series data determined to be data abnormality. The respective determination methods for the data abnormality and the machine abnormality will be described below in order.

<4: Data abnormality judgment>
<4-1: Data abnormality judgment by machine learning>
In general, it is largely based on experience that a human observes a waveform to determine normal / abnormal. Machine learning is a method of expressing this experience as a mathematical expression and performing it on a computer. The basic idea of the change detection method using machine learning is to create a normal distribution of a reference data group (hereinafter referred to as sample data), and the data obtained in the operation stage (hereinafter referred to as observation data) deviates from the normal distribution. It is to confirm whether or not it is.

  In performing the data abnormality determination, it is conceivable that the sample data is assumed to be all normal in terms of data, and that the sample data is mixed with sample data labeled as normal and abnormal in terms of data. However, when applying to aging of mechanical components, it is difficult to prepare abnormal sample data in advance, so it is realistic to assume that all sample data is normal.

  To determine that the data deviates from the normal distribution, a threshold value for data abnormality determination is set at the end of the normal distribution, and the observed data is separated from the center of the normal distribution by more than the data abnormality determination threshold value. You just need to make sure that

例えば、同一の駆動機械１３を２つのモータ１２で駆動するモータ駆動機構（特に図示せず）で、図３に示すように各モータ１２それぞれのトルク指令と出力速度を時系列データとして取得した場合（つまり自由度Ｍ＝変数の種類数＝２）には、時系列データＤは以下のようになる。ただし、Ｄの添え字は時刻を示す。

・・・・・・ <4-2: Time series data>
In the present embodiment, when obtaining a plurality of types of sample data and observation data, the data is obtained as time-series data D defined in an array as follows.

For example, when a motor drive mechanism (not particularly shown) that drives the same drive machine 13 with two motors 12 acquires the torque command and output speed of each motor 12 as time-series data as shown in FIG. For (that is, the degree of freedom M = the number of types of variables = 2), the time-series data D is as follows. However, the subscript of D indicates time.

...

<4-3: For Hotelling ^{T 2} Method>
In the present embodiment, applying the T ² method Hotelling as a change detection method using machine learning. T ² method Hotelling is one method of multivariate analysis to observe a plurality of types of change waveform data in parallel, carried out in the process the following (Step 1) to (step 6).

(Step 1) Determining the False Report Rate There are normal data and abnormal data in the data, but the index of how far out of the normal distribution the abnormal data is set is the false report rate α. For example, if it is considered that the false alarm rate is 1%, α = 0.01. In the probability statistics theory, if the false alarm rate is set to 0, all data becomes normal, and the false alarm rate α is not set to 0 in principle.

ただし、Γはガンマ関数を表し、次式で定義される。

(Step 2) Calculate the Chi-square distribution The chi-square distribution is calculated from the following equation with the degree of freedom M and the scale factor s = 1. The degree of freedom M is a parameter that specifies the number of independent sample data types (the number of variable types in the above-described multivariable analysis).

Here, Γ represents a gamma function and is defined by the following equation.

を満たすデータ異常判定しきい値ａ_ｔｈを算出する。 (Step 3) Calculate a Data Abnormality Determination Threshold From the false alarm rate α determined in the above (Step 1) and the chi-square distribution calculated in the above (Step 2),

The data abnormality determination threshold value _ath that satisfies is satisfied.

ただし、はｎ番目の種類の標本データである。 (Step 4) Calculate the sample mean and the sample covariance matrix From the sample data that is normal data, the sample mean μ (the hat is omitted in the notation in the sentence, the same applies hereinafter) and the sample covariance matrix Σ (the hat is omitted in the notation in the sentence) , Etc.) are calculated from the following equation.

Here, is the n-th type of sample data.

(Step 5) Calculate Mahalanobis distance Based on the sample mean μ and the sample covariance matrix 算出 calculated in the above (Step 4) and the detected observation data, the Mahalanobis distance a (x ′) is calculated from the following equation.

A data abnormality determination threshold a _th calculated in (Step 6) data abnormality determination threshold and the comparing the Mahalanobis distance (step 3), and the Mahalanobis distance calculated in (Step 5) a (x ') Compare. When the Mahalanobis distance a (x ′) exceeds the data abnormality determination threshold value a _th (a (x ′)> a _th ), the observation data used in the above (Step 5) becomes a data abnormal state. It is determined that there is.

As shown in FIG. 4, the chi-square distribution is a probability distribution in which the distribution changes for each degree of freedom M, and is suitable for application in multivariable analysis because of its so-called reproducibility. For example, when acquiring time-series data having two types of variables (torque command and output speed) as shown in FIG. 3, the degree of freedom is M = 2 and the chi-square shown by the solid line in FIG. The distribution is used. If the Mahalanobis distance a (x ′) is larger than the data abnormality determination threshold value a _th corresponding to the false alarm rate α in the chi-square distribution, data is added to the observation data used for calculating the Mahalanobis distance a (x ′). It can be considered that an abnormality has occurred. In other words, in a multivariable analysis in which the number of types of variables is two, the degree of multiple abnormality (the degree of deviation from normal) due to the combination of the two data is determined by the data abnormality determination threshold _ath . It can be determined by unitary comparison of the Mahalanobis distance a (x ′). Note that the use of the sample mean μ and the sample covariance matrix 時 に when calculating the Mahalanobis distance a (x ′) cancels out the influence of the correlation between the normal distributions of the two types of data. Even when acquiring time series data D when as shown in FIG. 3, it is also possible to apply the data abnormality determination Hotelling T ² method using the degree of freedom = 1, respectively each type of data independently.

<4-4: Specific data abnormality determination>
First, as a comparative example, a method of determining a data abnormality without using machine learning will be described.
(Advance preparation)
1: Acquire a plurality of normal data as sample data.
2: A normal distribution at each time is created from the sample data group.
3: Set a data abnormality determination threshold value for the normal distribution at each time.
(Data abnormality judgment)
1: Obtain observation data.
2: Add to normal distribution corresponding to acquisition time.
3: If the observed data exceeds the data abnormality determination threshold value set in the normal distribution, it is determined to be abnormal.

  In the method of the comparative example, it is necessary to create a normal distribution and a data abnormality determination threshold for each time, and further, it is necessary to calculate a normal distribution also from observation data. The calculation of the normal distribution requires the calculation of the average value and the standard deviation. However, since the calculation of the standard deviation is complicated, it is not realistic to perform the calculation at each time. Also, the data abnormality determination threshold value is different for each time because it is set for the normal distribution at each time.

Next, a case where machine learning is used to solve the problem of the comparative example will be described. The process is as follows by using machine learning.
(Advance preparation)
1: Acquire a plurality of normal data as sample data.
2: Calculate the sample mean μ and the sample covariance matrix か ら from the sample data group.
3: Calculate the data abnormality determination threshold _ath from the false alarm rate α and the chi-square distribution.
(Data abnormality judgment)
1: Obtain observation data.
2: Calculate Mahalanobis distance a (x ') for observation data.
3: If the Mahalanobis distance a (x ') exceeds the data abnormality determination threshold value _ath , it is determined that the data is abnormal.

As described above, in the method using machine learning, a sample mean μ, a sample covariance matrix Σ, and a Mahalanobis distance a (x ′) are calculated instead of calculating a normal distribution. Since these calculations are simple four arithmetic operations, even if they are sequentially calculated in a short cycle during the actual operation time of the motor drive mechanism 1 over a long period of time, a large load processing is not performed. Also, the data abnormality determination threshold value _ath is a complicated calculation formula, but it is a constant that does not depend on time, so that it only needs to be calculated once in advance.

<5: Machine abnormality judgment>
According to the above data abnormality determination, the presence / absence (that is, abnormal / normal) of the abnormal state seen on the data at the time when the time-series data is acquired can be binary-determined. However, it should not be determined that a machine abnormality has occurred in the entire machine system just because a data abnormality has been determined even once, as shown in the experimental results described later. Further, when a data abnormality occurs a plurality of times, it is possible to temporarily estimate the contents of the mechanical abnormality based on the mode of occurrence. In the present embodiment, based on the consideration that the frequency of occurrence of the data abnormality gradually increases as the aging progresses, if the frequency of occurrence of the data abnormality exceeds a predetermined value, the type of the aging is added to the motor drive mechanism 1. It is determined that a mechanical abnormality has occurred.

<6: Specific control flow>
An example of a specific control flow for determining a mechanical abnormality due to aging described above will be described in detail below. First, FIG. 5 is a flowchart illustrating a control procedure of a preparation process corresponding to the above-described preparation when using machine learning for data abnormality determination. This flowchart is executed by the CPU (corresponding to the first arithmetic unit; not particularly shown) of the setup PC 5 shown in FIG. 1 during the normal drive state of the motor drive mechanism 1 in which it is convinced that the data abnormality hardly occurs. I do. As the normal drive, for example, drive in a state (initial operation or test operation) in which the motor drive mechanism 1 can be confident that the motor drive mechanism 1 operates almost as designed in a state where sufficient adjustment has been performed after assembly and manufacture.

  First, in step S5, the CPU of the setup PC 5 determines the false alarm rate α. This may be arbitrarily determined by an input from a user, or may be determined by a value set in advance or a value calculated based on a predetermined method.

  Next, proceeding to step S10, the CPU of the setup PC 5 calculates a chi-square distribution with the number of types of variables being M degrees of freedom. In the present embodiment, since two types of time-series data of a torque command and an output speed are acquired for one motor 12, the degree of freedom M = 2.

Next, the process proceeds to step S15, where the CPU of the setup PC 5 calculates the data abnormality determination threshold _ath based on the false alarm rate α and the chi-square distribution.

  Next, proceeding to step S20, the CPU of the setup PC 5 starts normal driving of the motor drive mechanism 1 by motion control and motor drive control via the host controller 3 and the servo amplifier 2.

  Next, proceeding to step S25, the CPU of the setup PC 5 sends the time-series sample data of each variable (torque command and output speed of each axis) every predetermined time such as a system cycle via the servo amplifier 2 and the host controller 3. To get.

  Next, the process proceeds to step S30, and the CPU of the setup PC 5 determines whether the normal driving has been completed. If the normal driving has not been completed yet, the determination is not satisfied and the process returns to step S25 to repeat the same procedure.

  On the other hand, when the normal driving is completed, the determination is satisfied, and the routine goes to Step S35.

  In step S35, the CPU of the setup PC 5 stops the normal driving of the motor drive mechanism 1.

  Next, proceeding to step S40, the CPU of the setup PC 5 calculates a sample mean μ and a sample covariance matrix か ら from the time-series sample data group acquired in step S25. Then, this flow ends.

  According to the flow of the preparation processing described above, the preparation processing (steps 1 to 4) of the machine learning, in which the load of the calculation processing is particularly large, can be performed in advance by the setup PC 5 having a relatively high CPU power. Can be reduced.

  Next, FIG. 6 is a flowchart illustrating a control procedure of an aging deterioration determination process for performing data abnormality determination and machine abnormality determination. This flowchart is executed by the CPU (corresponding to the second arithmetic unit; not particularly shown) of the higher-level control device 3 shown in FIG. 1 during the driving state of the observation driving of the motor driving mechanism 1 where data abnormality may occur. As the observation driving, for example, driving in a state where the motor driving mechanism 1 is operated for a sufficiently long period (practical operation) is considered.

  First, in step S105, the CPU of the host control device 3 starts observation driving of the motor drive mechanism 1 by motion control and motor drive control via the servo amplifier 2.

  Next, the process proceeds to step S110, in which the CPU of the host control device 3 acquires, via the servo amplifier 2, time-series observation data of each variable (torque command and output speed of each axis) at predetermined time intervals such as a system cycle. .

  Next, the process proceeds to step S115, in which the CPU of the host control device 3 determines the Mahalanobis distance a (from the time series observation data group acquired in step S110) and the sample mean μ and the sample covariance matrix 算出 calculated in advance in step S40. x ′) is calculated.

Next, the process proceeds to step S120, where the CPU of the host control device 3 determines that the Mahalanobis distance a (x ′) calculated in step S115 is equal to the data abnormality determination threshold a _th (in FIG. (Abbreviated as “threshold”) is determined. In other words, it is determined whether or not the time-series observation data acquired in step S110 is in a data abnormal state. If the Mahalanobis distance a (x ') does not exceed the data abnormality determination threshold _ath , the determination is not satisfied and the routine goes to Step S125. In other words, it is considered that no data abnormality has occurred.

  In step S125, the CPU of the host control device 3 determines whether the observation driving has been completed. If the observation drive has not been completed yet, the determination is not satisfied and the process returns to step S110 to repeat the same procedure.

  On the other hand, when the observation driving is completed, the determination is satisfied, and the routine goes to Step S130.

  In step S130, the CPU of the host controller 3 stops the observation drive of the motor drive mechanism 1. Then, this flow ends.

On the other hand, is determined in the step S120, if the Mahalanobis distance a (x ') had exceeded the data abnormality determination threshold _{a th,} the determination is satisfied, the process proceeds to step S135. In other words, it is considered that a data abnormality has occurred.

  In step S135, the CPU of the higher-level control device 3 sets the abnormality determination frequency (the acquisition frequency of the time-series observation data determined to be abnormal) in the data abnormality determination performed a predetermined number of times immediately before in the past to a predetermined value (predetermined value). (Threshold) is determined. In other words, it is determined whether or not an aging mechanical abnormality has occurred. If the data abnormality determination frequency in the latest predetermined number of times is larger than the predetermined value, the determination is satisfied, and the routine goes to Step S140. In other words, it is considered that an aging mechanical abnormality has occurred.

  In step S140, the CPU of the host control device 3 transmits a determination result indicating that the motor drive mechanism 1 has deteriorated over time to the operator 4, and notifies the display unit or the like of the determination result. Then, the process proceeds to step S130.

  On the other hand, in the determination in step S135, if the data abnormality determination frequency in the latest predetermined number of times is less than or equal to the predetermined value, the determination is not satisfied, and the routine goes to step S125. In other words, it is considered that there is no aging mechanical abnormality.

  According to the flow of the aging deterioration determination processing described above, the determination processing of machine learning (steps 5 and 6) and the machine abnormality determination processing in which the load of the calculation processing is relatively small can be performed even by the host controller 3 having relatively low CPU power. As a result, the resource burden on the entire motor control system 100 can be reduced. Also, during the execution of the aging determination processing (during operation), the setup PC 5 does not perform any processing and becomes unnecessary, and thus the connection of the motor control system 100 is simplified by removing the connection from the host control device 3. It can improve the robustness.

  The procedure of steps S5, S10, S15, S40, S115, and S120 corresponds to the data abnormality determination section described in each claim, and the procedure of step S135 corresponds to the machine normality determination section described in each claim. Alternatively, it corresponds to a mechanical deterioration determination unit.

<7: Confirmation of data abnormality judgment by experiment>
<7-1: Application result for time axis waveform>
The validity of the data abnormality determination method according to the present embodiment is confirmed by applying to the time-axis waveform obtained in the actual test. In this actual test, a torque command and an output speed at a specific position of a vertically movable ball screw slider (not shown) are respectively detected 20 times as time-series sample data, and the same slider is used at different positions to change the torque command. And output speed were detected as time series observation data.

<7-1-1: When the processing is performed with the degree of freedom M = 1>
FIG. 7 shows a graph in which the time-series sample data and the time-series observation data of the output speed are plotted. The solid curve in the center curve is the time-series sample data, and the other white broken curve is the time-series observation data (similar in each of the following figures). Further, in the lower part, a graph indicating a data abnormality determination result in a single variable analysis (degree of freedom M = 1) with a false alarm rate α = 1% in a binary (High / Low) manner is also plotted. Since the trace data in this experiment is composed of 1000 points, the data abnormality judgment is calculated for each point. This takes into account that, even at the same speed, the position on the ball screw is different if the time is different, so that the characteristics as a mechanical system are different. FIG. 8 is a graph in which the time series sample data of the torque command, the time series observation data, and the determination result of the data abnormality are plotted instead of the output speed of FIG.

  First, as a first comparative example, when machine learning is not used, data abnormality determination based on a normal distribution as shown in FIGS. 9 to 12 is performed. 9 and 10 show the normal distribution and time-series observation data of the time-series sample data of the output speed at each of the time A1 and the time A2 in FIG. FIGS. 11 and 12 show the normal distribution and the time-series observation data of the time-series sample data of the torque command at each of the time B1 and the time B2 in FIG. In FIG. 9, the time-series observation data is assumed to be normal data because it exists inside the normal distribution. On the other hand, the time-series observation data in FIG. 10 is assumed to be abnormal data because it exists outside the normal distribution. Similarly, FIG. 11 is assumed to be normal data, and FIG. 12 is assumed to be abnormal data. As a basic idea when machine learning is not used, the comparison between normal distribution and time series observation data is performed for all data, but the actual processing requires a considerable computational load. It is not realistic because it becomes large.

In FIG. 7, 57 points (5.7%) of the 1000 time-series observation data are determined to be data abnormal, and in FIG. 8, 59 points (5.9%) are determined to be data abnormal. It is difficult to visually determine that there is a difference only by looking at the waveform curve. On the other hand, in FIGS. 13 and 14 corresponding to FIGS. 7 and 8, respectively, the Mahalanobis distance a (x ′) calculated in (Step 5) and the data abnormality determination threshold a _th calculated in (Step 3) are shown. Are shown in comparison. Looking at the graphs plotted in FIGS. 13 and 14, it can be clearly seen whether or not the time series observation data deviates from the time series sample data.

  FIGS. 15 and 16 show results that are easier to understand and correspond to FIGS. 7 and 8 when oscillating using the same ball screw slider. In FIGS. 7 and 8, the data abnormality determination result in which High / Low is repeated little by little indicates data abnormality (High) continuously.

<7-1-2: Case of processing with degree of freedom M = 2>
FIGS. 7 to 16 show the case where the degree of freedom M = 1 is set and the data abnormality is detected with respect to the individual signals of the output speed or the torque command. On the other hand, when performing multi-variable data abnormality detection by multivariable analysis with the degree of freedom M = 2, the data abnormality determination threshold a _th with the degree of freedom M = 2 in the above (Step 3). May be calculated and compared with the Mahalanobis distance a (x ′) calculated by the square root of the sum of the square of the output speed and the torque command.

17, FIG. 7 shows a graph comparing the Mahalanobis distance a (x ') and data abnormality determination threshold a _th in case the condition of FIG. 8 degrees of freedom M = 2. When FIG. 17 is compared with FIG. 13 and FIG. 14, the Mahalanobis distance a (x ′) is determined to be a data abnormality judgment at a time when the time-series observation data of either the output speed or the torque command deviates significantly from the time-series sample data. it can be seen that exceeds the value _{a th.}

<7-2: Result of application to frequency axis waveform>
In the above <7-1>, a data abnormality detection technique is applied to a time axis waveform (that is, a time domain). In the following, a method based on machine learning with a degree of freedom of M = 1 is applied to frequency characteristic data having a frequency axis on the horizontal axis. The movable slider performs frequency measurement 20 times on the side opposite to the motor of the ball screw, and acquires this as time-series sample data. Thereafter, the movable slider performed frequency measurement on the motor side of the ball screw, and obtained as time-series observation data. That is, the frequency analysis was performed on a predetermined number of the latest time-series data, and data abnormality in the frequency domain was determined.

  FIG. 18 shows a gain characteristic, and FIG. 19 shows a graph in which time series sample data and time series observation data are plotted on the frequency axis. In the lower part of each figure, the result of the data abnormality determination in which the false alarm rate α is set to 1% is also plotted in a binary manner (High / Low). For comparison, FIGS. 20 and 21 show graphs on the frequency axis of the gain characteristic and the phase characteristic of the time-series observation data when the mechanical analysis is performed at the same position as the time-series sample data. In FIG. 18, data abnormalities are detected at 178 points (44%) of the 401 data points, and in FIG. 19, data abnormalities are detected at 185 points (46%). 5%), and in FIG. 21, data abnormalities were detected at 34 points (8.5%), and a clear difference can be confirmed.

<8: Effect of the present embodiment>
As described above, according to the motor control system 100 of the present embodiment, the time series data relating to the input and output of the motor 12 during the driving of the motor driving mechanism 1 is acquired in the procedure of steps S25 and S110. In addition, the motor control system 100 determines the data abnormality of the time-series data by the procedure of step S5, step S10, step S15, step S40, step S115, and step S120. Further, the motor control system 100 determines the mechanical abnormality of the motor drive mechanism 1 based on the acquisition mode of the time-series data determined to be abnormal in the procedure of step S135. That is, in applying the predictive diagnosis to the motion-based mechanical control of the electric motor, the data abnormality and the mechanical abnormality are distinguished, and the data abnormality is determined based on the occurrence mode of the data abnormality (the acquisition mode of the time-series data indicating the data abnormality). Judge machine abnormality. As a result, a more effective, detailed and clear judgment can be made on the mechanical abnormality of the entire mechanical system irrespective of a minute change in the data abnormality. Determining the presence or absence of a mechanical abnormality in the motor drive mechanism 1 can be regarded as being equivalent to determining whether the motor drive mechanism 1 is normal in the machine at that time. For example, if the determination in the step 135 is not satisfied (No) and it is determined that the motor drive mechanism 1 is not in a mechanical abnormality, the motor drive mechanism 1 is in a mechanical normal state (not an abnormal state but a normal state). It can be considered that it was determined to be. In this way, in the motor control system 1 of the present embodiment, which can determine the normality of the machine, for example, by individually controlling the drive of the mass-produced motor drive mechanism 1 to determine the normality of the machine, the performance and function as designed can be obtained. Can be tested for identity. Alternatively, the soundness of whether the performance or function of the overhauled motor drive mechanism 1 is within an allowable range can be verified. Alternatively, by using the sample data obtained from one of the two different motor drive mechanisms 1 and the observation data obtained from the other to determine whether the other machine is normal, the performance between the two motor drive mechanisms 1 can be determined. It is also possible to verify the similarity of functions.

  Note that the acquisition mode of the time-series data of the data abnormality referred to when determining the machine abnormality is not limited to the acquisition frequency. In addition, various acquisition modes (judgment modes) such as an acquisition time, an acquisition frequency, and an acquisition combination of abnormal data can be applied according to the machine abnormality to be determined.

  In the present embodiment, in particular, the motor 12 is drive-controlled by feedback control based on the equation of motion, and the time-series data is at least one of a torque command input to the motor 12 and an output speed output by the motor 12. Includes one. Thereby, when the normal drive and the observation drive are performed by the same drive operation, appropriate time-series data can be easily obtained, and the operation state of the mechanical part of the motor drive mechanism 1 can be verified in detail. In determining the data abnormality, substantially the same result can be obtained even if the output position is obtained as time-series data instead of the output speed.

  Further, in the present embodiment, in particular, by determining the type of machine abnormality, it is clear what kind of machine abnormality has occurred in the mechanical system, and the user can easily deal with the machine abnormality.

  In the present embodiment, in particular, the aging of the motor drive mechanism 1 is determined as the type of mechanical abnormality in the procedure of step S135. This makes it clear that the user only needs to take measures against aging in improving the machine abnormality, and the convenience is improved.

  Further, in the present embodiment, in particular, in the procedure of step S135, when the acquisition frequency of the time-series observation data determined to be data abnormal exceeds a predetermined value, it is determined that the mechanical abnormality of aging has occurred. As a result, it is possible to determine aged mechanical errors with high reliability.

  It should be noted that the aging may be determined by another criterion based on the consideration that the occurrence frequency of the data abnormality gradually increases as the aging progresses. For example, when the frequency of acquisition of time-series observation data determined to be data abnormal tends to increase each time the motor drive mechanism 1 performs the observation drive, it may be determined that an aging mechanical abnormality has occurred. Specifically, every time an observation drive is performed, the acquisition frequency of the time-series observation data determined to be abnormal is calculated, the history is recorded, and the comparison is made between the most recent observation drives in the past. If it is determined that there is, it is sufficient to determine that a mechanical abnormality due to aging has occurred.

  In addition, if the causal relationship between the characteristics of the machine abnormality and the determination mode of the data abnormality is understood, it is possible to determine the type of the machine abnormality other than the above-described aging deterioration. For example, as shown in FIGS. 15 and 16, the oscillation of the motor drive mechanism 1 may be determined as the type of mechanical abnormality. Thereby, it becomes clear that the user has to take measures against the oscillation in improving the mechanical abnormality, and the convenience is improved.

  As a criterion for determining the oscillation, there is a method for determining that a mechanical error has occurred in the case where the time-series data determined to be abnormal is continuously obtained for a predetermined period of time. Thereby, the motor control system 100 can determine the oscillation with high accuracy.

  In addition, there is also a method of determining that an oscillation mechanical abnormality has occurred when it is determined that both the torque command and the output speed are abnormal. This also allows the motor control system 100 to determine the oscillation with high accuracy. Further, other types based on the combination of normal / abnormal on the data respectively determined by the torque command and the output speed with the degree of freedom M = 1, that is, based on the acquisition combination (acquisition mode) of the time-series data determined to be abnormal Can be determined. Specifically, as shown in FIG. 22, when the torque command is abnormal data and the output speed is normal, the torque is varied so as to maintain the output speed against a large frictional force. Therefore, it can be determined that there is a mechanical abnormality for disturbance suppression. When the torque command is normal and the output speed is abnormal, it can be considered that the output speed fluctuates while the torque is maintained, that is, it can be determined that there is a mechanical abnormality due to mechanical shaking. Oscillation may occur as a result of aging, while oscillation may occur due to an installation error or an adjustment error in the new driving machine 13. For this reason, the occurrence mode of the data abnormality can also help to estimate the cause of the occurrence of the mechanical abnormality.

Also, particularly in this embodiment, it determines the data abnormality by T ² method Hotelling. This makes it possible to determine data abnormality by highly reliable multivariable analysis.

Further, in the present embodiment in particular, the T ² method described above Hotelling, it calculates the sample mean μ and the sample covariance matrix Σ based on series sample data when obtained in the normal driving. Further, the motor control system 100 calculates the Mahalanobis distance a (x ′) based on the time series observation data acquired during the observation driving together with the sample mean μ and the sample covariance matrix Σ. Then, the motor control system 100 determines the data abnormality of the time-series observation data by comparing the Mahalanobis distance a (x ′) with the data abnormality determination threshold _ath . Accordingly, since it can perform machine learning Hotelling T ² method in the so-called "supervised learning", thereby improving the reliability of the data abnormality determination. Note that machine learning may be used in so-called “unsupervised learning” by clustering or the like.

  Further, in the present embodiment, in particular, by judging the data abnormality in the time domain, it is suitable for examining a machine abnormality in consideration of a displacement or an arrangement such as an output position of the motor 12.

  As shown in FIGS. 18 to 21, data abnormality may be determined in the frequency domain. In this case, it is suitable for examining a mechanical abnormality in consideration of a component type, a cause, and the like.

  In the present embodiment, in particular, as shown in FIGS. 7, 8, and 13 to 21, the time axis waveform or the frequency axis waveform of the time series data and the data abnormality determination result are displayed on the same screen of the operator 4. By superimposing and displaying the information, it is possible to greatly contribute to estimating the type, location, or cause of the machine abnormality.

  Further, in the present embodiment, in particular, the CPU of the setup PC 5 executes the preparation process of FIG. 5 and the CPU of the host control device 3 executes the aging determination process of FIG. As described above, by performing the preparation process and the aging deterioration determination process in different CPUs, the processing load on each CPU can be reduced, and the processing speed and reliability of the data abnormality determination during observation driving can be improved.

  In the present embodiment, in particular, the setup PC 5 that executes the preparation process can be disconnected from the connection with the observation controller (and the higher-level control device 3) at the time of the observation drive (when the aging deterioration determination process is executed). Hardware resources of the entire system at the time can be reduced, and robustness can be improved.

  It should be noted that the sharing of each processing in each CPU such as the preparation processing and the aging deterioration determination processing, the data abnormality determination and the mechanical abnormality determination is not limited to the above embodiment. For example, the servo amplifier 2, the host control device 3, and the CPUs of the setup PC 5 may be integrated in any combination and may be executed in any sharing.

<9: Modification>
The embodiments of the disclosure are not limited to the above, and various modifications can be made without departing from the spirit and technical idea thereof. For example, it is also useful to acquire data other than the torque command and the output speed (output position) as time-series data, and determine a data abnormality. For example, as shown in FIG. 23 corresponding to FIG. 2, the output of the disturbance observer 27 may be acquired as time-series data to determine a data abnormality. The disturbance observer 27 in the illustrated example estimates and outputs a disturbance applied to the motor 12 based on the torque command and the output speed. Accordingly, when normal driving and observation driving are performed by different driving operations, appropriate time-series data excluding the influence of acceleration can be obtained.

  Alternatively, when the output speed (output position) output by the motor 12 is constant, a speed deviation may be obtained as time-series data to determine a data abnormality as shown in FIG. Thus, when normal driving and observation driving are performed by different driving operations, appropriate time-series data excluding the influence of speed can be obtained.

  Further, in the above embodiment, the case where the host control device 3 performs the position control of inputting the position command to the servo amplifier 2 has been described as an example. Applicable. In this case, the servo amplifier 2 only needs to include a speed control feedback loop including a subtractor, a speed controller, a current controller, and a speed converter.

  Although not particularly shown, the motor 12 included in the motor drive mechanism 1 may be a linear linear motor. In this case, a linear scale is used instead of the encoder 11, and a thrust command is used instead of the torque command. Is obtained as time-series data.

  In the above description, when there is a description such as “vertical”, “parallel”, and “plane”, the description is not strictly meaning. That is, the terms “vertical”, “parallel”, and “plane” mean tolerances and errors in design and manufacturing, and mean “substantially vertical”, “substantially parallel”, and “substantially plane”. .

  Further, in the above description, when there is a description such as “identical”, “same”, “equal”, “different”, and the like in terms of external dimensions, size, shape, position, and the like, the description is not strictly meaning. In other words, the terms “same”, “equal”, and “different” mean that tolerances and errors in design and manufacture are allowed, and that “substantially the same”, “substantially the same”, “substantially the same”, Is different. "

  Further, in addition to those already described above, the methods according to the above-described embodiment and each modified example may be appropriately combined and used. In addition, although not illustrated one by one, the above-described embodiment and each modified example are implemented with various changes added thereto without departing from the gist thereof.

Reference Signs List 1 motor drive mechanism 2 servo amplifier 3 host controller 4 operator 5 setup PC
DESCRIPTION OF SYMBOLS 11 Encoder 12 Motor 13 Driving machine 21 Subtractor 22 Position control part 23 Subtractor 24 Speed control part 25 Current control part 26 Speed conversion part 27 Disturbance observer 100 Motor control system

Claims (4)

A motor control system that drives and controls a motor that drives a motor drive mechanism,
A data abnormality is determined based on a comparison between a predetermined data abnormality determination threshold set in advance by normally driving the motor and a Mahalanobis distance calculated based on time series detection data detected by observing and driving the motor. A data abnormality determining unit for determining,
Machine deterioration determining unit that determines aging of the motor drive mechanism based on the frequency of occurrence of the data abnormality,
Whether the normal drive and the observation drive are the same drive operation, and the time series detection data is at least one of a torque command, a motor speed, and a motor position,
Or, the normal drive and the observation drive are different drive operations, and the time series detection data is a disturbance observer estimated value, at least one of speed deviation,
A motor control system, characterized in that: The mechanical deterioration determination unit,
2. The method according to claim 1, further comprising: determining that aging has occurred even when the frequency of acquisition of the time-series detection data determined to be abnormal is increasing each time the motor drive mechanism is observed and driven. Motor control system. The data abnormality determination unit includes:
3. The motor control system according to claim 1, wherein data abnormality is determined in a time domain. The data abnormality determination unit includes:
3. The motor control system according to claim 1, wherein a data abnormality is determined in a frequency domain.

Images