Motor control device and its automatic adjustment method

Abstract

To estimate a mechanical system parameter with high accuracy even when the rigidity of the machine is low.SOLUTION: A motor control device that controls machinery includes an identification unit that estimates a moment of inertia and viscous friction, a speed command generation unit that generates a speed command obtained by integrating an acceleration command, a difference signal calculation unit that calculates a difference signal from an acceleration command with a time difference, and an identification section determination unit that determines a valid section for operating the identification unit on the basis of the difference signal and operates the identification unit in the valid section. The motor is driven on the basis of the speed command generated by the speed command generation unit, and the identification unit estimates the moment of inertia and viscous friction.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a motor control device and an automatic adjustment method thereof.

Background Art

[0002] In order to assemble a motor and a motor control device into an industrial machine and operate it as desired, appropriate adjustment of the motor control device according to the characteristics of the industrial machine equipped with the motor is required. A technique for automatically performing this adjustment can reduce the human and time costs involved in the adjustment and is industrially useful.

[0003] Regarding the automatic adjustment technology of motor control devices, a method has been conventionally proposed in which, after automatically grasping the characteristics of an industrial machine, the motor control device is automatically adjusted based on those characteristics. The characteristics of the industrial machine at this time are physical characteristics related to control performance, such as the moment of inertia related to the motor shaft, viscous friction, Coulomb friction, static friction, resonance characteristics, etc., of the industrial machine equipped with the motor.

[0004] For example, Patent Document 1 provides a method for online estimating the moment of inertia and the viscous friction coefficient in consideration of Coulomb friction even when the motor rotational angular velocity is low. In Patent Document 1, Coulomb friction is regarded as a constant value disturbance, differential processing is performed on the observed value of the motor torque and the observed value of the rotational angular velocity, and after removing the influence of the constant value disturbance, the estimation of the moment of inertia and the viscous friction coefficient is carried out.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0006] In Patent Document 1, differential processing is used. However, in differential processing, even if high-frequency noise superimposed on the observed value of motor torque and the observed value of motor rotational angular velocity is emphasized and a low-pass filter or the like is applied to remove the high-frequency noise, it is not necessarily advantageous for the purpose of highly accurate estimation of the moment of inertia and the viscous friction coefficient (hereinafter, these may be referred to as mechanical system parameters). In particular, when the motor rotational angular velocity is low and noise strongly affects, it is hard to say that it is a desirable process.

[0007] Also, in Patent Document 1, it is made possible to cope with the case where the motor rotational angular velocity is low. However, when the motor torque and the change in the motor torque are not sufficient, the influence of noise on the observed value of the motor torque becomes relatively strong, and there is a concern about deterioration in the estimation accuracy of the mechanical system parameters. If the motor torque and the change in the motor torque are made sufficiently large to ensure the estimation accuracy, in the case where the rigidity of the industrial machine is low, mechanical vibration of several to several tens of hertz is excited due to the change in the motor torque, and this becomes a low-frequency motor torque disturbance and superimposes on the observed value of the motor torque, which is a factor deteriorating the estimation accuracy of the mechanical system parameters. That is, in an industrial machine where mechanical vibration of several to several tens of hertz with low rigidity superimposes on the observed value of the motor torque, when the motor torque and the change in the motor torque are small, it is affected by noise and the estimation accuracy decreases, and when the motor torque and the change in the motor torque are large, it is affected by mechanical vibration, and there is a trade-off relationship that the estimation accuracy decreases, which is an issue.

[0008] An object of the present invention is to highly accurately estimate the moment of inertia and the viscous friction even when the rigidity of the machine is low.

Means for Solving the Problem

[0009] As a preferable example of the present invention, it is a motor control device for controlling a machine, An identification unit that estimates the moment of inertia and viscous friction, a speed command generation unit that generates a speed command obtained by integrating an acceleration command, a differential signal calculation unit that calculates a differential signal from the acceleration commands with a time difference, and an identification interval determination unit that determines an effective interval for operating the identification unit based on the differential signal and operates the identification unit in the effective interval. The motor is driven based on the speed command generated by the speed command generation unit, and the identification unit is a motor control device that estimates the moment of inertia and the viscous friction.

Advantages of the Invention

[0010] According to the present invention, even when the rigidity of the machine is low, the moment of inertia and viscous friction can be accurately estimated.

Brief Description of the Drawings

[0011]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Modes for Carrying Out the Invention

[0012] Hereinafter, embodiments to which the present invention is applied will be described with reference to the drawings. In each figure, components having the same functions are given the same numbers, and the description thereof will be omitted.

Embodiment

[0013] FIG. 1 is a diagram showing the first basic configuration of Example 1. The motor control device 19 of this embodiment has a configuration other than the machine 15 with the motor assembled, which consists of the motor 13 and the machine 14 in FIG. 1. The motor control device 19 includes a current controller 12 and a speed controller 11 for controlling the motor 13 assembled to the machine 14, a speed command generator 1 for providing a speed command to the speed controller 11, a torque detector 10, and a speed detector 9. Based on the observed value of the motor torque and the observed value of the motor rotational angular velocity provided by each of them, an identifier 5 for estimating the moment of inertia and the viscous friction coefficient, which are the mechanical characteristics of the machine 15 with the motor 13 assembled, and an identification interval discriminator 4 for operating the identifier 5 based on the acceleration command provided by the acceleration command generator 2 included in the speed command generator 1. The speed detector 9 and the torque detector 10 of the motor control device 19 are configured by hardware. The speed command generator 1, the identification interval discriminator 4, the identifier 5, the speed controller 11, and the current controller 12 are configured as software in which a processor reads a program stored in a storage device such as a memory and executes each function.

[0014] The speed command generator 1 consists of an acceleration command generator 2 and an integrator 3. The speed command generator 1 generates a speed command 17 by the integrator 3 performing an integration process on the acceleration command generated by the acceleration command generator 2.

[0015] The speed detector 9 detects the rotational angular velocity of the motor shaft of the motor 13, outputs it as an observed value of the motor rotational angular velocity, and based on the speed command 17 and the observed value of the motor rotational angular velocity, the speed controller 11 calculates a command value (motor torque command) of the motor torque that the motor 13 applies to the machine 14. Note that the speed detector 9 calculates an observed value of the motor rotational angular velocity from, for example, the rotational position of the motor shaft provided by an encoder attached to the motor 13 and the like.

[0016] The current controller 12 provides a voltage command to the motor 13 so that the motor torque that the motor 13 applies to the machine 14 follows the motor torque command.

[0017] The torque detector 10 detects the motor torque applied by the motor 13 and calculates an observed value of the motor torque. Note that the torque detector 10 detects, for example, the current of the motor 13 with a current sensor and multiplies it by a torque multiplier to calculate the motor torque and the like.

[0018] The identifier 5 consists of a sequential identifier 6, a differentiator 7, and a differentiator 8. The sequential identifier 6 periodically and sequentially estimates the moment of inertia and the viscous friction coefficient based on the output of the differentiator 7 and the output of the differentiator 8. The timing for performing the identification is based on an instruction from the identification interval discriminator 4. The identifier 5 identifies a model of the mechanical characteristics of the machine 15 and estimates mechanical system parameters based on the identified model.

[0019] Although not shown in FIG. 1, the motor control device 19 may include an automatic adjuster that automatically adjusts the control gains of the speed controller and / or the position controller based on the moment of inertia and the viscous friction coefficient estimated by the sequential identifier 6.

[0020] FIG. 2 is a diagram showing the configuration of a modified example of Example 1 including an automatic adjuster. The automatic adjuster 21 calculates an appropriate control gain of the speed controller 11 for the machine 15 equipped with the motor 13 based on the information 22 of the moment of inertia and the viscous friction coefficient calculated by the identifier 5, and sets the control gain in the speed controller 11.

[0021] Assume that the speed controller 11 is, for example, a PI (Proportional-Integral) controller, and the desired responsiveness is set by the user. In that case, the auto-tuner 21 calculates the P gain and the I gain for achieving the desired responsiveness based on the moment of inertia and the viscous friction coefficient, and sets the result to the speed controller 11. Thereby, it becomes possible to provide the motor control device 19 capable of automatically tuning the speed controller 11.

[0022] The configuration and operation of the identifier 5 will be described in detail based on FIG. 3. The sequential identifier 6 includes a mechanical system parameter sequential identification calculator 33, low-pass filters (hereinafter sometimes abbreviated as LPFs) 35 to 37, and an acceleration calculator 34.

[0023] The speed detector 9 and the torque detector 10 each include a downsampler 31 and a downsampler 32. The speed detector 9 calculates an observed value of the motor rotational angular velocity used for identification via the downsampler 31. The torque detector 10 calculates an observed value of the motor torque used for identification via the downsampler 32.

[0024] If the identification period of the identifier 5 is TID and the control period of the speed controller 11 is TSP, In recent years, due to the background that high-speed control response is desired, TID > TSP may be sufficient. Therefore, the downsamplers 31 and 32 play a role of adjusting and resampling the signal period so that they become the observed values of the motor rotational angular velocity and the observed values of the motor torque in the period of the identification period TID, respectively.

[0025] If TSP is several tens of microseconds, TID may be TID > 10 × TSP, and it is assumed that the identification period TID is set sufficiently long compared to TSP.

[0026] Note that the downsamplers 31 and 32 may employ, for example, a decimation filter or the like.

[0027] When the machine 14 is operating at a non-zero speed, generally, Coulomb friction affects the observed value of the motor torque and the observed value of the motor rotational angular velocity as a constant torque disturbance. Here, Coulomb friction is a type of dynamic friction. It refers to a constant friction force that acts to prevent the movement of the machine during operation, independent of the moving speed, after exiting the static friction region and starting to move.

[0028] The identifier 5 employs the differentiator 7 and the differentiator 8 for the purpose of removing this influence. This is because taking the difference of the motor torque or the like with a differentiator that takes into account the time difference can remove the influence of Coulomb friction, which is a constant friction force.

[0029] The differentiators 7 and 8 will be described with reference to FIGS. 4 and 5. FIG. 4 is a diagram for explaining a linear difference filter. FIG. 5 is a diagram showing the frequency characteristics of a linear difference filter and a differentiator. The differentiators 7 and 8 are linear difference filters shown in FIG. 4 and the following mathematical formula (1).

[0030]

Equation

[0031] FIG. 5 shows the frequency characteristics (gain) of the linear difference filter with A = 1 in combination with the frequency characteristics of the differentiator. The dotted line, dashed line, and solid line are the frequency characteristics (gain) of the differentiator, the linear difference filter with n = 10, and the linear difference filter with n = 50, respectively.

[0032] As can be understood from FIG. 5, even a differentiator can remove the influence of Coulomb friction. However, due to the characteristic that the gain increases significantly in the high-frequency range, various noises superimposed on the observed value of the motor torque and the observed value of the motor rotational angular velocity are emphasized, leading to an undesirable decrease in the signal-to-noise ratio (SNR) of the signal used for identification.

[0033] On the other hand, a linear difference filter has the characteristic that when A = 1, the gain becomes up to twice as large in the high-frequency range, and the larger n (or n×TID) is, the more the output signal contains low-frequency components. Therefore, it is possible to remove the influence of Coulomb friction while suppressing a decrease in the SNR of the signal used for identification.

[0034] Generally, generating a significant motor torque can improve the SNR of the signal used for identification. However, even when the motor torque is not significant, the linear difference filter has an advantage in that it can remove the influence of Coulomb friction without causing a significant decrease in the SNR compared to differential processing.

[0035] On the other hand, the linear difference filter has the disadvantage that it must always keep the past value of the input signal x for a time difference τ. When performing calculations using a digital arithmetic device such as a microcontroller, as the positive integer n related to the time difference τ increases, the memory capacity for storing past signals increases, and arithmetic resources are required for memory management.

[0036] In this embodiment, this disadvantage is alleviated by the downsampler 31 and the downsampler 32. That is, by the downsampler 31 and the downsampler 32 with a sufficiently large TID set, the time difference τ can be increased without making the positive number n extremely large. As a result, it is possible to avoid an increase in memory capacity and arithmetic resources for memory management and remove the influence of Coulomb friction without causing a significant decrease in the SNR.

[0037] Another drawback of the linear difference filter is that the response delay is significant. Since the influence of the initial state of the linear difference filter continues to remain in the output of the linear difference filter until the time difference τ elapses after receiving the first input, the accurate difference signal with respect to the input signal is output after the time difference τ. Therefore, in the identifier 5, it is desirable to perform identification using an accurate difference signal.

[0038] In this embodiment, this drawback is solved by the identification interval determiner 4. Although details will be described later, this is because the sequential identifier 6 of the identifier 5 is not operated as long as the influence of the initial state of the linear difference filter remains.

[0039] The sequential identifier 6 in FIG. 3 operates at the identification period TID. Also, it operates only in the time interval during which it receives the operation instruction of the identification interval determiner 4.

[0040] Let the output of the differentiator 7 be s(t) and the output of the differentiator 8 be T(t). At this time, the sequential identifier 6 receives s(t) and T(t), and the acceleration calculator 34 outputs a(t) by differentiating s(t). Note that the differentiation process in the digital arithmetic unit is realized by approximate differentiation or the like.

[0041] s(t), a(t), and T(t) are each processed by the LPF 35, LPF 36, and LPF 37, respectively, and the results are input to the mechanical system parameter sequential identification arithmetic unit 33.

[0042] Specifically, the mechanical system parameter sequential identification arithmetic unit 33 is calculated by sequential processing (identification period TID) according to the following mathematical formulas (2) to (6). However, for the sake of simplifying the symbols, the signals after being processed by the LPFs 35 to 37 are denoted as s(t), a(t), and T(t), respectively.

[0043]

Equation

[0044]

Equation

[0045] [Number]

[0046] [Number]

[0047] [Number]

[0048] Note that J(k) and D(k) are the estimated values of the moment of inertia and the viscous friction coefficient at the k-th step, and θ(k) is the estimated value vector combining them. Here, the k-th step refers to the k-th calculation step executed by the mechanical system parameter sequential identification calculator 33. Also, P(0) is the initial value of the sequential identification, and λ is the forgetting factor, which is set such that 0 < λ ≤ 1. λ is a parameter that determines the degree to which the influence of past signals is reflected in the current identification. The closer λ is to 1, the more the influence of past signals is reflected.

[0049] LPFs 35 to 37 are provided to remove the noise superimposed on s(t), a(t), and T(t). For example, the adoption of an LPF of a first-order lag system shown in the following mathematical formula (7) can be considered.

[0050] [Number] However, s and ω are the Laplace operator and the cut-off frequency [rad / s] of the LPF, respectively.

[0051] By cascading LPFs of a first-order lag system or setting the cut-off frequency ω low, the influence of noise can be further reduced. Note that LPFs 35 to 37 all have the same filter characteristics so that the phases of the respective signals used for identification are aligned. Also, for the sake of convenience, mathematical formula (7) is expressed in a continuous-time system, but for a digital calculator, it is discretized and implemented.

[0052] When the rigidity of the machine 15 with the motor 13 assembled is low, in the case of significant motor torque or an operation with a change in motor torque, the vibration of the machine is excited, and the influence of the vibration becomes a torque disturbance and is superimposed on the observed value of the motor torque.

[0053] The vibration of the machine caused by low rigidity generally occurs at several Hz to several tens of Hz. Therefore, the torque disturbance caused by this vibration is also at several Hz to several tens of Hz.

[0054] From the perspective of improving the SNR, if the motor torque and the change in motor torque are made significant to be advantageous for identification, the vibration of the machine with low rigidity will be excited, and this influence will be superimposed on the observed value of the motor torque as a torque disturbance, resulting in a decrease in the SNR.

[0055] Therefore, considering a machine with low rigidity, it is desirable that the cut-off characteristics of the LPFs 35 to 37 are set sufficiently low so as to remove the torque disturbance caused by machine vibration. However, in this case, there is a risk of removing up to the band required for identification, and it is not realistic to set the cut-off characteristics of the LPFs 35 to 37 sufficiently low.

[0056] In order to accurately estimate the mechanical system parameters even for such a machine with low rigidity, in this embodiment, a speed command generator 1 is provided. The speed command generator 1 generates a smooth speed command that does not excite the vibration of the machine based on the acceleration command generator 2.

[0057] The acceleration command generator 2 of this embodiment is such that the maximum value, minimum value, and zero point of the acceleration command are connected by a combination of functions of the first order (linear) or higher, and has a function capable of designing the acceleration command so that the speed command obtained by integrating it generates a smooth one that does not include the dominant frequency component that excites the machine.

[0058] In addition, the acceleration command generator 2 has a function of holding a pre-designed acceleration command and generating and outputting an acceleration command based on this. The design of the acceleration command may be arbitrarily performed by the user, or a pre-designed one may be selected and used.

[0059] FIG. 6 is a diagram showing a specific example of the relationship between the acceleration command (vertical axis) and time (horizontal axis) in the acceleration command generator 2 of this embodiment. The upper part of FIG. 6 is a diagram showing the zero points 61 and 64, the maximum point 62, and the minimum point 63 of the acceleration command. The lower part of FIG. 6 is a diagram showing the case where the zero points 61 and 64, the maximum point 62, and the minimum point 63 (however, in this embodiment, each is a point group) are connected by a linear function.

[0060] The zero point 61 and the maximum point 62 are connected by linear functions 65 and 66 having different slopes and lengths, and the zero point 64 and the minimum point 63 are the same. The maximum point 62 and the minimum point 63 are connected by a linear function 67. Also, each point group of the zero point 61, the zero point 64, the maximum point 62, and the minimum point 63 is a case where the points within the point group are connected by a linear function.

[0061] The reason for configuring the zero point 61 and the maximum point 62 with two linear functions having different slopes and lengths is for smoothness and shortening of the moving distance (the number of rotations of the motor [rad]).

[0062] The speed command is the integral of the acceleration command, and the position command is the integral of the speed command. Therefore, if the acceleration command is smooth, the speed command is also smooth, and the smaller the area of the acceleration command, the shorter the moving distance.

[0063] Note that Coulomb friction is assumed to be a constant torque disturbance, that is, a constant torque disturbance independent of position. Since this assumption is considered to have a higher probability of holding as the moving distance is shorter, it is desirable from the viewpoint of removing Coulomb friction that the moving distance be shorter. Also, in view of the case where there are restrictions on the operating range of the machine, it is desirable that it can be identified even with a short moving distance from the viewpoint of the perceived user usability.

[0064] Figure 7 is a numerical example regarding the design of the acceleration command in Figure 6. The acceleration command 70 is designed as shown in Figure 6, and the speed command 71 is obtained by integrating the acceleration command 70.

[0065] Figure 8 is a diagram showing the frequency characteristics (gain) of the speed command 71 in Figure 7. From Figure 8, it can be seen that in the frequency band of about 2 [Hz] or more, there is no sufficient gain, and it is a smooth speed command that does not contain the dominant frequency component that excites the machine.

[0066] Figure 9 is a diagram for explaining the identification interval discriminator 4. As shown in Figure 9, the identification interval discriminator 4 consists of a differentiator 92 and a valid interval discriminator 93. The differentiator 92 is a linear difference filter with the same time difference τ (= n × TID) and the same filter gain A as the differentiators 7 and 8. The valid interval discriminator 93 determines the interval effective for identification based on the difference signal of the acceleration command calculated by the differentiator 92, and operates the identifier 5.

[0067] One of the conditions for the valid interval discriminator 93 to operate the identifier 5 is related to the response delay of the differentiator 92. The sequential identifier has a forgetting factor λ, and it is desirable to perform identification after the influence of the initial states of the differentiators 7 and 8 has disappeared in order to reflect past signals in the identification.

[0068] Since the time difference τ of the differentiator 92 is set to the same value as the differentiators 7 and 8 that process the signals used by the identifier 5 for identification, one of the conditions for the valid interval discriminator 93 to operate the identifier 5 for significant identification is that the difference signal output by the differentiator 92 has elapsed the time difference τ from the initial time. This is defined as condition (A).

[0069] For example, in the speed command shown in Figure 7, when the initial time is 0 seconds and the time difference τ is 1 second, the interval from 1 to 4 seconds is the interval where condition (A) is satisfied.

[0070] Figure 10 shows the difference signal 1001 obtained by processing the acceleration command 70 in Figure 7 with the differentiator 92 having a time difference τ = 1 second and its absolute value signal 1002.

[0071] The signal used for identification preferably has a larger signal gain from the viewpoint of the signal-to-noise ratio. Also in the mechanical parameter sequential identification calculator based on the differential signal, higher gain of the differential signal allows for higher expected identification accuracy.

[0072] Therefore, one of the conditions for the valid interval determiner 93 to operate the identifier 5 is that the absolute value of the differential signal obtained by processing the acceleration command with the differentiator 92 is equal to or greater than a predetermined value AX, or the average value thereof is equal to or greater than a predetermined value AAX. This is defined as condition (B).

[0073] If the interval where the differential signal of the acceleration command becomes zero continues, theoretically the identification condition is not desirable. This is because the interval where the differential signal of the acceleration command becomes zero is the interval where the acceleration command is constant, and in this case, the mechanical characteristics related to acceleration are removed by the differential process.

[0074] Therefore, one of the conditions for the valid interval determiner 93 to operate the identifier 5 is that the interval where the differential signal of the acceleration command is non-zero. This is defined as condition (C).

[0075] Note that although the absolute value signal 1002 shown in FIG. 10 has a timing where the positive and negative of the differential signal of the acceleration command are inverted, note that it is not a continuous interval where the differential signal of the acceleration command becomes zero.

[0076] The valid interval determiner 93 operates the identifier 5 in the interval that satisfies conditions (A) to (C) based on the acceleration command generated by the acceleration command generator 2.

[0077] When the predetermined value AX in FIG. 10 is 100, the valid interval determiner 93 determines that the interval from approximately 1.6 to 3.65 seconds is a valid interval for identification and operates the identifier 5. The identifier 5 performs identification in this interval and completes the estimation of the mechanical system parameters.

[0078] Note that for the automatic adjuster 21 shown in FIG. 2, the estimation result of the final mechanical system parameters used for the adjustment of the controller may be the estimation result of one time, or may be the average value of the estimation results of multiple times.

[0079] According to the motor control device of this embodiment, even for a machine with low rigidity, it does not excite the vibration of the machine, and the signal-to-noise ratio of the signal used for identification is made good, and identification in a significant section for identification can be realized.

[0080] As a result, it is possible to provide a motor control device that can accurately estimate mechanical system parameters without exciting the vibration of the machine even for a machine with low rigidity, and automatically adjust the motor controller based on this estimation result. It is also possible to provide a motor control device that can avoid mechanical damage and deterioration of the perceived user usability caused by the excitation and generation of mechanical vibration.

[0081] Near zero speed, the effects of various non-linear phenomena such as Coulomb friction, viscous friction, and the Stribeck effect are superimposed on the speed response to the motor torque. For this reason, it is not desirable to continue the sequential identification assuming that the mechanical system is linear, as shown in Formulas (2) to (6), in a section including the timing when the speed becomes zero. In particular, when the time difference τ of the differentiator 92 is set long, the time including the section where the speed of the differential signal becomes zero tends to increase in principle. Therefore, it is necessary to operate the identifier 5 while being aware of the relationship between the speed command and the time difference τ, in other words, the acceleration command that is the basis of the speed command and the time difference τ.

[0082] Therefore, as one of the conditions for the valid section determiner 93 to operate the identifier 5, the differential signal output by the differentiator 92 is from a section that does not include the timing when the speed command becomes zero. This is defined as condition (D). For this determination, the valid section determiner 93 may be configured to refer to the speed command.

[0083] Also, in this embodiment, three different linear functions were used in the design of the acceleration command, but combinations of four or more may be used, or functions of the second order or higher may be used.

[0084] In addition, the sequential identifier 6 employs a mechanical parameter sequential identification arithmetic unit 33 that takes s(t), a(t), and T(t) as inputs, but an identifier that takes s(t) and T(t) as inputs may also be employed.

[0085] Also, although the identifier 5 employs a sequential type, a batch type identifier that performs calculations non-sequentially may also be employed.

Embodiment

[0086] FIG. 11 is a diagram showing Embodiment 2. As shown in FIG. 11, this embodiment is an embodiment of the motor control device 1100 when the command generation and the configuration of the control system are different from those in Embodiment 1. Specifically, a position command generator 1103, a position controller 1101, and a first integrator 1102 are added. The position command generator 1103 includes an acceleration command generator 2, a second integrator 3, and a third integrator 1104. It is assumed that the components given the same numbers as those in FIG. 1 have common functions in FIG. 11.

[0087] The position command generator 1103 generates a position command. The position controller 1101 generates a speed command based on the position command and the observed value of the motor rotation position obtained by integrating the observed value of the motor rotation angular velocity so that the observed value of the motor rotation angular velocity follows the position command.

[0088] Even in this position control system, the machine operates based on the position command generated based on the speed command without exciting the vibration of the machine designed based on the desired acceleration command. Therefore, the same level of estimation accuracy as that in Embodiment 1 is expected in this embodiment as well.

[0089] As a result, it is possible to provide a motor control device that can estimate mechanical system parameters with high accuracy without exciting the vibration of the machine even for a machine with low rigidity, and automatically adjusts the motor controller based on this estimation result.

[0090] In addition, it is possible to provide a motor control device that automatically adjusts a motor controller capable of avoiding mechanical damage and a decrease in tactile user-friendliness caused by the excitation and generation of mechanical vibration.

[0091] In this embodiment, the automatic adjuster 21 shown in FIG. 2 may perform gain adjustment not only of the speed controller but also of the position controller.

Embodiment

[0092] This embodiment is an embodiment directed to the speed control system of an AC servo motor having a cascade configuration shown in FIG. 13. An embodiment for automatically adjusting the speed control system of an AC servo motor will be described.

[0093] A permanent magnet synchronous motor 137 that operates on three-phase alternating current and is assembled to a low-rigidity machine 1313 detects the rotational position of the motor with an encoder 139, calculates an observed value of the motor rotational angular velocity with a speed calculator 1311, and based on a speed command 1314 and the observed value of the motor rotational angular velocity, a speed controller 132 calculates a current command. The motor current of the permanent magnet synchronous motor 137 is detected for its current value by a current sensor 138 and is converted into an observed value of the dq-axis current by a first coordinate converter (coordinate converter from three-phase to dq-axis) 1310. Based on the current command and the observed value of the dq-axis current, a current controller 133 calculates a dq-axis voltage command.

[0094] The dq-axis voltage command is converted into a three-phase voltage command by a second coordinate converter (coordinate converter from dq-axis to three-phase) 134, and this is converted into a voltage pulse command by a PWM (Pulse Width Modulation) device 135. An inverter 136 supplies a three-phase voltage to the permanent magnet synchronous motor 137 based on the voltage pulse command. The speed controller 132, the current controller 133, the first coordinate converter 1310, the second coordinate converter 134, the PWM device 135, and the speed calculator 1311 may be configured as software in which a processor reads a program stored in a storage device such as a memory and executes each function.

[0095] FIG. 12 is a diagram showing the processing steps of the automatic adjustment process of Example 3. For the speed control system of the AC servo motor in FIG. 13, the steps of the automatic adjustment process of this embodiment are roughly composed of a speed command generation step 1201, an identification interval determination step 1202, an identification step 1203, and a controller adjustment step 1204, as shown in FIG. 12.

[0096] The speed command generation step 1201 includes an acceleration command generation step. In the acceleration command generation step, an acceleration command is generated in which the maximum value, minimum value, and zero point of the command are combined by a combination of functions of the first order (linear) or higher, and the speed command obtained by integrating the acceleration command is generated so as to be smooth and not include the dominant frequency component that excites the machine. This is the speed command generation step 1201. Specific examples of the acceleration command designed and generated in the acceleration command generation step are FIGS. 6 and 7.

[0097] In the identification interval determination step 1202, a determination is made on the identification interval for estimating the mechanical system parameters based on the acceleration command generated in the acceleration command generation step. The determination as to whether it is a significant interval for carrying out the identification is made based on the satisfaction of the conditions (A) to (C) shown in Example 1. Also, the condition (D) shown in Example 1 may be taken into account.

[0098] In the identification step 1203, based on the speed command generated in the speed command generation step 1201, the permanent magnet synchronous motor 137 and the low-rigidity machine 1313 that are the control objects are driven, and from the observed value of the motor torque calculated from the current sensor 138 and the observed value of the motor rotational angular velocity calculated from the encoder 139, identification is carried out in the interval determined to be significant for identification in the identification interval determination step 1202, and estimation of the mechanical system parameters is carried out. The identification at this time is carried out using sequential identification means after preprocessing the observed value of the motor torque and the observed value of the motor rotational angular velocity with a linear difference filter of the time difference τ. Examples of the sequential identification means when calculating the acceleration positively include Mathematical Formulas (2) to (6). Also, the linear difference filter is Mathematical Formula (1).

[0099] As shown in FIG. 12, the speed command generation step 1201 to the identification step 1203 may be repeated a predetermined number of times. Based on the estimated values (for multiple times) of the mechanical system parameters obtained in the identification step 1203, the controller adjustment step 1204 automatically adjusts the control gain of the speed controller 132 so as to achieve the responsiveness desired by the user.

[0100] As a result, even for a machine with low rigidity, it is possible to accurately estimate the mechanical system parameters without exciting the vibration of the machine, and to automatically adjust the motor control device shown in FIG. 13 based on this estimation result. Also, it is possible to automatically adjust the motor control device that can avoid mechanical damage and deterioration of the perceived user-friendliness caused by the excitation and generation of mechanical vibration.

[0101] FIG. 14 is a diagram showing the processing steps of the automatic adjustment process when the motor control device constitutes a position control system. Even when the motor control device constitutes a position control system, as shown in FIG. 14, by generating a position command in the position command generation step 1401, it is possible to automatically adjust the motor control device constituting the position control system.

[0102] Specifically, the position command generation step 1401 generates a position command by integrating the speed command generated in the speed command generation step 1201.

[0103] Note that the low-rigidity machine 1313 is, for example, a take-out machine that takes out machine parts as shown in FIG. 15 from a machine tool. The upper part of FIG. 15 is a side view of the take-out machine, and the lower part of FIG. 15 is a top view of the take-out machine.

[0104] The automatic adjustment of the motor control device in this embodiment is effective for a machine whose mechanical characteristics do not change significantly as the machine moves. Even if the mechanical characteristics change as the machine moves, it is possible to estimate the mechanical system parameters, which are the piecewise mechanical characteristics, even for a low-rigidity machine by dividing the moving section into pieces and identifying them. The controller adjustment step 1204 may include an automatic adjuster that automatically and appropriately adjusts the control gain of the controller based on the estimation results of the piecewise mechanical system parameters. [Explanation of symbols]

[0105] 1...speed command generator, 2...acceleration command generator, 4...identification interval judger, 5...identifier, 6...sequential identifier, 7, 8...differential calculator, 11...speed controller, 21...automatic adjuster

Claims

1. A motor control device for controlling a machine, an identification unit that estimates the moment of inertia and viscous friction, a speed command generation unit that generates a speed command obtained by integrating an acceleration command, a differential signal calculation unit that calculates a differential signal from the acceleration commands with a time difference, an identification section determination unit that determines a valid section for operating the identification unit based on the differential signal and operates the identification unit in the valid section, driving a motor based on the speed command generated by the speed command generation unit, a motor control device in which the identification unit estimates the moment of inertia and the viscous friction.

2. In the motor control device according to claim 1, the identification unit has a second differential signal calculation unit that calculates a second differential signal with a time difference for the observed value of the torque of the motor and the observed value of the rotational angular velocity of the motor, a motor control device that estimates the moment of inertia and the viscous friction based on the differential signal.

3. In the motor control device according to claim 2, the differential signal calculation unit or the second differential signal calculation unit is a linear differential filter with a time difference as a parameter, and the time difference is set based on the acceleration command and its frequency characteristics.

4. In the motor control device according to claim 1, the identification section determination unit regards a section in which the differential signal of the acceleration command is non-zero as a valid section and operates the identification unit.

5. In the motor control device according to claim 1, the identification section determination unit regards a section in which the absolute value of the differential signal of the acceleration command or the time average value of the absolute value of the differential signal of the acceleration command is equal to or greater than a predetermined value as a valid section and operates the identification unit.

6. In the motor control device according to claim 1, the speed command generation unit generates a speed command that does not include a dominant frequency component for exciting the machine, obtained by integrating the acceleration command in which the maximum value, the minimum value, and the zero point are connected by a combination of functions of the first order or higher.

7. In the motor control device according to claim 1, has a speed control unit, and has an automatic adjustment unit that adjusts the control gain of the speed control unit based on the estimated values of the moment of inertia and the viscous friction obtained by the identification unit.

8. In the motor control device according to claim 1, A position command generation unit that generates a position command by integrating the speed command generated by the speed command generation unit. A motor control device having a position control unit that controls the position of the machine so as to follow the position command. **Claim 9** An automatic adjustment method for a motor control device that controls a machine, comprising: A speed command generation step of generating a speed command obtained by integrating an acceleration command; A step of driving a motor based on the speed command; A differential signal calculation step of calculating a differential signal from the acceleration commands with a time difference; An identification step of estimating the moment of inertia and viscous friction; Based on the differential signal, determining an effective section for executing the identification step; An identification section determination step of causing the identification step to be executed in an effective section; An automatic adjustment step of adjusting a control gain based on the estimated values of the moment of inertia and the viscous friction. An automatic adjustment method for a motor control device having these steps. **Claim 10** In the automatic adjustment method for a motor control device according to Claim 9, The identification step includes: A second differential signal calculation step of calculating a second differential signal with a time difference for the observed value of the torque of the motor and the observed value of the rotational angular velocity of the motor; An automatic adjustment method for a motor control device that estimates the moment of inertia and the viscous friction based on the second differential signal. **Claim 11** In the automatic adjustment method for a motor control device according to Claim 10, The differential signal calculation step or the second differential signal calculation step: Performs a linear difference process using the time difference as a parameter; The time difference is set based on the acceleration command and its frequency characteristics. An automatic adjustment method for a motor control device. **Claim 12** In the automatic adjustment method for a motor control device according to Claim 9, The identification section determination step: Regards a section in which the differential signal is non-zero as an effective section and operates the identification step. An automatic adjustment method for a motor control device. **Claim 13** The automatic adjustment method for a motor control device according to Claim 9, comprising: The identification section determination step: Regards a section in which the absolute value of the differential signal of the acceleration command or the time average value of the absolute value of the differential signal of the acceleration command is equal to or greater than a predetermined value as an effective section; An automatic adjustment method for a motor control device that operates the identification step. **Claim 14** The automatic adjustment method for a motor control device according to Claim 9, comprising: The speed command generation step: An automatic tuning method for a motor control device that generates a speed command that does not include a dominant frequency component that excites the machine, obtained by integrating the acceleration command in which the maximum value, the minimum value, and the zero point are connected by a combination of functions of the first order or higher.

Images