JP2005245088A - Motor controller - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a motor controller capable of computing the frequency of vibration generated when a motor drives a load machine. <P>SOLUTION: The motor controller comprises: a difference computing portion that determines a difference in speed feedback signal every control sampling time and generates a false acceleration signal; an adaptive filter that is fed with a false acceleration signal and outputs a filter factor; and a frequency computing portion that is fed with a filter factor and determines the frequency of vibration generated between a motor and a load machine. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a motor control device capable of calculating the frequency of vibration generated when a motor drives a load machine.

As a conventional vibration frequency calculation device and method for a motor control device, there is, for example, Patent Document 1.
FIG. 7 is a block diagram of one embodiment of the invention disclosed in Patent Document 1. In FIG. In FIG. 7, reference numeral 701 denotes an amplifier that gives a position loop gain, 702 an amplifier that gives a speed loop gain, 703 a motor, 704 a pulse generator, 705 a machine, 706 a drive mechanism, and 707 a machine position detector. In the command system to which the machine position command P _cmd is input, a sweep signal generator 709 is connected to an adder that takes a difference from the machine position signal. In this generator 709, from a low frequency to a high frequency (about 1 Hz to 100 Hz). A sine wave with a constant amplitude is generated and the position command is output. On the other hand, a branching into a position feedback loop having a machine position detector 707 is provided with a frequency analyzer 710. The frequency analyzer 710 performs frequency analysis on the machine position signal, and the machine position is determined with respect to the resonance frequency, that is, the input amplitude. The frequency which becomes extremely large is obtained. Specifically, it can also be realized by performing a fast Fourier transform with computer software that can be realized by connecting a plurality of bandpass filters in parallel and monitoring the output signal. A cut-off frequency variable notch filter 711 is inserted in the subsequent stage of the amplifier providing the speed loop gain. This cut-off frequency variable notch filter 711 changes the constant so that only the resonance frequency obtained by the frequency analyzer 710 is removed from the command system. As an actual operation, a sweep signal in which the frequency is gradually changed with a constant amplitude is applied as a position command from the sweep signal generator 709 in the same state as that at the time of machining, for example, a state in which a workpiece is placed on a table. . Then, the resonance frequency is obtained by the frequency analyzer 11 based on the sweep signal. On the other hand, since the notch filter 711 removes the resonance frequency from the command in accordance with the machine characteristics in the same state as during machining, it can prevent machine resonance even if an arbitrary position command is given during actual machining.

Further, as a second conventional technique, for example, there is Patent Document 2.
FIG. 8 is a block diagram of one embodiment of the invention disclosed in Patent Document 2. In FIG. In FIG. 8, 809 is a main filter capable of restricting the passage of frequency components according to the filter coefficient f by inputting the filter coefficient f, and 810 is a filter coefficient setting means for setting the filter coefficient f and outputting it to the main filter 809. is there. A speed detection value ω _r output from a speed detector 802 serving as a speed detection unit is input 812, and a frequency component of a predetermined frequency or higher superimposed on the speed detector ω _r , that is, an electric motor 801, a coupler 803, A high-pass filter that passes the resonance frequency component d of the mechanical system including the load 804, 813 is a reference signal generation unit that outputs a reference signal r including a frequency component equal to or lower than a predetermined frequency, and 815 is a reference that is output by the reference signal generation unit 813 An adding means for adding the signal r and the resonance frequency component d of the mechanical system that has passed through the high-pass filter 812; 816 inputs the output signal x of the adding means 815 and the reference signal r and restricts the passage of the frequency component d. It is an adaptive filter that calculates the filter coefficient f and outputs it to the main filter 809. The basic configuration of this adaptive filter is a block diagram referring to 7.2.1 of Non-Patent Document 3, as shown in page 9, line 8 of Patent Document 2.

As the actual operation, as described from page 28, line 28 of Patent Document 9, the high-pass filter in the filter coefficient setting means 810 has the cutoff frequency set to the speed control bandwidth f _b , A speed ω _r is input and a mechanical resonance frequency component d on which the ω _r is superimposed is output. The reference signal generating means 813 generates a reference signal r including the same frequency component as the speed control band f _b, and the adder 815 adds the output d of the high-pass filter 812 and the reference signal r and outputs a signal x. An adaptive filter is a filter that changes its characteristics to generate a signal that best suits the purpose when an input signal is applied, and changes its characteristics following changes in the nature of the input signal. Therefore (see 7.1.1 of Non-Patent Document 3), the adaptive filter 816 sequentially adjusts its own filter coefficient so that the output y is closest to the reference signal r that is the target signal with respect to the input x. Since the input x includes the reference signal r and the vibration component d due to mechanical resonance, the adaptive filter 816 allows the frequency band component of the reference signal r to pass through its own filter coefficient, and the vibration component d due to mechanical resonance. frequency components having suppress filter, namely the center frequency f _c is to adjust the filter coefficients so that the notch filter that matches the mechanical resonant frequency f _p. Further, the main filter 809 has notch filter characteristics because its filter coefficient is set to the same value as that of the adaptive filter 816 by the input from the adaptive filter 816. As described above, the main filter 809 automatically adjusts so as to have a characteristic that suppresses a mechanical resonance frequency component that is a vibration excitation source included in the output τ ₁ ^* of the operational amplification unit 806.

Japanese Patent Laid-Open No. 5-346813 (FIG. 1 on page 3) Japanese Patent Publication No. 2504307 (FIG. 1 on page 9) S. Hekin, “Introduction to Adaptive Filters”, Modern Engineering, 1987, (Chapter 4) Takashi Tanizaki “Theory of Digital Signal Processing 3” Corona, 1986 (Part 8) Takebe Miki “Digital Filter Design”, Tokai University Press, 1986

  However, in the first prior art, a frequency is calculated by applying a sweep signal that gradually changes the frequency from the outside to the input to excite mechanical vibration, and then performing a Fourier transform or frequency analysis on the machine position signal. It was. For this reason, it was impossible to calculate the frequency during normal operation (when cutting or simply moving the table without inputting the sweep signal), and it was necessary to switch to the above sweep signal and give the frequency once. . This can be used when the frequency is calculated as described above during a test run, but if the frequency changes due to aging, etc., the above operation must be performed to measure the frequency again, and the frequency is measured. It took a long time. In addition, there is a limitation of a machine that can input a sweep signal, a signal including a multi-frequency signal, or the like as a command input. In order to excite the machine vibration, it is necessary to increase the amplitude of the input. If the amplitude is increased, the machine is greatly shaken, which may damage the machine. In particular, giving a sweep signal of 1 Hz to 100 Hz depends on the control band, so it produces a loud sound near the mechanical resonance and the machine shakes. Furthermore, adjusting the best according to the work piece means that the work piece is placed on the table and the work piece is not placed on the table (no load) because there is no work and no vibration is generated. There is a problem that adjustment is necessary before driving in each state.

The second prior art is a device that inputs a signal including mechanical vibrations to an adaptive filter and uses the coefficient corrected by the successive correction means of the adaptive filter as the coefficient of the main filter. Patent Document 2 on page 10, lines 22 to 42 describes the inside of the adaptive filter. If the filter part inside the adaptive filter is an FIR filter, the order of the filter needs to be increased if the sampling time is shortened. If the main filter uses the same coefficient, it is necessary to increase the filter order by the same amount. In such a case, since the delay of the control system becomes large, it is possible to make the control system unstable. Further, when the filter portion inside the adaptive filter is an IIR filter, the filter order can be fixed to the second order even if the sampling time is shortened, but as described in Chapter 8 p165 to 166 of Non-Patent Document 2. In the case of a recursive adaptive digital filter (IIR type), there is a description that the convergence and optimality of the solution and the stability of the filter are not compensated. In addition, it is a problem common to the above two filters. When there are a plurality of vibrations, it is not known which vibration should be removed. It becomes an adaptive filter that does not converge forever, and there is a problem that vibration cannot be removed.
The present invention has been made in view of such problems, and during normal operation in which a sweep signal or the like is not input (when cutting or simply moving the table without inputting a sweep signal) or a plurality of vibration frequencies are present. There is provided a motor control device capable of calculating the frequency of vibration generated when a motor drives a load machine even if it exists.

The invention according to claim 1 is a speed control unit that outputs a torque command by controlling a difference between a speed command and a speed feedback signal, a current control unit that receives the torque command and controls a motor current, and a position detection In a motor control device that is configured by a speed detection unit that generates a speed feedback signal from a position feedback signal from the means and controls a motor based on a command from a host system and an output from a position or speed detection unit, the speed feedback signal is A difference calculation unit that generates a pseudo acceleration signal by taking a difference every control sampling time, an adaptive filter that inputs the pseudo acceleration signal and outputs a filter coefficient, and is generated between the motor and the load machine using the filter coefficient as an input. A frequency calculation means for determining the frequency of vibration is provided.
According to a second aspect of the present invention, in the motor control apparatus according to the first aspect, the adaptive filter includes an FIR filter having a filter coefficient that can be corrected, and coefficient correction means for sequentially updating the filter coefficient of the FIR filter, and the FIR filter The order of 2 is an integer multiple of 2 and 2 and the coefficient correction means corrects the output of the FIR filter to be a predetermined value or less and outputs from the coefficient correction means when the output of the FIR filter becomes a predetermined value or less. Output filter coefficients.
According to a third aspect of the present invention, in the motor control device according to the first aspect, when the order of the FIR filter capable of correcting the filter coefficient is 2, the frequency calculation means has the formula f = √ {(2+ (h ₁ / H ₀ )) / (2- (h ₁ / h ₀ ))} * (2 / T _s ) * (1 / 2π)
The operation is performed using. Here, h ₁ and h ₀ are the coefficients of the FIR filter, and T _s is the sampling time.
According to a fourth aspect of the present invention, in the motor control device according to the first aspect, when the order of the FIR filter capable of correcting the coefficient is 4, the frequency calculation means has the formula m = h ₁ / h ₀
n = (h ₂ −2h ₀ ) / h ₀
a ₁ = (m + √ (m ² −4n)) / 2
a ₂ = (m−√ (m ² −4n)) / 2
f ₁ = √ {(2 + a ₁ ) / (2-a ₁ )} * (1 / T _s ) * (1 / 2π)
f ₂ = √ {(2 + a ₂ ) / (2-a ₂ )} * (1 / T _s ) * (1 / 2π)
The operation is performed using.
According to a fifth aspect of the present invention, there is provided the motor control device according to the first aspect, wherein a plurality of adaptive filters are coupled in series, and a switching means is provided after the filter coefficients of the adaptive filter are output. Each output filter coefficient is calculated.

  ADVANTAGE OF THE INVENTION According to this invention, the motor control apparatus which can calculate the frequency of the vibration generate | occur | produced when a motor drives a load machine can be provided. It can be executed during normal operation (when cutting or simply moving the table without inputting a sweep signal), and the vibration frequency can be measured in a short time because no Fourier transform is used.

  Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing a motor control device for realizing a frequency calculation method according to an embodiment of the present invention.
In FIG. 1, 2 is a speed control unit, 3 is a current control unit, 4 is a motor, 5 is a load machine, 6 is an encoder as position detection means, 7 is a speed detection unit, 9 is a difference calculation unit, and 10 is an adaptive filter. , 12 are frequency calculation units. 1 is a speed command, 8 is a speed feedback signal, and 7 is a filter coefficient.

  Next, the operation will be described. The speed control unit 2 performs control calculation processing on the difference between the speed command 1 and the speed feedback signal 8 and outputs a torque command. The current control unit converts the torque command into a current command and controls the current flowing through the motor. When current flows through the motor, the motor generates torque and rotates according to the load and moment of inertia. The encoder 6 is mechanically coupled to the motor 4 and generates a pulse corresponding to the rotational angle position of the motor. Although not shown, the encoder pulses are accumulated by a position counter, and the output of the position counter becomes a position feedback signal. The speed detection unit 7 generates a speed feedback signal by taking the time difference of the position feedback signal for each control sampling period and dividing it by the sampling time. A load 5 is coupled to the motor, and the motor is configured to drive the load. The difference calculation unit 9 calculates a difference between the speed feedback signals for each sampling time and outputs a pseudo acceleration signal. The adaptive filter 10 receives the pseudo acceleration signal and outputs a filter coefficient. The frequency calculation unit 12 inputs a filter coefficient and outputs a vibration frequency f.

Next, the frequency calculation unit 12 will be described with reference to FIG.
FIG. 2 shows the internal structure of the adaptive filter 10 of FIG. In FIG. 2, FIR filter 101 may coefficient correction, outputs the pseudo acceleration signal output from the difference computing unit 9 of FIG. 1 from the result and the coefficient correction means 100 through the sampling time delayer z ^-1 The product-sum operation is performed using the coefficients to be output to the coefficient correction means 100. The coefficient calculation means 100 corrects the coefficients h0, h1, and h2 so that the output of the FIR filter is within a predetermined value. The coefficient correction means is executed by a least square (LMS) algorithm as shown in (4.16) in Chapter 4, p96 of Non-Patent Document 1. Further, if this least square algorithm is changed to a normalized LMS algorithm, the coefficient accuracy is increased and the coefficient is rapidly converged, so that the measurement accuracy of the vibration frequency is increased. This normalized LMS algorithm is shown in Equation (4.67) in Chapter 4, p120 of Non-Patent Document 1. The FIR filter can be expressed by a mathematical formula as shown in formula (1). h ₀ , h ₁ and h ₂ are coefficients of the FIR filter, and z ⁻¹ represents a delay device. z- ² similarly represents two delay devices.

Next, a simulation in which frequency calculation is performed with the configuration of the first embodiment is shown. The simulation of the configuration in FIG. 1 was executed when the control sampling time of the speed control was set to 125 μs and the resonance frequency between the motor and the machine was 500 Hz. FIG. 5 shows the frequency characteristics of the FIR filter made by the adaptive filter. FIG. 5 shows −30 dB at a frequency lower than the notch center frequency and 0 dB at a high frequency. Of the coefficients h ₀ , h ₁ , h ₂ of the FIR filter at this time, h ₀ , h ₁ are input to the equation (2). In the frequency calculation unit, equation (2) is programmed by software to calculate the frequency. In Expression (2), f represents the calculated vibration frequency, and Ts represents the sampling time of the adaptive filter. When the FIR filter coefficient (h ₀ , h ₂ = 0.1840, h ₁ = −0.34) when FIG. 5 is obtained is executed by the frequency calculation unit (formula (2)), 506.5 Hz is obtained. It was. It can be seen that the vibration frequency is calculated by executing claims 1 and 2.

As another example, FIG. 6 shows the frequency characteristics obtained as a result of evaluating the case where the resonance frequency of the motor and the machine is 100 Hz. When the FIR filter coefficients (h ₀ , h ₂ = 0.1673, h ₁ = −0.3336) at this time are input to the frequency calculation unit, 100 Hz is calculated. From this result, it can be seen that the vibration frequency is calculated by executing claims 1 and 2.

FIG. 3 shows the inside of the adaptive filter of FIG. 1, but the order of the FIR filter 201 that can modify the coefficients is the fourth order, and the coefficient correcting means 200 corrects the five coefficients. Five coefficients are input to the frequency calculation unit 12 in FIG. In the frequency calculation unit, equation (3) is programmed by software, and the frequency is calculated. In equation (3), h ₀ , h ₁ , and h ₂ represent FIR filter coefficients, and the coefficients are composed of five of h ₀ to h ₄ , of which h ₀ to h ₂ are used. a ₁ and a ₂ are intermediate variables, Ts is a sampling time of the adaptive filter, and f ₁ and f ₂ are vibration frequencies. This example is applied when there are two vibration frequencies, and two frequencies f1 and f2 are calculated.

G (z) = h ₀ + h ₁ z ⁻¹ + h ₂ z ⁻² (1)
f = √ {(2+ (h ₁ / h ₀ )) / (2- (h ₁ / h ₀ ))} * (2 / T _s ) * (1 / 2π)
... (2)
m = h ₁ / h ₀
n = (h ₂ −2h ₀ ) / h ₀
a ₁ = (m + √ (m ² −4n)) / 2
a ₂ = (m−√ (m ² −4n)) / 2
f ₁ = √ {(2 + a ₁ ) / (2-a ₁ )} * (1 / T _s ) * (1 / 2π)
f ₂ = √ {(2 + a ₂ ) / (2-a ₂ )} * (1 / T _s ) * (1 / 2π)
... (3)

Example 3 is devised to measure the vibration frequency even when there are three or more vibrations, and is shown in FIG. Drawing numbers 1 to 9 are the same as those described in FIG. 1, 300 adaptive filters 1, 301 adaptive filters 2 and 302 adaptive filters 3 are connected in series, and the outputs of adaptive filters 1 to 3 are as follows. FIR filter outputs 306 and 307 and coefficient coefficients of the adaptive filter 1 within each adaptive filter, the filter coefficient 303 of the adaptive filter 1, the filter coefficient 304 of the adaptive filter 2, and the filter coefficient 305 of the adaptive filter 3 are switching means 308. The frequency calculation unit 309 calculates the frequency of each adaptive filter, and displays the frequency on a display device having a display function 310, for example. If the FIR filter capable of correcting the coefficients in the adaptive filter is limited to the second order and the calculation of claim 2 is performed, the same number of vibration frequencies can be measured for a plurality of vibrations.

  Since the vibration frequency is known by the frequency calculation means, it can be applied to a control device that suppresses vibration using the vibration frequency.

The block diagram of the control apparatus which implement | achieves the frequency calculation which shows 1st Example of this invention The block diagram which shows the internal operation | movement of the adaptive filter of Claim 2 of this invention The block diagram which shows the internal operation | movement of the adaptive filter of Claim 3 of this invention The block diagram of the control apparatus which shows Claim 4 of this invention Frequency characteristics of adaptive filter according to embodiment 1 (when vibration frequency is 500 Hz) Frequency characteristics of adaptive filter according to embodiment 1 (when vibration frequency is 100 Hz) The figure which shows the Example of patent document 1 The figure which shows the Example of patent document 2

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Speed command 2 Speed control part 3 Current control part 4 Motor 5 Load 6 Encoder 7 Speed detection part 8 Speed feedback signal 9 Difference calculation part 10 Adaptive filter 11 Filter coefficient 12 Frequency calculation part 100 Coefficient correction means 101 FIR filter 200 Coefficient correction means 201 FIR filter 300 Adaptive filter 1
301 Adaptive filter 2
302 Adaptive filter 3
303 Filter Coefficient 304 Filter Coefficient 305 Filter Coefficient 306 FIR Filter Output 307 FIR Filter Output 308 Switching Unit 309 Frequency Calculation Unit 310 Display 701 Amplifier 702 that Provides Position Loop Gain Amplifier 703 Motor that Provides Speed Loop Gain 703 Motor 704 Pulse Generator 705 Machine 706 Drive machine 707 Machine position detector 708 Differentiator 709 Sweep signal generator 710 Frequency analyzer 711 Frequency variable notch filter 801 Electric motor 802 Speed detector 803 Linkage 804 Load 805 Subtraction means 806 Operation amplification means 808 Torque control means 809 Main Filter 810 Filter coefficient setting means 812 High pass filter 813 Reference signal generation means 815 Addition means 816 Adaptive filter

Claims (5)

A speed control unit that outputs a torque command by controlling the difference between the speed command and the speed feedback signal, a current control unit that controls the motor current using the torque command as an input, and a speed feedback from the position feedback signal from the position detection means In a motor control device configured by a speed detection unit that generates a signal and controlling a motor based on a command and position from a host system or an output from the speed detection unit,
A difference calculation unit that generates a pseudo acceleration signal by taking a difference between the speed feedback signals at each control sampling time;
An adaptive filter that receives the pseudo acceleration signal and outputs a filter coefficient;
A motor control apparatus comprising frequency calculating means for obtaining a frequency of vibration generated between a motor and a load machine using the filter coefficient as an input. The adaptive filter comprises an FIR filter having a filter coefficient that can be corrected, and coefficient correction means for sequentially updating the filter coefficient of the FIR filter,
The order of the FIR filter is 2 or more and an integer multiple of 2;
2. The coefficient correction unit according to claim 1, wherein the coefficient correction means corrects an output of the FIR filter to be a predetermined value or less, and outputs a filter coefficient when the output of the FIR filter becomes a predetermined value or less. Motor control device. When the order of the FIR filter capable of correcting the filter coefficient is 2, the frequency calculation means uses the equation f = √ {(2+ (h ₁ / h ₀ )) / (2- (h ₁ / h ₀ ))} * (2 / T _s ) * (1 / 2π)
The motor control apparatus according to claim 1, wherein the calculation is performed using
Here, h ₁ and h ₀ are the coefficients of the FIR filter, and T _s is the sampling time. When the order of the FIR filter capable of correcting the filter coefficient is 4, the frequency calculation means uses the equation m = h ₁ / h ₀
n = (h ₂ −2h ₀ ) / h ₀
a ₁ = (m + √ (m ² −4n)) / 2
a ₂ = (m−√ (m ² −4n)) / 2
f ₁ = √ {(2 + a ₁ ) / (2-a ₁ )} * (1 / T _s ) * (1 / 2π)
f ₂ = √ {(2 + a ₂ ) / (2-a ₂ )} * (1 / T _s ) * (1 / 2π)
The motor control apparatus according to claim 1, wherein the calculation is performed using   The adaptive filter includes a plurality of adaptive filters coupled in series, and provided with a switching means after outputting the filter coefficients of the adaptive filter, and the frequency calculating means calculates each filter coefficient output from the adaptive filter. Item 2. The motor control device according to Item 1.

Images