JP2587701B2 - Oscillation detection of servo system and automatic adjustment of speed loop gain - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a system for detecting oscillation of a servo system and adjusting a speed loop gain of a servo system in a machine driven by a servomotor such as a robot or a feed shaft of a machine tool. .

2. Description of the Related Art FIG. 11 is a block diagram of a speed loop in a servomotor. In the block diagram, k1 is an integral gain, k2 is a proportional gain, P is a parameter, and when the parameter P is "1", When the proportional integral (PI) control is performed and the parameter P is set to “0”, the integral proportional (IP) control is performed, and is a parameter for determining the PI control and the IP control. Kt is the torque constant, and Jm is the inertia of the entire system. Vc is a speed command signal, and Vt is a speed signal. The block diagram of the speed loop shown in FIG. 11 is generally known, and a detailed description thereof will be omitted.

This speed control loop is a secondary system as is clear from the block diagram of FIG. And the cutoff frequency
Once fn (Hz) and damping constant ξ are determined, the integral gain k
1. The proportional gain k2 is determined by the following equations (1) and (2).

k1 = (Jm / Kt) · (2πfn) ² … (1) k2 = (Jm / Kt) · 2ξ · (2πfn) ² … (2) The speed loop gain determined by the integral gain k1 and the proportional gain k2 is If it is low, the servomotor will not rotate due to the low gain even if the servomotor is sent for one pulse, and the servomotor will rotate only after a few pulses of the movement command, causing a feed phenomenon or causing disturbance. Also, the speed of the servomotor causes low frequency undulations. On the other hand, if the speed loop gain is too high, vibration occurs due to system resonance or the like at the time of normal speed, for example, in the case of a machine tool during cutting feed or rapid feed.

Therefore, conventionally, the integral gain k1,
In order to determine the proportional gain k2, the cutoff frequency fn and the damping constant ξ are determined by trial and error, and set so that the integral gain k1 and the proportional gain k2, which should be within the range that does not cause oscillation due to system resonance, etc., become large. doing.

Problems to be Solved by the Invention As described above, the adjustment of the speed loop gain (k1, k2) must be performed for each machine by trial and error so that transmission due to system resonance or the like does not occur. Requires a lot of trouble and time.

Therefore, a first object of the present invention is to provide a detection method for automatically detecting system oscillation.

A second object of the present invention is to provide a speed loop gain automatic adjustment method capable of automatically determining a speed loop gain.

Means for Solving the Problems The position command is set to zero, and a speed command having a predetermined offset value is given according to the position deviation value to cause vibration in the servo system, and the actual speed change or position of the servo motor obtained at this time is obtained. A change in the deviation amount is detected, and a frequency analysis of the actual speed change or the change in the position deviation amount is performed. Alternatively, the change in the differential value is obtained by further differentiating the change in the actual speed or the change in the position deviation amount, and the frequency analysis of the change in the differential value is performed.
The frequency of the maximum amplitude is obtained from the frequencies of the amplitudes obtained by the frequency analysis. Then, the speed loop gain is increased, and the above-mentioned frequency is sequentially obtained. When the frequency suddenly rises extremely greatly with respect to the fluctuation up to that time, the state at that time is detected as the oscillation of the servo system. Further, when the frequency band of the speed loop is low (the speed loop gain is small), the offset value is increased, and the offset value is decreased as the frequency band increases, thereby detecting the oscillation of the servo system.

Further, when adjusting the speed loop gain, the speed loop gain is automatically adjusted sequentially in accordance with the above-described method until the frequency of the maximum amplitude reaches a predetermined range with respect to a preset reference frequency. Is determined as the gain of the speed loop, so that the optimum speed loop gain can be automatically determined.

When the position command is set to zero and a speed command having a predetermined offset value is given in accordance with the position deviation value, a speed command having an offset value is input to the speed loop due to a small amount of position deviation, and the servo motor reduces friction. Defeat and move.
That is, a stick step is performed. As a result, the position deviation amount increases in the reverse direction, so that a speed command in the reverse direction is output, and the servo motor rotates in the reverse direction this time. This is repeated, and the servo motor will vibrate.
At this time, a change in the actual speed of the servomotor or a change in the position deviation value is detected, and a frequency analysis is performed to detect a frequency at which the amplitude becomes maximum. Then, when the speed loop gain is sequentially increased, the frequency of the maximum amplitude is greatly increased as compared with the increase of the frequency due to resonance of the servo system including the machine driven by the servomotor. By detecting this very large rising point, the oscillation of the servo system can be detected.

However, in the above method, the resonance frequency cannot be detected while the resonance is weak, and the system oscillates only when the resonance frequency becomes dominant and the amplitude of the resonance frequency becomes the maximum when the system starts full-scale resonance. Not detectable.

Generally, the resonance frequency of the mechanical system is several times higher than the vibration frequency generated by the control system itself of the servo system. Suppose the basic frequency of the control system is f0, and the resonance frequency of the machine is mf
Actual speed change of servo motor when set to 0 (actual speed waveform)
Is represented by the following equation (3).

v (t) = A1sin (2πf0t) + A2sin (2πmf0t + ρ) (3) where ρ is a phase delay.

On the other hand, when the above equation (3) is differentiated to obtain a differential signal ω (t), the following equation (4) is obtained.

ω (t) = dv (t) / dt = 2πf0A1cos (2πf0t) + 2πmf0tA2cos (2πmf0t + ρ) (4) Therefore, the amplitude A1 of the fundamental frequency f0 and the machine speed A1 at the actual speed change obtained by the above equation (3) Resonance frequency mf
While the ratio of the amplitude A2 of 0 is A2 / A1, the amplitude ratio of the differential signal ω (t) expressed by the equation (2) is mA2 / A1, and m
Double. Therefore, rather than analyzing the actual speed change (position deviation change) by frequency analysis, the value obtained by differentiating the actual speed change (position deviation change), that is, the frequency analysis of the acceleration change, to obtain the frequency of the maximum amplitude is more speed loop. Oscillation of the system due to an excessive increase in gain can be detected at an early stage where the influence of the oscillation is weak.

Further, when the offset value is small, when the frequency band of the speed loop is low (when the speed loop gain is small), the servomotor vibrates within the backlash of the mechanical system, and the friction torque of the mechanical system becomes smaller. If the offset value is large and the frequency band of the speed loop is high, the torque command value may be saturated due to the torque limit, the motor may generate heat, or a large mechanical force may be generated in the mechanical system. The above problem is solved by reducing the offset value in proportion to the increase of the frequency band of the speed loop since the target vibration is applied.

When adjusting the gain of the speed loop, a desired reference frequency empirically obtained by a system such as a speed loop and a position loop of a servo system is set, and the oscillation of the servo system is detected. In the same manner as in the above, the speed loop gain is sequentially changed and automatically adjusted so that the frequency having the maximum detected amplitude reaches the same as or near the desired reference frequency, and enters the same or within a predetermined range. And the speed loop gain at that time is automatically determined as the speed loop gain of the speed control loop.

Example Hereinafter, an example of the present invention will be described.

FIG. 1 is a block diagram of a position loop including a speed loop in the present embodiment.
This is the point that has been added.

In FIG. 1, 1 is a proportional term in the position loop,
Kp is the position gain, 2 is the function generator, 3 is the integral term of the speed loop, k1 is the integral gain, 4 is the proportional term of the velocity loop, k2 is the proportional gain, 5 is the parameter P term, and PI control In this case, the value is "1", and in the case of IP control, it is "0". 6 and 7 are the terms of the transfer function of the servo motor,
Is the torque constant, and Jm is the inertia of the entire system. Also,
Reference numeral 8 denotes an integral term for converting the speed Vt into the position Pr.

The proportional term 1 is the position command Pc and the actual position of the servomotor.
The difference from Pr, that is, the position deviation amount is amplified by the gain Kp and output as the speed command Vc (in).
The speed command Vc (in) is input to the speed loop as a speed command. In the present invention, when the speed loop gain is adjusted, the speed command Vc (in) is input to the function generator 2 and the function generator 2 executes the speed command Vc (in). Is converted into a speed command Vc (ou
t) and output to the speed loop. In the speed loop, from the difference between the speed command Vc (out) and the speed signal (actual speed of the servo motor) Vt, that is, the value obtained by integrating the speed deviation amount by the integration term 3, when the parameter P is "1", the speed signal Vt
The value obtained by multiplying the value obtained by subtracting the speed command value Vc (out) from the control signal by the proportional gain is output to the servo motor as a torque command. When the parameter P is "0", the integral term The value obtained by multiplying the speed signal Vt by the proportional gain k2 is subtracted from the integrated value output from 3 and the value is output as a torque command to drive the servomotor.

As shown in FIG. 2, the function generator 2 has an offset value ± A with a position deviation amount of zero and an input speed command Vc (i
A function generator set so that the output speed command Vc (out) increases or decreases with a proportionality constant K as n) increases in the plus or minus direction, as shown in FIG. When the input Vc (in) is near zero (position deviation amount is near zero), the input Vc (out) greatly changes with a slight change in the input Vc (in), and the other parts are the input Vc (in).
Even if c (in) changes, the amount of change in output Vc (out) is small.
The function generator 2 having such an input / output relationship is configured.

That is, the input / output relationship of the function generator is expressed by the following (5)
Equation (6) is obtained.

When Vc (in) ≧ 0 Vc (out) = K · Vc (in) + A (5) When Vc (in) <0 Vc (out) = K · Vc (in) −A (6) Then, in the position loop at the time of gain adjustment shown in FIG. 1, the servomotor always vibrates when the position command Pc is zero. That is, even if the position command Pc is “0”, the position deviation amount slightly varies. Therefore, as shown in FIG. 2, if the position deviation amount slightly fluctuates from “0”, the speed command output from the function generator 2 to the speed control loop
Vc (out) varies greatly, and the servomotor rotates. If the servomotor rotates, the position deviation increases in the reverse direction, and the function generator 2 outputs the speed command Vc in the reverse direction.
(Out) is output, and the servomotor rotates in the opposite direction this time. This is repeated, causing the servo motor to vibrate.

The frequency of the vibration of the servomotor is high if the band frequency (cutoff frequency fn) of the speed control loop is high, because the response of the speed control loop is fast.

Therefore, the vibration of the servo motor, that is, the speed signal Vt
Is detected, and this frequency also includes other frequency parts due to friction or the like, so if the main vibration component, that is, the frequency component having the largest amplitude is extracted by frequency analysis such as Fourier transform, this frequency It is assumed that the main vibration component has responded to the speed command Vc (out). If the resonance does not occur in the mechanical system, the frequency with the maximum amplitude is the speed loop gain.
It increases continuously when K1 and K2 are increased sequentially. But,
When resonance occurs in the mechanical system, the frequency having the maximum amplitude suddenly fluctuates much more greatly than the fluctuation of the frequency due to the influence of the resonance frequency. Therefore, a point in time at which this frequency fluctuates greatly is detected, and this time is detected as oscillation of the servo system.

FIG. 3 is a flowchart of a process for each speed loop processing cycle in a servo system oscillation detection mode performed by a processor (hereinafter, referred to as a CPU) of a digital servo circuit that controls a servo system by software.

First, in a numerical controller for controlling a machine tool, a robot, or the like, a cutoff frequency fn is set to a set value fn0, a damping constant に is set to a set value ξ0 as initial parameters, and a parameter P is set to “0”. The CPU of the digital servo circuit performs the calculations of the above equations (1) and (2) to initialize the speed loop gains K1 and K2, and initializes a flag F and an index i to be described later to "0". . When the mode is set to the servo system oscillation detection mode, the CPU of the servo circuit executes the processing of FIG. 3 every speed loop processing cycle.

First, it is determined whether or not the flag F is "1" (step 10).
0) If it is "0", the speed command Vc (in) output from the position loop 1 is read (step 101). Since the position command Pc is "0" in the servo system oscillation detection mode, the speed command
Vc (in) is “0” or a speed command Vc (ou
Since t) has the offset value A, it takes a positive or negative value. Therefore, it is determined whether or not the read speed command Vc (in) is equal to or greater than zero (step 102). If the read speed command Vc (in) is equal to or greater than zero, the speed command Vc (out) to be output to the speed loop is output by performing the calculation of equation (5). (Step 103), if negative (6)
The expression is calculated and output (step 104). That is, the processing of steps 102 to 104 is the processing of the function generator 2 in FIG. Next, the actual speed Vt is read, the actual speed Vt is stored in the memory Mi at the address indicated by the index i, and the index i is incremented by "1" (steps 105 to 107). Then, it is determined whether or not the value of the index i is larger than the set value N (step 108).
A normal speed loop process is performed based on the speed command Vc (out) obtained in step (step 110), and the process of the speed loop process cycle ends. In the speed loop process, a torque command value is calculated, and then, a current loop process and the like are performed in the same manner as in the related art. A drive current flows through the servo motor, and the servo motor is driven.

The set value N is used for a frequency analysis described later.
This defines the number of data required for FFT processing.

The CPU repeats the above processing every speed loop processing cycle. For example, it is determined in step 102 that Vc (in) ≧ 0, and Vc (out) = K · Vc (in) + A as a speed command.
Is output, the servo motor rotates forward due to the offset value A, and as a result, the position deviation becomes negative,
The speed command Vc (in) output from the position loop 1 is negative. As a result, in the next cycle, the process of step 104 is performed, and the speed command Vc (out) to the speed loop is (K · Vc (in)
-A), a current flows so as to reverse the servo motor, and the servo motor and the machine vibrate. Then, the actual speed Vt detected for each speed loop period is stored in the memory (step 10).
5,106). When the value of the index i exceeds the set value N (step 108), the CPU sets the flag F to "1" (step 109). As a result, in the subsequent speed loop processing cycle, the flow proceeds from step 100 to step 101. Then, only the normal speed loop processing is performed.

On the other hand, the CPU of the digital servo circuit executes the fourth oscillation detection processing in the remaining time other than the cycle of the position loop processing, velocity loop processing, current loop processing, and the like for servo control.

First, it is determined whether or not the flag F is set to "1" (step 200). If not, no processing is performed. On the other hand, in step 109 of the speed loop process shown in FIG. 3, the flag F is set to "1" and the memories M0 to M
When (n + 1) actual speed data is stored in n, the actual speed data is read out and frequency analysis is performed.
In this embodiment, the FFT (Farst
Frequency analysis is performed by Fourier Transform (Fast Fourier Transform), and the amplitude c (f) of each frequency f _Hz component is obtained (step 201). Note that the FFT requires a predetermined number of data for frequency analysis, and the number of data must be a power of two. Therefore, the value of N set in the processing of FIG. 3 is a number required for frequency analysis of the number of data M0 to Mn, and the number is set to be a power of two.

Of the amplitude c (f) obtained in step 201, the frequency f _{Hz of the} maximum amplitude is obtained as fmax (step 202), and the frequency fmax of the maximum amplitude obtained in the previous process is stored in the storage register R (f) as " 0 ”or not (step 20).
3). Since the register R (f) is initially set to "0" in the initial setting, the process proceeds to step 205, where the frequency fmax obtained in the current process is stored in the register R (f), and the currently set value is stored. The set value fs is added to the existing cutoff frequency fn to obtain a new cutoff frequency fn (step 206).
Then, based on the frequency fn, the calculations of the equations (1) and (2) are performed to determine and set the gains K1 and K2 of the new speed loop (step 207).
Is set to “0”, and the current oscillation detection processing ends.

Since the flag F is set to "0", the processing of steps 100 to 110 in FIG. 3 is started again in the speed loop processing cycle, and the data of the actual speed Vt by the new speed loop gain is stored in the memories M0 to Mn. Will be stored.

Then, when the index i becomes "N" and the data is stored in the memories M0 to Mn and the flag F is set to "1", the processing of steps 200 to 208 in FIG. Since the frequency fmax is obtained and the previous frequency fmax is already stored in the register R (f) this time, step 203
The process further proceeds to step 204, and it is determined whether or not the frequency fmax obtained this time is not less than m times (m = 2 to 3) the previous frequency fmax (step 204). As described above, the speed loop gain K
When K2 rises and the mechanical system resonates, the effect of this resonance
The frequency fmax of the maximum amplitude greatly increases, and becomes m times as large as the frequency fmax of the previous cycle.
Since the frequency fmax does not increase drastically, it does not become m times, and the process proceeds to step 205, where the speed loop gain is increased and the index i and the flag F are set to "0" (step 206). ~ 208), again the third step 100 ~ 110
Start the process.

Hereinafter, the above-described processing is repeated, and if a frequency fmax that is m times or more than the previous frequency fmax is detected in step 204, the gain of the speed loop K1, K2, The cutoff frequency fn and the detection frequency fmax during oscillation are displayed, and the servo system oscillation mode processing ends.

In the above-described servo system oscillation detection method, even if the mechanical system causes resonance, the influence of the resonance is small while the resonance is weak,
The maximum amplitude is not detected corresponding to the resonance frequency. Therefore, as described above, the change in the acceleration is obtained by differentiating the change in the actual speed, and the frequency analysis is performed on the change in the acceleration (acceleration waveform) to detect the oscillation of the servo system. Then, as described in the equation (4), the oscillation of the system can be detected even at the stage where the resonance is weak.

For example, for a mechanical system with a mechanical resonance of 120 Hz (period is about 8 msec), adjust the gain of the speed loop to 25 Hz.
When the vibration is performed at a cycle of 40 msec, the obtained actual speed change (actual speed waveform) is as shown in FIG. On the other hand, the change of the differential value obtained by differentiating the velocity (velocity differential waveform) is as shown in FIG. 9, and it is understood that the change is dominant in the resonance frequency 120 Hz differential waveform, and the detection becomes easier.

At this time, the processing shown in FIG. 5 is executed for each processing of the speed loop processing cycle of the processor of the digital servo circuit in place of step 106 in FIG.

That is, stick slip processing is performed (step 10
(3, 104) After reading the actual speed Vt (step 105), the actual speed read in the previous cycle stored in the register R (v) is subtracted from the read actual speed Vt to obtain the acceleration signal a (i) (step 105). 300). Note that the register R (v) is set to “0” by default. Next, the read actual speed Vt
Is stored in the register R (v), and the acceleration signal a (i) calculated in step 300 is stored in the memory Mi (step 301,
302), the processing from step 107 onward in FIG. 3 is performed. Thus, the acceleration signals a (i) are stored in the memories M0 to Mn.

Next, in the oscillation detection processing, substantially the same processing as the processing shown in FIG. 4 is performed. The difference is that step 208
i, the flag F is set to "0" and the register R
(V) is also the only point set to “0”.

In this manner, the frequency analysis is performed on the acceleration change (acceleration waveform), and the oscillation of the servo system is detected as a null.
Oscillation can be detected more quickly than when oscillation is detected by a change in speed (change in position deviation).

Further, with respect to the offset value A, when the offset value A is small, vibration may occur during the backlash of the mechanical system or the friction torque of the mechanical system may be so large that stick slip may not occur. . This phenomenon is particularly likely to occur when the gain of the speed loop is low because the rise of the torque command is slow. Also, when the offset value A is large, the amplitude of the speed and the position becomes large. Especially, in a state where the gain of the speed loop is high, the torque command is saturated by the torque limit, or the heat generation of the motor or large mechanical Vibration here is not preferable.

Looking at the maximum value of the torque required to realize the stick slip, a stick slip is caused at the same offset value A, and the maximum torque value required at that time is experimentally obtained as shown in FIG. It increases almost in proportion to the frequency band of the speed loop. This is a cause of saturation of the torque command, generation of motor heat, and generation of large mechanical vibration in a high gain state where the frequency band of the speed loop is increased.

Therefore, in order to solve the above problem, when the frequency band of the speed loop is low, the offset value A is increased,
When it becomes higher, the offset value A may be reduced.
Specifically, the actual frequency band of the speed loop at a certain point in time is the frequency fmax of the maximum amplitude appearing at the actual speed of the motor when the stick slip is caused in the processing shown in FIGS. Therefore, the frequency fmax is regarded as the band of the velocity loop, and in this embodiment, A (fmax) = B /
The offset value A is changed as shown in FIG. 10 as fmax. Note that B is a fixed value, and an estimated value is set as an initial value of fmax.

In this case, among the processes shown in FIGS. 3 and 4, the servo value is set as an initial value before the process shown in FIG. 3 is started, and the offset value A of the initial value is determined. Unlike the system oscillation detection process, the processes in FIG. 3 are all the same. Then, in the oscillation detection process shown in FIG. 4, after the step 207, the process of step 400 shown in FIG. 6 is added, and the offset value A is changed to B / fmax every time the frequency fmax of the maximum amplitude is obtained. Then, the process proceeds to step 208. As a result, as the gain of the speed loop increases, the offset value A decreases, and the above-described problem is solved. In addition, instead of step 106 in FIG. 3, steps 300 to 302 in FIG. Perform processing and calculate the maximum amplitude frequency fm from the acceleration change (acceleration waveform).
Of course, even when ax is obtained, the offset value A may be updated.

Next, using the above-described servo system oscillation detection method,
A method for automatically setting the speed loop gain of the servo system to an optimum value will be described.

Switch the numerical controller to the automatic gain adjustment mode, and set the cutoff frequency fn to fn0 and the damping constant
Is set to ξ0, the parameter P is set to “0”, and a desired reference frequency fa empirically obtained by a system such as a speed loop and a position loop is set. The CPU of the digital servo circuit calculates the speed loop gains K1 and K2 by performing the calculations of the equations (1) and (2) from the above initial values, and initializes the index i and the flag F to "0". When a start command is input, the CPU starts the processing shown in FIG. 3, and writes the actual speed Vt of the motor in the memory in the same number as the number set for each speed loop processing cycle, similarly to the oscillation detection method of the servo system. When the data of the actual speed Vt is stored in the memories M0 to Mn and the flag F is set to "1" (step 109), the CPU starts the gain adjustment processing shown in FIG.

That is, it is determined whether the flag F is "1" (step
500), if not “1”, no gain adjustment processing is performed,
If it is “1”, the memory M
FFT analysis is performed on the actual speed data stored in 0 to Mn to determine the amplitude c (f) of the frequency component (step 201). Next, the maximum frequency f of the detected amplitudes is expressed as fmax
(Step 502).

Then, it is determined whether or not the absolute value of the difference between the obtained frequency fmax and the set reference frequency fa is smaller than a threshold ε (about 1 Hz) set for gain adjustment (step 503). If not smaller, the cutoff frequency fn currently set to the detection frequency fm from the reference frequency fa
Setting parameter α (about 0.5 to 0.7) minus ax
Are added to obtain a new cutoff frequency (step 504). That is, new fn = (old fn) + α (f
a-fmax).

Then, the new cutoff frequency fn is compared with the upper limit value flim of the speed loop band determined by the set control system (step 505). If the new cutoff frequency fn is small, the new cutoff frequency fn is set to (1) ), Gains K1 and K of the new velocity loop by calculating equation (2).
2 is set (step 508), and the index i and the flag F are set to “0”. By setting the flag F to “0”, the gain adjustment processing is performed only in step 500.

On the other hand, in the speed loop processing cycle, step 100 in FIG.
The flag F is determined to be "0", so that the processing of step 101 and subsequent steps is started again, and stick slip is performed under the new speed loop gains K1 and K2, and the data of the actual speed Vt is stored in each speed loop process. Is stored in the memory Mi. When the set number (N + 1) of data is stored in the memories M0 to Mn and the flag F is set to "1" (step 10).
9) Then, the gain adjustment processing shown in FIG. 7 is started again.

Thereafter, the above-described processing is repeated, and the speed loop gain
K1 and K2 are sequentially updated and become larger. When the cutoff frequency fn is equal to or higher than the upper limit value flim of the speed loop band (step 505), it is determined whether or not the parameter P is “0” (step 506). 1 ”, set the speed control loop to PI control,
Further, the cut-off frequency fn is set to an initial value fn0 (step 507), and the same processing is repeated again.

If it is determined in step 503 that the absolute value of the difference between the detection frequency fmax and the reference frequency fa is smaller than the threshold ε during the repetition of the above processing, the gain adjustment is completed, and the integration gain k1 and the proportional gain k2 currently stored are determined. And the value of the parameter P are determined to this value, and the adjustment of the speed loop gain is ended. At this time, although not shown in the flowchart, the determined gains K1 and K2, the parameter P, and the current cutoff frequency fn may be displayed on a display device to notify the end of the gain adjustment.

Also, during the repetition of the above processing, when the absolute value of the difference between the detection frequency fmax and the reference frequency fa does not become smaller than the threshold value ε, and the parameter P is determined to be “1” in step 506, the cutoff frequency fn is changed to the speed loop band. (Step 510), the gains k1 and k2 are determined based on the cutoff frequency (step 511), and the adjustment of the speed loop gain is completed.

In this manner, the integral gain k1, the proportional gain k2, and the parameter P are determined, and if speed control is performed based on the determined values, desirable speed control with a large gain without vibration is performed.

If the setting of the reference frequency fa is poor in this gain adjustment processing, the absolute value of the difference between the maximum amplitude frequency fmax and the reference frequency fa becomes smaller than the threshold ε as the gains k1 and k2 of the speed loop increase. Before that, it is conceivable that the mechanical system oscillates. Therefore, Step 502 and Step 5
During step 03, as in the oscillation detection method, step 203 in FIG.
The processing of steps 204 and 209 may be inserted to detect mechanical system oscillation and notify that the set value of the reference frequency fa is bad.

Also, in this speed loop gain adjustment method,
Similar to the servo system oscillation detection method, the processing shown in FIG. 5 may be performed instead of step 106, and the frequency fmax of the maximum amplitude may be obtained by a change in acceleration. Further, the offset value changing processing shown in FIG. (Step 400) to Step 508
And step 509.

In each of the above embodiments, the actual speed of the motor is detected. However, since the position command is zero, the position deviation amount may be detected.

It should be noted that the process of the function generator 2 in FIG. 1 is not performed in the case of the ordinary servo motor position and speed control other than the servo system oscillation detection mode and the gen-in automatic adjustment mode.

According to the present invention, the oscillation of the servo system can be automatically detected, and the speed loop gain is automatically determined.
It does not require time and effort, and can be set to always have an equivalent gain (determined by the reference frequency fa). Also, the optimum speed loop gain can be obtained even if the mechanical system changes over time or inertia changes. Can be easily obtained.

[Brief description of the drawings]

FIG. 1 is a block diagram of a position loop at the time of servo system oscillation detection and automatic adjustment of a speed loop gain in one embodiment of the present invention. FIG. 2 is an explanatory diagram of an input / output relationship of a function generator. FIG. 3 is a flowchart showing processing in a speed loop processing cycle in one embodiment of the servo system oscillation detection system and the speed loop gain automatic adjustment system of the present invention. FIG. 4 is a flowchart showing the oscillation detection system embodiment of the present invention. FIG. 5 is a flowchart of an oscillation detection process, FIG. 5 is a flowchart of a process executed in place of step 106 in FIG. 3, FIG. 6 is a flowchart of a process added when an offset value is changed, and FIG. 8 and 9 are flow charts of the gain adjustment processing in the embodiment of the speed loop gain automatic adjustment method of FIG.
Diagram showing the change in actual speed (velocity waveform) and the differential change in speed (velocity differential waveform) when vibrated at 5 Hz. Fig. 10 shows the maximum torque value and the speed loop required to cause stick slip. And FIG. 11 is a block diagram of a speed loop. 2 ... Function generator, k1 ... Integral gain, k2 ... Proportional gain, P ... Parameter, fa ... Reference frequency, ε ... Threshold, α ... Parameter, flim ... Velocity loop bandwidth upper limit, fmax: Detection frequency, fn: Cutoff frequency.

 ──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-58-107080 (JP, A) JP-A-58-133188 (JP, A) JP-A-60-3716 (JP, A) 82078 (JP, A) JP-A-60-162492 (JP, A) JP-A-60-148392 (JP, A) JP-A-62-110491 (JP, A) JP-A-62-119603 (JP, A) JP-A-63-706 (JP, A) JP-A-63-37408 (JP, A) JP-A-63-129875 (JP, A) JP-A-63-205863 (JP, A) JP-A-63-284602 (JP, A)

Claims (6)

(57) [Claims] 1. A position command is set to zero, and a speed command having a predetermined offset value is given in accordance with the position deviation value to cause vibration in a servo system. The actual speed change or position deviation amount of the servo motor obtained at this time is obtained. And the frequency of the maximum amplitude obtained by performing a frequency analysis of the actual speed change or the change in the position deviation amount suddenly increases the speed loop gain until the frequency of the maximum amplitude fluctuates very greatly with respect to the previous change. A servo system oscillation detection method, wherein the oscillation of the servo system is detected by detecting that the frequency has increased very greatly. 2. The method according to claim 1, wherein the position command is set to zero, and a speed command having a predetermined offset value is given in accordance with the position deviation value to cause a vibration in the servo system. Of the actual speed change or the change in the position deviation amount to obtain a change in the differential value, and the frequency of the maximum amplitude obtained by performing the frequency analysis of the change in the differential value suddenly changes The servo system is characterized in that the oscillation of the servo system is detected by increasing the speed loop gain until the frequency fluctuates significantly, and suddenly detecting that the frequency has increased very greatly. Oscillation detection method. 3. The servo system according to claim 1, wherein the offset value is increased when the frequency of the maximum amplitude obtained by performing the frequency analysis is low, and the offset value is decreased as the frequency increases. Oscillation detection method. 4. A method of causing a servo system to vibrate by setting a position command to zero and giving a speed command having a predetermined offset value in accordance with the position deviation value. And the frequency of the actual speed change or the change in the position deviation amount is analyzed to determine the frequency of the maximum amplitude, and the speed loop gain is automatically adjusted so that the frequency reaches a set predetermined range with respect to the set reference frequency. A speed loop gain automatic adjustment method for a servo system in which the speed loop gain at the time of adjustment and reaching a predetermined range is set as the gain of the speed loop. 5. The servo system is caused to vibrate by giving a position command to zero and giving a speed command having a predetermined offset value in accordance with the position deviation value. Is detected, the change in the differential value is obtained by differentiating the actual speed change or the change in the position deviation amount, the frequency analysis of the change in the differential value is performed, the frequency of the maximum amplitude is obtained, and the frequency is set. The speed loop gain is automatically adjusted in order to reach the set reference range with respect to the set reference frequency, and the speed loop gain of the servo system is used as the speed loop gain when the speed loop gain reaches the set predetermined range. Adjustment method. 6. The servo system according to claim 4, wherein the maximum amplitude obtained by performing the frequency analysis is increased when the frequency is low, and the offset value is decreased as the frequency increases. Speed loop gain automatic adjustment method.