JPH02261083A - Oscillation detection and automatic control of speed loop gain in servo system - Google Patents

Abstract

PURPOSE:To detect the oscillation of a system by setting a position command to zero, giving a velocity command provided with an offset value corresponding to a variation value in a position, and obtaining a point suddenly increasing the maximum amplitude when an actual speed variation at the time of oscillating the servo system is analyzed. CONSTITUTION:A proportional clause 1 amplifies a difference between a position command Pc and an actual position Pr of a servomotor by a gain Kp, and a speed command Vc(in) is outputted. In case of a normal position loop, the Vc(in) is inputted to a speed loop as the speed command, and in case of speed loop gain adjustment, the Vc(in) outputted to the speed loop as a speed command Vc(out) by a function generator 2. In the speed loop, a difference between an actual speed Vt of the servomotor and the speed command Vc(out) is integrated 3 or amplified 4 to generate a torque command. Until the resonance of a mechanical system generates, gains K1 and K2 boost successively, and when the mechanical resonance generates, the amplitude becomes suddenly large. According to the constitution, the oscillation of the servo system is easily detected and, at the same time, the speed loop gain is automatically decided easily.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a method for detecting oscillation of a servo system and adjusting a speed loop gain of a servo system in a machine driven by a servo motor such as a feed axis of a machine tool or a robot. .

Conventional technology FIG. 11 is a block diagram of a speed loop in a servo motor. In the block diagram, 1 is an integral gain, k2 is a proportional gain, and P is a parameter, and the parameter P is set to "1". When parameter P is set to 0, proportional integral (PI) control is performed.
) control is performed, and is a parameter that determines this PI control and IP control. Kt is a torque constant, and Jm is the inertia of the entire system. Further, Vc is a speed command signal, and Vt is a speed signal. The block diagram of the speed loop shown in FIG. 11 is well known in boat fishing, and a detailed explanation thereof will be omitted.

This speed control loop is a secondary system, as is clear from the blower □ diagram in FIG. Then, when the cutoff frequency fn (Hz) and the damping constant ξ are determined, the integral gain kl and the proportional gain 2 are expressed by the following equations (1) and (2).
Determined by the formula.

kl= (Jm/Kt) ・(2πfn) 2 =
lI) k2= (Jm/Kt) ・2ξ・(2w f
n) = 12) This integral gain kl, the proportional gain is 2
If the speed loop gain, which is determined by arise or
It is also vulnerable to external disturbances, and the speed of the servo motor produces low-frequency undulations. On the other hand, if the speed loop gain is too high, vibrations will occur due to resonance of the system at normal speeds, for example, in the case of a machine tool, during cutting feed or rapid feed.

Therefore, conventionally, the integral gain kl of the velocity loop gain,
To determine the proportional gain 2, cutoff frequency f
Determine n and the damping constant ξ by trial and error, and adjust the integral gain kl as much as possible without causing oscillation due to system resonance, etc.
The proportional gain is set so that 2 becomes larger.

Problems that the invention aims to solve As mentioned above, the velocity loop gain (kl.

Adjustment of k2) must be performed for each machine by trial and error to prevent oscillations due to system resonance, etc., and it takes a lot of effort and time to determine the speed loop gain.

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

A second object of the present invention is to provide a speed loop gain automatic adjustment method that can automatically determine the speed loop gain.

Means for Solving the Problem By setting the position command to zero and giving a speed command with a predetermined offset value according to the position deviation value, vibration is caused in the servo system, and the actual speed change or position of the servo motor obtained at this time is A change in the amount of deviation is detected, and a frequency analysis of the actual speed change or change in the positional deviation amount is performed. Alternatively, the actual speed change or the position deviation amount change is further differentiated to obtain a change in the differential value, and a frequency analysis of this differential value change is performed. Among the frequencies of each amplitude obtained in the above frequency analysis, the frequency of the maximum amplitude is determined. Then, the speed loop gain is increased to sequentially obtain the above-mentioned frequencies, and when the frequency suddenly increases significantly compared to the previous fluctuations, it is detected, and the state at that time is detected as oscillation of the servo system. moreover,
When the frequency band of the speed loop is low (velocity loop gain is small), the offset value is increased, and as the frequency band becomes higher, the offset value is decreased to detect oscillation of the servo system.

In addition, when adjusting the speed loop gain, the speed loop gain is automatically adjusted in sequence until the maximum amplitude frequency reaches a predetermined range with respect to the preset reference frequency using the method described above. The speed loop gain at the time when the speed loop gain is reached is determined as the gain of the speed loop, so that the optimum speed loop gain can be automatically determined.

When the action position command is set to zero and a speed command with a predetermined offset value is given according to the position deviation value, a speed command with an offset value is input to the speed loop due to the slight position deviation amount, and the servo motor is affected by friction. Win and move on. In other words, stick step. As a result, the positional deviation amount increases in the opposite direction, so a speed command in the opposite direction is output, and the servo motor now rotates in the opposite direction. This is repeated, causing the servo motor to vibrate. At this time, the actual speed change or position deviation value change of the servo motor is detected, and frequency analysis is performed to detect the frequency at which the amplitude is maximum. Then, as the speed loop gain is gradually increased, the frequency of the maximum amplitude increases much more than the previous increase in frequency due to resonance of the servo system including the machine driven by the servo motor. By detecting this very large rising point, servo system oscillation can be detected.

However, with the above method, the resonant frequency cannot be detected while the resonance is weak, and the system oscillates only when the system fully resonates, the resonant frequency becomes dominant, and the amplitude of the resonant frequency reaches its maximum. Undetectable.

Generally, the resonance frequency of a mechanical system is several times higher than the vibration frequency generated by the servo control system itself.

Assuming that the fundamental frequency of the control system is fO1 and the resonance frequency of the machine is m times mfO, the actual speed change (actual speed waveform) of the servo motor is expressed by the following equation (3).

v(t)=A1s in (2rfOt)+A2s i
n (2πmfOt+ρ) where ρ is a phase delay.

On the other hand, when the above equation (3) is differentiated to obtain the differential signal ω(1), the following equation (4) is obtained.

ω(t) = dv(D, /dt = 2πfOAlcos (2πfOt) + 2πmfOt
A2cos (2πmfQ t+ρ)・・・・・・(4
) Therefore, while the ratio of the amplitude A1 of the fundamental frequency fO and the amplitude A2 of the machine's resonance frequency mfO in the actual speed change determined by the above equation (3) is A2/Al,
The amplitude ratio of the differential signal ω(1) shown in equation (2) is m
A2/A1, which is multiplied by m. Therefore, rather than frequency-analyzing the actual speed change (positional deviation change), it is better to calculate the frequency of the maximum amplitude by frequency-analyzing the value obtained by differentiating this actual speed change (positional deviation change), that is, the acceleration change. Oscillation in the system due to an excessive rise in gain can be detected at an early stage when the influence of the oscillation is weak.

Furthermore, if the above offset value is small, when the frequency band of the speed loop is low (when the speed loop gain is small)
In some cases, the servo motor may vibrate in the mechanical system's balaclava or sosh, or the friction torque of the mechanical system may be larger and no vibration may occur.On the other hand, if the offset value is large, the frequency band of the speed loop may be high. When this occurs, the torque command value becomes saturated due to the torque limit, heat is generated in the motor, or large mechanical vibrations are given to the mechanical system. Therefore, as the frequency band of the speed loop increases, the above offset value should be reduced in proportion. This solves the above problem.

In addition, when adjusting the gain of the speed loop, a method of detecting the oscillation of the servo system is to set a desirable reference frequency that has been empirically determined by the speed loop, position loop, etc. of the servo system. Perform a stick step in the same manner as above, and automatically adjust the speed loop gain by sequentially changing the speed loop gain so that the frequency of the detected maximum amplitude reaches the same as or near the desired reference frequency, and is the same or within a predetermined range. Then, the speed loop gain at that time is automatically determined as the speed loop gain of the speed control loop.

Example An embodiment of the present invention will be described below.

FIG. 1 is a block diagram of a position loop including a velocity loop in this embodiment. The difference from the block diagram of a normal position loop in a servo motor is that a function generator 2 is added.

In Figure 1, 1 is the proportional term in the position loop, and K
p is the position gain, 2 is the function generator, 3 is the integral term of the velocity loop, kl is the integral gain, 4 is the proportional term of the velocity loop, k2 is the proportional gain, 5 is the term of the parameter P,
It is "1" for PI control and "0" for IP control. Further, 6 and 7 are terms of the transfer function of the servo motor, Kt is the torque constant, and Jm is the inertia of the entire system. Further, 8 is an integral term that converts the velocity Vt into the position Pr.

Proportional term 1 amplifies the difference between the position command Pc and the actual position Pr of the servo motor, that is, the position deviation amount, by a gain Kp and outputs it as a speed command V c (in). If it is a normal position loop, This speed command V c (in) is input to the speed loop as a speed command, but when adjusting the speed loop gain of the present invention, it is input to the function generator 2, and the function generator 2 generates the speed command V c (in). in) is converted as described later and output as a speed command Vc (out), which is output to the speed loop.

In the speed loop, if the parameter P is "1", the speed signal is calculated from the difference between the speed command V c (out) and the speed signal (actual speed of the servo motor) vt, that is, the value obtained by integrating the speed deviation amount using the integral term 3. From Vt to speed command value V c (out
) is multiplied by the proportional gain, and that value is output as a torque command to the servo motor. Also, if parameter P is "0", the integral value output from integral term 3 is The value obtained by multiplying the proportional gain by 2 is subtracted from the speed signal Vt, and the resulting value is output as a torque command to drive the servo motor.

As shown in FIG. 2, the function generator 2 has a positional deviation of zero and an offset value of ±A, and has an input speed command V c
As shown in FIG. Input V c (in) of generator 2
is near zero (positional deviation amount is near zero), the input V
The output V c (ou
t) changes greatly, and in other parts, even if the input Vc (in) changes, the amount of change in the output V c (out) is small. A function generator 2 having such an input/output relationship is constructed.

In other words, the input/output relationship of the function generator is expressed by the following equation (5):
This becomes equation (6).

When V c (in)≧0, V c (out) = K * V c (in) + A
-----(5) When Vc (in) < 0, V c (out) = K ・V c (in) −A
-.-= (6) Then, in the position loop during gain adjustment shown in FIG. 1, the servo motor always causes vibration when the position command Pc is zero. That is, position command P
Even when c is "0", there is slight variation in the amount of positional deviation.

Therefore, as shown in Fig. 2, if the positional deviation amount changes from "0" even slightly, the speed command V c (out) output from the function generator 2 to the speed control loop will fluctuate greatly, and the servo motor will rotate. When the servo motor rotates, the positional deviation increases in the opposite direction, and the function generator 2 outputs a speed command V c (out) in the opposite direction.
is output, and the servo motor now rotates in the opposite direction.

This is repeated, causing the servo motor to vibrate.

The frequency of the vibration of the servo motor becomes higher because the higher the band frequency (cutoff frequency fn) of the speed control loop is, the faster the response of the speed control loop is.

Therefore, the vibration of the servo motor, that is, the speed signal Vt
This frequency includes other frequency parts due to friction, etc., so if the main vibration component, that is, the frequency component with the largest amplitude, is extracted by frequency analysis such as Fourier transform, this frequency can be detected. The main vibration component is the speed command V c (o
ut). If resonance does not occur in the mechanical system, the frequency with the maximum amplitude is the speed loop gain Kl.

When K2 is raised sequentially, it increases continuously. However, when resonance occurs in a mechanical system, the frequency with the maximum amplitude suddenly changes to a much larger extent than the previous frequency fluctuation due to the influence of the resonance frequency. Therefore, the point in time when this frequency fluctuates greatly is detected, and this time is detected as oscillation of the servo system.

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

First, the cutoff frequency fn is set to a set value fnO and the damping constant ξ is set to a set value ξ0 as initial parameters in a numerical control device that controls machine tools, robots, etc.
When the parameter P is set to 10'', the CPU of the digital servo circuit calculates the above equations (1) and (2) to set the speed loop gain to 1. , perform the initial settings of 2,
A flag F and an index i, which will be described later, are initially set to "0". Then, when the servo system oscillation detection mode is set, the CPU of the servo circuit executes the process shown in FIG. 3 for each speed loop processing station period.

First, step 100 determines whether the flag F is "1" or not.
), the speed command V c (in) output from the position loop 1 is read as "0" (step 101). In the servo system oscillation detection mode, since the position command Pc is "0", the speed command V c (in) is "0", or since the speed command V c (out) has an offset value A as described later, it is correct. Or it takes a negative value. Therefore, it is determined whether the read speed command Vc (in) is greater than or equal to zero (step 102), and if it is greater than or equal to zero, the velocity command V c (out) to be output to the velocity loop is calculated by equation (5) and output. If the value is negative (step 103), the calculation of equation (6) is performed and output (step 104). That is, the processing of steps 102 to 104 is performed by the function generator 2 in FIG.
This is the process. 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 10).
7). Then, it is determined whether the value of the index i is greater than the set value N (step 108), and if not, the speed command V c (o
ut), normal speed loop processing is performed (step 110), and the processing of the speed loop processing cycle is ended.

In the speed loop processing, a torque command value is calculated, and then, as in the conventional case, current loop processing and the like are performed, a drive current flows to the servo motor, and the servo motor is driven.

Note that the above set value N is F for frequency analysis to be described later.
This defines the number of data required for FT processing.

The CPU repeats the above processing every speed loop processing period. For example, in step 102, Vc(in)≧0
It is determined that V c (out) =
When -V c (in) + A is output to , the servo motor rotates forward due to the offset value A, and as a result, the position deviation becomes negative, and the speed command V c (in) output from position loop 1 ) is negative. As a result, in the next cycle, the process of step 104 is performed, and the speed command V c (out) to the speed loop becomes (K-V c (in)
-A), current will flow to reverse the servo motor, and the servo motor and machine will vibrate. Then, the actual speed Vt detected for each speed loop cycle is stored in the memory (step 1
05°106). Then, 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, steps 100 to 101
, and only normal speed loop processing is performed.

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

First, it is determined whether the flag F is set to "1" (step 200); if it is not set, 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 memory M
When (n+1) pieces of actual speed data are stored in O-Mn, this actual speed data is read out and frequency analysis is performed. In this example, FF of software that is already commercially available is used.
T(First FouyiI!t Tr2nslot
Frequency analysis is performed by m: fast Fourier transform) to obtain the amplitude c (1) of each frequency fH component (step 201). Note that 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 process of FIG. 3 is set so that the number of data MO to Mn is the number necessary for frequency analysis, and the number is a power of two.

Among the amplitudes c (f) obtained in step 201, the frequency fH□ of the maximum amplitude is obtained as fmax (step 2
02L It is determined whether or not the frequency fmax of the maximum amplitude obtained in the previous process is set to "0" in the storage register R(1) (step 203). Since register R(1) is initially set to "0", the process moves to step 205, stores the frequency f m a x obtained in this process in register R(1), and stores the current value. The set value fs is added to the set cutoff frequency fn to obtain a new cutoff frequency fn (step 206), and the calculations of equations (1) and (2) are performed based on this frequency fn,
Find and set 2 for the gain Kl of the new speed loop (
step 207), and sets the index i and flag F to "0".
” and completes the current oscillation detection process.

Since the flag F is set to "0", steps 100 to 110 in FIG. 3 are performed again in the speed loop processing cycle.
The process starts, and the data of the actual speed Vt based on the new speed loop gain is stored in the memories MO to Mn.

Then, when the index i becomes rNJ and data is stored in the memories MO to Mn, and the flag F is set to "1",
The process of steps 200 to 208 in FIG. 4 is started again,
The maximum amplitude frequency f m aX is calculated, and since the previous frequency f max has already been stored in the register R(f), the process proceeds from step 203 to step 204, where the frequency f max calculated this time is m of the previous frequency f max. It is determined whether it is more than twice (m2 to 3) (step 204).

As mentioned above, when the speed loop gains Kl and K2 increase and the mechanical system resonates, due to the influence of this resonance, the maximum amplitude frequency fmax increases greatly and becomes m times the frequency fmax of the previous cycle, but the resonance If it does not occur,
Since the frequency fmax does not increase dramatically, it will not be multiplied by m, and the process moves to step 205, where the speed loop gain is increased and the index i and flag F are set to "0" as described above. 208), the third processing of steps 100 to 110 is started again.

Hereinafter, the above-described process is repeated, and when a frequency fmax that is m times or more higher than the previous frequency fmax is detected in step 204, the gain Kl of the speed loop is displayed as oscillation of the servo system on the display device of the numerical control device. ．． Cutoff frequency fn during oscillation.

The detected frequency fmax is displayed, and the servo system oscillation mode processing ends.

In the servo system oscillation detection method described above, even if the mechanical system causes resonance, as long as the resonance is weak, the influence of the resonance is small, and the maximum amplitude is not detected in correspondence with the resonance frequency. Therefore, as described above, the change in acceleration is obtained by differentiating the change in actual speed, and the frequency is analyzed using this change in acceleration (acceleration waveform) to detect oscillation of the servo system. In this way, as explained in equation (4), system oscillation can be detected even when the resonance is weak.

For example, if a mechanical system with mechanical resonance of 120 Hz (period: approximately 8 msec) is vibrated at 25 Hz (period: 40 msec) by adjusting the speed loop gain, the resulting actual speed change will be (Actual speed waveform)
is as shown in Figure 8. On the other hand, the change in the differential value obtained by differentiating the velocity (velocity differential waveform) is as shown in FIG. 9, and it can be seen that the differential waveform with the resonant frequency of 120H2 is dominant and easier to detect.

At this time, the process shown in FIG. 5 is executed in place of step 106 in FIG. 3 for each speed loop processing cycle of the processor of the digital servo circuit.

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 an acceleration signal a (
i) is determined (step 300). Note that 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.
Steps 301, 302) and steps 107 onward in FIG. 3 are performed. In this way, memory MO~M
The acceleration signal a (i) will be stored in n.

Next, in the oscillation detection process, the process shown in FIG. 4 and the inter-path process are performed. The difference is that in step 208, index i and flag F are set to "0" and register R (
v) is also the only point that is set to "0".

In this way, frequency analysis is performed on acceleration changes (acceleration waveforms), and oscillations of the servo system are detected, which is faster than detecting oscillations based on speed changes (position deviation changes). .

Furthermore, regarding the offset value A, if the offset value A is small, vibration may occur during backlash in the mechanical system, or the friction torque of the mechanical system may be so large that it may not be possible to cause stick slip. . This phenomenon is particularly likely to occur when the speed loop gain is low because the torque command rises slowly. Also, if the offset value A is large, the amplitude of the speed 9 position will be large, and especially when the gain of the speed loop is high, the torque command may become saturated due to the torque limit, or the motor may generate heat or a large mechanical problem may occur in the mechanical system. This is not desirable as it gives vibration.

Looking at the maximum value of torque required to achieve stick slip, if stick slip is caused with the same offset value A and the maximum torque value required at that time is experimentally determined, as shown in Figure 10. It increases approximately in proportion to the frequency band of the velocity loop. This is the cause of saturation of the torque command, generation of motor heat generation, and large mechanical vibration in a high gain state where the frequency band of the speed loop is high.

Therefore, in order to solve the above problem, the offset value A may be increased when the frequency band of the speed loop is low, and the offset value A may be decreased when the frequency band is high.

Specifically, the actual frequency band of the speed loop at a certain point is the frequency fmax of the maximum amplitude that appears at the actual speed of the motor when stick slip is caused by the processing shown in FIGS. 1 and 3. Therefore, the frequency fmax is considered as the speed loop band, and in this example, A (
fmax) -B/fmax, and the offset value A is changed as shown in FIG.

Note that B is a fixed value, and an estimated value is set as the initial value of fmax.

In this case, among the processes shown in FIGS. 3 and 4, the estimated value fmax is set as an initial setting before starting the process shown in FIG. 3, and the offset value A of the initial value is determined. Unlike the system oscillation detection process, all the processes in FIG. 3 are the same. In the oscillation detection process shown in FIG. 4, after step 207, step 400 shown in FIG.
, and each time the maximum amplitude frequency fmax is calculated, the offset value A is changed to B/fmax.
x, and proceed to step 208.

As a result, as the velocity loop gain increases,
The offset value A becomes smaller and the above-mentioned problem is resolved. Note that steps 300 to 302 in FIG. 5 are performed instead of step i-06 in FIG. 3 to change the acceleration (acceleration waveform). The frequency of maximum amplitude from fmax
Of course, even in the case where the offset value A is determined, the offset value A may be updated.

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

The numerical control device is switched to gain automatic adjustment mode, and the cutoff frequency fn is set to fn[l.

Let the damping constant ξ be ξ0. The parameter P is set to "0" and a desirable reference frequency fa, which has been empirically determined by a system such as a velocity loop and a position loop, is set. The CPU of the digital servo circuit calculates equations (1) and (2) from the above initial values, sets the speed loop gain Kl to 2, and sets the index i and flag F to 10.
”. Then, when the start command is input, the CPU starts the process shown in Fig. 3, and writes the actual speed Vt of the motor in the memory every speed loop processing cycle, similar to the oscillation detection method of the servo system. When the data of the actual speed Vt is stored in the memories MO to Mn and the flag F is set to "1" (step 109), the CPU starts the gain adjustment process shown in FIG.

In other words, in step 5 it is necessary to determine whether flag F is "1" or not.
00), if it is not “1”, gain adjustment processing is not performed, and “
1", the actual speed data stored in the memories MO to Mn is analyzed by FFT and the amplitude C(() of each frequency component is determined (step 2).
01). Next, the frequency f of the maximum amplitude among the detected amplitudes is determined as fmax (step 502).

Then, it is determined whether the absolute value of the difference between the obtained frequency fmax and the set reference frequency fa is smaller than the threshold value ε (approximately IHz) set for gain adjustment (step 503). If it is not smaller, set parameter α to the value obtained by subtracting detection frequency fmax from reference frequency fa to the currently set cutoff frequency fn.
(approximately 0.5 to 0.7) is added to obtain a new cutoff frequency (step 504). That is, new fn=(old fn) +a (fa-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), and the new cutoff frequency fn is
is small, this new cutoff frequency fn
1) Calculate equation (2) and set the gain Kl of the new velocity loop to 2. Step 508) Index i,
Set flag F to "0". By setting the flag F to "0", the gain adjustment process proceeds to step 5.
00 is only processed.

On the other hand, in the speed loop processing cycle, step 100 in FIG.
Since the flag F is determined to be "0", the process from step 101 onwards is restarted, and a new speed loop gain Kl,
2, a step slip is performed and the actual speed Vt
data is stored in the memory Mi for each speed loop process. Then, the set number (N+1) of data is stored in the memory MO.
~Mn, and when the flag F is set to "1" (step 109), the gain adjustment process shown in FIG. 7 is started again.

Thereafter, the above-mentioned process is repeated, and the velocity loop gain K
1. 2 is updated sequentially and becomes larger. Then, the cutoff frequency fn is the upper limit value fAim of the velocity loop band.
When the parameter P becomes ``0'' (step 505), the parameter P becomes ``0''.
'' or not (step 506), if rOJ, set the value of parameter P to 11'', set the speed control loop to PI control, and set the cutoff frequency fn to the initial value f.
Set it to nO (step 507) and repeat the same process as described above.

While repeating the above processing, if it is determined in step 503 that the absolute value of the difference between the detected frequency fmax and the reference frequency fa is smaller than the threshold value ε, the gain adjustment is completed and the integral gain kl currently stored.

The proportional gain is set to 2 and the value of the parameter P is determined to this value, and the adjustment of the speed loop gain is completed. Furthermore, at this time,
Although not shown in the flowchart, the determined gain K1. 2. The parameter P and the current cutoff frequency fn may be displayed on the display device to notify the end of the gain adjustment.

Also, while repeating the above process, the absolute value of the difference between the detection frequency fmax and the reference frequency faO does not become smaller than the threshold ε,
If the parameter P is determined to be "1" in step 506, the cutoff frequency fn is set to the upper limit value fl of the velocity loop band.
step 510), and determine the gain kl to 2 according to this cutoff frequency (step 51).
1) Finish adjusting the speed loop gain.

In this way, the integral gain kl, the proportional gain 2, and the parameter P are determined, and when speed control is performed using the determined values, desirable speed control that does not generate vibration and has a large gain is performed.

Note that in this gain adjustment process, if the reference frequency fa is incorrectly set, the speed loop gain kl.

As k2 increases, the frequency of maximum amplitude fmax
A case can be considered in which the mechanical system oscillates before the absolute value of the difference between and the reference frequency fa becomes smaller than the threshold value ε. Therefore,
Similar to the oscillation detection method, steps 203, 204, and 209 in FIG. 4 are inserted between step 502 and step 503 to detect mechanical system oscillation and set the reference frequency fa
It may also be possible to notify that the set value of is incorrect.

Also, in this speed loop gain adjustment method, similarly to the servo system oscillation detection method, the process shown in FIG. Furthermore, the offset value changing process (step 400) shown in FIG. 6 may be inserted between step 508 and step 509.

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

It should be noted that, of course, in the case of normal servo motor position and speed control other than the servo system oscillation detection mode and the automatic gain adjustment mode, the processing of the function generator 2 in FIG. 1 is not performed.

Effects of the Invention In the present invention, the oscillation of the servo system can be automatically detected, and the speed loop gain is also automatically determined, which requires more time and effort than the conventional method of determining it by trial and error. In addition, it can be set to always have an equivalent gain (determined by the reference frequency fa), and the optimum speed loop gain can be easily obtained even if the mechanical system changes over time or inertia changes. I can do it.

[Brief explanation of drawings]

FIG. 1 is a block diagram of the position loop when detecting servo system oscillation and automatically adjusting the speed loop gain in an embodiment of the present invention. FIG. 2 is an explanatory diagram of the input/output relationship of the function generator. FIG. 4 is a flowchart showing the processing in the speed loop processing cycle in an embodiment of the servo system oscillation detection method and speed loop gain automatic adjustment method of the present invention. FIG. Flowchart of detection processing; FIG. 5 is a flowchart of processing executed in place of step 106 in FIG. 3; FIG. 6 is a flowchart of processing added when changing the offset value; FIG. 7 is a flowchart of processing executed in place of step 106 in FIG. 3; Flowcharts of the gain adjustment process in the embodiment of the speed loop gain automatic adjustment method, Figures 8 and 9 show the change in actual speed (velocity waveform) when a mechanical system with 120Hz mechanical resonance is vibrated at 25Hz. ) and a diagram showing differential changes in speed (velocity differential waveform). FIG. 11 is a block diagram of the velocity loop. 2...Function generator, kl...Integral gain, k2...
・Proportional gain, P...parameter, fa...reference frequency, ε...threshold, α...parameter, fI! im
... Upper limit value of the speed loop band, fmax... Detection frequency, fn... Cutoff frequency. Figure 5 Skup 107

Claims (6)

[Claims] (1) By setting the position command to zero and giving a speed command with a predetermined offset value according to the position deviation value, vibration is generated in the servo system, and the resulting change in the actual speed of the servo motor or the change in position deviation amount The speed loop gain is suddenly increased until the frequency of the maximum amplitude obtained by detecting the actual speed change or the change in the position deviation amount suddenly changes significantly compared to the previous fluctuation.
A method for detecting oscillation of a servo system, characterized in that oscillation of the servo system is detected by detecting a very large increase in the frequency. (2) By setting the position command to zero and giving a speed command with a predetermined offset value according to the position deviation value, vibration is generated in the servo system, and the resulting change in the actual speed of the servo motor or the change in position deviation amount Detect the change in the actual speed or position deviation amount to obtain a change in the differential value, perform frequency analysis of the change in the differential value, and suddenly the frequency of the maximum amplitude obtained by Oscillation of the servo system, characterized in that the oscillation of the servo system is detected by increasing the speed loop gain until the frequency fluctuates extremely greatly and detecting a sudden and extremely large increase in the frequency. Detection method. (3) Oscillation detection in a servo system according to claim 1 or 2, wherein the offset value is increased when the frequency of the maximum amplitude obtained by frequency analysis is low, and the offset value is decreased as the frequency increases. method. (4) By setting the position command to zero and giving a speed command with a predetermined offset value according to the position deviation value, vibration is generated in the servo system, and the resulting change in the actual speed of the servo motor or the change in position deviation amount is detected, the frequency of the actual speed change or position deviation amount change is analyzed to find the frequency of the maximum amplitude, and the speed loop gain is automatically adjusted sequentially so that the frequency reaches the set predetermined range with respect to the set reference frequency. , an automatic speed loop gain adjustment method for a servo system in which the speed loop gain at the time when the speed loop gain reaches a predetermined setting range is set as the gain of the speed loop. (5) By setting the position command to zero and giving a speed command with a predetermined offset value according to the position deviation value, vibration is generated in the servo system, and the resulting change in the actual speed of the servo motor or the change in position deviation amount is detected, the change in the differential value is obtained by differentiating the change in the actual speed or the change in the position deviation amount, the frequency of the change in the differential value is analyzed to find the frequency of the maximum amplitude, and the standard for which this frequency is set is determined. A speed loop gain automatic adjustment method for a servo system in which the speed loop gain is automatically adjusted sequentially so that the frequency is within a predetermined range, and the speed loop gain at the time the gain reaches the predetermined range is set as the gain of the speed loop. (6) The speed loop of the servo system according to claim 4 or 5, wherein the offset value is increased when the frequency of the maximum amplitude obtained by frequency analysis is low, and the offset value is decreased as the frequency becomes higher. Automatic gain adjustment method.