Motor control device

Abstract

The motor control device includes: a vibration extraction filter that extracts a plurality of vibrations superimposed in a response of a motor control system from a rotational speed response or a rotational position response of a motor; a sequential frequency estimation section that sequentially estimates a frequency of each of the plurality of vibrations obtained from the vibration extraction filter and outputs an estimation result of the frequency as a frequency estimation value sequence; a vibration damping control section that processes an instruction input to the motor control system to suppress a vibration generated at an end portion of a machine mounted on the motor; and a vibration damping control design section that adjusts the vibration damping control section based on the frequency estimation value sequence.

Description

Technical Field

[0001] This invention relates to electric motor control technology. Background Technology

[0002] In recent years, the field of Factory Automation (FA) has seen a desire to optimize and adjust motor control systems to shorten cycle time and improve production efficiency. In semi-closed-loop motor control systems that control machinery, when the rigidity of the machinery mounted on the motor is low, the ends of the machinery (hereinafter referred to as the mechanical ends) typically vibrate at low frequencies of several Hz to 100 Hz due to the resonant / anti-resonant characteristics of the machinery. This results in positioning time delays and prevents the reduction of cycle time. Vibration damping control is usually required in such cases.

[0003] Vibration reduction control is one of the adjustment elements of an electric motor control system. A technology that automatically adjusts this control in a short time and optimally without manual intervention can help shorten cycle time and improve production efficiency. Patent Document 1 proposes a method for automatically adjusting vibration reduction control.

[0004] Patent document 1 discloses an automatic adjustment technology for vibration reduction control in an electric motor control system that observes vibrations caused by multiple mechanical resonances based on the response of the electric motor and that vibrations occur at the mechanical end.

[0005] Specifically, it includes: an estimation signal processing unit that extracts a single vibration from multiple superimposed vibrations observed based on the motor response; a resonance characteristic estimation processing unit that estimates vibration characteristics (frequency, etc.) based on the extracted single vibration; a feedforward control unit that suppresses vibrations at the mechanical end caused by multiple mechanical resonances; and a vibration reduction control setting unit that adjusts the vibration reduction control unit, i.e., the feedforward control unit, based on the estimated vibration characteristics. The estimation signal processing unit extracts a single vibration in a predetermined order from multiple vibrations observed when the machine is driven in a predetermined drive mode. The resonance characteristic estimation processing unit estimates the vibration characteristics of the extracted single vibration. The vibration reduction control setting unit adjusts the feedforward control unit based on the estimated vibration characteristics. This adjustment is repeated under the predetermined drive mode until each of the multiple vibrations is estimated, thereby forming an automatic adjustment for vibration reduction control.

[0006] Existing technical documents

[0007] Patent documents

[0008] Patent Document 1: WO2012 / 056868 Summary of the Invention

[0009] The technical problem that the invention aims to solve

[0010] In Patent Document 1, multiple vibrations need to be excited according to a specified driving mode. The vibration reduction control unit repeatedly adjusts the vibration reduction control unit, i.e., the feedforward control unit, under the specified driving mode until the vibration characteristics of all multiple vibrations are estimated. Therefore, there is a problem that the adjustment takes time.

[0011] The purpose of this invention is to provide an electric motor control device that, when vibration caused by mechanical resonance is observed based on the response of the electric motor and vibration occurs at the mechanical end, adjusts the vibration damping control unit in a short time / real-time to suppress the vibration at the mechanical end.

[0012] Technical means to solve the problem

[0013] As a preferred example of the present invention, the motor control device includes: a vibration extraction filter that extracts multiple vibrations from the rotational speed response or rotational position response of the motor, which are superimposed on the response of the motor control system due to multiple mechanical resonances; a successive frequency estimation unit that estimates the frequency of each of the multiple vibrations obtained from the vibration extraction filter one by one, and outputs the frequency estimation results as a frequency estimation value sequence; a vibration reduction control unit that processes the instructions input to the motor control system to suppress vibrations generated at the ends of machinery mounted on the motor; and a vibration reduction control design unit that adjusts the vibration reduction control unit based on the frequency estimation value sequence.

[0014] Invention Effects

[0015] According to the present invention, the vibration damping control unit can be adjusted in a short time / real-time to automatically suppress the vibration of the mechanical end. Attached Figure Description

[0016] Figure 1 This is a diagram illustrating Embodiment 1 of an FB control system applied to an electric motor.

[0017] Figure 2 This is a graph representing the frequency characteristics of a notch filter.

[0018] Figure 3 This is a diagram representing the successive frequency estimation part.

[0019] Figure 4 This is a graph representing the calculation of the frequency estimation sequence in the successive frequency estimation section.

[0020] Figure 5 This is a diagram showing the processing by the vibration reduction control design department.

[0021] Figure 6 This is a diagram illustrating the application of Embodiment 1 in a motor position FB control system.

[0022] Figure 7 This is a diagram illustrating the application of Embodiment 1 in a motor speed FB control system. Detailed Implementation

[0023] The embodiments will now be described with reference to the accompanying drawings. In the drawings, components with common functions are labeled with the same reference numerals and their descriptions are omitted. Furthermore, "feedback" will sometimes be abbreviated as "FB", "notch filter" as "NF", "low-pass filter" as "LPF", "high-pass filter" as "HPF", and "band-pass filter" as "BPF".

[0024] Example 1

[0025] Figure 1 This diagram illustrates the structure when the automatic adjustment unit 2 of Embodiment 1 is applied to a typical semi-closed-loop electric motor FB control system. In a typical electric motor FB control system that does not include the automatic adjustment unit 2, the operating quantity of the FB controller 13 is provided to the electric motor 14 to control the controlled object machine 15 in such a way that the deviation between the command r(t) and the output y(t) of the electric motor 14 is zero.

[0026] Specifically, the output y(t) is the rotational position of the motor. The motor's rotational position is measured using a sensor (e.g., an encoder), and the deviation from the rotational position command r(t) is calculated using adder / subtractor 16. The FB controller 13 processes this deviation as the position error. Additionally, a device for driving the motor 14 (an inverter, etc.) and a controller for controlling the speed and current of the motor 14 are provided upstream of the motor 14. Figure 1 These have been omitted.

[0027] Figure 1 In the FB control system, when the rigidity of the mechanical parts mounted on the rotating shaft of the motor is low, the ends of the mechanical parts (hereinafter referred to as the mechanical ends) will typically vibrate at low frequencies of several Hz to 100 Hz due to the resonance / anti-resonance characteristics of the mechanical parts, resulting in positioning time delay.

[0028] In this case, a vibration damping control unit 5 can be used to suppress vibration at the mechanical end and shorten the positioning time. The vibration damping control unit 5 processes the rotational position command r(t) to be input into the motor control system in order to suppress vibration at the mechanical end mounted on the motor.

[0029] In this embodiment, the vibration reduction control unit 5 is a notch filter (hereinafter referred to as the actual notch filter). Figure 2This represents the frequency characteristics of the actual notch filter. By matching the notch frequency of the actual notch filter with the frequency of the vibration at the mechanical end, the frequency components that can excite the vibration at the mechanical end can be removed from the rotational position command r(t). The vibration reduction control unit 5 can generate a rotational position command rs(t) that will not excite the vibration at the mechanical end by processing the rotational position command r(t).

[0030] In the mechanical response, due to the multiple resonance / anti-resonance characteristics of the machine, multiple vibrations with different frequencies may superimpose. Moreover, for each vibration, the amplitude and decay time are usually different due to the specific resonance / anti-resonance characteristics.

[0031] When multiple vibrations superimposed occur in the mechanical response, applying a practical notch filter to each vibration can suppress them. When N vibrations with frequencies f1 to fN are superimposed on the mechanical end, N stages of practical notch filters with notch frequencies fk (k = 1…N) can be cascaded. However, using multiple stages of practical notch filters increases the delay in the rotational position command r(t), resulting in situations where the positioning time cannot be shortened.

[0032] In this case, to shorten the positioning time, the vibration control unit 5 may not use an actual notch filter but instead employ other vibration control methods based on the controlled object model. Furthermore, when using other vibration control methods, at least the frequency of vibration at the mechanical end must be known.

[0033] The automatic adjustment unit 2 can automatically adjust the vibration reduction control unit 5 without prior information about the controlled object machinery 15, including the successive frequency estimation unit 3, the vibration reduction control design unit 4, the vibration reduction control unit 5, and the vibration extraction filter 6.

[0034] In addition, in this embodiment, the automatic adjustment unit 2 aims to suppress the vibration of the mechanical end at low frequencies of several Hz to 100 Hz, and the vibration of the mechanical end at frequencies above 100 Hz is not the target of suppression.

[0035] Vibration extraction filter 6 extracts multiple vibrations superimposed on the response of the motor control system due to multiple mechanical resonances from the rotational speed response or rotational position response of the motor. Figure 1 In this process, its role is to remove the steady-state response component of the rotational position response y(t) and extract only the vibration component related to the vibration of the mechanical end, and output it as yd(t).

[0036] In this embodiment, the vibration extraction filter 6 is a bandpass filter that allows only signals in the specified frequency band fL to fH to pass through. Alternatively, a combination of a high-pass filter and a low-pass filter can also be used.

[0037] In the case of multiple vibrations superimposed in the mechanical end response, multiple vibrations are also superimposed in the rotational position response y(t). To extract multiple vibrations, the vibration extraction filter 6 sets the frequency band fL~fH to be sufficiently wide. As a result, multiple vibrations are observed as yd(t).

[0038] The successive frequency estimation unit 3 estimates the frequencies of the multiple vibrations yd(t) obtained from the vibration extraction filter 6 one by one, and outputs the frequency estimation results as a frequency estimation value sequence f(k) composed of 1-dimensional sequence data.

[0039] For ease of explanation in this embodiment, it is assumed that yd(t) is observed as a superposition of two vibrations: yd1(t) with frequency f1, maximum amplitude M1, and duration T1, and yd2(t) with frequency f2, maximum amplitude M2, and duration T2 (where f1...). <f2≤100[Hz])。

[0040] Successive frequency estimation unit 3, as shown Figure 3 As shown, it includes a notch filter 31, an adaptive notch filter 34, a vibration detection unit 32, and a convergence determination unit 33. The notch filter 31 is a vibration removal unit that removes only a specified frequency component. The adaptive notch filter 34 is an adaptive estimation unit that estimates the frequency of vibration successively based on the vibration waveform and outputs the estimated value as a frequency estimation value signal successively. The convergence determination unit 33 determines whether the frequency estimation value signal has converged and outputs the frequency estimation value of the frequency estimation value signal when it is determined to have converged as a frequency estimation value sequence.

[0041] The notch filter 31 has the following characteristics: Figure 2 As shown in the frequency characteristics, the notch frequency Nf[Hz] of the notch filter 31 can be changed to remove frequency components based on the frequency estimation value sequence f(k)[Hz] output by the convergence judgment unit 33.

[0042] Notch filter 31 removes only the notch frequency Nf [Hz] from multiple vibrations yd(t), and the signal after removing the notch frequency Nf from the multiple vibrations yd(t) is taken as the residual vibration and output as yr(t).

[0043] In the initial state when the automatic adjustment unit 2 starts adjustment, the notch frequency Nf [Hz] of the notch filter 31 is set to a value that is large enough compared to the upper limit of 100 [Hz] of the vibration reduction frequency band of the mechanical end undertaken by the vibration reduction control unit, so that the vibrations of frequencies f1 and f2 will not be removed by the notch filter 31, but will be output to the residual vibration yr(t).

[0044] After the automatic adjustment unit 2 starts adjustment, whenever k in the frequency estimation sequence f(k) (k = 1...) output by the convergence judgment unit 33 is updated, the notch frequency Nf [Hz] of the notch filter 31 is updated to Nf = f(k). After the automatic adjustment unit 2 completes adjustment, the notch frequency Nf [Hz] is set to a value that is sufficiently large compared to 100 [Hz], and the system returns to its initial state.

[0045] If the vibration detection unit 32 observes a vibration that is above a specified amplitude and lasts for a specified time in the residual vibration yr(t), it determines that vibration has been detected; otherwise, it determines that vibration has not been detected.

[0046] The adaptive notch filter 34 and the convergence determination unit 33 operate only when the vibration detection unit 32 determines that vibration has been detected.

[0047] If the amplitude of one of the two vibrations of yd(t) is 0 (e.g., M2 = 0) and the notch frequency Nf[Hz] of the notch filter 31 is f1, ideally no vibration will be observed in the remaining vibration yr(t), and the vibration detection unit 32 will determine that no vibration has been detected. In this case, the adaptive notch filter 34 and the convergence determination unit 33 will not start working.

[0048] The adaptive notch filter 34 successively estimates the frequency of the principal vibration of the residual vibration yr(t). The adaptive notch filter 34, which has a notch filter and an adaptive algorithm, uses the following (Equation 1) and (Equation 2) with excellent computational cost.

[0049] □ Notch filter

[0050]

[0051] □ Adaptive Algorithm

[0052] a(t+1)=a(t)-μ(a(t))·L(e(t))·sgn(x(t-1)) (Formula 2)

[0053] Where a(t) represents the notch frequency (but the unit is not [Hz]), rL represents the notch width, μ represents the update gain used to adjust the update amount, L(Y) represents the constraint processing of Y, sgn represents the sign function, x(t) represents the notch filter state, and e(t) represents the notch filter output.

[0054] The adaptive notch filter 34 includes the notch filter shown in Equation 1, and the notch frequency a(t) of the notch filter in Equation 1 is adaptively adjusted by the adaptive algorithm in Equation 2.

[0055] Specifically, the adaptive algorithm in (Equation 2) works repeatedly so that the notch filter in (Equation 1) can remove the main vibration of the residual vibration yr(t) of the input. When the notch filter in (Equation 1) completes the removal of the main vibration of the residual vibration yr(t), a(t) is the frequency of the main vibration.

[0056] Therefore, the adaptive notch filter 34 can be called a successive frequency estimation unit for the residual vibration yr(t), and a(t) is the frequency estimation signal for the residual vibration yr(t).

[0057] Furthermore, a(t) exhibits nonlinear characteristics with respect to frequency [Hz], especially in the low-frequency range below 100 [Hz] where it displays strong nonlinearity. Specifically, there is a trend where the absolute value of the change in a(t) per unit frequency change [Hz] is smaller at lower frequencies. Therefore, if the update gain μ is fixed when estimating in the low-frequency range, the update gain may become too large, and the adaptive algorithm in (Equation 2) may oscillate and fail to converge.

[0058] Therefore, in this embodiment, the update gain is set to μ(a(t)) to address this problem, so that it changes accordingly with the frequency estimation signal. As an example, μ(a(t)) is represented by the following (Equation 3).

[0059]

[0060] Where a1 is the design frequency and μa is the basic update gain.

[0061] Let the appropriate update gain at the design frequency a1 be μa. According to μ(a(t)) in (Equation 3), when the frequency a(t) < a1, the update gain can be considered to be of the same magnitude as μa. Therefore, when the frequency a(t) < a1, the problem of the adaptive algorithm in (Equation 2) vibrating and failing to converge can be solved.

[0062] Furthermore, to reduce computational costs, the data for μ and a(t) in Equation 3 can be stored in tabular form, similar to MAP. Alternatively, μ(a(t)) can be represented by other functions that have similar effects to Equation 3.

[0063] If the adaptive notch filter 34 converges a(t), it is considered that the notch filter of (Equation 1) has completed the removal of the main vibration of the residual vibration yr(t), and it can be judged that the frequency estimation of the main vibration of the residual vibration yr(t) has been completed.

[0064] The convergence determination unit 33 is a unit used to determine whether the frequency estimation signal a(t) has converged. The convergence determination unit 33 receives the frequency estimation signal a(t) and determines whether a(t) has converged. However, as mentioned above, the frequency estimation signal a(t) has a nonlinear characteristic with respect to frequency [Hz], so it is not easy to determine whether it has converged based on a(t).

[0065] The convergence determination unit 33 has a transformation unit that transforms a(t) into frequency units [Hz], thereby transforming a(t) into f(t) [Hz]. Therefore, the convergence determination unit 33 determines whether a(t) has converged based on f(t) [Hz].

[0066] Specifically, if the variation of f(t) within a specified time is below a specified value [Hz], it is determined that a(t) has converged. The f(t) at this point is then kept as the k-th (k=1…) value, and convergence determination continues sequentially. Each time convergence is determined, the k count is incremented and kept at f(k). As a result, the convergence determination unit 33 outputs f(k) (k=1…) as a sequence of frequency estimates.

[0067] During the predetermined time from the start of adjustment by the automatic adjustment unit 2 to the end, the successive frequency estimation unit 3 is executed, and the frequency estimation value sequence f(k) (k=1……) for the predetermined time period can be obtained from the successive frequency estimation unit 3.

[0068] exist Figure 4 The text indicates the calculation status of the frequency estimation sequence f(k) in the successive frequency estimation unit 3. Figure 4 In the diagram, the frequencies of each vibration yd1(t) and yd2(t) are f1 = 10 [Hz] and f2 = 20 [Hz]. For the vibration yd(t) obtained by superimposing the two vibrations, during the adjustment time of the automatic adjustment unit 2, the dashed line represents the frequency estimation unit 3 when it operates, and the solid line represents the frequency estimation value sequence f(k) obtained by making a convergence judgment on f(t).

[0069] In addition, Figure 4 In the diagram, the output yr(t) of the notch filter 31 is yr(t) = yd1r(t) + yd2r(t), and yd1r(t) and yd2r(t) are shown respectively.

[0070] At the convergence judgment points k=1 and 2, the adaptive notch filter 34 estimates the frequency f1 of the main vibration component of vibration yd(t), i.e., vibration yd1(t). f(1) and f(2) are applied to the notch filter 31. Therefore, at k=1 and 2, the amplitude of yd1r(t) decreases, and the main vibration component of vibration yd(t) becomes yd2(t).

[0071] Therefore, at k=3 and 4, the adaptive notch filter 34 estimates the frequency f2 of the vibration yd2(t), and f(3) and f(4) are applied to the notch filter 31, and the main vibration component of the vibration yd(t) becomes yd1(t).

[0072] The adaptive notch filter 34 estimates the frequency of yd1(t) to obtain f(5). After k=5, the vibration yd(t) completely decays and becomes a state where vibration does not occur. Therefore, the vibration detection unit 32 judges that no vibration is detected, and stops the operation of the adaptive notch filter 34 and the convergence judgment unit 33.

[0073] Thus, the successive frequency estimation unit 3 of this embodiment is characterized in that, even when more than one decaying vibration is superimposed and observed as yd(t), it can preferentially and alternately estimate the frequency of the main vibration at high speed.

[0074] When the amplitude, frequency, decay (duration) time, and number of superpositions of vibrations are unknown, conventional estimation methods, such as recursive least squares estimation, may not be able to accurately estimate the frequency of each vibration. This is because, for example, when two vibrations yd1 and yd2 are superimposed to obtain yd(t) and the decay time is T1 < T2, it is a two-vibration system in the range of 0 to T1 but a single-vibration system in the range of T1 to T2. This becomes a time-varying estimation problem where the frequency of the vibration system varies with the unknown decay (duration) times T1 and T2.

[0075] In contrast, the successive frequency estimation unit 3 has the advantage of being able to accurately and quickly estimate the vibration frequency even when the decay (duration) time and the number of vibration superpositions are unknown.

[0076] The vibration reduction control design unit 4 adjusts the vibration reduction control unit 5 based on the frequency estimation value sequence. In this embodiment, the vibration reduction control design unit 4 performs unsupervised clustering.

[0077] Specifically, the frequency estimation sequence f(k) (k=1...) obtained during the specified adjustment time of the automatic adjustment unit 2 is regarded as point cluster data, and unsupervised clustering such as k-means is performed on it. The number of clusters obtained is taken as the superposition number N of vibrations, and the average value of each cluster is taken as the frequency estimation value fk (k=1...N) of each vibration. In addition, if the centroid distance between each cluster is less than a specified value, these clusters are merged.

[0078] The vibration control design unit 4 sends the output F, which includes the superposition number N of vibrations and the estimated frequency fk (k = 1...N) of each vibration, to the vibration control unit 5. The vibration control unit 5 sets the actual notch filter with notch frequency fk as N stages in series, thereby completing the processing of the automatic adjustment unit 2.

[0079] Figure 5Four pairs of vibration reduction control design sections are shown. Figure 4 The processing of f(k) is shown. Cluster 1 consists of f(1), f(2) and f(5), and cluster 2 consists of f(3) and f(4). Since there are 2 clusters, the superposition number N of vibrations is 2. The frequency of each vibration is the average value of each cluster (f(1)+f(2)+f(5)) / 3 and (f(3)+f(4)) / 2.

[0080] Furthermore, regarding f(1) after the automatic adjustment has just begun, since the notch filter 31 is in an inactive state, the superposition of vibrations is more significant, so there is a tendency for the frequency estimation value to be less accurate. Therefore, it is also possible to add a process of removing the first half of the frequency estimation value sequence f(k) (k=1……) before clustering.

[0081] In addition to the unsupervised clustering described above, the vibration reduction control design department 4 can also consider using the following algorithm when the number of vibration superpositions N is 1 to 2. The following algorithm is superior to the k-means method in terms of processing cost.

[0082] S1: Calculate the average value Ma of the point cluster data, and generate two clusters C1 and C2 based on whether it is larger or smaller than Ma.

[0083] S2: Calculate the average value Mc1 of C1 and the average value Mc2 of C2, and calculate Te = |Mc1 - Mc2|.

[0084] S3: When Te is above the specified value, the number of superpositions of vibrations is N = 2, and the estimated frequencies of each vibration are f1 = Mc1 and f2 = Mc2. When Te is below the specified value, the number of superpositions of vibrations is N = 1, and the estimated frequency of the vibration is f1 = Ma.

[0085] Regarding the specified adjustment time of the automatic adjustment unit 2, if vibration can be observed in yd(t) and the frequency estimation value sequence f(k) can be obtained when the controlled object machine 15 operates once, then it can be the time consumed by the controlled object machine 15 operating only once.

[0086] In addition, if it is necessary to obtain more frequency estimation value sequences f(k) (k=1……) to improve the estimation accuracy of vibration frequency, the time required for the controlled mechanical 15 to perform multiple actions can be set as the specified adjustment time of the automatic adjustment unit 2.

[0087] In this embodiment, when vibrations caused by one or more resonance / anti-resonance characteristics of the controlled object are observed based on the response of the motor control system, and vibrations occur at the mechanical end, it is not necessary to know the number of the above-mentioned mechanical resonances in advance.

[0088] In Patent Document 1, multiple resonance characteristics of the vibration damping control unit, i.e., the feedforward control unit, are repeatedly adjusted from the vibration damping control setting unit. According to this embodiment, in the initial state of starting the adjustment of the vibration damping control unit, the vibration damping control unit 5 does not process the commands input to the motor control system. After the vibration damping control design unit 4 completes the adjustment of the vibration damping control unit 5 based on the frequency estimation value sequence, the vibration damping control unit 5 begins to process the commands input to the motor control system. This embodiment provides an automatic adjustment method for motor control capable of adjusting the vibration damping control unit 5 in a short time / real-time to suppress vibrations at the mechanical end, and a motor control device equipped with an automatic adjustment unit 2.

[0089] In addition, in this embodiment, the automatic adjustment unit 2 uses the motor rotation position y(t) as input. However, from the viewpoint of easily extracting vibration components, the input of the automatic adjustment unit 2 can also be the motor rotation position deviation, the output of the FB controller 13, the deviation of the motor rotation speed, the output of the speed FB controller, or the response of the motor rotation speed, etc.

[0090] Alternatively, the width of the actual notch filter can be adjusted based on the variance of the cluster generated by the vibration reduction control design unit 4. This is because it is difficult to expect high accuracy in frequency estimation when the variance of the cluster is large, so it is sometimes effective to set the width of the actual notch filter to be larger in order to remove frequency components more effectively.

[0091] Furthermore, the notch filter 31 can employ a multi-stage structure arranged in series, with the frequency estimation sequence f(k) appropriately applied to each notch frequency. Thus, even when a superposition of numerous vibrations is observed in yd(t), the frequency estimation accuracy can still be expected. However, in this case, it is necessary to separately manage which notch filter f(k) is applied to; this is an important point to note.

[0092] In addition, the automatic adjustment unit 2 can also have the function of adjusting the gain of the FB controller 13. This is because the vibration damping control unit 5 is adjusted by the automatic adjustment unit 2. When it is effective, the vibration of the mechanical end is suppressed. Even if the FB control gain is higher, the mechanical end can be controlled and positioned without vibration.

[0093] Figure 1The motor control device, including the automatic adjustment unit 2, the adder / subtractor 16, and the FB controller 13, has a CPU (Central Processing Unit) which is omitted from the illustration. For each processing unit, including the vibration extraction filter 6, the successive frequency estimation unit 3, the vibration reduction control design unit 4, and the vibration reduction control unit 5, the automatic adjustment unit 2, the adder / subtractor 16, and the FB controller 13, the CPU reads and executes the program to perform the processing of each processing unit. Each processing unit can also be constructed entirely or partially using hardware such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array).

[0094] According to this embodiment, in the semi-closed-loop structure of the electric motor FB control system, when vibration occurs at the mechanical end, the vibration reduction control unit can be adjusted in a short time / real-time to suppress the vibration at the mechanical end without needing to know the amount of mechanical resonance in advance.

[0095] Example 2

[0096] Example 2 is an example of applying Example 1 to a motor control device. Descriptions of items identical to those in Example 1 are omitted. Figure 6 As shown, this is to Figure 1 The automatic adjustment unit 2 shown is an example of an automatic regulator 622 applied to a semi-closed-loop position FB control system of an AC servo motor. Figure 6 The control system, excluding the automatic regulator 622, is a commonly known semi-closed-loop position FB control system.

[0097] The semi-closed-loop position FB control system of the AC servo motor, such as Figure 6 As shown, it includes an adder / subtractor 612, an adder / subtractor 614, a position controller 61, a speed controller 62, a current controller 63, a first coordinate transformer 64 that performs coordinate transformation from the dq coordinate system to the three-phase coordinate system, a second coordinate transformer 610 that performs coordinate transformation from the three-phase coordinate system to the dq coordinate system, a PWM output device 65 that inputs three-phase voltage commands and outputs PWM pulses, an inverter (power converter) 66 with switching elements, a current detector 68, a position / speed calculation unit 611, an automatic regulator 622, an encoder 69 that measures the motor speed, a motor 67, and a machine 613 driven by the motor.

[0098] The position / speed calculation unit 611 calculates the rotational position / speed of the motor based on the output of the encoder 69, and the calculated motor rotational position or motor rotational speed 616 is input to the automatic regulator 622.

[0099] In order to avoid exciting vibration at the end (mechanical end) of the machine 613, the vibration damping control unit 5 of the automatic regulator 622 processes the rotation position command 615 and outputs the processed rotation position command 617. The automatic regulator 622 plays the role of automatically adjusting the vibration damping control unit 5 appropriately based on the rotation position 616 of the motor.

[0100] The machining rotational position command 617 output from the automatic regulator 622 and the motor rotational position 616 from the position / speed calculation unit 611 are input to the adder / subtractor 614. The difference between the machining rotational position command 617 and the motor rotational position 616 is input to the position controller 61 as a position command.

[0101] The position controller 61 outputs a speed command 618, and the adder / subtractor 612 outputs the difference between the speed command 618 and the motor rotation speed 619 to the speed controller 62.

[0102] The position controller 61 uses the post-machining rotational position command 617 as the command to implement position FB control with the speed / current control system of the minor loop system and the integral element as the controlled objects.

[0103] Assuming the inertia number of mechanical 613 is 1, and considering mechanical 613 and the rotor of the electric motor as elastically coupled, the controlled object mechanical can be regarded as a dual-inertial system of mechanical 613 and the rotor of the electric motor coupled through springs and shock absorbers. The controlled object mechanical has frequency characteristics including one set of resonance / anti-resonance characteristics.

[0104] Furthermore, when the inertia number of mechanism 613 is 2, and it is considered that both sides are elastically coupled with one side being elastically coupled to the rotor of the motor, or when it is considered that the two inertia of mechanism 613 are elastically coupled to the rotor of the motor respectively, the controlled object mechanism can be regarded as a three-inertia system with frequency characteristics including two sets of resonance / anti-resonance characteristics. An example of such a system is, for instance, a dual-inertia servo motor mechanism with movable inertial bodies mounted on a low-stiffness platform.

[0105] As shown in Embodiment 1, the automatic regulator 622 does not require prior investigation of the resonance / anti-resonance characteristics of the machine. Even when multiple vibrations are observed superimposed in the response of the motor control system due to vibration at the mechanical end, it can still appropriately and automatically adjust the vibration damping control unit 5 in a short time / real-time to suppress the vibration at the mechanical end. Therefore, in this embodiment, the automatic regulator 622 can also appropriately and automatically adjust the vibration damping control unit 5 in a short time / real-time to suppress the vibration at the mechanical end, just like in Embodiment 1.

[0106] According to this embodiment, in such Figure 6In the semi-closed-loop position FB control system of the AC servo motor shown, when vibration is observed at the mechanical end due to one or more resonance / anti-resonance characteristics of the controlled object's machinery based on the response of the motor control system, the number of such mechanical resonances does not need to be specifically known in advance by setting the automatic adjuster 622. Furthermore, according to this embodiment, the vibration damping control unit 5 can be appropriately and automatically adjusted in a short time / real-time to suppress the vibration at the mechanical end.

[0107] In addition, the automatic regulator 622, such as Figure 7 As shown, it can also be applied to the semi-closed-loop speed FB control system of AC servo motors. Figure 7 Removed from Figure 6 The position controller 61. This is because, in the speed FB control system, vibrations caused by more than one resonance / anti-resonance characteristic of the controlled object's machinery can also be observed according to the speed response 716, which may result in vibrations at the mechanical end.

[0108] Figure 7 In order to avoid exciting vibration at the end (mechanical end) of the machine 613, the vibration damping control unit 5 of the automatic regulator 622 processes the speed command 715 and outputs the processed speed command 717. The adder / subtractor 612 outputs the difference between the processed speed command 717 and the motor rotation speed 716 from the position / speed calculation unit 611 to the speed controller 62.

[0109] The automatic adjuster 622, using the speed response 716, can also automatically adjust the vibration damping control unit 5 as shown in Embodiment 1. Therefore, even if... Figure 7 The semi-closed-loop speed FB control system of the AC servo motor shown can achieve the above-mentioned mechanical resonance quantity without special prior knowledge by setting the automatic regulator 622.

[0110] Furthermore, in Figure 7 In the semi-closed-loop speed FB control system of the AC servo motor, the vibration damping control unit 5 can also be automatically adjusted appropriately in a short time / real time to suppress the vibration of the mechanical end.

[0111] According to this embodiment, in the semi-closed-loop position FB control system or the semi-closed-loop speed FB control system of the AC servo motor, when vibration occurs at the mechanical end, the vibration reduction control unit can be adjusted in a short time / real-time to suppress the vibration at the mechanical end without needing to know the amount of mechanical resonance in advance.

[0112] In addition to electric motor control devices, the above embodiments can also be applied to semiconductor testing devices, main electric motor control devices for electric vehicles, electric power steering devices, etc.

[0113] Explanation of reference numerals in the attached figures

[0114] 2……Automatic adjustment unit, 3……Successive frequency estimation unit, 4……Vibration reduction control design unit, 5……Vibration reduction control unit, 6……Vibration extraction filter, 13……FB controller, 14……Motor, 15……Controlled object machinery, 31……Notch filter, 32……Vibration detection unit, 33……Convergence judgment unit, 34……Adaptive notch filter.

Claims

1. A motor control device, characterized in that, include: A vibration extraction filter extracts multiple vibrations from the rotational speed response or rotational position response of an electric motor, which are superimposed on the response of the electric motor control system due to multiple mechanical resonances. The successive frequency estimation unit estimates the frequency of each of the plurality of vibrations obtained from the vibration extraction filter one by one, and outputs the frequency estimation results as a sequence of frequency estimation values. The vibration damping control unit processes the instructions input to the motor control system to suppress vibrations generated at the ends of the machinery mounted on the motor. and The vibration reduction control design unit adjusts the vibration reduction control unit based on the frequency estimation value sequence, wherein... The vibration reduction control design department, The frequency estimation sequence over a specified time period is processed as point group data. The point cluster data is divided into multiple clusters using unsupervised clustering. The number of clusters obtained from the division is used as the vibration superposition number of the multiple vibrations. The vibration frequencies of the multiple vibrations are estimated based on the point group data belonging to each cluster, and these are used as vibration frequency estimates. Based on the superposition number and the estimated vibration frequency, the vibration reduction control unit is adjusted. Specifically, the point group data is divided into two clusters based on the average value of the point group data, and the average value of each cluster is calculated. If the absolute value of the difference between the calculated average values ​​is above a predetermined value, the vibration superposition number is set to 2; otherwise, it is set to 1. When the vibration superposition number is 2, the average value of each cluster is used as the vibration frequency estimate; when the vibration superposition number is 1, the average value of the point group data is used as the vibration frequency estimate.

2. The motor control device as described in claim 1, characterized in that: The successive frequency estimation unit includes: The vibration removal unit removes only components of a specified frequency. The adaptive estimation unit estimates the frequency of vibration successively based on the vibration waveform. Convergence decision part; and Vibration detection department, among which, The vibration removal unit processes the plurality of vibrations and outputs the signal after removing a predetermined frequency component from the plurality of vibrations as the remaining vibration signal. The adaptive estimation unit successively estimates the frequency of the dominant vibration of the remaining vibration signal, and outputs the estimated value as a frequency estimation signal successively. The convergence determination unit determines whether the frequency estimation signal has converged, and outputs the frequency estimation value of the frequency estimation signal when convergence is determined as the frequency estimation value sequence. The vibration removal unit changes the specified frequency components to be removed based on the frequency estimation value sequence. The vibration detection unit determines whether continuous vibration occurs in the remaining vibration signal. When vibration is detected, the adaptive estimation unit and the convergence determination unit are activated.

3. The motor control device as described in claim 2, characterized in that: The vibration removal section is a notch filter. The adaptive estimation unit is an adaptive notch filter. The convergence determination unit successively receives the frequency of the main vibration of the plurality of vibrations estimated successively by the adaptive notch filter as the frequency estimation signal, and outputs the frequency estimation sequence. The notch filter changes its notch frequency based on the output frequency estimation sequence.

4. The motor control device as described in claim 3, characterized in that: In the initial state of starting the adjustment of the vibration reduction control unit, The notch frequency of the notch filter is set to be above the upper limit of the vibration reduction frequency band of the end of the machine, which is managed by the vibration reduction control unit.

5. The motor control device as described in claim 1, characterized in that: In the initial state of starting the adjustment of the vibration reduction control unit, The vibration reduction control unit does not process the commands input to the motor control system. After the vibration reduction control design unit completes the adjustment of the vibration reduction control unit based on the frequency estimation value sequence, the vibration reduction control unit begins to process the instructions input to the motor control system.

6. The motor control device as described in claim 3, characterized in that: The adaptive notch filter adjusts the update gain in each successive frequency update based on the current frequency estimate.

7. An automatic adjustment method for motor control, characterized in that: Extract multiple vibrations from the motor's rotational speed or rotational position response, which are superimposed on the response of the motor control system due to multiple mechanical resonances. The frequencies of the extracted vibrations are estimated one by one, and the frequency estimation results are output as a sequence of frequency estimates. Adjustments are made based on the frequency estimation sequence, thereby processing the instructions input to the motor control system to suppress vibrations generated at the mechanical ends mounted on the motor. Specifically, the frequency estimation sequence over a specified time period is processed as point group data. The point cluster data is divided into multiple clusters using unsupervised clustering. The number of clusters obtained from the division is used as the vibration superposition number of the multiple vibrations. The vibration frequencies of the multiple vibrations are estimated based on the point group data belonging to each cluster, and these are used as vibration frequency estimates. Based on the superposition number and the estimated vibration frequency, the processing of the command is adjusted. Specifically, the point group data is divided into two clusters based on the average value of the point group data, and the average value of each cluster is calculated. If the absolute value of the difference between the calculated average values ​​is above a predetermined value, the vibration superposition number is set to 2; otherwise, it is set to 1. When the vibration superposition number is 2, the average value of each cluster is used as the vibration frequency estimate; when the vibration superposition number is 1, the average value of the point group data is used as the vibration frequency estimate.

8. The automatic adjustment method for motor control as described in claim 7, characterized in that: In the initial state, the commands input to the motor control system are not processed. After the adjustment of the command is completed based on the frequency estimation value sequence, the command input to the motor control system is processed.

9. The automatic adjustment method for motor control as described in claim 7, characterized in that: The successive estimation of the frequencies of the multiple vibrations is performed through a combination of a notch filter, an adaptive notch filter, and a convergence judgment of the adaptive notch filter.

Images