Sweep injection method in motor control
By superimposing an adaptive chirped excitation signal under the motor control closed loop, the runaway risk and mode identification failure problem in frequency sweep testing of complex mechanical systems are solved, and safe, accurate online identification and non-destructive testing of resonant frequencies are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANGHAI MITSUBISHI ELEVATOR CO LTD
- Filing Date
- 2026-03-26
- Publication Date
- 2026-06-19
Smart Images
Figure CN122247288A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of motor control technology, and specifically to a sweep frequency injection method in motor control. Background Technology
[0002] Frequency sweep excitation, as an effective means of identifying the frequency response characteristics of a system, has been maturely applied in fields such as servo control. However, existing technologies have significant limitations. Chinese Patent Document 1 (Application No. CN202010875711.9) discloses a frequency sweep test using an open-loop control mode, directly inputting the frequency sweep signal as a torque command into the system and disconnecting the normal closed-loop control. This leads to a risk of runaway when applied to complex mechanical systems with high inertia and variable loads (such as heavy-duty robots and large lifting equipment), and makes it impossible to achieve safe online testing when the equipment is operating normally. Although Chinese Patent Document 2 (Application No. CN202411529630.8) emphasizes "non-destructive," its concept focuses on avoiding mechanical damage and does not address the problem of intrusive interference to the stable control loop during the frequency sweep process. In addition, existing methods typically inject a frequency sweep signal with a fixed amplitude, which cannot adapt to conditions with drastic load changes. Under light loads, excessive excitation may cause nonlinear vibrations, while under heavy loads, insufficient excitation may lead to mode identification failure. There is a lack of a mechanism to adaptively adjust the excitation amplitude based on the real-time load torque. In terms of response detection, existing technologies generally rely on feedback from drive motor encoders, which can only reflect local characteristics of the transmission chain and cannot effectively capture vibration modes at the actuator or load end far from the drive end, making it difficult to comprehensively evaluate the dynamic performance of multi-degree-of-freedom coupled mechanical systems.
[0003] Therefore, how to achieve safe, accurate, and online identification of the resonant frequency of complex mechanical systems while maintaining stable closed-loop operation is a current technical challenge. Summary of the Invention
[0004] To solve the above technical problems, the present invention provides a sweep frequency injection method in motor control. When the driver is in normal closed-loop operation, a chirped excitation signal with adaptive amplitude variation is generated in real time according to preset sweep frequency parameters, and the chirped excitation signal is superimposed on the basic current command of the current loop.
[0005] Preferably, the frequency sweep injection method in motor control includes the following steps: Step S1: Configure the frequency sweep parameters; the frequency sweep parameters include: start frequency, end frequency, total frequency sweep time, and amplitude coefficient of the frequency sweep signal; Step S2: Deploy the sensor; the sensor is used to collect vibration signals from the motor system. Step S3: The motor system enters closed-loop operation; Step S4: Generate a chirped excitation signal with adaptive value changes in real time according to the preset sweep frequency parameters, and superimpose the chirped excitation signal onto the basic current command of the current loop; Step S5: Collect vibration signals from the motor system, and perform preprocessing and specific filtering on the vibration signals; Step S6: Perform frequency domain conversion and peak identification on the filtered vibration signal; Step S7: Record the frequency and amplitude corresponding to the peak value to identify the resonant frequency of the motor system.
[0006] Preferably, the chirped excitation signal satisfies the formula: ;in For chirping excitation signal, The amplitude coefficient of the swept frequency signal. The current command value for the q-axis is calculated in real time by the current loop, where k is the rate of frequency increase. Let t be the initial angular frequency of the injected sweep, and t be time.
[0007] Preferably, the chirped excitation signal satisfies the formula: Among them, For chirping excitation signal, The amplitude coefficient of the swept frequency signal. The current command value for the q-axis is calculated in real time by the current loop, where t is time. Represents the segment number. This represents the start time of the segment. This represents the end time of the segment. This represents the highest frequency during that period. This represents the minimum frequency during that period.
[0008] Preferably, the chirped excitation signal satisfies the formula: ;in For chirping excitation signal, The amplitude coefficient of the swept frequency signal. The current command value for the q-axis is calculated in real time by the current loop, and T is the total sweep time. The starting frequency for the frequency sweep. t is the termination frequency of the sweep, and t is the time.
[0009] Preferably, the chirped excitation signal satisfies the formula: ;in For chirping excitation signal, The amplitude coefficient of the swept frequency signal. The current command value for the q-axis is calculated in real time by the current loop, and T is the total sweep time. The starting frequency for the frequency sweep. t is the termination frequency of the sweep, and t is the time.
[0010] Preferably, the base current command of the current loop is synthesized by the output of the speed controller and the feedforward signal.
[0011] Preferably, the specific filter is adapted to different filter types according to the characteristics of the sensor measurement point location and the load.
[0012] Preferably, when the sensor is mounted on a motor bearing housing or a load coupling, the specific filter is adapted to a high-pass filter or a band-pass filter.
[0013] Preferably, when the sensor is installed at the load end far from the motor, the specific filter is adapted to a low-pass filter or a band-pass filter.
[0014] Preferably, when the purpose of frequency sweeping is to verify whether there is resonance in a known preset narrow frequency band, the specific filter is adapted to a narrow bandpass filter centered on the preset frequency band.
[0015] Those skilled in the art should understand that the above is merely an example. In practical applications, the appropriate filter type can be flexibly selected according to the specific characteristics of the measurement point, as long as it conforms to the analysis logic of "location-characteristic-filter matching".
[0016] The frequency sweep injection method in motor control of this invention fundamentally avoids the risk of system runaway caused by traditional open-loop frequency sweeping. Because the main control loop is always operational, the system maintains a stable operating state throughout the entire frequency sweep process, making online modal testing possible for large, high-value, or high-safety-requirement equipment (such as heavy-duty robots and large rotary mechanisms). Simultaneously, small-amplitude excitation can constrain the system response within the linear range, avoiding nonlinear vibration or mechanical damage caused by excessive excitation, thus achieving truly "non-destructive" testing. Attached Figure Description
[0017] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments: Figure 1 This is a schematic diagram of the sweep frequency injection method steps in motor control of Example 1; Figure 2 A schematic diagram of a specific motor system structure for implementing the frequency sweep injection method in motor control of Example 1; Figure 3 This is a schematic diagram of the sweep frequency signal waveform; Figure 4 This is a schematic diagram of vibration signal data. Detailed Implementation
[0018] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can fully understand other advantages and technical effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through different specific embodiments, and the details in this specification can also be applied based on different viewpoints, with various modifications or changes made without departing from the overall design concept of the invention. It should be noted that, unless otherwise specified, the following embodiments and features can be combined with each other. The following exemplary embodiments of the present invention can be implemented in many different forms and should not be construed as being limited to the specific embodiments set forth herein. It should be understood that these embodiments are provided to make the disclosure of the present invention thorough and complete, and to fully convey the technical solutions of these exemplary embodiments to those skilled in the art. Example
[0019] This embodiment provides a sweep frequency injection method for a single motor system. In this method, when the driver is in normal closed-loop operation, a chirped excitation signal with adaptive amplitude variation is generated in real time according to preset sweep frequency parameters, and the chirped excitation signal is superimposed on the basic current command of the current loop.
[0020] The following section provides a detailed explanation using a typical permanent magnet synchronous motor drive system.
[0021] like Figure 1 As shown, the following steps exemplify the detailed steps of the frequency sweep injection method in motor control: Step S1: Configure the frequency sweep parameters; the frequency sweep parameters include: start frequency, end frequency, total frequency sweep time, and amplitude coefficient of the frequency sweep signal. At the same time, you can choose between linear sweep frequency and logarithmic sweep frequency modes depending on the purpose of the test.
[0022] Step S2, deploy sensors; the sensors are used to collect vibration signals from the motor system; for example, vibration sensors are deployed at key mechanical locations such as the motor shaft extension end and load coupling; or the encoder is confirmed to be available in preparation for collecting vibration or speed response signals from the system.
[0023] Step S3: The motor system enters closed-loop operation (speed loop and current loop operate in closed loop). Step S4: Generate a chirped excitation signal with adaptively varying amplitude values in real time according to preset sweep frequency parameters. The chirped excitation signal is then superimposed onto the base current command of the current loop. The "closed-loop superposition" mechanism ensures that the excitation signal is only a small-amplitude disturbance, and the main control system always remains stable. The motor runs under the drive of this composite command, and the system synchronously and at high speed records the waveform of the injected excitation signal and the response data from the vibration sensor and encoder.
[0024] Step S5: Collect vibration signals from the motor system and perform preprocessing and specific filtering on the vibration signals; that is, apply targeted bandpass filtering based on the sensor position and target frequency band to optimize the signal-to-noise ratio.
[0025] Step S6: Perform frequency domain transformation (Fast Fourier Transform, FFT) and peak identification on the filtered vibration signal; Step S7: Record the frequency and amplitude corresponding to the peak value to identify the resonant frequency of the motor system.
[0026] Figure 2 A specific motor system diagram is shown for implementing the frequency sweep injection method in the motor control of this embodiment. The motor system mainly includes three functional modules: a frequency sweep injection module, a motor drive control system, and the motor and the controlled object.
[0027] The frequency sweep injection module independently generates the excitation signal for frequency response testing. Its core function is to generate a chirped excitation signal with continuously varying frequency. Specifically, this module calculates and outputs a sine wave signal in real time based on preset start and end frequencies and sweep time. (This is the chirped excitation signal, sometimes referred to as the swept-frequency injection signal below; the two are the same.) Its frequency increases linearly with time or according to a specific pattern. This signal is the swept-frequency excitation current that will be injected into the system later.
[0028] Specifically, the frequency sweep injection signal can be described as
[0029] The amplitude coefficient of the sweep frequency signal represents the ratio of the injected signal amplitude to the current torque current. The advantage of this approach is that, for different load conditions, there is no need to repeatedly modify the embedded program; only the parameters need to be changed.
[0030] This is the command value of the q-axis current (torque current) calculated in real time by the current loop. This ensures that the strength of the injected signal is proportional to the actual load torque, guaranteeing an effective excitation amplitude under different loads.
[0031] The core of the sweep frequency signal can include various forms, the most common being angular frequency. Over time Linear increase, i.e. Where k represents the rate at which the frequency increases. This represents the initial angular frequency of the injected sweep. A detailed formula can be written as follows: .
[0032] For example, if the sweep frequency f is set to continuously scan from a preset lower limit (e.g., 30Hz) to a preset upper limit (e.g., 300Hz) for 30 seconds, then it can be designed. That is, the initial sweep frequency is 30Hz, and the frequency increase rate is 9Hz / s.
[0033] In addition to the linear increase method mentioned above, a step-like or piecewise function can also be injected to sweep key frequency points instead of sweeping the entire frequency range. The frequency sweep injection signal can be described as follows: ; Among them, For chirping excitation signal, The amplitude coefficient of the swept frequency signal. The current command value for the q-axis is calculated in real time by the current loop, where t is time. Represents the segment number. This represents the start time of the segment. This represents the end time of the segment. This represents the maximum frequency during that period, and this represents the minimum frequency during that period.
[0034] For example, if a system has known vibration frequencies within the 30~40Hz and 90~100Hz ranges, then a frequency sweep injection signal can be designed as follows: .
[0035] Additionally, the injected sweep frequency signal can be classified into linear sweep frequency and logarithmic sweep frequency, depending on the actual needs.
[0036] Linear frequency sweep refers to a signal whose frequency f increases linearly with time t, i.e. , where K is a constant sweep rate (Hz / s). Its time-domain waveform is as follows: Figure 3 As shown on the left, the frequency changes uniformly within the same time period. The advantage of linear frequency sweep is that the excitation time of each frequency component is the same throughout the entire swept frequency band. Therefore, after Fast Fourier Transform (FFT) analysis, the frequency resolution of the entire frequency axis is uniform. This makes the peak clarity of all frequency bands in the spectrogram consistent, which is particularly suitable for scenarios that require global and uniform observation of modal distribution within the entire frequency band (such as 0-300Hz).
[0037] Logarithmic frequency sweep refers to a signal whose frequency f increases exponentially with time t, which is represented by a linear change in frequency on a logarithmic scale. Its mathematical relationship can be expressed as: Where T is the total sweep time. The starting frequency for the frequency sweep. This is the termination frequency of the frequency sweep. Its time-domain waveform is as follows: Figure 3As shown on the right, logarithmic frequency sweeps are slow in the low-frequency region (due to longer dwell time at each low-frequency point) but fast in the high-frequency region. This results in significantly higher frequency resolution and signal-to-noise ratio in the low-frequency band compared to the high-frequency band. For most mechanical systems, the accuracy requirements for identifying key low-frequency rigid body modes or main structural modes (usually located in the low-frequency band) are higher. Logarithmic frequency sweeps naturally optimize low-frequency excitation, and the dynamic characteristics of many mechanical systems are observed more naturally on a logarithmic scale, with modal distributions often exhibiting more uniform spacing on logarithmic coordinates.
[0038] Specifically, the logarithmic sweep frequency injection signal can be described as
[0039] To save computation time in embedded chips, when the frequency sweep range is narrow, or when only the low-frequency sweep results are considered, the above equation can be simplified to the following form.
[0040] The system adopts a classic three-ring structure based on magnetic field orientation control, which is the existing technology basis of this embodiment, but its key improvement lies in the injection method of the sweep frequency signal.
[0041] Speed loop processing: given reference speed With the actual feedback speed from the motor encoder The error signal is compared and adjusted by a speed PI controller.
[0042] Feedforward and Command Synthesis: To improve response speed, a feedforward circuit is often added. The output of the speed controller and the feedforward signal are combined to synthesize the basic q-axis current command. This current command directly corresponds to the motor's output torque.
[0043] Core Improvement: Safe Superposition of Sweep Signals: The key innovation of this invention lies in ensuring uninterrupted normal closed-loop control. This is achieved by injecting the excitation current generated by the sweep signal into the module. At this moment, it is superimposed on the base current command. Above, a new total current setpoint is formed. And excitation current The amplitude depends on the torque current. (Usually accounting for a low percentage of 1% to 5%), this superposition point is located at the input end of the current loop and is the core to ensure safety and effectiveness.
[0044] Current Loop and Power Drive: Combined Total Current Command With the actual current of the feedback The comparison is performed, and the error signal is precisely adjusted by the current PI controller to output the dq axis voltage command. The voltage command is then inversely transformed by Clark to generate a three-phase voltage command. Finally, the inverter drives the motor to run according to this instruction.
[0045] An encoder mounted on the motor shaft monitors the motor speed in real time, forming a speed closed-loop feedback W. Simultaneously, to comprehensively analyze the mechanical system's response, a vibration sensor (such as an accelerometer) can be installed on the motor shaft extension or the load coupling. This vibration signal is compared with the injected sweep frequency excitation signal. These data are recorded together and used as a data source for subsequent frequency response analysis and identification of the system's resonant frequency. A vibration sensor is attached to the motor base to analyze the vibration of the motor base.
[0046] After completing the injection of the sweep frequency signal and the acquisition of synchronous data, the core task lies in analyzing the acquired vibration response signal to accurately extract the system's modal parameters. The analysis method of this invention includes the following key steps: The acquired data undergoes preprocessing and specific filtering. The raw vibration signals typically contain high-frequency noise, DC drift, and environmental interference. Preprocessing begins with removing the DC component to eliminate sensor zero drift. Then, the core specific filtering stage begins. This step does not apply a uniform filter, but rather selects a filter based on the core target frequency range of the measurement (starting from the sweep frequency). and termination frequency (Definition) and the physical characteristics sensitively reflected by the installation location of the vibration sensor, and apply targeted bandpass filtering.
[0047] Specifically, if the sensor is mounted on the motor bearing housing or load coupling to identify high-frequency torsional or bending resonances in the drivetrain (typically above 100Hz), a high-pass or band-pass filter (e.g., 80Hz-1000Hz) can be used to effectively filter out low-frequency mechanical vibrations and environmental interference, highlighting the high-frequency resonance peaks. If the sensor is mounted at the load end, away from the motor, to identify the overall low-frequency rigid body modes of the system (typically below 30Hz), a low-pass or band-pass filter (e.g., 0.5Hz-50Hz) should be used to suppress motor electromagnetic noise and high-frequency structural vibrations, ensuring the clarity of the low-frequency modes. If the frequency sweep is intended to verify the existence of resonance in a known narrow frequency band (e.g., 90-110Hz), a narrow band-pass filter centered on that frequency band can be used directly to maximize the signal-to-noise ratio of that band.
[0048] For the vibration spectra obtained from each measuring point after filtering and FFT analysis, a combination of an automatic peak detection algorithm and manual interpretation is used to identify significant peaks in the spectrum. FFT analysis is then performed on vibration data within a certain time period before and after the peak (e.g., 1 second before and after the peak center) to determine the accurate vibration frequency and amplitude corresponding to that peak. The corresponding vibration frequency is the system's resonant frequency.
[0049] For example Figure 4 As shown in the figure, this is a frequency sweep experiment on the motor body. The frequency sweep injection amplitude is 1~60Hz. The vibration data is low-pass filtered at 100Hz, and the peak points in the vibration direction are analyzed. The frequency and amplitude of the peak points are marked in the figure, which helps to determine the resonant frequency of the system. There are obvious peaks near 42Hz and 50Hz.
[0050] The present invention has been described in detail above through specific embodiments and examples, but these are not intended to limit the invention. Many modifications and improvements can be made by those skilled in the art without departing from the principles of the invention, and these should also be considered within the scope of protection of the present invention.
Claims
1. A sweep frequency injection method for motor control, characterized in that, When the driver is in normal closed-loop operation, a chirped excitation signal with adaptive amplitude variation is generated in real time according to the preset sweep frequency parameters, and the chirped excitation signal is superimposed on the basic current command of the current loop.
2. The frequency sweep injection method in motor control according to claim 1, characterized in that, Includes the following steps: Step S1: Configure the frequency sweep parameters; the frequency sweep parameters include: start frequency, end frequency, total frequency sweep time, and amplitude coefficient of the frequency sweep signal; Step S2: Deploy the sensor; the sensor is used to collect vibration signals from the motor system. Step S3: The motor system enters closed-loop operation; Step S4: Generate a chirped excitation signal with adaptive value changes in real time according to the preset sweep frequency parameters, and superimpose the chirped excitation signal onto the basic current command of the current loop; Step S5: Collect vibration signals from the motor system, and perform preprocessing and specific filtering on the vibration signals; Step S6: Perform frequency domain conversion and peak identification on the filtered vibration signal; Step S7: Record the frequency and amplitude corresponding to the peak value to identify the resonant frequency of the motor system.
3. The frequency sweep injection method in motor control according to claim 2, characterized in that, The chirped excitation signal satisfies the following formula: ;in For chirping excitation signal, The amplitude coefficient of the swept frequency signal. The current command value for the q-axis is calculated in real time by the current loop, where k is the rate of frequency increase. Let t be the initial angular frequency of the injected sweep, and t be time.
4. The frequency sweep injection method in motor control according to claim 2, characterized in that, The chirped excitation signal satisfies the following formula: ; Among them, For chirping excitation signal, The amplitude coefficient of the swept frequency signal. The current command value for the q-axis is calculated in real time by the current loop, where t is time. Represents the segment number. This represents the start time of the segment. This represents the end time of the segment. This represents the highest frequency during that period. This represents the minimum frequency during that period.
5. The frequency sweep injection method in motor control according to claim 2, characterized in that, The chirped excitation signal satisfies the following formula: ;in For chirping excitation signal, The amplitude coefficient of the swept frequency signal. The current command value for the q-axis is calculated in real time by the current loop, and T is the total sweep time. The starting frequency for the frequency sweep. t is the termination frequency of the sweep, and t is the time.
6. The frequency sweep injection method in motor control according to claim 2, characterized in that, The chirped excitation signal satisfies the following formula: ;in For chirping excitation signal, The amplitude coefficient of the swept frequency signal. The current command value for the q-axis is calculated in real time by the current loop, and T is the total sweep time. The starting frequency for the frequency sweep. t is the termination frequency of the sweep, and t is the time.
7. The frequency sweep injection method in motor control according to claim 2, characterized in that, The basic current command of the current loop is synthesized by the output of the speed controller and the feedforward signal.
8. The frequency sweep injection method in motor control according to claim 2, characterized in that, Depending on the sensor measurement point location and load characteristics, the specific filter is adapted to different filter types.
9. The frequency sweep injection method for motor control according to claim 8, characterized in that, When the sensor is mounted on a motor bearing housing or load coupling, the specific filter is adapted to a high-pass filter or a band-pass filter.
10. The frequency sweep injection method in motor control according to claim 8, characterized in that, When the sensor is installed at the load end far from the motor, the specific filter is adapted to a low-pass filter or a band-pass filter.
11. The frequency sweep injection method in motor control according to claim 8, characterized in that, When the purpose of frequency sweeping is to verify whether there is resonance in a known preset narrow frequency band, the specific filter is adapted to a narrow bandpass filter centered on the preset frequency band.