Permanent magnet synchronous motor vector control method and system

Abstract

The application discloses a kind of permanent magnet synchronous motor vector control method and system, method includes: according to each current harmonic distortion rate and its corresponding preset speed instruction value, construct speed-distortion response sequence, according to the amplitude variation trend between adjacent elements in the sequence, identify candidate resonance speed point;Judge whether the number of candidate resonance speed point is greater than the preset number threshold;If not greater than, extract the target resonance point corresponding to the maximum distortion change rate and divide the response sequence into multiple response subintervals according to the point;According to the distortion gradient characteristics between elements in each subinterval, identify the current load inertia;According to the current load inertia identification value, adjust the proportional gain and integral time constant of speed loop PI controller.The application utilizes the resonance characteristics of current harmonic distortion rate with speed variation to realize the online identification of load inertia and the adaptive adjustment of PI parameter, without additional signal injection, without interrupting the normal operation of motor, with the advantages of non-disturbance identification and adaptive adjustment.

Description

Technical Field

[0001] This invention belongs to the field of permanent magnet synchronous motor control technology, and particularly relates to a vector control method and system for permanent magnet synchronous motors. Background Technology

[0002] Permanent magnet synchronous motors (PMSMs) are widely used in high-end manufacturing fields such as electric vehicles, industrial servo drives, and aerospace due to their high efficiency, high power density, and excellent control performance. Vector control, as the mainstream control strategy for PMSMs, decomposes the stator current into excitation components (d-axis) and torque components (q-axis) and adjusts them independently, thereby achieving decoupled control of motor torque and flux linkage.

[0003] In a permanent magnet synchronous motor (PMSM) vector control system, the proportional gain and integral time constant of the speed loop PI controller have a decisive impact on the system's dynamic response performance and steady-state accuracy. However, during actual operation, the load inertia of a PMSM changes significantly with varying operating conditions; for example, the load inertia of an industrial robot can change several times when switching between different workstations. When the load inertia changes and the PI controller parameters are not adjusted accordingly, it can lead to performance degradation problems such as increased speed overshoot, increased response delay, and decreased disturbance rejection capability, and in severe cases, even system oscillation.

[0004] Existing load inertia identification methods mainly include the following categories: One approach is the model-reference adaptive inertia identification method, which estimates inertia in real time by constructing an error signal between a reference model and an adjustable model. This type of method relies on an accurate mathematical model of the motor, is sensitive to model errors and parameter changes, and its identification accuracy drops sharply under low-speed or light-load conditions.

[0005] The second method is the inertia identification method based on disturbance observers, which indirectly calculates the inertia by observing load torque disturbances. This type of method requires the design of complex observer structures, involves a large amount of computation, and there is an inherent contradiction between the convergence speed of the observer and the identification accuracy.

[0006] Thirdly, there is the inertia identification method using signal injection. This involves injecting a sinusoidal or square wave excitation signal into the speed loop or current loop, and identifying the inertia by analyzing the system's response to the injected signal. This method requires interrupting the normal operation of the motor, and the injected signal itself can disturb the motor's output torque, affecting the quality of normal motor operation.

[0007] Therefore, there is a need for a method that can identify the load inertia online by utilizing the inherent current harmonic characteristics of the system during normal motor operation, and adaptively adjust the parameters of the speed loop PI controller based on the identification results. Summary of the Invention

[0008] This invention provides a vector control method and system for a permanent magnet synchronous motor, which is used to solve at least one of the above-mentioned technical problems.

[0009] In a first aspect, the present invention provides a vector control method for a permanent magnet synchronous motor, comprising: The q-axis current feedback signal of the permanent magnet synchronous motor under multiple preset speed commands is obtained, and the current harmonic distortion rate corresponding to each preset speed command is extracted from each q-axis current feedback signal according to the preset amplitude-frequency feature extraction strategy. Based on the current harmonic distortion rate and its corresponding preset speed command value, a speed-distortion response sequence is constructed, and at least one candidate resonant speed point is identified based on the amplitude variation trend between adjacent elements in the speed-distortion response sequence. Determine whether the number of candidate resonant speed points is greater than a preset threshold. If the number is not greater than a preset threshold, then the target resonance point corresponding to the maximum distortion rate is extracted from the candidate resonance speed points, and the speed-distortion response sequence is divided according to the target resonance point to obtain at least one response sub-interval. Based on the distortion gradient characteristics between elements within the at least one response sub-interval, a preset rotational inertia identification strategy is used to identify the load inertia of the permanent magnet synchronous motor, thereby obtaining the current load inertia identification value. Based on the current load inertia identification value, adjust the proportional gain and integral time constant of the permanent magnet synchronous motor speed loop PI controller.

[0010] Secondly, the present invention provides a vector control system for a permanent magnet synchronous motor, comprising: The extraction module is configured to acquire the q-axis current feedback signal of the permanent magnet synchronous motor under multiple preset speed commands, and extract the current harmonic distortion rate corresponding to each preset speed command from each q-axis current feedback signal according to the preset amplitude-frequency feature extraction strategy. The identification module is configured to construct a speed-distortion response sequence based on each current harmonic distortion rate and its corresponding preset speed command value, and to identify at least one candidate resonant speed point based on the amplitude variation trend between adjacent elements in the speed-distortion response sequence. The judgment module is configured to determine whether the number of candidate resonant speed points is greater than a preset number threshold. The partitioning module is configured to extract the target resonance point corresponding to the maximum distortion rate from the candidate resonance speed points if the number is not greater than a preset threshold, and partition the speed-distortion response sequence according to the target resonance point to obtain at least one response sub-interval. The identification module is configured to identify the load inertia of the permanent magnet synchronous motor based on the distortion gradient characteristics between elements in the at least one response sub-interval, using a preset rotational inertia identification strategy, and obtain the current load inertia identification value. The adjustment module is configured to adjust the proportional gain and integral time constant of the permanent magnet synchronous motor speed loop PI controller based on the current load inertia identification value.

[0011] Thirdly, an electronic device is provided, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps of the permanent magnet synchronous motor vector control method of any embodiment of the present invention.

[0012] Fourthly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor performs the steps of the permanent magnet synchronous motor vector control method according to any embodiment of the present invention.

[0013] The vector control method and system for permanent magnet synchronous motors disclosed in this application have the following beneficial effects: By utilizing the physical law that the inherent harmonic components in the q-axis current of a permanent magnet synchronous motor change with the load inertia under vector control operation, the load inertia change is sensed by extracting the harmonic distortion rate of the q-axis current under different speed commands. No additional excitation signal is required, realizing true non-disruptive online inertia identification. Compared with the traditional signal injection method, this invention does not generate any additional torque pulsation and does not affect the normal operation quality of the motor. The current harmonic distortion rate is classified into inverter switching frequency-related components, rotor cogging harmonic components, and broadband noise components according to their physical sources. Adaptive weighting coefficients related to rotational speed are set for the three types of components, so that the extracted distortion rate has consistent sensitivity to load inertia changes at different rotational speeds. Compared with the traditional total harmonic distortion rate calculation method, the extracted distortion rate can more accurately reflect the resonance peak characteristics in the speed-distortion response curve, providing a high-quality data foundation for subsequent identification of resonance speed points. Candidate resonant speed points are identified by analyzing the inflection point where the sign of the differential slope changes from positive to negative in the speed-distortion response sequence. Furthermore, candidate points are simplified by filtering with the vertical axis and merging with the horizontal axis. This resonant point identification method based on the second-order change trend is more robust than directly finding local maxima and can effectively distinguish between true mechanical resonance peaks and spurious peaks caused by measurement noise. An interval centroid discrimination strategy based on the principal feature vector is adopted. By calculating the jump value between the principal distortion reference value and the preceding distortion reference value and the significance of the distortion gradient of the principal interval, it automatically determines whether to use the distortion value of the principal interval or the weighted average distortion value of all intervals to query the inertia identification result. This adaptive selection mechanism improves the adaptability and accuracy of inertia identification in different scenarios of single-peak resonance and multi-peak resonance. The proportional gain and integral time constant of the speed loop PI controller are adjusted based on the ratio of the identified current load inertia value to the nominal no-load inertia value. The adjustment logic is simple and clear, requiring no complex parameter tuning algorithm, making engineering implementation easy, and ensuring consistent dynamic performance of the speed loop across the entire load range. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 A flowchart of a vector control method for a permanent magnet synchronous motor provided in an embodiment of the present invention; Figure 2 This is a structural block diagram of a permanent magnet synchronous motor vector control system provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Please see Figure 1 The diagram shows a flowchart of a vector control method for a permanent magnet synchronous motor according to this application.

[0018] like Figure 1 As shown, the vector control method for permanent magnet synchronous motors specifically includes the following steps: Step S101: Obtain the q-axis current feedback signal of the permanent magnet synchronous motor under multiple preset speed commands, and extract the current harmonic distortion rate corresponding to each preset speed command from each q-axis current feedback signal according to the preset amplitude-frequency feature extraction strategy.

[0019] In this step, each preset speed command among the plurality of preset speed commands is taken as the current command. The current command is applied to the permanent magnet synchronous motor and a preset stable waiting time is continuously performed. After the stable waiting time ends, the time-domain waveform data of the q-axis current feedback signal within the preset sampling time is continuously collected at a preset sampling frequency to obtain the q-axis current time-domain sequence corresponding to the current command. Using the fundamental frequency corresponding to the current command as a reference frequency, an adaptive notch filter is used to separate the fundamental component sequence and the residual component sequence from the q-axis current time domain sequence. The fundamental frequency is calculated based on the number of pole pairs of the permanent magnet synchronous motor and the rotational speed value of the current command. Using the inverter switching frequency and its integer multiples as the center frequency, the frequency band components within a preset bandwidth range at each center frequency are extracted from the residual component sequence and merged into an inverter switching frequency related component sequence. Using the rotor cogging harmonic frequency and its integer multiples as the center frequency, the frequency band components within a preset bandwidth range at each center frequency are extracted from the residual component sequence and merged into the rotor cogging harmonic component sequence. The remaining part of the residual component sequence after removing the inverter switching frequency related component sequence and the rotor cogging harmonic component sequence is taken as the broadband noise component sequence. Calculate the first effective value of the inverter switching frequency related component sequence, the second effective value of the rotor cogging harmonic component sequence, and the third effective value of the broadband noise component sequence, and simultaneously calculate the fourth effective value of the fundamental component sequence; The first distortion rate is obtained by dividing the first effective value of the inverter switching frequency related component sequence by the second effective value of the fundamental component sequence; the second effective value of the rotor cogging harmonic component sequence is obtained by dividing the fourth effective value of the fundamental component sequence; and the third effective value of the broadband noise component sequence is obtained by dividing the fourth effective value of the fundamental component sequence. Obtain the rotational speed value corresponding to the current instruction, and determine the first weighting coefficient, the second weighting coefficient, and the third weighting coefficient based on the rotational speed value; Calculate the sum of the first distortion factor multiplied by the first weighting coefficient, the second distortion factor multiplied by the second weighting coefficient, and the third distortion factor multiplied by the third weighting coefficient, and use the summation result as the current harmonic distortion factor corresponding to the current instruction; Iterate through all preset speed commands to obtain the current harmonic distortion rate corresponding to each preset speed command.

[0020] In one specific embodiment, the numerical range and distribution of multiple preset speed commands are first determined. These preset speed commands increase from a starting speed to a ending speed according to a preset step size. The starting speed is set to 10% of the rated speed of the permanent magnet synchronous motor, the ending speed is set to 100% of the rated speed, and the preset step size is set to 5% of the rated speed. This generates a total of 19 preset speed command values. In practical applications, the step size can be adjusted according to the motor's speed range and accuracy requirements. A smaller step size results in higher accuracy for subsequent resonance point identification, but increases calibration time.

[0021] Each preset speed command from a plurality of preset speed commands is taken as the current command and applied to the permanent magnet synchronous motor. The corresponding torque current command is output through the speed loop PI controller to drive the motor to the target speed. After applying the current command, a preset stabilization waiting time is maintained. The stabilization waiting time is determined based on the mechanical time constant of the permanent magnet synchronous motor, and is usually taken as 3 to 5 times the mechanical time constant to ensure that the motor speed transition process is completely completed and enters a steady state. For conventional industrial permanent magnet synchronous motors, the stabilization waiting time is 1 to 3 seconds.

[0022] After a stable waiting period, the time-domain waveform data of the q-axis current feedback signal within the preset sampling time is continuously acquired at a preset sampling frequency. The sampling frequency is no less than 10 times the switching frequency of the permanent magnet synchronous motor inverter to ensure accurate sampling of harmonic components related to the switching frequency. The preset sampling time is an integer multiple of the fundamental period of the permanent magnet synchronous motor, typically 10 to 20 fundamental periods, to ensure the frequency resolution of the frequency domain analysis. The acquired data is used as the q-axis current time-domain sequence corresponding to the current command.

[0023] Preprocessing is performed on the q-axis current time-domain sequence: the DC bias component is removed, i.e., the arithmetic mean of the time-domain sequence is calculated and the mean is subtracted from each sampling point; a Hanning window is applied to the debiased sequence to reduce spectral leakage.

[0024] Using the fundamental frequency corresponding to the current command as the reference frequency, an adaptive notch filter separates the fundamental component sequence and the residual component sequence from the preprocessed q-axis current time-domain sequence. The adaptive notch filter adopts a second-order IIR notch structure, with its notch frequency tracking the reference frequency in real time and a notch bandwidth coefficient of 0.05. The reference frequency is calculated based on the number of pole pairs of the permanent magnet synchronous motor and the speed value of the current command, using the formula: f1 = p × n / 60, where p is the number of pole pairs and n is the current speed command value (in rpm). After adaptive notch filtering, the component with the same frequency as the fundamental frequency in the output signal is filtered out, and the remaining part is the residual component sequence.

[0025] In the residual component sequences, extraction is performed according to different physical sources: With the inverter switching frequency f sw Using the second and third harmonics as center frequencies, frequency band components within a preset bandwidth range at each center frequency are extracted from the spectrum of the residual component sequence. The preset bandwidth is ±2% of the switching frequency to cover the sidebands caused by modulation changes. All extracted frequency band components are reconstructed in the time domain and merged into an inverter switching frequency-related component sequence. This component mainly originates from nonlinear factors such as inverter dead-time effect and asymmetric diode voltage drop; its amplitude is affected by the load current and has a weak relationship with the motor mechanical state.

[0026] With rotor cogging harmonic frequency f slot =Z×f r Its 2nd and 3rd harmonic frequencies are the center frequencies, where Z is the number of rotor slots of the permanent magnet synchronous motor, and f r The rotor's mechanical rotation frequency (f) r =n / 60). Extract frequency band components within a preset bandwidth range at each center frequency from the spectrum of the residual component sequence. The preset bandwidth is ±1% of the cogging harmonic frequency. Reconstruct all extracted frequency band components in the time domain and merge them into a rotor cogging harmonic component sequence. This component originates from the magnetic permeability harmonics between the rotor permanent magnet magnetic field and the stator core slots. Its amplitude is affected by the air gap magnetic flux density distribution and the rotor eccentricity, and is indirectly related to the load inertia change.

[0027] The remaining portion of the residual component sequence after removing the previously extracted inverter switching frequency-related component sequence and rotor cogging harmonic component sequence is taken as the broadband noise component sequence. This component includes measurement noise, quantization error, and other unclassified non-periodic random components.

[0028] Calculate the first RMS value of the inverter switching frequency-related component sequence, the second RMS value of the rotor cogging harmonic component sequence, and the third RMS value of the broadband noise component sequence, and simultaneously calculate the fourth RMS value of the fundamental component sequence. The RMS value is calculated by squaring the amplitude of each sampling point in the sequence, taking the arithmetic mean, and then taking the square root.

[0029] The first distortion factor is obtained by dividing the first effective value of the inverter switching frequency-related component sequence by the fourth effective value of the fundamental component sequence; the second distortion factor is obtained by dividing the second effective value of the rotor cogging harmonic component sequence by the fourth effective value of the fundamental component sequence; and the third distortion factor is obtained by dividing the third effective value of the broadband noise component sequence by the fourth effective value of the fundamental component sequence. All three distortion factors are dimensionless ratios, representing the relative amplitude of the three types of components relative to the fundamental component.

[0030] Obtain the rotational speed corresponding to the current command, and determine the first, second, and third weighting coefficients based on this value. All three coefficients are values ​​greater than 0 and less than 1, and their sum equals 1. The first weighting coefficient monotonically decreases with increasing rotational speed, following a pattern of 0.4 in the low-speed range (<30% of rated speed), decreasing to 0.1 at the rated speed, with linear interpolation used in between. The second weighting coefficient monotonically increases with increasing rotational speed, following a pattern of 0.1 in the low-speed range, increasing to 0.7 at the rated speed, with linear interpolation used in between. The third weighting coefficient remains constant at 0.2 across the entire rotational speed range. The design of these weighting coefficients is based on the fact that at low speeds, the switching frequency component and noise component account for a large proportion. As the rotational speed increases, the contribution of rotor cogging harmonics to distortion gradually becomes dominant. Therefore, increasing the second weighting coefficient and decreasing the first weighting coefficient ensures that the extracted distortion rate maintains consistent sensitivity to changes in mechanical state at different rotational speeds.

[0031] Calculate the sum of the first distortion factor multiplied by the first weighting coefficient, the second distortion factor multiplied by the second weighting coefficient, and the third distortion factor multiplied by the third weighting coefficient, and use the summation result as the current harmonic distortion factor corresponding to the current command.

[0032] Iterate through all preset speed commands, and repeat the above steps of signal acquisition, filtering and separation, classification and extraction, effective value calculation and weighted summation for each preset speed command to obtain the current harmonic distortion rate corresponding to each preset speed command.

[0033] This step categorizes the non-fundamental components of the q-axis current time-domain signal into three physical sources: inverter switching frequency-related, rotor cogging harmonic-related, and broadband noise. Adaptive weighting coefficients are then set for different speed ranges, achieving uniform sensitivity of the current harmonic distortion rate to load inertia changes across a wide speed range. Traditional total harmonic distortion (THD) treats all harmonic components equally. As speed increases, inverter switching frequency-related harmonics weaken due to filter attenuation, while rotor cogging harmonics strengthen due to increased mechanical frequency. This results in the THD value being dominated by different physical factors at different speeds, failing to reflect the unified harmonic variation caused by load inertia changes. By using categorized weighting, the weight of switching frequency components is reduced and the weight of cogging harmonic components is increased at low speeds, and vice versa at high speeds. This ensures that the final output distortion rate highlights the cogging harmonic component variation closely related to mechanical dynamics across the entire speed range, suppressing interference from switching frequency components and random noise components related to electrical nonlinearity. This establishes a stable monotonic mapping relationship between the distortion rate and load inertia.

[0034] Step S102: Based on the current harmonic distortion rate and its corresponding preset speed command value, construct a speed-distortion response sequence, and identify at least one candidate resonant speed point based on the amplitude variation trend between adjacent elements in the speed-distortion response sequence.

[0035] In this step, each preset speed command value is used as the horizontal axis value, and the current harmonic distortion rate corresponding to each preset speed command value is used as the vertical axis value. All data points are arranged in ascending order of horizontal axis values ​​to obtain the speed-distortion response sequence. The speed-distortion response sequence contains M data points arranged in ascending order of horizontal axis. Each data point contains a horizontal axis value and a vertical axis value, and M is the total number of preset speed commands. Starting from the second data point in the speed-distortion response sequence and ending at the last data point, the forward differential slope of each data point is calculated sequentially. All the calculated forward differential slopes are arranged in order of the horizontal coordinate to obtain the differential slope sequence. The forward differential slope is the difference between the vertical coordinate value of the current data point and the vertical coordinate value of the previous data point, divided by the difference between the horizontal coordinate value of the current data point and the horizontal coordinate value of the previous data point. Starting from the second differential slope in the differential slope sequence and ending at the last differential slope, determine sequentially whether the sign of the current differential slope is opposite to the sign of the previous differential slope; If the opposite is true and the sign of the current differential slope is negative, then mark the x-coordinate value of the current data point corresponding to the current differential slope as a candidate resonant speed point and add it to the candidate resonant speed point set; if the opposite is true and the sign of the current differential slope is positive, then do not mark it. Traverse the differential slope sequence and add all x-coordinate values ​​that satisfy the condition of opposite signs and negative current differential slope to the candidate resonant speed point set; Calculate the arithmetic mean of the ordinate values ​​of all candidate resonant speed points in the candidate resonant speed point set, and remove candidate resonant speed points whose ordinate values ​​are less than the arithmetic mean from the candidate resonant speed point set; When the difference between the abscissa values ​​of any two candidate resonant speed points in the candidate resonant speed point set is less than the preset merging distance threshold, the candidate resonant speed point with the larger ordinate value among the two candidate resonant speed points is retained, and the candidate resonant speed point with the smaller ordinate value among the two candidate resonant speed points is deleted. The set of candidate resonant speed points after vertical axis filtering and horizontal axis merging is used as the final identified candidate resonant speed points.

[0036] In one specific embodiment, the preset speed command values ​​obtained in step S101 are paired with their corresponding current harmonic distortion rates to form data point pairs. Using each preset speed command value as the abscissa and the corresponding current harmonic distortion rate as the ordinate, all data points are arranged in ascending order of their abscissa values ​​to obtain a speed-distortion response sequence. The speed-distortion response sequence contains M data points arranged in ascending order of their abscissas. Each data point contains an abscissa value x(i) and a ordinate value y(i), where i = 1, 2, ..., M, and M is the total number of preset speed commands. This sequence describes the complete trend of the current harmonic distortion rate changing with the speed command.

[0037] Starting from the second data point in the speed-distortion response sequence and ending at the last data point, the forward differential slope of each data point is calculated sequentially. For the i-th data point (i from 2 to M), the formula for calculating its forward differential slope k(i) is: k(i)=[y(i)-y(i-1)] / [x(i)-x(i-1)]. The physical meaning of the forward differential slope is: the rate of change of current harmonic distortion rate between two adjacent preset speed commands, and its unit is distortion rate units / rpm. Arrange all the calculated forward differential slopes in order of the horizontal axis to obtain the differential slope sequence K=[k(2),k(3),…,k(M)], and the length of this sequence is M-1.

[0038] Starting from the second differential slope k(3) in the differential slope sequence and ending at the last differential slope k(M), the following judgments are performed sequentially: Obtain the sign of the current differential slope k(i) and the sign of the previous differential slope k(i-1), and determine whether they are opposite. Opposite signs mean one is positive and the other is negative. If k(i) and k(i-1) have opposite signs and the current differential slope k(i) has a negative sign, then mark the x-coordinate value x(i) of the current data point (i.e., the i-th data point) corresponding to the current differential slope as a candidate resonant speed point and add it to the candidate resonant speed point set C. If k(i) and k(i-1) have opposite signs but the current differential slope k(i) has a positive sign, then no marking is performed. Traverse the differential slope sequence and add all x-coordinate values ​​that satisfy the condition of "opposite signs and the current differential slope is negative" to the candidate resonant speed point set C.

[0039] The physical meaning of the differential slope changing from positive to negative is that as the rotational speed gradually increases, the current harmonic distortion rate changes from an increasing trend to a decreasing trend; that is, the curve of distortion rate versus rotational speed reaches a local maximum at this point. In the vector control system of a permanent magnet synchronous motor, when the speed command value approaches the resonant frequency of the speed loop closed-loop transfer function, the harmonic components in the current loop output current are amplified by mechanical resonance, and the distortion rate reaches its peak. As the rotational speed continues to increase, the resonance effect weakens, and the distortion rate decreases. Therefore, the turning point of the differential slope changing from positive to negative accurately indicates the location of the resonant speed point.

[0040] After initially marking the candidate resonant speed points, the ordinate selection step is performed. The arithmetic mean of the ordinate values ​​(i.e., current harmonic distortion rates) corresponding to all candidate resonant speed points in the candidate resonant speed point set C is calculated and denoted as y. mean Iterate through each candidate resonant speed point in the set C of candidate resonant speed points. If the ordinate value of that point is less than y... mean If the candidate resonant rotation speed point is not significant, then that candidate point is removed from set C. The ordinate screening step is used to eliminate candidate points with insignificant distortion rate peaks. In actual measurements, due to noise and measurement errors, the differential slope may change sign due to small fluctuations at non-resonant points, generating false candidate points. These false points usually have low distortion rate amplitudes and can be effectively eliminated by ordinate average screening, retaining only true resonant peak points with distortion rates significantly higher than the average level.

[0041] After the ordinate filtering is completed, the abscissa merging step is performed. The set of candidate resonant speed points C is traversed. If the difference between the abscissa values ​​of any two candidate resonant speed points in the set is less than the preset merging distance threshold D, then... merge If these two candidate points are not found, they are considered duplicate detection results belonging to the same resonance peak. In this case, the candidate resonance speed point with the larger ordinate value is retained, while the candidate resonance speed point with the smaller ordinate value is deleted. The distance threshold D is then merged. merge The step size is determined based on the preset speed command, typically 1.2 times the preset step size. For example, if the preset step size is 5% of the rated speed, then the merging distance threshold is 6% of the rated speed. The horizontal coordinate merging step is used to handle multiple adjacent candidate points generated near the same resonance peak due to data fluctuations. Because differential slope detection is sensitive to local fluctuations, the slope may oscillate repeatedly between positive and negative in the top region of the resonance peak, resulting in the marking of multiple candidate points with similar horizontal coordinates. By setting the merging distance threshold, only the candidate point with the largest distortion rate is retained, ensuring that each resonance peak contributes only one candidate resonance speed point.

[0042] The set of candidate resonant speed points after vertical axis filtering and horizontal axis merging is taken as the final identified candidate resonant speed points and output to step S103 for quantity judgment.

[0043] This step achieves accurate identification of the resonant rotation speed point by using a detection strategy that changes the sign of the differential slope from positive to negative, rather than employing a simple local maximum search method. This approach offers the following technical advantages: The forward differential slope directly reflects the monotonic trend of distortion rate with rotational speed. At the resonance point, distortion first increases and then decreases, and the slope sign undergoes a deterministic change from positive to negative; this characteristic is unaffected by the absolute amplitude of the distortion rate. Even if the amplitude of the resonance peak changes under different load conditions, the change pattern of the slope sign remains stable. Compared to methods that directly find the maximum distortion rate, slope sign detection is less sensitive to amplitude fluctuations and can still effectively identify the resonance point location in scenarios with weak resonance peaks caused by light loads or small inertia changes.

[0044] The ordinate screening step uses the arithmetic mean of the distortion rates of all candidate points in the candidate resonant speed point set as the threshold to eliminate candidate points with distortion rates below the average level. In actual measurements, signal noise may cause sporadic sign changes in the differential slope at non-resonant points, but the distortion rate amplitudes of these spurious peaks are usually low. Using the average value as the screening threshold is adaptive, eliminating the need for a preset fixed amplitude threshold. It can automatically adjust the screening criteria according to the distortion rate levels of different motors under different operating conditions, avoiding the problem of poor adaptability of fixed thresholds under different speed ranges or load conditions.

[0045] The horizontal coordinate merging step combines candidate points with similar horizontal coordinates into a single candidate point by using a preset merging distance threshold. This effectively addresses the issue of duplicate marking caused by slope oscillations at the top of the resonance peak. The merging distance threshold is determined based on the preset rotational speed command step size, ensuring the rationality of the merging rules under different sampling densities. Candidate points with larger vertical coordinates are retained, ensuring that the position with the most significant distortion rate peak is preserved as a representative candidate point after merging, thus improving the reliability of target resonance point selection in subsequent steps.

[0046] The set of candidate resonant speed points directly serves as the input for quantity judgment in step S103 and target resonant point extraction in step S104, and its quality directly affects the accuracy of the entire adaptive adjustment process of control parameters. Through two-stage purification of vertical axis filtering and horizontal axis merging, the final output set of candidate resonant speed points has a high signal-to-noise ratio and positional accuracy, effectively avoiding the interference of false resonant points on subsequent stages, while ensuring that true resonant points are not missed, providing reliable basic data for load inertia identification.

[0047] Step S103: Determine whether the number of candidate resonant speed points is greater than a preset number threshold.

[0048] In this step, if the number exceeds a preset threshold, it is determined that there is a global mismatch between the proportional gain and integral time constant of the speed loop PI controller of the permanent magnet synchronous motor across the entire speed range. In this case, the proportional gain of the speed loop PI controller is reduced by a first step, while the integral time constant is increased by a second step, and subsequent steps are terminated. The first and second step lengths are predetermined based on the rated parameters and stability margin of the permanent magnet synchronous motor. In one specific embodiment, the first step length is 15% of the current proportional gain value, and the second step length is 15% of the current integral time constant value. After completing the above adjustment of the proportional gain and integral time constant, the adjusted proportional gain and integral time constant values ​​are written into the parameter register of the speed loop PI controller, enabling the speed loop PI controller to perform speed closed-loop control according to the adjusted parameters. Subsequently, steps S104, S105, and S106 are terminated, meaning that the response sub-interval division, load inertia identification, and parameter fine-tuning based on the identification results are no longer executed to avoid erroneous identification results caused by an abnormal number of candidate resonant speed points affecting the normal operation of the control system. Simultaneously, a global mismatch message is generated for the speed loop PI controller parameters, which is output via the permanent magnet synchronous motor's upper-level monitoring system or human-machine interface to prompt maintenance personnel to check the motor's mechanical connection status and controller parameter settings.

[0049] Step S104: If the number is not greater than a preset threshold, extract the target resonance point corresponding to the maximum distortion rate from the candidate resonance speed points, and divide the speed-distortion response sequence according to the target resonance point to obtain at least one response sub-interval.

[0050] In this step, the set of candidate resonant speed points is obtained, and it is determined whether the set of candidate resonant speed points is an empty set; When the set of candidate resonant speed points is empty, all data points in the speed-distortion response sequence are output as a response sub-interval; When the set of candidate resonant speed points is not empty, the differential slope value corresponding to each candidate resonant speed point in the set of candidate resonant speed points is extracted from the differential slope sequence. The absolute value of all extracted differential slope values ​​is calculated, and the candidate resonant speed point corresponding to the differential slope value with the largest absolute value is taken as the target resonant point. Using the x-coordinate value of the target resonance point as the dividing boundary, all data points in the speed-distortion response sequence whose x-coordinate value is less than that of the target resonance point are divided into forward response sub-intervals, all data points whose x-coordinate value is greater than that of the target resonance point are divided into backward response sub-intervals, and data points whose x-coordinate value is equal to that of the target resonance point are assigned to the backward response sub-intervals. The number of data points contained in each of the forward response sub-intervals and the backward response sub-intervals is counted separately. When the number of data points contained in the forward response sub-interval is less than the preset minimum data point threshold, the forward response sub-interval is merged into the backward response sub-interval and the forward response sub-interval is deleted. When the number of data points contained in the backward response sub-interval is less than the minimum data point number threshold, the backward response sub-interval is merged into the forward response sub-interval, and the backward response sub-interval is deleted. All remaining response sub-intervals after merging are output as the divided response sub-intervals, and each response sub-interval contains at least the minimum data point threshold number of data points.

[0051] In one specific embodiment, the candidate resonant speed point set C finally identified in step S102 is first obtained, and it is determined whether the candidate resonant speed point set C is an empty set.

[0052] When the candidate resonant speed point set C is empty, it indicates that no inflection point where the differential slope changes from positive to negative was detected across the entire speed range of the speed-distortion response sequence. That is, the current harmonic distortion rate changes monotonically with increasing speed, and there is no significant resonant peak. At this time, all M data points in the speed-distortion response sequence are output as a response sub-interval, and the process jumps to step S105.

[0053] When the candidate resonant speed point set C is not empty, the differential slope value corresponding to each candidate resonant speed point in the candidate resonant speed point set C is extracted from the differential slope sequence K calculated in step S102. The differential slope value corresponding to the candidate resonant speed point x(i) is k(i), which is the forward differential slope at that point. The absolute value of all extracted differential slope values ​​is calculated to obtain the absolute value of the distortion rate of change corresponding to each candidate resonant speed point. The candidate resonant speed point with the largest absolute value of the distortion rate of change is taken as the target resonant point x. target The physical meaning of the largest absolute value of the distortion rate of change is that at the candidate resonant speed point, the current harmonic distortion rate changes with the speed the fastest, that is, the falling edge of the distortion rate curve is the steepest, indicating that the mechanical resonance effect corresponding to this point is the strongest, and it is the optimal dividing boundary for dividing the response sub-interval.

[0054] With the target resonance point x targetThe x-coordinate value is used as the dividing boundary to divide the speed-distortion response sequence into two parts: a forward response sub-interval and a backward response sub-interval. Specifically, all data points in the speed-distortion response sequence with x-coordinate values ​​less than x_target are assigned to the forward response sub-interval; all data points with x-coordinate values ​​greater than x_target are assigned to the backward response sub-interval; and data points with x-coordinate values ​​equal to x_target are assigned to the backward response sub-interval. This division method assigns the points with the steepest descent of the distortion rate to the backward sub-interval, ensuring that the complete response curve shape at the dividing point is not disrupted.

[0055] Count the number of data points contained in each of the forward and backward response subintervals, and denote it as N. front and N back Obtain the preset minimum data point threshold N. min N min The accuracy requirements for calculating statistical characteristics such as the interval average distortion value, interval distortion volatility, and interval distortion gradient value in subsequent step S105 are determined. For statistical quantities such as mean, standard deviation, and gradient, the statistical results are unstable and the reliability is reduced when the sample size is too small. In a specific embodiment, the minimum data point threshold N is... min We set the value to 3 to ensure that each response sub-interval contains at least 3 data points, thus giving the statistical features basic reliability.

[0056] The current response sub-interval contains N data points. front Less than N min When necessary, the forward response sub-intervals are merged into the backward response sub-intervals, and the forward response sub-intervals are deleted. The physical basis for the merging operation is that when there are too few data points in the forward sub-interval, the amount of information contained in the sub-interval is insufficient to independently extract statistically significant distortion features. Merging it into the backward sub-interval can increase the number of samples in the backward sub-interval and improve the reliability of subsequent statistical features.

[0057] When the backward response sub-interval contains N data points back Less than N min When the backward response sub-interval is merged into the forward response sub-interval, the backward response sub-interval is deleted. The physical basis for the merging operation is the same as that for the forward sub-interval: when the backward sub-interval has too few data points, it is merged into the forward sub-interval to ensure the reliable calculation of statistical features.

[0058] When the number of data points contained in both the forward and backward response sub-intervals is not less than Nmin, the two response sub-intervals remain independent and no merging operation is performed.

[0059] The remaining response sub-intervals after merging are output as the divided response sub-intervals, each containing at least the minimum data point threshold Nmin data points. The output results will be used as input to step S105 to calculate the distortion feature vector of each response sub-interval.

[0060] This step extracts the candidate resonant speed point with the largest absolute value of the distortion rate of change as the dividing boundary, and divides the speed-distortion response sequence into independent response sub-intervals. This achieves the differentiation and processing of different variation patterns of current harmonic distortion rate in different speed regions, and has the following technical effects: The absolute value of the differential slope directly reflects the rate of decrease of the distortion rate at the candidate resonant speed point. The larger the absolute value of the distortion rate of change, the faster the distortion rate decays with the change of speed at that point, that is, the speed loop's ability to suppress current harmonics changes significantly at that speed point. Selecting the candidate point with the largest absolute value of the distortion rate of change as the target resonant point ensures that the segmentation boundary is located at the position of the most drastic change in the distortion rate response curve, making the distortion rate change trend within the divided forward and backward sub-intervals more consistent, thereby improving the representativeness of the distortion feature vectors of each interval in step S105.

[0061] When the candidate resonant rotational speed point set is empty, all data points are treated as a single response sub-interval, avoiding invalid calculations on the empty dataset in subsequent steps. In scenarios without resonant peaks, the current harmonic distortion rate monotonically changes with rotational speed, exhibiting no segmented characteristics. Therefore, treating the entire response sequence as a whole for feature extraction is reasonable and efficient. A merging strategy based on a minimum data point threshold ensures that each response sub-interval has a sufficient number of samples for statistical feature calculation. In subsequent step S105, the calculation of the interval average distortion value, interval distortion volatility, and interval distortion gradient value all depend on the data points contained within the sub-interval. When the number of data points within a sub-interval is insufficient, the representativeness of the mean decreases, the reliability of the standard deviation declines, and the calculation bias of the gradient increases. By setting a minimum data point threshold, it is ensured that each sub-interval has an effective statistical scale before entering step S105.

[0062] By assigning data points with x-coordinates equal to the target resonant point to the backward response sub-interval instead of the forward response sub-interval, consistency in the continuity processing at the segmentation boundaries is ensured. Since the interval average distortion value and interval distortion gradient value are calculated using all data points within the sub-interval in subsequent steps, the backward sub-interval, including the boundary points, can fully preserve the variation characteristics of the distortion rate after the peak point, avoiding information loss or duplicate calculations caused by unclear boundary point attribution.

[0063] The merging operation of the forward and backward response sub-intervals ensures a reasonable number of sub-intervals. When a sub-interval lacks sufficient data points, it is merged into another sub-interval, avoiding instability in statistical feature calculations caused by excessively small factor intervals, and preventing inconsistencies in the input data dimensions of the inertia identification model in subsequent step S105. The final output response sub-intervals are at least one in number and at most related to the number of candidate resonant speed points. Each sub-interval has sufficient sample density and a clear speed range division, providing high-quality sub-interval data for the accurate extraction of interval distortion feature vectors and accurate identification of load inertia in step S105.

[0064] Step S105: Based on the distortion gradient characteristics between elements in the at least one response sub-interval, a preset rotational inertia identification strategy is used to identify the load inertia of the permanent magnet synchronous motor, and the current load inertia identification value is obtained.

[0065] In this step, the arithmetic mean of the ordinate values ​​of all data points in a certain response sub-interval obtained after division is calculated as the interval average distortion value, the standard deviation of the ordinate values ​​of all data points in a certain sub-interval is calculated as the interval distortion fluctuation, and the quotient obtained by dividing the difference between the ordinate values ​​of the data points with the largest and smallest abscissa values ​​in a certain sub-interval by the difference between the abscissa values ​​is calculated as the interval distortion gradient value. The interval average distortion value, interval distortion volatility, and interval distortion gradient value of each response sub-interval are combined to form the distortion feature vector of the response sub-interval. All distortion feature vectors are arranged in ascending order according to the median value of the x-coordinate of the corresponding response sub-interval to obtain the distortion feature vector sequence. Calculate the vector magnitude of each distorted feature vector in the distorted feature vector sequence, mark the distorted feature vector with the largest vector magnitude as the main feature vector, and mark the response sub-interval where the main feature vector is located as the main response sub-interval. The interval average distortion value of the main response sub-interval is obtained as the main distortion reference value, the interval average distortion value of the adjacent response sub-interval before the main response sub-interval is obtained as the preceding distortion reference value, and the difference between the main distortion reference value and the preceding distortion reference value is calculated as the preceding-main interval distortion jump value. Calculate the ratio between the interval distortion gradient value of the main response sub-interval and the arithmetic mean of the interval distortion gradient values ​​of all response sub-intervals in the distortion feature vector sequence, and use the ratio as the significance of the main interval distortion gradient. When the preceding-main interval distortion jump value is greater than the preset jump threshold and the main interval distortion gradient significance is greater than the preset significance threshold, the inertia value corresponding to the main distortion reference value is read as the current load inertia identification value by referring to the preset distortion-inertia mapping curve based on the main distortion reference value. When the preceding-main interval distortion jump value is not greater than the jump threshold or the main interval distortion gradient significance is not greater than the significance threshold, the weighted average of the interval average distortion values ​​of all response sub-intervals is calculated. The weight is the proportion of the number of data points contained in each response sub-interval to the total number of data points. Based on the weighted average and the distortion-inertia mapping curve, the inertia value corresponding to the weighted average is read as the current load inertia identification value.

[0066] In one specific embodiment, all response sub-intervals obtained after the division in step S104 are acquired, and the following calculations are performed on each response sub-interval: For a given response sub-interval, extract all data points contained within that sub-interval. Let the sub-interval contain L data points, each containing an abscissa value x(l) and a ordinate value y(l), where l = 1, 2, ..., L. Calculate the arithmetic mean of the ordinate values ​​of all data points within the sub-interval to obtain the interval average distortion value.

[0067] The interval average distortion value characterizes the overall level of current harmonic distortion rate within the response sub-interval, reflecting the average influence of load inertia on current harmonic content within the speed range.

[0068] Calculate the standard deviation of the ordinate values ​​of all data points within the sub-interval to obtain the interval distortion volatility. The interval distortion volatility characterizes the degree of dispersion of the distortion rate data around its mean within the response sub-interval. The greater the volatility, the more drastic the change in distortion rate within the speed range, and the higher the sensitivity of the distortion rate to changes in speed. The smaller the volatility, the more stable the distortion rate within the interval.

[0069] Within this sub-interval, find the data point with the largest x-coordinate value and the data point with the smallest x-coordinate value, and denote them as x and x respectively. max and x min The corresponding ordinate value is denoted as y. max and y min The interval distortion gradient value D is obtained by dividing the difference in the ordinate between the data point with the largest and smallest abscissa by the difference in the abscissa. grad The calculation formula is: D grad =(y max -y min ) / (x max -x min ), The interval distortion gradient value characterizes the overall trend and rate of change of the distortion rate with rotational speed within the response sub-interval. A positive gradient value indicates that the distortion rate increases with increasing rotational speed, while a negative gradient value indicates that the distortion rate decreases with increasing rotational speed. The absolute value of the gradient reflects the degree of drastic change.

[0070] The interval average distortion value D is calculated for each response sub-interval. avg Interval distortion volatility D std and interval distortion gradient value D grad The distorted feature vector V=[D] is arranged in a preset order to form the response sub-interval. avg D std D grad The distortion feature vector comprehensively describes the distribution characteristics of the distortion rate within the response sub-interval from three dimensions: central tendency, dispersion, and trend of change.

[0071] Arrange the distorted feature vectors of all response sub-intervals in ascending order according to the median value of the corresponding x-coordinate of the sub-interval, to obtain the distorted feature vector sequence V. seq .

[0072] Calculate each distortion feature vector V in the distortion feature vector sequence. p For vectors (p=1,2,…,P), the distorted eigenvector with the largest magnitude is designated as the principal eigenvector, denoted as V. main The response sub-interval containing the main feature vector is marked as the main response sub-interval. The vector magnitude comprehensively reflects the overall magnitude of the average amplitude of the distortion rate, the degree of discrete fluctuation, and the magnitude of the change gradient within the sub-interval. The sub-interval with the largest magnitude is the one with the most significant distortion features.

[0073] The interval average distortion value of the main response sub-interval is obtained as the main distortion baseline value, denoted as D. main The average distortion value of the adjacent response sub-intervals preceding the main response sub-interval (i.e., the response sub-interval with the smallest difference where the median value of the x-coordinate is less than that of the main response sub-interval) is obtained as the preceding distortion baseline value, denoted as D. prev When the main response sub-interval is the first sub-interval in the distorted feature vector sequence, that is, when there is no preceding adjacent response sub-interval, the interval average distortion value of the main response sub-interval itself is used as the preceding distortion reference value.

[0074] Calculate the primary distortion baseline value D main Compared with the preceding distortion baseline value D prev The difference between them yields the preceding-principal interval distortion jump value, calculated using the following formula: J main =D main -D prev , The preceding-main interval distortion jump value characterizes the magnitude of the step change in the average distortion rate when entering the main response sub-interval from the upstream adjacent interval. The larger the jump value, the more significant the distortion level of the main response sub-interval has changed relative to its preceding interval, that is, a significant change in physical state has occurred within the rotational speed range covered by the sub-interval.

[0075] The ratio of the interval distortion gradient value of the main response sub-interval to the arithmetic mean of the interval distortion gradient values ​​of all response sub-intervals in the distortion feature vector sequence is used to obtain the significance of the main interval distortion gradient. The significance of the main interval distortion gradient characterizes the prominence of the rate of change of distortion rate with rotational speed within the main response sub-interval relative to the average rate of change of distortion rate across all sub-intervals. A significance greater than 1 indicates that the rate of change of distortion rate within this sub-interval is higher than the average level, and the distortion feature changes most dramatically within this interval.

[0076] The preceding-principal interval distortion jump value J main Compared with the preset jump threshold J threshold The comparison was performed, and the significance of the distortion gradient S in the main interval was also compared. main Compared with the preset significance threshold S threshold Comparison, jump threshold J threshold The significance threshold S is determined based on the statistical distribution of the distortion jump values ​​in the preceding-main interval obtained from multiple consecutive measurements of the permanent magnet synchronous motor under calibration conditions. Typically, it is taken as twice the mean jump value under calibration conditions. threshold The value is determined based on the statistical distribution of the significance of the main interval distortion gradient obtained from multiple consecutive measurements of the permanent magnet synchronous motor under calibration conditions, and is usually taken as 1.5.

[0077] Current order - main interval distortion jump value J main Greater than the preset jump threshold J threshold And the significance of the distortion gradient S in the main interval main Greater than the preset significance threshold S threshold When this occurs, it indicates that the main response sub-interval has experienced a significant step increase in distortion level relative to its preceding sub-interval, and the distortion rate within this sub-interval is significantly higher than the average level. In this case, the rotational speed range covered by the main response sub-interval is determined to be the concentrated region of resonance effect caused by load inertia changes, and its interval average distortion value best represents the distortion level corresponding to the current load inertia. At this time, based on the main distortion reference value D... main Compare with the preset distortion-inertia mapping curve, and read the value of D. main The corresponding inertia value is used as the current load inertia identification value J. load .

[0078] Current order - main interval distortion jump value J main Not greater than the jump threshold J threshold Or the significance of the distortion gradient S in the main interval main Not greater than the significance threshold S thresholdWhen the distortion level is not significantly different between the main response sub-interval and its preceding sub-interval, or the distortion rate within the main sub-interval is not significantly higher than the average level, the distortion characteristics of the main sub-interval are not highly representative. In this case, the weighted average distortion value of all response sub-intervals is used for inertia identification. The weighted average of the interval average distortion values ​​of all response sub-intervals is calculated, with the proportion of data points contained in each response sub-interval to the total number of data points as the weight. The proportion of data points in each response sub-interval to the total number of data points reflects the weight of that sub-interval within the overall scan range. More data points mean a wider rotational speed range covered by that sub-interval, and its contribution to the weighted average is greater. Based on the weighted average D... weighted Compare the distortion-inertia mapping curve with D. weighted The corresponding inertia value is used as the current load inertia identification value J. load .

[0079] The final current load inertia identification value J load The output is sent to step S106 for adjusting the proportional gain and integral time constant of the speed loop PI controller.

[0080] This step constructs a three-dimensional distortion feature vector containing the interval average distortion value, interval distortion volatility, and interval distortion gradient value. This comprehensively characterizes the distribution features of the current harmonic distortion rate within each response sub-interval from three dimensions: central tendency, dispersion, and trend. The distortion feature vector contains richer interval information than a single average distortion rate. The interval distortion volatility reflects the stability of the distortion rate data within the sub-interval, while the interval distortion gradient value reflects the dynamic trend of the distortion rate with rotational speed. These three dimensions complement each other, forming a complete description of the distortion characteristics of the response sub-intervals, providing a sufficient data foundation for subsequent selection of the principal feature vector and inertia identification.

[0081] By selecting the principal eigenvector and principal response sub-interval using the maximum vector magnitude criterion, the system can automatically identify the key interval with the most significant distortion characteristics when multiple response sub-intervals exist. When changes in load inertia cause significant mechanical resonance, the distortion rate near the resonance peak exhibits a high average value, large fluctuations, and a steep gradient descent. Its vector magnitude is the most prominent among all sub-intervals, and the maximum magnitude criterion can accurately pinpoint this key interval. When the distortion rate response curve is relatively flat, the differences in eigenvectors among the sub-intervals are not significant, and the representativeness of the principal eigenvector decreases. In this case, the system automatically switches to a weighted average mode through dual judgments of a jump threshold and a significance threshold, achieving adaptive selection of inertia identification strategies under different scenarios.

[0082] The comparison between the distortion jump value of the preceding-main interval and the preset jump threshold is used to determine whether the main response sub-interval has a significant step increase in distortion level relative to its preceding interval. The comparison between the distortion gradient significance of the main interval and the preset significance threshold is used to determine whether the distortion change rate within the main response sub-interval is significantly higher than the average level. The simultaneous satisfaction of the above two conditions means that the distortion characteristics within the sub-interval are the result of an abrupt change from the preceding interval to the current interval, rather than a slow and gradual trend. This abrupt change characteristic is physically consistent with the resonance effect caused by the change in load inertia and is a sign that the interval can effectively represent the current load inertia.

[0083] The distortion-inertia mapping curve is established through multiple known inertia experiments during the factory calibration phase. It reflects the correspondence between the distortion characteristics and inertia values ​​of each response sub-interval of a specific model of permanent magnet synchronous motor under different load inertia. During identification, the corresponding inertia value can be quickly obtained by querying the mapping curve based on the main distortion reference value or the weighted average value. It does not require iterative calculation, has high real-time performance, and is suitable for online implementation of embedded controllers.

[0084] Step S106: Adjust the proportional gain and integral time constant of the permanent magnet synchronous motor speed loop PI controller according to the current load inertia identification value.

[0085] In this step, the nominal value of the no-load inertia of the permanent magnet synchronous motor is obtained. The nominal value of the no-load inertia is the measured value of the rotor's own rotational inertia under the factory calibration state of the permanent magnet synchronous motor. Calculate the ratio between the current load inertia identification value and the no-load inertia nominal value, and use the ratio as the inertia ratio; Obtain the current proportional gain value and the current integral time constant value of the speed loop PI controller; When the inertia ratio is greater than 1, the proportional gain of the speed loop PI controller is adjusted to the current proportional gain value multiplied by the inertia ratio, and the integral time constant of the speed loop PI controller is adjusted to the current integral time constant value multiplied by the inertia ratio. When the inertia ratio is less than or equal to 1, the proportional gain of the speed loop PI controller is adjusted to the current proportional gain value, and the integral time constant of the speed loop PI controller is adjusted to the current integral time constant value.

[0086] In summary, the method of this application acquires the q-axis current feedback signal of a permanent magnet synchronous motor under multiple preset speed commands, extracts the current harmonic distortion rate corresponding to each preset speed command, constructs a speed-distortion response sequence based on each current harmonic distortion rate and its corresponding preset speed command value, identifies candidate resonant speed points based on the amplitude variation trend between adjacent elements in the sequence, determines whether the number of candidate resonant speed points is greater than a preset threshold, if not, extracts the target resonant point corresponding to the maximum distortion rate and divides the response sequence into multiple response sub-intervals based on this point, identifies the current load inertia based on the distortion gradient characteristics between elements in each sub-interval, and adjusts the proportional gain and integral time constant of the speed loop PI controller based on the identified current load inertia value. Utilizing the resonant characteristics of the current harmonic distortion rate changing with speed, online identification of load inertia and adaptive adjustment of PI parameters are achieved without additional signal injection, without interrupting normal motor operation, and have the advantages of non-disruptive identification and adaptive adjustment.

[0087] Please see Figure 2 The diagram shows a structural block diagram of a permanent magnet synchronous motor vector control system according to this application.

[0088] like Figure 2 As shown, the permanent magnet synchronous motor vector control system 200 includes an extraction module 210, an identification module 220, a judgment module 230, a division module 240, an identification module 250, and an adjustment module 260.

[0089] The extraction module 210 is configured to acquire the q-axis current feedback signal of the permanent magnet synchronous motor under multiple preset speed commands, and extract the current harmonic distortion rate corresponding to each preset speed command from each q-axis current feedback signal according to a preset amplitude-frequency feature extraction strategy; the identification module 220 is configured to construct a speed-distortion response sequence based on each current harmonic distortion rate and its corresponding preset speed command value, and identify at least one candidate resonant speed point based on the amplitude change trend between adjacent elements in the speed-distortion response sequence; the judgment module 230 is configured to determine whether the number of candidate resonant speed points is greater than a preset number threshold; the division module 2 40, configured to extract the target resonant point corresponding to the maximum distortion rate of change from the candidate resonant speed points if the number is not greater than a preset threshold, and divide the speed-distortion response sequence according to the target resonant point to obtain at least one response sub-interval; identification module 250, configured to identify the load inertia of the permanent magnet synchronous motor according to the distortion gradient characteristics between elements in the at least one response sub-interval, using a preset rotational inertia identification strategy, to obtain the current load inertia identification value; adjustment module 260, configured to adjust the proportional gain and integral time constant of the speed loop PI controller of the permanent magnet synchronous motor according to the current load inertia identification value.

[0090] It should be understood that Figure 2 The modules and references described in the document Figure 1 The steps described in the text correspond to those in the method described above. Therefore, the operations, features, and corresponding technical effects described above also apply to the method described in the text. Figure 2 The various modules in the document will not be described in detail here.

[0091] In other embodiments, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor performs the permanent magnet synchronous motor vector control method in any of the above method embodiments. In one embodiment, the computer-readable storage medium of the present invention stores computer-executable instructions, which are configured as follows: The q-axis current feedback signal of the permanent magnet synchronous motor under multiple preset speed commands is obtained, and the current harmonic distortion rate corresponding to each preset speed command is extracted from each q-axis current feedback signal according to the preset amplitude-frequency feature extraction strategy. Based on the current harmonic distortion rate and its corresponding preset speed command value, a speed-distortion response sequence is constructed, and at least one candidate resonant speed point is identified based on the amplitude variation trend between adjacent elements in the speed-distortion response sequence. Determine whether the number of candidate resonant speed points is greater than a preset threshold. If the number is not greater than a preset threshold, then the target resonance point corresponding to the maximum distortion rate is extracted from the candidate resonance speed points, and the speed-distortion response sequence is divided according to the target resonance point to obtain at least one response sub-interval. Based on the distortion gradient characteristics between elements within the at least one response sub-interval, a preset rotational inertia identification strategy is used to identify the load inertia of the permanent magnet synchronous motor, thereby obtaining the current load inertia identification value. Based on the current load inertia identification value, adjust the proportional gain and integral time constant of the permanent magnet synchronous motor speed loop PI controller.

[0092] Computer-readable storage media may include a stored program area and a stored data area, wherein the stored program area may store an operating system and an application program required for at least one function; the stored data area may store data created based on the use of the permanent magnet synchronous motor vector control system, etc. Furthermore, the computer-readable storage medium may include high-speed random access memory, and may also include memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, the computer-readable storage medium may optionally include memory remotely disposed relative to the processor, and these remote memories may be connected to the permanent magnet synchronous motor vector control system via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0093] Figure 3 This is a schematic diagram of the structure of the electronic device provided in the embodiment of the present invention, such as... Figure 3 As shown, the device includes a processor 310 and a memory 320. The electronic device may also include an input device 330 and an output device 340. The processor 310, memory 320, input device 330, and output device 340 can be connected via a bus or other means. Figure 3 Taking a bus connection as an example, the memory 320 is the computer-readable storage medium described above. The processor 310 executes various server functions and data processing by running non-volatile software programs, instructions, and modules stored in the memory 320, thereby implementing the permanent magnet synchronous motor vector control method described in the above embodiment. The input device 330 can receive input digital or character information and generate key signal inputs related to user settings and function control of the permanent magnet synchronous motor vector control system. The output device 340 may include a display screen or other display device.

[0094] The aforementioned electronic device can execute the method provided in the embodiments of the present invention, and has the corresponding functional modules and beneficial effects for executing the method. Technical details not described in detail in this embodiment can be found in the method provided in the embodiments of the present invention.

[0095] In one implementation, the above-described electronic device is used in a permanent magnet synchronous motor vector control system for a client application, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to: The q-axis current feedback signal of the permanent magnet synchronous motor under multiple preset speed commands is obtained, and the current harmonic distortion rate corresponding to each preset speed command is extracted from each q-axis current feedback signal according to the preset amplitude-frequency feature extraction strategy. Based on the current harmonic distortion rate and its corresponding preset speed command value, a speed-distortion response sequence is constructed, and at least one candidate resonant speed point is identified based on the amplitude variation trend between adjacent elements in the speed-distortion response sequence. Determine whether the number of candidate resonant speed points is greater than a preset threshold. If the number is not greater than a preset threshold, then the target resonance point corresponding to the maximum distortion rate is extracted from the candidate resonance speed points, and the speed-distortion response sequence is divided according to the target resonance point to obtain at least one response sub-interval. Based on the distortion gradient characteristics between elements within the at least one response sub-interval, a preset rotational inertia identification strategy is used to identify the load inertia of the permanent magnet synchronous motor, thereby obtaining the current load inertia identification value. Based on the current load inertia identification value, adjust the proportional gain and integral time constant of the permanent magnet synchronous motor speed loop PI controller.

[0096] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0097] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A vector control method for a permanent magnet synchronous motor, characterized in that, include: The q-axis current feedback signal of the permanent magnet synchronous motor under multiple preset speed commands is obtained, and the current harmonic distortion rate corresponding to each preset speed command is extracted from each q-axis current feedback signal according to the preset amplitude-frequency feature extraction strategy. Based on the current harmonic distortion rate and its corresponding preset speed command value, a speed-distortion response sequence is constructed, and at least one candidate resonant speed point is identified based on the amplitude variation trend between adjacent elements in the speed-distortion response sequence. Determine whether the number of candidate resonant speed points is greater than a preset threshold. If the number is not greater than a preset threshold, then the target resonance point corresponding to the maximum distortion rate is extracted from the candidate resonance speed points, and the speed-distortion response sequence is divided according to the target resonance point to obtain at least one response sub-interval. Based on the distortion gradient characteristics between elements within the at least one response sub-interval, a preset rotational inertia identification strategy is used to identify the load inertia of the permanent magnet synchronous motor, thereby obtaining the current load inertia identification value. Based on the current load inertia identification value, adjust the proportional gain and integral time constant of the permanent magnet synchronous motor speed loop PI controller.

2. The vector control method for a permanent magnet synchronous motor according to claim 1, characterized in that, The step of extracting the current harmonic distortion rate corresponding to each preset speed command from each q-axis current feedback signal according to the preset amplitude-frequency feature extraction strategy includes: Taking each of the plurality of preset speed commands as the current command, the current command is applied to the permanent magnet synchronous motor and a preset stable waiting time is continuously maintained. After the stable waiting time ends, the time-domain waveform data of the q-axis current feedback signal within the preset acquisition time is continuously acquired at a preset sampling frequency to obtain the q-axis current time-domain sequence corresponding to the current command. Using the fundamental frequency corresponding to the current command as a reference frequency, an adaptive notch filter is used to separate the fundamental component sequence and the residual component sequence from the q-axis current time domain sequence. The fundamental frequency is calculated based on the number of pole pairs of the permanent magnet synchronous motor and the rotational speed value of the current command. Using the inverter switching frequency and its integer multiples as the center frequency, the frequency band components within a preset bandwidth range at each center frequency are extracted from the residual component sequence and merged into an inverter switching frequency related component sequence. Using the rotor cogging harmonic frequency and its integer multiples as the center frequency, the frequency band components within a preset bandwidth range at each center frequency are extracted from the residual component sequence and merged into the rotor cogging harmonic component sequence. The remaining part of the residual component sequence after removing the inverter switching frequency related component sequence and the rotor cogging harmonic component sequence is taken as the broadband noise component sequence. Calculate the first effective value of the inverter switching frequency related component sequence, the second effective value of the rotor cogging harmonic component sequence, and the third effective value of the broadband noise component sequence, and simultaneously calculate the fourth effective value of the fundamental component sequence; The first distortion rate is obtained by dividing the first effective value of the inverter switching frequency related component sequence by the second effective value of the fundamental component sequence; the second effective value of the rotor cogging harmonic component sequence is obtained by dividing the fourth effective value of the fundamental component sequence; and the third effective value of the broadband noise component sequence is obtained by dividing the fourth effective value of the fundamental component sequence. Obtain the rotational speed value corresponding to the current instruction, and determine the first weighting coefficient, the second weighting coefficient, and the third weighting coefficient based on the rotational speed value; Calculate the sum of the first distortion factor multiplied by the first weighting coefficient, the second distortion factor multiplied by the second weighting coefficient, and the third distortion factor multiplied by the third weighting coefficient, and use the summation result as the current harmonic distortion factor corresponding to the current instruction; Iterate through all preset speed commands to obtain the current harmonic distortion rate corresponding to each preset speed command.

3. The vector control method for a permanent magnet synchronous motor according to claim 1, characterized in that, The step of constructing a speed-distortion response sequence based on each current harmonic distortion rate and its corresponding preset speed command value, and identifying at least one candidate resonant speed point based on the amplitude variation trend between adjacent elements in the speed-distortion response sequence, includes: Each preset speed command value is used as the horizontal axis value, and the current harmonic distortion rate corresponding to each preset speed command value is used as the vertical axis value. All data points are arranged in ascending order of horizontal axis values ​​to obtain the speed-distortion response sequence. The speed-distortion response sequence contains M data points arranged in ascending order of horizontal axis. Each data point contains a horizontal axis value and a vertical axis value, and M is the total number of preset speed commands. Starting from the second data point in the speed-distortion response sequence and ending at the last data point, the forward differential slope of each data point is calculated sequentially. All the calculated forward differential slopes are arranged in order of the horizontal coordinate to obtain the differential slope sequence. The forward differential slope is the difference between the vertical coordinate value of the current data point and the vertical coordinate value of the previous data point, divided by the difference between the horizontal coordinate value of the current data point and the horizontal coordinate value of the previous data point. Starting from the second differential slope in the differential slope sequence and ending at the last differential slope, determine sequentially whether the sign of the current differential slope is opposite to the sign of the previous differential slope; If the opposite is true and the sign of the current differential slope is negative, then mark the x-coordinate value of the current data point corresponding to the current differential slope as a candidate resonant speed point and add it to the candidate resonant speed point set; if the opposite is true and the sign of the current differential slope is positive, then do not mark it. Traverse the differential slope sequence and add all x-coordinate values ​​that satisfy the condition of opposite signs and negative current differential slope to the candidate resonant speed point set; Calculate the arithmetic mean of the ordinate values ​​of all candidate resonant speed points in the candidate resonant speed point set, and remove candidate resonant speed points whose ordinate values ​​are less than the arithmetic mean from the candidate resonant speed point set; When the difference between the abscissa values ​​of any two candidate resonant speed points in the candidate resonant speed point set is less than the preset merging distance threshold, the candidate resonant speed point with the larger ordinate value among the two candidate resonant speed points is retained, and the candidate resonant speed point with the smaller ordinate value among the two candidate resonant speed points is deleted. The set of candidate resonant speed points after vertical axis filtering and horizontal axis merging is used as the final identified candidate resonant speed points.

4. The vector control method for a permanent magnet synchronous motor according to claim 3, characterized in that, The step involves extracting the target resonance point corresponding to the maximum distortion rate from the candidate resonance speed points, and dividing the speed-distortion response sequence according to the target resonance point to obtain at least one response sub-interval, including: Obtain the set of candidate resonant speed points and determine whether the set of candidate resonant speed points is an empty set; When the set of candidate resonant speed points is empty, all data points in the speed-distortion response sequence are output as a response sub-interval; When the set of candidate resonant speed points is not empty, the differential slope value corresponding to each candidate resonant speed point in the set of candidate resonant speed points is extracted from the differential slope sequence. The absolute value of all extracted differential slope values ​​is calculated, and the candidate resonant speed point corresponding to the differential slope value with the largest absolute value is taken as the target resonant point. Using the x-coordinate value of the target resonance point as the dividing boundary, all data points in the speed-distortion response sequence whose x-coordinate value is less than that of the target resonance point are divided into forward response sub-intervals, all data points whose x-coordinate value is greater than that of the target resonance point are divided into backward response sub-intervals, and data points whose x-coordinate value is equal to that of the target resonance point are assigned to the backward response sub-intervals. The number of data points contained in each of the forward response sub-intervals and the backward response sub-intervals is counted separately. When the number of data points contained in the forward response sub-interval is less than the preset minimum data point threshold, the forward response sub-interval is merged into the backward response sub-interval and the forward response sub-interval is deleted. When the number of data points contained in the backward response sub-interval is less than the minimum data point number threshold, the backward response sub-interval is merged into the forward response sub-interval, and the backward response sub-interval is deleted. All remaining response sub-intervals after merging are output as the divided response sub-intervals, and each response sub-interval contains at least the minimum data point threshold number of data points.

5. The vector control method for a permanent magnet synchronous motor according to claim 1, characterized in that, The step of identifying the load inertia of the permanent magnet synchronous motor using a preset rotational inertia identification strategy based on the distortion gradient characteristics between elements within the at least one response sub-interval, and obtaining the current load inertia identification value, includes: The arithmetic mean of the ordinate values ​​of all data points in a certain response sub-interval obtained after division is used as the interval average distortion value. The standard deviation of the ordinate values ​​of all data points in a certain sub-interval is used as the interval distortion fluctuation. The quotient obtained by dividing the difference between the ordinate values ​​of the data points with the largest and smallest abscissa values ​​in a certain sub-interval by the difference between the abscissa values ​​is used as the interval distortion gradient value. The interval average distortion value, interval distortion volatility, and interval distortion gradient value of each response sub-interval are combined to form the distortion feature vector of the response sub-interval. All distortion feature vectors are arranged in ascending order according to the median value of the x-coordinate of the corresponding response sub-interval to obtain the distortion feature vector sequence. Calculate the vector magnitude of each distorted feature vector in the distorted feature vector sequence, mark the distorted feature vector with the largest vector magnitude as the main feature vector, and mark the response sub-interval where the main feature vector is located as the main response sub-interval. The interval average distortion value of the main response sub-interval is obtained as the main distortion reference value, the interval average distortion value of the adjacent response sub-interval before the main response sub-interval is obtained as the preceding distortion reference value, and the difference between the main distortion reference value and the preceding distortion reference value is calculated as the preceding-main interval distortion jump value. Calculate the ratio between the interval distortion gradient value of the main response sub-interval and the arithmetic mean of the interval distortion gradient values ​​of all response sub-intervals in the distortion feature vector sequence, and use the ratio as the significance of the main interval distortion gradient. When the preceding-main interval distortion jump value is greater than the preset jump threshold and the main interval distortion gradient significance is greater than the preset significance threshold, the inertia value corresponding to the main distortion reference value is read as the current load inertia identification value by referring to the preset distortion-inertia mapping curve based on the main distortion reference value. When the preceding-main interval distortion jump value is not greater than the jump threshold or the main interval distortion gradient significance is not greater than the significance threshold, the weighted average of the interval average distortion values ​​of all response sub-intervals is calculated. The weight is the proportion of the number of data points contained in each response sub-interval to the total number of data points. Based on the weighted average and the distortion-inertia mapping curve, the inertia value corresponding to the weighted average is read as the current load inertia identification value.

6. The vector control method for a permanent magnet synchronous motor according to claim 1, characterized in that, The step of adjusting the proportional gain and integral time constant of the permanent magnet synchronous motor speed loop PI controller based on the current load inertia identification value includes: Obtain the nominal value of the no-load inertia of the permanent magnet synchronous motor, wherein the nominal value of the no-load inertia is the measured value of the rotor's own rotational inertia under the factory calibration state of the permanent magnet synchronous motor. Calculate the ratio between the current load inertia identification value and the no-load inertia nominal value, and use the ratio as the inertia ratio; Obtain the current proportional gain value and the current integral time constant value of the speed loop PI controller; When the inertia ratio is greater than 1, the proportional gain of the speed loop PI controller is adjusted to the current proportional gain value multiplied by the inertia ratio, and the integral time constant of the speed loop PI controller is adjusted to the current integral time constant value multiplied by the inertia ratio. When the inertia ratio is less than or equal to 1, the proportional gain of the speed loop PI controller is adjusted to the current proportional gain value, and the integral time constant of the speed loop PI controller is adjusted to the current integral time constant value.

7. A vector control system for a permanent magnet synchronous motor, characterized in that, include: The extraction module is configured to acquire the q-axis current feedback signal of the permanent magnet synchronous motor under multiple preset speed commands, and extract the current harmonic distortion rate corresponding to each preset speed command from each q-axis current feedback signal according to the preset amplitude-frequency feature extraction strategy. The identification module is configured to construct a speed-distortion response sequence based on each current harmonic distortion rate and its corresponding preset speed command value, and to identify at least one candidate resonant speed point based on the amplitude variation trend between adjacent elements in the speed-distortion response sequence. The judgment module is configured to determine whether the number of candidate resonant speed points is greater than a preset number threshold. The partitioning module is configured to extract the target resonance point corresponding to the maximum distortion rate from the candidate resonance speed points if the number is not greater than a preset threshold, and partition the speed-distortion response sequence according to the target resonance point to obtain at least one response sub-interval. The identification module is configured to identify the load inertia of the permanent magnet synchronous motor based on the distortion gradient characteristics between elements in the at least one response sub-interval, using a preset rotational inertia identification strategy, and obtain the current load inertia identification value. The adjustment module is configured to adjust the proportional gain and integral time constant of the permanent magnet synchronous motor speed loop PI controller based on the current load inertia identification value.

8. An electronic device, characterized in that, include: At least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by a processor, it implements the method described in any one of claims 1 to 6.

Images