Control system and method suitable for direct current brushless motor

Abstract

The invention discloses a direct current brushless motor control system and method, and the system comprises an air gap data collection module which is used for obtaining air gap angle distribution data through a radial displacement collection device; the torque ripple characteristic analysis module is used for determining torque ripple characteristic data in different angle intervals according to the three-phase current data and a motor back electromotive force constant; the air gap state prediction module is used for predicting air gap angle distribution data in the corresponding section and determining the air gap state in the corresponding section; the control instruction correction module is used for predicting the torque correction amount of the corresponding section and generating motor control instruction correction data; and the correction effect evaluation module is used for comparing the variation of the torque ripple in each section before and after correction, and evaluating the suppression effect of the motor control instruction correction data on the torque ripple. According to the method, the torque ripple control of the brushless DC motor is converted from integral compensation to sectionalized fine control, and an effective technical means is provided for stable operation of the motor under complex working conditions.

Description

Technical Field

[0001] This invention relates to the field of mechatronics and condition monitoring technology, and in particular to a control system and method for a brushless DC motor. Background Technology

[0002] Brushless DC motors are widely used in industrial automation, electric vehicles, smart home appliances, and precision drives due to their high efficiency, high power density, and low maintenance costs. As application scenarios increasingly demand smooth operation, control precision, and reliability, the torque fluctuation problem generated by brushless DC motors during operation has gradually attracted attention. Torque fluctuations not only cause vibration and noise but can also lead to accelerated fatigue of mechanical components, decreased control precision, and in severe cases, even affect the long-term stable operation of the system. Research shows that the air gap state between the motor rotor and stator has a significant impact on the electromagnetic force distribution and torque output. Due to manufacturing errors, assembly deviations, shaft deformation, and structural changes caused by long-term operation, brushless DC motors often exhibit varying degrees of air gap non-uniformity during actual operation. Existing technologies for addressing the air gap state mostly focus on static detection or overall evaluation, lacking a description of the air gap state's characteristics changing with angle throughout the complete mechanical cycle, and especially lacking effective means to correlate the air gap state with torque fluctuations at the mechanical angle segment level. Furthermore, when modeling and predicting motor operating states using historical operating data or intelligent algorithms, it is often unavoidable to collect abnormal data with severe air gap morphology disturbances. If such abnormal data is directly used in model training, it can easily lead to model parameter shifts, reducing the stability and reliability of prediction results. Current technologies typically lack a method to identify and quantify the overall air gap consistency at the mechanical cycle level, making it difficult to effectively distinguish and process data with severe air gap morphology disturbances. Further, in terms of torque compensation strategies, existing methods often employ a uniform or fixed compensation approach, applying the same or approximately the same torque correction amount throughout the entire mechanical cycle. This approach does not fully consider the differences in air gap state and torque fluctuation characteristics within different mechanical angle segments, easily leading to insufficient or excessive compensation in local segments, making it difficult to achieve refined, segmented torque control. Therefore, there is an urgent need for a method that can establish the correlation between air gap state and torque fluctuation at the mechanical angle segment level, and on this basis, effectively filter abnormal air gap morphology data, while simultaneously achieving refined torque compensation control based on segmented air gap states, to improve the targeting, stability, and reliability of DC brushless motor torque control. Summary of the Invention

[0003] To address the problems existing in the prior art, the present invention provides a control system and method suitable for DC brushless motors.

[0004] The first aspect of this invention provides a control system for a brushless DC motor, mainly comprising: The air gap data acquisition module is used to acquire radial displacement data, three-phase current data and rotor mechanical angle data through the radial displacement acquisition device and motor operation control system of the brushless DC motor, and to map the radial displacement data to the corresponding angle range to obtain air gap angle distribution data. The torque fluctuation characteristic analysis module is used to calculate the electromagnetic torque at each time point based on the three-phase current data and the motor back EMF constant information, and to determine the torque fluctuation characteristic data in different angle intervals. The air gap state prediction module is used to predict the air gap angle distribution data in the corresponding section based on the section identification results, torque fluctuation amplitude sequence, air gap consistency fluctuation factor and motor operating parameter data, and to determine the air gap state in the corresponding section. The control command correction module is used to predict the torque correction amount of the corresponding section based on the section air gap state data and torque fluctuation characteristic data, and map the torque correction amount to the three-phase current correction amount to generate motor control command correction data. The correction effect evaluation module is used to load motor control command correction data through the motor drive system, compare the changes in torque fluctuations in each section before and after correction, and evaluate the suppression effect of motor control command correction data on torque fluctuations.

[0005] A second aspect of the present invention provides a control method for a brushless DC motor, mainly comprising: The radial displacement acquisition device and motor operation control system of the brushless DC motor are used to acquire radial displacement data, three-phase current data and rotor mechanical angle data, and the radial displacement data is mapped to the corresponding angle range to obtain air gap angle distribution data. Based on the three-phase current data and the motor back EMF constant information, the electromagnetic torque at each time point is calculated, and the torque fluctuation characteristic data in different angle intervals are determined. Based on the section identification results, torque fluctuation amplitude sequence, air gap consistency fluctuation factor and motor operating parameter data, predict the air gap angle distribution data in the corresponding section and determine the air gap state in the corresponding section. Based on the section air gap state data and torque fluctuation characteristic data, the torque correction amount of the corresponding section is predicted, and the torque correction amount is mapped to the three-phase current correction amount to generate motor control command correction data. By loading motor control command correction data into the motor drive system, the changes in torque fluctuations in each section before and after correction are compared to evaluate the suppression effect of the motor control command correction data on torque fluctuations.

[0006] Furthermore, the radial displacement acquisition device and motor operation control system of the brushless DC motor acquire radial displacement data, three-phase current data, and rotor mechanical angle data, and map the radial displacement data to the corresponding angle range to obtain air gap angle distribution data, including: Based on the structural dimensions of the inner circle of the stator and the outer circle of the rotor of the brushless DC motor, radial displacement acquisition devices are arranged at multiple preset angle positions in the circumferential direction of the stator to acquire radial displacement data corresponding to each angle during rotor rotation. Through the motor operation control system, three-phase current data and rotor mechanical angle data are acquired, and the radial displacement data, three-phase current data, and rotor mechanical angle data are time-stamped and synchronized based on a unified clock source. Based on the rotor mechanical angle data, the radial displacement data is mapped to the corresponding angle interval to form air gap angle distribution data, which includes radial displacement information under different mechanical angles. By filtering and fundamental frequency separation processing of the three-phase current data, a three-phase current dataset corresponding to the air gap angle distribution data is formed, and the air gap angle distribution data and three-phase current data are stored in the motor operation monitoring database.

[0007] Furthermore, the step of calculating the electromagnetic torque at each time point based on the three-phase current data and the motor back EMF constant information, and determining the torque fluctuation characteristic data within different angle intervals, includes: Based on the three-phase current data and the motor back EMF constant information, the electromagnetic torque at each time point is calculated using the equivalent torque calculation formula. The electromagnetic torque data is continuously sampled and numerically smoothed to remove high-frequency noise components. The electromagnetic torque data is reconstructed according to the rotor mechanical angle and synchronously mapped with the air gap angle distribution data to establish a one-to-one correspondence between torque fluctuation and air gap state. Statistical analysis methods are used to calculate the torque fluctuation amplitude in different angle intervals, forming torque fluctuation characteristic data in that angle interval, which is then stored in the motor operation monitoring database.

[0008] Further, the step of predicting the air gap angle distribution data within the corresponding section based on the section identification results, torque fluctuation amplitude sequence, air gap consistency fluctuation factor, and motor operating parameter data, and determining the air gap state within the corresponding section, includes: Based on the detection method of zero-crossing point and abrupt change in current slope of three-phase current, the timing of current phase switching is identified, and the identified commutation timing is mapped to the mechanical angle domain to determine the start and end angles of each commutation interval. Based on the identified commutation timing, within a complete mechanical cycle, the torque fluctuation characteristic data is divided into multiple commutation and non-commutation intervals. Historical data of torque fluctuation characteristic data, motor operating parameter data, and air gap angle distribution data are obtained from the motor operation monitoring database. Combined with rotor mechanical angle data, the air gap consistency fluctuation factor within the current mechanical cycle is determined, and data with severe air gap morphology disturbances are removed. For each... For a given angle interval, the corresponding torque fluctuation amplitude sequence is extracted. Based on the segment identification results, torque fluctuation amplitude sequence, air gap consistency fluctuation factor, motor operating parameter data, and historical data of air gap angle distribution, a recurrent neural network is used to train the model and construct a segment air gap state prediction model. Based on the real-time acquired segment identification, torque fluctuation amplitude sequence, air gap consistency fluctuation factor, and motor operating parameter data, an air gap dynamic prediction model is used to predict the air gap angle distribution data within the corresponding segment. Based on the predicted air gap angle distribution data, the segment air gap state prediction result is determined, including the mean, rate of change, and fluctuation range of the air gap deviation.

[0009] This also includes determining the air gap consistency fluctuation factor within the current mechanical cycle by combining rotor mechanical angle data, and removing data with severe air gap morphology disturbances, specifically including: Based on the real-time output data of each angular displacement sensor, an air gap angle distribution function synchronized with the rotor mechanical angle is established. Used to describe the rotor angles within a complete mechanical cycle. The corresponding air gap value changes, where the angle Let be the angular variable of the rotor over one complete mechanical cycle, defined in the interval [0, 2π]; based on the air gap angle distribution function. The air gap consistency fluctuation factor formula is used to calculate the average air gap value within the current period. Determine the air gap consistency fluctuation factor within the current period. The extreme point angle pairs are marked and stored in the motor operation monitoring database. and angles and angle The air gap value below, The average value of the air gap in the current period is used. If the air gap consistency fluctuation factor in the current period is greater than the preset factor threshold, it is determined that there is a serious air gap morphology disturbance in the current period, and the data with serious air gap morphology disturbance is removed from the training data.

[0010] Further, the step of predicting the torque correction amount for the corresponding section based on the section air gap state data and torque fluctuation characteristic data, and mapping the torque correction amount to the three-phase current correction amount to generate motor control command correction data includes: By analyzing the motor drive system's operating logs, we obtain segment air gap state data, torque fluctuation characteristic data, and corresponding torque correction data from historical torque correction events. A recurrent neural network is used for model training to construct a torque correction prediction model. Based on the current motor segment air gap state and real-time torque command, we predict the torque correction for the corresponding segment. According to the motor's electromagnetic characteristic parameters, we map the torque correction to a three-phase current correction and perform continuous processing on the three-phase current correction to eliminate numerical abrupt changes caused by segment switching. Finally, we combine the processed current corrections according to the segment sequence to generate motor control command correction data for a complete mechanical cycle.

[0011] Furthermore, the step of loading motor control command correction data through the motor drive system, comparing the changes in torque fluctuations in each segment before and after correction, and evaluating the suppression effect of the motor control command correction data on torque fluctuations includes: The corrected motor control command data is loaded into the motor drive system. Radial displacement data, three-phase current data, and rotor mechanical angle data are collected during the corrected operating state, and air gap angle distribution data and torque fluctuation data are obtained. The newly obtained air gap angle distribution data and torque fluctuation data are divided into segments, and the changes in torque fluctuation in each segment before and after correction are compared. By comparing the torque fluctuations in different segments, the suppression effect of the corrected motor control command data on torque fluctuation is evaluated. If the suppression effect does not meet expectations, the corrected motor control command data is adjusted until the suppression effect meets expectations.

[0012] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects: This invention provides a control system and method suitable for brushless DC motors. By jointly analyzing radial displacement, current, and rotor angle data at the mechanical angle segment level, this invention establishes a correlation between air gap angle distribution and torque fluctuation, transforming the analysis and control of torque fluctuations from a traditional holistic or time-domain approach to a segmented, angle-dependent, refined approach. By introducing an air gap consistency fluctuation factor, this invention identifies and eliminates abnormal data with severe air gap morphology disturbances at the mechanical cycle level, improving the stability and reliability of the air gap state prediction and torque correction model. Based on this, by combining the air gap state and torque fluctuation characteristics within different segments, the invention predicts segment-level torque correction amounts, generates corresponding motor control command correction data, and achieves closed-loop adjustment by comparing and evaluating the torque fluctuations of each segment before and after correction. This effectively improves the targeting, stability, and long-term operational reliability of brushless DC motor torque fluctuation suppression. The control method for brushless DC motors provided by this invention enables refined perception and segmented control of the motor's air gap state and torque fluctuations, significantly improving the targeting and stability of torque fluctuation suppression, thereby ensuring the smooth, efficient, and reliable operation of brushless DC motors under complex operating conditions. Attached Figure Description

[0013] Figure 1 This is a flowchart of a DC brushless motor control system according to the present invention; Figure 2 This is a flowchart of a control method for a brushless DC motor according to the present invention. Detailed Implementation

[0014] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0015] like Figure 1 This embodiment of a control system for a brushless DC motor may specifically include: The air gap data acquisition module is used to acquire radial displacement data, three-phase current data, and rotor mechanical angle data through the radial displacement acquisition device and motor operation control system of the brushless DC motor, and to map the radial displacement data to the corresponding angle range to obtain air gap angle distribution data.

[0016] The torque fluctuation characteristic analysis module is used to calculate the electromagnetic torque at each time point based on the three-phase current data and the motor back EMF constant information, and to determine the torque fluctuation characteristic data in different angle intervals.

[0017] The air gap state prediction module is used to predict the air gap angle distribution data in the corresponding section based on the section identification results, torque fluctuation amplitude sequence, air gap consistency fluctuation factor and motor operating parameter data, and to determine the air gap state in the corresponding section.

[0018] The control command correction module is used to predict the torque correction amount of the corresponding section based on the section air gap state data and torque fluctuation characteristic data, and to map the torque correction amount into the three-phase current correction amount to generate motor control command correction data.

[0019] The correction effect evaluation module is used to load motor control command correction data through the motor drive system, compare the changes in torque fluctuations in each section before and after correction, and evaluate the suppression effect of motor control command correction data on torque fluctuations.

[0020] like Figure 2 This embodiment provides a control method for a brushless DC motor, which may specifically include: Step S101: The radial displacement data, three-phase current data and rotor mechanical angle data are acquired through the radial displacement acquisition device and motor operation control system of the brushless DC motor, and the radial displacement data is mapped to the corresponding angle range to obtain the air gap angle distribution data.

[0021] Based on the structural dimensions of the stator inner circle and rotor outer circle of the brushless DC motor, radial displacement acquisition devices are arranged at multiple preset angle positions along the stator circumference to acquire radial displacement data corresponding to each angle during rotor rotation. Through the motor operation control system, three-phase current data and rotor mechanical angle data are acquired, and the radial displacement data, three-phase current data, and rotor mechanical angle data are timestamped and synchronized based on a unified clock source. Based on the rotor mechanical angle data, the radial displacement data is mapped to the corresponding angle interval to form air gap angle distribution data, which includes radial displacement information under different mechanical angles. By filtering and fundamental frequency separation of the three-phase current data, a three-phase current dataset corresponding to the air gap angle distribution data is constructed, and the air gap angle distribution data and three-phase current data are stored in the motor operation monitoring database.

[0022] For example, consider a DC brushless motor with a rated speed of 3000 rpm, a stator inner diameter of 100 mm, and a rotor outer diameter of 99.6 mm. Radial displacement acquisition devices are positioned at four angular locations along the stator circumference: 0°, 90°, 180°, and 270°, to measure the radial displacement change of the rotor relative to the stator. When the motor operates at a constant speed of 1500 rpm, the time required for the rotor to complete one revolution is approximately 40 ms. During this time, the radial displacement data measured by the device at the 0° position is... 0.015mm to The radial displacement varies between 0.010 mm, while the radial displacement data measured by the radial displacement acquisition device located at 180° varies between +0.020 mm and +0.025 mm, with the ideal center of the stator inner circle as the reference zero point. Simultaneously, three-phase current data is acquired through the motor operation control system, with the effective values ​​of phase A, B, and C currents being 5.2A, 5.0A, and 5.1A, respectively. Rotor angle data is acquired through a rotor mechanical angle encoder with a resolution of 0.1°. All radial displacement data, three-phase current data, and rotor mechanical angle data are timestamped based on the same system clock. For example, at timestamp 12.000 ms, the rotor mechanical angle is 135.0°, and the corresponding radial displacement data is... 0.012mm, +0.022mm, 0.014mm and +0.020mm. Based on the rotor mechanical angle data, the radial displacement data collected at each time point is mapped to the corresponding mechanical angle range. For example, radial displacement data in the range of 134.9° to 135.1° is assigned to the 135° angle range, thus forming air gap angle distribution data. The air gap angle distribution data represents the actual radial offset of the rotor relative to the stator at each mechanical angle position. When filtering the three-phase current data, high-frequency noise components above 1kHz are removed, and the fundamental components of the three-phase current are extracted to ensure consistency with the corresponding air gap angle distribution data in time and angle. Finally, the data record containing the mechanical angle, corresponding radial displacement value, and corresponding three-phase current fundamental components is stored in the motor operation monitoring database.

[0023] Step S102: Calculate the electromagnetic torque at each time point based on the three-phase current data and the motor back EMF constant information, and determine the torque fluctuation characteristic data within different angle intervals.

[0024] Based on three-phase current data and motor back EMF constant information, the electromagnetic torque at each time point is calculated using the equivalent torque calculation formula. The electromagnetic torque data is continuously sampled and numerically smoothed to remove high-frequency noise components. The electromagnetic torque data is reconstructed according to the rotor mechanical angle and synchronously mapped with the air gap angle distribution data to establish a one-to-one correspondence between torque fluctuations and air gap states. Statistical analysis methods are used to calculate the torque fluctuation amplitude within different angle intervals, forming torque fluctuation characteristic data for those angle intervals, which is then stored in the motor operation monitoring database.

[0025] For example, in a motor operation monitoring test, the sampling system continuously collected three-phase current data at a sampling frequency of 10kHz. At a certain moment t=0.105s, the measured three-phase currents were as follows: Phase A current was 8.2A, Phase B current was... 4.1A, C-phase current is At 4.1A, the three-phase current meets the current balance condition. The back EMF constant of the motor has been determined by factory calibration as Ke = 0.85 N·m / A. Using the equivalent torque calculation method, which equates the three-phase current to the effective current component that generates torque, and assuming constant flux linkage and ignoring the influence of magnetic saturation, the equivalent electromagnetic torque at this moment can be calculated as Te = 0.85 × 8.2 = 6.97 N·m. As time progresses, approximately 1000 discrete points of electromagnetic torque are continuously obtained within a time window of 0.1s to 0.2s. Due to the high-frequency noise introduced by current sampling and inverter switching operations, the original torque curve is superimposed with jitter components with an amplitude of approximately ±0.2 N·m and a frequency in the range of several kilohertz. Therefore, numerical smoothing with a sliding window width of 5ms is applied to this torque sequence to retain the low-frequency variation trend of the torque and filter out high-frequency disturbances unrelated to the inertia of the mechanical system, resulting in a smoothed electromagnetic torque curve whose effective fluctuation range converges to approximately ±0.05 N·m. Subsequently, based on the rotor mechanical angle information measured by the encoder, the torque data in the time domain is reconstructed into the angle domain. For example, if the rotor completes a full 360° rotation within this time period, each torque sampling point can be mapped to the corresponding rotor mechanical angle position. Simultaneously, the air gap angle distribution data has been obtained from finite element simulation or online magnetic field sensors and represented using the same angle reference. Therefore, under the same mechanical angle coordinates, the air gap eccentricity state at a certain angle, such as 45°, can be matched one-to-one with the corresponding electromagnetic torque value, thus establishing a direct correspondence between torque fluctuation and air gap state. Further, the entire 360° mechanical angle is divided into several angle intervals, such as 30° per interval. Statistical analysis is performed on the torque fluctuation amplitude within each interval. The maximum torque value is calculated to be 7.05 N·m and the minimum to be 6.90 N·m within the 0°–30° interval. Therefore, the torque fluctuation amplitude within this interval is defined as 0.15 N·m and used as the torque fluctuation characteristic quantity for this angle interval. After repeating the above calculations for all angle intervals, a complete set of angle interval-torque fluctuation characteristic data is formed. Finally, these characteristic data, along with timestamps, motor speeds, and operating condition identifiers, are stored in the motor operation monitoring database.

[0026] Step S103: Based on the section identification results, torque fluctuation amplitude sequence, air gap consistency fluctuation factor and motor operating parameter data, predict the air gap angle distribution data in the corresponding section and determine the air gap state in the corresponding section.

[0027] Based on the detection methods of zero-crossing points and sudden changes in current slope of three-phase currents, the timing of current phase switching is identified, and the identified commutation times are mapped to the mechanical angle domain to determine the starting and ending angles of each commutation interval. According to the identified commutation times, within a complete mechanical cycle, the torque fluctuation characteristic data is divided into multiple commutation and non-commutation intervals. Historical data on torque fluctuation characteristics, motor operating parameters, and air gap angle distribution are obtained from the motor operation monitoring database. Combined with rotor mechanical angle data, the air gap consistency fluctuation factor within the current mechanical cycle is determined, and data with severe air gap morphology disturbances are removed. For each angle interval, the corresponding torque fluctuation amplitude sequence is extracted. Based on the segment identification results, torque fluctuation amplitude sequence, air gap consistency fluctuation factor, motor operating parameter data, and historical air gap angle distribution data, a recurrent neural network is used for model training to construct a segment air gap state prediction model. Based on the real-time acquired section identifiers, torque fluctuation amplitude sequences, air gap consistency fluctuation factors, and motor operating parameter data, the air gap dynamic prediction model is used to predict the air gap angle distribution data within the corresponding section. Based on the predicted air gap angle distribution data, the predicted air gap state of the section is determined, including the mean, rate of change, and fluctuation range of the air gap deviation.

[0028] For example, during the operation of a brushless DC motor with a rated speed of 1500 r / min, the monitoring system collects three-phase current signals at a sampling frequency of 20 kHz. By jointly detecting the zero-crossing point and the sudden change in current slope of phase A current, three obvious phase switching moments are identified within a certain mechanical cycle, occurring at times of 0.012 s, 0.024 s, and 0.036 s, respectively. The moment when the phase A current changes from positive to negative and the absolute value of the slope suddenly increases from 120 A / s to 480 A / s is used as the commutation criterion. Combining the rotor mechanical angle information provided by the encoder, after mapping the above commutation moments to the mechanical angle domain, the three commutation intervals can be determined to correspond to 58°–62°, 178°–182°, and 298°–302°, respectively. The remaining angle ranges are defined as non-commutation intervals. Subsequently, within the complete 360° mechanical cycle, the obtained torque fluctuation characteristic data was segmented. For example, in the non-commutation intervals of 0°–58°, 62°–178°, 182°–298°, and 302°–360°, smoothed torque fluctuation amplitude sequences were extracted, with values ​​concentrated in the range of 0.08–0.15 N·m. However, in the commutation intervals, due to the influence of current reconstruction and magnetic field redistribution, the torque fluctuation amplitude significantly increased, reaching a maximum of 0.32 N·m. Next, historical data from the past 30 days under the same speed of 1500 r / min and similar load rated torque of approximately 7 N·m were retrieved from the motor operation monitoring database. This data included torque fluctuation characteristics, motor stator temperature (approximately 65°C), bus voltage (540V), and air gap angle distribution information. After aligning with the current rotor mechanical angle, the air gap consistency fluctuation factor within the current mechanical cycle was calculated to be 0.08. This factor characterizes the overall deviation of the air gap angle distribution from the historical normal state. Since the air gap consistency fluctuation factor is 0.18, which is less than the system's preset air gap consistency fluctuation factor threshold of 0.30, it is considered that there is no serious air gap morphology disturbance. Therefore, the data for this period is deemed valid and retained. However, some previous consistency fluctuation factors reached 0.42, which is greater than the preset air gap consistency fluctuation factor threshold of 0.30. Therefore, the historical period data for these periods was automatically removed and not included in subsequent modeling. Subsequently, each 30° mechanical angle was used as an analysis segment. Within the 120°–150° segment, the torque fluctuation amplitude sequence corresponding to this segment was extracted. The average value was 0.11 N·m, and the standard deviation was 0.02 N·m. The commutation identification results, torque fluctuation amplitude sequence, air gap consistency fluctuation factor of 0.18, motor operating parameters, and historical air gap angle distribution data for this segment were used as input samples and fed into a recurrent neural network for model training to construct a segment air gap state prediction model.After model training is completed, during real-time operation, when the real-time torque fluctuation amplitude sequence of the same 120°–150° segment within the current cycle is obtained, and the average value slightly increases to 0.14 N·m, the consistency fluctuation factor increases to 0.26, and the corresponding operating parameters are obtained, the established segment air gap dynamic prediction model is used to predict the air gap angle distribution in this segment. The results show that the average air gap deviation in this segment is 0.21 mm, which is significantly larger than the historical normal average of 0.12 mm. Its rate of change is about 0.015 mm / °, and the fluctuation range reaches ±0.06 mm. Based on this, the system determines that there is a gradually aggravating local air gap eccentricity trend in this segment, and outputs and stores the air gap state prediction results of this segment as key diagnostic information.

[0029] Specifically, by combining rotor mechanical angle data, the air gap consistency fluctuation factor within the current mechanical cycle is determined, and data with severe air gap morphology disturbances are removed.

[0030] Based on the real-time output data of each angular displacement sensor, an air gap angle distribution function synchronized with the rotor mechanical angle is established. Used to describe the rotor angles within a complete mechanical cycle. The corresponding air gap value changes, where the angle Let be the angular variable of the rotor over one complete mechanical cycle, defined in the interval [0, 2π]. Based on the air gap angle distribution function... The air gap consistency fluctuation factor formula is used to calculate the average air gap value within the current period. Determine the air gap consistency fluctuation factor within the current period. The extreme point angle pairs are marked and stored in the motor operation monitoring database. and angles and angle The air gap value below, This represents the average air gap value within the current period. If the air gap consistency fluctuation factor within the current period exceeds a preset threshold, it is determined that there is a severe air gap morphology disturbance within that period, and data with severe air gap morphology disturbances are removed from the training data.

[0031] For example, during the online operation monitoring of a brushless DC motor, six circumferentially arranged radial displacement sensors collected air gap data synchronized with the rotor's mechanical angle in real time throughout a complete mechanical cycle, and constructed an air gap angle distribution function based on this data. ,in The angle of the rotor within one mechanical cycle is defined in the interval [0, 2π]. Within this cycle, the average air gap value for the current cycle is obtained by statistically calculating the air gap data at all angular positions. It is 0.50mm, while at the angle 60 The air gap value measured at the location It is 0.62mm, at the angle 210 The air gap value measured at the location The value is 0.38 mm. These two angles correspond to a maximum and a minimum point in the air gap distribution within this period. The formula for the air gap consistency fluctuation factor is used. Determine the air gap consistency fluctuation factor within the current period. The calculated air gap consistency fluctuation factor is 0.24. If the preset air gap consistency fluctuation factor threshold is 0.30, and the calculated air gap consistency fluctuation factor is 0.24, which does not exceed the threshold, it is determined that the overall air gap distribution within this mechanical cycle still maintains good consistency. The data for this cycle is retained and used for training the air gap state prediction model for subsequent sections. However, if the calculated air gap consistency fluctuation factor in other cycles is greater than the preset air gap consistency fluctuation factor threshold, it is determined that there is a significant structural disturbance between some angle pairs in the air gap within this cycle. This is manifested as excessive air gap differences and significant angle disturbance factor gains, which poses a physical risk of causing high-frequency torque fluctuations or commutation mismatch. Therefore, the complete data sample within the current cycle will be marked as a high-disturbance cycle and removed from the subsequent air gap prediction model training set to avoid this abnormal disturbance pattern misleading the model parameter learning. At the same time, the angle pairs where the extreme points are located are marked and stored in the motor operation monitoring database for subsequent abnormal trend tracking and control command fine-tuning reference.

[0032] Step S104: Based on the section air gap state data and torque fluctuation characteristic data, predict the torque correction amount for the corresponding section, map the torque correction amount to the three-phase current correction amount, and generate motor control command correction data.

[0033] By analyzing the motor drive system's operating logs, segment air gap state data, torque fluctuation characteristic data, and corresponding torque correction amount data from historical torque correction events are obtained. A recurrent neural network is used for model training to construct a torque correction amount prediction model. Based on the current motor segment air gap state and real-time torque command, the torque correction amount for the corresponding segment is predicted. According to the motor's electromagnetic characteristic parameters, the torque correction amount is mapped to a three-phase current correction amount, and the three-phase current correction amount is continuously processed to eliminate numerical abrupt changes caused by segment switching. The processed current correction amounts are combined according to the segment sequence to generate motor control command correction data for a complete mechanical cycle.

[0034] For example, during the long-term operation of a DC brushless motor with a rated torque of 8 N·m and a rated speed of 1500 r / min, the motor drive system operation log records a large amount of historical torque correction event data. This includes segment air gap status data divided into sections of 30° mechanical angle, torque fluctuation characteristic data within the corresponding sections, and the actual effective torque correction amount at that time. Taking one type of historical data as an example, when the rotor mechanical angle is within the 120°–150° range, the historical air gap status shows that the average air gap deviation in this section is 0.18 mm, with a fluctuation range of ±0.05 mm. The corresponding torque fluctuation characteristic data shows that the torque fluctuates around 6.8 N·m within this section, with a maximum value of 6.95 N·m and a minimum value of 6.65 N·m, and the torque fluctuation amplitude is approximately 0.30 N·m. Under these operating conditions, to suppress torque fluctuations, the historical control strategy actually applied a torque correction of +0.45 N·m, meaning that the target torque in this section was corrected from the original command of 6.8 N·m to approximately 7.25 N·m. Using air gap state parameters, torque fluctuation characteristics, and corresponding torque corrections for sections similar to those described above as input samples, a recurrent neural network was used for model training. This enabled the model to learn the temporal relationship between the air gap state and the torque correction requirement as the section changes, thus constructing a torque correction prediction model. After model training was completed, in a new operating cycle, when real-time monitoring showed that the average air gap deviation in the same 120°–150° section increased to 0.22 mm and the fluctuation range expanded to ±0.07 mm, while the real-time torque command was 7.0 N·m and the predicted amplitude of the torque fluctuation characteristics was approximately 0.35 N·m, the torque correction prediction model output that the required torque correction for this section was +0.6 N·m. Subsequently, based on the motor's electromagnetic characteristic parameters, such as a torque constant of 0.9 N·m / A, the control system converts the torque correction into a three-phase equivalent current correction, yielding a required current increment of approximately 0.67 A. Since the predicted current corrections for adjacent sections are 0.55 A and 0.60 A, respectively, to avoid introducing sudden current changes during section switching, the three-phase current correction is processed continuously, ensuring a smooth transition within the 120°–150° range. Finally, the continuously processed current corrections for each section are combined in a 0°–360° sequence to form the motor control command correction data for the complete mechanical cycle. This correction data is then superimposed on the original current command, achieving continuous compensation and stable control of torque fluctuations throughout the entire mechanical cycle. This also includes calculating the extreme value of the air gap consistency fluctuation factor for each motor operating cycle, marking the corresponding angle pair, and cross-comparing the angle range covered by this pair with the identified commutation angle intervals in the current cycle. If the extreme value of the air gap consistency fluctuation factor is found to significantly overlap with the reversal operation area, it is considered that there is a high risk of disturbance within that cycle.For this high-disturbance angle segment, an enhanced control strategy is triggered, executing control command updates at a higher frequency or with smaller angle steps within this segment. Simultaneously, when generating the control command correction sequence, the current correction within this segment is enhanced, or the timing of the correction response is shifted forward to more effectively combat torque fluctuations caused by magnetic disturbances or commutation mismatch. If, over multiple consecutive cycles, GCUF extreme points are found to cluster near a fixed angle segment, and the disturbance amplitude remains high for an extended period, historical fluctuation data for this region is extracted, the disturbance trend sequence is reconstructed, and this structural anomaly is fed back to the control correction model training module for targeted parameter optimization, thereby improving the model's prediction accuracy and control stability in this segment.

[0035] Step S105: Load motor control command correction data through the motor drive system, compare the changes in torque fluctuations in each section before and after correction, and evaluate the suppression effect of motor control command correction data on torque fluctuations.

[0036] The corrected motor control command data is loaded into the motor drive system. Radial displacement data, three-phase current data, and rotor mechanical angle data are collected during the corrected operating state, and air gap angle distribution data and torque fluctuation data are obtained. The newly obtained air gap angle distribution data and torque fluctuation data are divided into segments, and the changes in torque fluctuation within each segment before and after correction are compared. By comparing the torque fluctuations in different segments, the suppression effect of the corrected motor control command data on torque fluctuation is evaluated. If the suppression effect does not meet expectations, the corrected motor control command data is adjusted until the suppression effect meets expectations.

[0037] For example, after generating a round of motor control command correction data, the control system loads this correction data into the motor drive system, causing the motor to enter the correction operation state while maintaining a speed of 1500 r / min and a load torque of approximately 7 N·m. In this state, the radial displacement sensor collects data with the same sampling configuration as before, obtaining new air gap angle distribution data. The results show that the air gap value is mainly distributed between 0.19–0.28 mm within a complete 360° mechanical cycle, which is significantly converged compared to the distribution range of 0.15–0.35 mm before correction. At the same time, new torque fluctuation data is calculated using three-phase current data and back EMF constant. Subsequently, the air gap angle distribution data and torque fluctuation data under the corrected operating state were processed according to the established segmentation rules. For example, if each 30° mechanical angle is still considered as a segment, in the 120°–150° segment before correction, the torque fluctuation amplitude was 0.35 N·m, the maximum torque was approximately 7.18 N·m, and the minimum torque was approximately 6.83 N·m. After applying the correction control command, the recalculated maximum torque in the same segment was approximately 7.05 N·m, and the minimum torque was approximately 6.92 N·m, with the corresponding torque fluctuation amplitude decreasing to 0.13 N·m. In the adjacent 150°–180° segment, the torque fluctuation amplitude also decreased from 0.28 N·m before correction to 0.12 N·m. A comparison of the torque fluctuation amplitudes in each section before and after correction revealed that the torque fluctuation decreased by more than 50% in most sections. However, in the 300°–330° section, the torque fluctuation remained at 0.22 N·m, only slightly improved from the pre-correction 0.26 N·m, falling below the pre-set evaluation standard of reducing the torque fluctuation to below 0.15 N·m. Therefore, it was determined that the suppression effect in this section did not meet expectations. Based on this evaluation result, the control system readjusted the motor control command correction data for the 300°–330° section. For example, the current correction amount for this section was increased by approximately 0.15 A, and the current transition slope at the section boundary was re-smoothed. Subsequently, the updated control command was reloaded, and the above acquisition and evaluation process was repeated until the recalculated torque fluctuation amplitude in this section decreased to 0.14 N·m, meeting the expected suppression standard. This completed the closed-loop verification and adjustment of the motor control command correction data.

[0038] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the concept of this application. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.

Claims

1. A control system for a brushless DC motor, characterized in that, The system includes: The air gap data acquisition module is used to acquire radial displacement data, three-phase current data and rotor mechanical angle data through the radial displacement acquisition device and motor operation control system of the brushless DC motor, and to map the radial displacement data to the corresponding angle range to obtain air gap angle distribution data. The torque fluctuation characteristic analysis module is used to calculate the electromagnetic torque at each time point based on the three-phase current data and the motor back EMF constant information, and to determine the torque fluctuation characteristic data in different angle intervals. The air gap state prediction module is used to predict the air gap angle distribution data in the corresponding section based on the section identification results, torque fluctuation amplitude sequence, air gap consistency fluctuation factor and motor operating parameter data, and to determine the air gap state in the corresponding section. The control command correction module is used to predict the torque correction amount of the corresponding section based on the section air gap state data and torque fluctuation characteristic data, and map the torque correction amount to the three-phase current correction amount to generate motor control command correction data. The correction effect evaluation module is used to load motor control command correction data through the motor drive system, compare the changes in torque fluctuations in each section before and after correction, and evaluate the suppression effect of motor control command correction data on torque fluctuations.

2. A method for controlling a brushless DC motor, applied to a brushless DC motor control system as described in claim 1, characterized in that, The method includes: The radial displacement acquisition device and motor operation control system of the brushless DC motor are used to acquire radial displacement data, three-phase current data and rotor mechanical angle data, and the radial displacement data is mapped to the corresponding angle range to obtain air gap angle distribution data. Based on the three-phase current data and the motor back EMF constant information, the electromagnetic torque at each time point is calculated, and the torque fluctuation characteristic data in different angle intervals are determined. Based on the section identification results, torque fluctuation amplitude sequence, air gap consistency fluctuation factor and motor operating parameter data, predict the air gap angle distribution data in the corresponding section and determine the air gap state in the corresponding section. Based on the section air gap state data and torque fluctuation characteristic data, the torque correction amount of the corresponding section is predicted, and the torque correction amount is mapped to the three-phase current correction amount to generate motor control command correction data. By loading motor control command correction data into the motor drive system, the changes in torque fluctuations in each section before and after correction are compared to evaluate the suppression effect of the motor control command correction data on torque fluctuations.

3. The method according to claim 2, wherein, The radial displacement acquisition device and motor operation control system of the brushless DC motor acquire radial displacement data, three-phase current data, and rotor mechanical angle data, and map the radial displacement data to the corresponding angle range to obtain air gap angle distribution data, including: Based on the structural dimensions of the stator inner circle and rotor outer circle of the brushless DC motor, radial displacement acquisition devices are arranged at multiple preset angle positions in the stator circumference to acquire radial displacement data corresponding to each angle during rotor rotation. Through the motor operation control system, three-phase current data and rotor mechanical angle data are acquired, and the radial displacement data, three-phase current data, and rotor mechanical angle data are time-stamped and synchronized based on a unified clock source. Based on the rotor mechanical angle data, the radial displacement data is mapped to the corresponding angle interval to form air gap angle distribution data, which includes radial displacement information under different mechanical angles. By filtering and fundamental frequency separation processing of the three-phase current data, a three-phase current dataset corresponding to the air gap angle distribution data is formed, and the air gap angle distribution data and three-phase current data are stored in the motor operation monitoring database.

4. The method according to claim 2, wherein, The process involves calculating the electromagnetic torque at each time point based on three-phase current data and motor back EMF constant information, and determining torque fluctuation characteristic data within different angle intervals, including: Based on the three-phase current data and the motor back EMF constant information, the electromagnetic torque at each time point is calculated using the equivalent torque calculation formula. The electromagnetic torque data is continuously sampled and numerically smoothed to remove high-frequency noise components. The electromagnetic torque data is reconstructed according to the rotor mechanical angle and synchronously mapped with the air gap angle distribution data to establish a one-to-one correspondence between torque fluctuation and air gap state. Statistical analysis methods are used to calculate the torque fluctuation amplitude in different angle intervals, forming torque fluctuation characteristic data in that angle interval, which is then stored in the motor operation monitoring database.

5. The method according to claim 2, wherein, The step of predicting the air gap angle distribution data within the corresponding section based on the section identification results, torque fluctuation amplitude sequence, air gap consistency fluctuation factor, and motor operating parameter data, and determining the air gap state within the corresponding section, includes: Based on the detection method of zero-crossing point and abrupt change in current slope of three-phase current, the timing of current phase switching is identified, and the identified commutation timing is mapped to the mechanical angle domain to determine the start and end angles of each commutation interval. Based on the identified commutation timing, within a complete mechanical cycle, the torque fluctuation characteristic data is divided into multiple commutation and non-commutation intervals. Historical data of torque fluctuation characteristic data, motor operating parameter data, and air gap angle distribution data are obtained from the motor operation monitoring database. Combined with rotor mechanical angle data, the air gap consistency fluctuation factor within the current mechanical cycle is determined, and data with severe air gap morphology disturbances are removed. For each... For a given angle interval, the corresponding torque fluctuation amplitude sequence is extracted. Based on the segment identification results, torque fluctuation amplitude sequence, air gap consistency fluctuation factor, motor operating parameter data, and historical data of air gap angle distribution, a recurrent neural network is used to train the model and construct a segment air gap state prediction model. Based on the real-time acquired segment identification, torque fluctuation amplitude sequence, air gap consistency fluctuation factor, and motor operating parameter data, an air gap dynamic prediction model is used to predict the air gap angle distribution data within the corresponding segment. Based on the predicted air gap angle distribution data, the segment air gap state prediction result is determined, including the mean, rate of change, and fluctuation range of the air gap deviation.

6. The method according to claim 5, wherein, The process involves combining rotor mechanical angle data to determine the air gap consistency fluctuation factor within the current mechanical cycle, and removing data exhibiting severe air gap morphology disturbances, including: Based on the real-time output data of each angular displacement sensor, an air gap angle distribution function synchronized with the rotor mechanical angle is established. Used to describe the rotor angles within a complete mechanical cycle. The corresponding air gap value changes, where the angle Let be the angular variable of the rotor over one complete mechanical cycle, defined in the interval [0, 2π]; based on the air gap angle distribution function. The air gap consistency fluctuation factor formula is used to calculate the average air gap value within the current period. Determine the air gap consistency fluctuation factor within the current cycle. The extreme point angle pairs are marked and stored in the motor operation monitoring database. and angles and angle The air gap value below, The average value of the air gap in the current period is used. If the air gap consistency fluctuation factor in the current period is greater than the preset factor threshold, it is determined that there is a serious air gap morphology disturbance in the current period, and the data with serious air gap morphology disturbance is removed from the training data.

7. The method according to claim 2, wherein, The step of predicting the torque correction amount for the corresponding section based on the section air gap state data and torque fluctuation characteristic data, and mapping the torque correction amount to the three-phase current correction amount to generate motor control command correction data includes: By analyzing the motor drive system's operating logs, we obtain segment air gap state data, torque fluctuation characteristic data, and corresponding torque correction data from historical torque correction events. A recurrent neural network is used for model training to construct a torque correction prediction model. Based on the current motor segment air gap state and real-time torque command, we predict the torque correction for the corresponding segment. According to the motor's electromagnetic characteristic parameters, we map the torque correction to a three-phase current correction and perform continuous processing on the three-phase current correction to eliminate numerical abrupt changes caused by segment switching. Finally, we combine the processed current corrections according to the segment sequence to generate motor control command correction data for a complete mechanical cycle.

8. The method according to claim 2, wherein, The process of loading motor control command correction data through the motor drive system, comparing the changes in torque fluctuations in each segment before and after correction, and evaluating the suppression effect of the motor control command correction data on torque fluctuations includes: The corrected motor control command data is loaded into the motor drive system. Radial displacement data, three-phase current data, and rotor mechanical angle data are collected during the corrected operating state, and air gap angle distribution data and torque fluctuation data are obtained. The newly obtained air gap angle distribution data and torque fluctuation data are divided into segments, and the changes in torque fluctuation in each segment before and after correction are compared. By comparing the torque fluctuations in different segments, the suppression effect of the corrected motor control command data on torque fluctuation is evaluated. If the suppression effect does not meet expectations, the corrected motor control command data is adjusted until the suppression effect meets expectations.

Images