Parameter self-tuning control method of permanent magnet synchronous motor

Abstract

The invention provides a parameter self-tuning control method for a permanent magnet synchronous motor, and relates to the technical field of synchronous motor control, and the method comprises the steps: firstly, monitoring a dynamic operation data set, including stator current, rotor flux linkage, rotating speed fluctuation quantity and load change characteristic parameters, of the permanent magnet synchronous motor in real time; secondly, feature relevance mapping is carried out based on multiple parameter combinations in the dynamic operation data set, initial parameter adjustment sets are generated, and each initial parameter adjustment combination comprises the mapping relation between the motor operation state and the target response index; dividing a plurality of parameter adjustment subsets according to the parameter correlation degree, configuring a dynamic error range, extracting a positive adjustment parameter from the first subset corresponding to the first operation state, and extracting a negative adjustment parameter from the second subset with the low correlation degree; and finally, training a dynamic parameter adjustment model by utilizing the positive adjustment parameters and the negative adjustment parameters, generating an optimized real-time parameter setting instruction set, realizing intelligent self-setting of the parameters of the permanent magnet synchronous motor, and improving the running performance of the motor.

Description

Technical Field

[0001] The present invention relates to the technical field of synchronous motor control, and in particular to a control method for self-tuning parameters of a permanent magnet synchronous motor. Background Art

[0002] Permanent magnet synchronous motors have been widely used in many fields such as industrial automation, electric vehicles, aerospace, etc. due to their advantages of high efficiency, energy saving, and high power density. However, the performance of permanent magnet synchronous motors is highly dependent on their precise parameter settings, and the accuracy of motor parameters directly affects the operating efficiency, stability, and control accuracy of the motor.

[0003] In the prior art, the parameter setting of permanent magnet synchronous motors usually adopts relatively traditional and fixed methods. On the one hand, some methods mainly rely on manual experience to manually adjust the motor parameters. Operators need to rely on their long-term accumulated experience to set parameters according to the approximate operating conditions of the motor. This method is not only inefficient, but also requires a very high level of professionalism from operators. The parameters adjusted by different operators vary greatly, making it difficult to ensure that the motor is always in the best operating state. In addition, when the operating conditions of the motor change, manual adjustments often fail to respond in time, resulting in a decrease in motor performance.

[0004] On the other hand, some existing technologies use a method of presetting fixed parameter values, and do not dynamically adjust the parameters during the operation of the motor. This static parameter setting method cannot adapt to the operation requirements of the motor under different load, speed and other conditions. For example, when the load changes suddenly or the speed needs to be adjusted quickly, the motor with fixed parameters may experience response delays, increased oscillations, or even loss of step, which seriously affects the normal operation of the motor and the stability of the system.

[0005] There are also some parameter adjustment methods based on simple models. Although they can make limited adjustments based on some single parameters of the motor (such as monitoring only the stator current), they ignore the correlation and coupling relationship between multiple parameters during the operation of the motor. The operation of a permanent magnet synchronous motor is a complex process in which multiple parameters interact with each other. Factors such as stator current, rotor flux, speed fluctuation, and load change interact with each other. Considering only a single parameter cannot fully reflect the actual operating status of the motor, and it is difficult to achieve accurate parameter adjustment. Summary of the invention

[0006] In view of the above-mentioned problems, in combination with the first aspect of the present invention, an embodiment of the present invention provides a control method for self-tuning parameters of a permanent magnet synchronous motor, the method comprising:

[0007] Real-time monitoring of the dynamic operation data set of the permanent magnet synchronous motor, wherein the dynamic operation data set includes stator current, rotor flux, speed fluctuation and load change characteristic parameters;

[0008] Performing feature correlation mapping based on multiple parameter combinations in the dynamic operation data set to generate an initial parameter adjustment set, wherein each parameter combination includes a mapping relationship between the motor operation state and the target response index;

[0009] Divide a plurality of parameter adjustment subsets according to the parameter correlation in the initial parameter adjustment set, configure a dynamic error range for each parameter adjustment subset, and extract at least one positive adjustment parameter from the first parameter adjustment subset corresponding to the first operating state, wherein the positive adjustment parameter satisfies a response convergence condition within the dynamic error range;

[0010] Extracting at least one negative adjustment parameter from a second parameter adjustment subset, wherein a parameter correlation between the second parameter adjustment subset and the first parameter adjustment subset is lower than a preset correlation threshold;

[0011] The dynamic parameter adjustment model is trained using the plurality of the first operating states and the corresponding positive adjustment parameters and negative adjustment parameters to generate an optimized real-time parameter setting instruction set.

[0012] On the other hand, an embodiment of the present invention also provides a control system for self-tuning parameters of a permanent magnet synchronous motor, including a processor and a machine-readable storage medium, wherein the machine-readable storage medium is connected to the processor, the machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

[0013] Based on the above aspects, the embodiment of the present application first monitors the comprehensive dynamic operation data set of the permanent magnet synchronous motor in real time, covering stator current, rotor flux, speed fluctuation and load change characteristic parameters. Compared with the traditional monitoring method that only focuses on a few single parameters, it can capture the operating status of the motor under various complex working conditions in an all-round and accurate manner, providing a rich and detailed data basis for subsequent parameter adjustment, greatly improving the ability to understand and grasp the operating status of the motor, and helping to promptly discover potential problems and make targeted adjustments.

[0014] The method of generating the initial parameter adjustment set by mapping the feature correlation of multiple parameter combinations in the dynamic operation data set breaks through the limitations of the traditional fixed parameter mapping mode. The unique mapping relationship between the motor operating state and the target response index contained in each parameter combination enables the initial parameter adjustment set to flexibly adapt to the diverse needs of the motor under different operating states, innovatively constructing a parameter adjustment space that is more in line with the actual operating laws of the motor, and providing a data basis for the subsequent fine adjustment of parameters.

[0015] The strategy of dividing multiple parameter adjustment subsets according to the parameter correlation in the initial parameter adjustment set and configuring a dynamic error range for each subset is extremely creative. This method changes the previous setting mode of a unified and fixed error range, and can dynamically and accurately define the error range of each subset according to the intrinsic connection between the parameters and the actual operating characteristics, thereby achieving differentiated and personalized control of parameter adjustment in different operating scenarios. In particular, the positive adjustment parameters that meet the response convergence conditions within the dynamic error range are extracted from the first parameter adjustment subset corresponding to the first operating state, ensuring that the parameter adjustment under a specific operating state can be carried out in the direction of optimizing motor performance and fast convergence response, effectively improving the stability and response speed of the motor in this state.

[0016] Negative adjustment parameters are extracted from the second parameter adjustment subset whose correlation with the parameters of the first parameter adjustment subset is lower than a preset correlation threshold. This cross-subset parameter extraction method based on correlation difference can comprehensively consider the interaction between the motor operating parameters and adjust the parameters from different dimensions, avoiding the one-sidedness and limitations that may be caused by relying solely on a single subset parameter adjustment, further enriching the strategies and means of parameter adjustment, and achieving more comprehensive and detailed optimization of motor parameters.

[0017] Finally, the dynamic parameter adjustment model is trained using multiple first operating states, corresponding positive adjustment parameters and negative adjustment parameters to generate an optimized real-time parameter adjustment instruction set, giving the motor adaptive and intelligent parameter adjustment capabilities. The dynamic parameter adjustment model can continuously learn and optimize parameter adjustment strategies according to the changing operating states and parameter adjustment requirements, and generate accurate instruction sets in real time, so that the permanent magnet synchronous motor can maintain optimal operating performance under different working conditions, significantly improving the control accuracy, efficiency and reliability of the motor. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of the execution flow of the control method for self-tuning parameters of a permanent magnet synchronous motor provided in an embodiment of the present invention.

[0019] Figure 2 It is a schematic diagram of the hardware architecture of a control system for self-tuning parameters of a permanent magnet synchronous motor provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0020] The present invention will be described in detail below with reference to the accompanying drawings. Figure 1 1 is a flow chart of a method for controlling the parameters of a permanent magnet synchronous motor provided by an embodiment of the present invention. The method for controlling the parameters of a permanent magnet synchronous motor provided by an embodiment of the present invention is described in detail below.

[0021] Step S110, real-time monitoring of a dynamic operation data set of the permanent magnet synchronous motor, wherein the dynamic operation data set includes stator current, rotor flux, speed fluctuation and load change characteristic parameters.

[0022] In this embodiment, a permanent magnet synchronous motor in an industrial production workshop is considered, which is used to drive a conveyor belt. The permanent magnet synchronous motor runs continuously to ensure that the conveyor belt stably transports production materials at a predetermined speed. In this scenario, the real-time monitoring system starts to work. The stator current is an important electrical parameter of the motor operation, which is measured by a high-precision current sensor installed on the stator winding of the motor. For example, during the normal operation of the motor, the current sensor continuously collects data on the stator current, and a series of current values ​​may be obtained. These values ​​change dynamically with factors such as slight changes in the motor load and fluctuations in the power supply voltage. The monitoring of the rotor flux is assisted by a special flux observer. Since the size and direction of the rotor flux have an important influence on the torque generation and operating efficiency of the motor, the flux observer accurately measures and records the relevant data of the rotor flux. For example, when the load of the conveyor belt changes, the amplitude and phase of the rotor flux may change accordingly, and the flux observer can capture these changes in time. The monitoring of the speed fluctuation is also critical. In the conveyor belt system, the speed sensor installed on the motor shaft is responsible for measuring the speed of the motor. The conveyor belt may cause fluctuations in motor load due to factors such as uneven distribution of materials and changes in friction of mechanical transmission parts, which in turn causes fluctuations in speed. The speed sensor will promptly feed back the information of these speed fluctuations to the monitoring system. In terms of load change characteristic parameters, the load changes are obtained by installing load sensors in the conveyor belt system or indirectly by using the input power of the motor. For example, when the conveyor belt begins to load materials, the load gradually increases, and the load sensor will detect characteristic parameters such as the rate of increase of the load and the peak value of the load. These stator current, rotor flux, speed fluctuation and load change characteristic parameters constitute a complete dynamic operation data set, which provides basic data support for subsequent motor performance optimization and control.

[0023] Step S120, performing feature correlation mapping based on multiple parameter combinations in the dynamic operation data set to generate an initial parameter adjustment set, wherein each parameter combination includes a mapping relationship between the motor operation state and the target response index.

[0024] Let's continue to take the conveyor belt permanent magnet synchronous motor in the industrial production workshop as an example. For each set of collected parameter combinations, it contains the operating state of the motor at a certain moment (for example, the stator current, rotor flux and speed fluctuation under a specific load) and the corresponding target response indicators (such as the output torque and efficiency of the motor). First, the following processing is performed for each set of parameter combinations: extract the stator current fluctuation sequence, rotor flux phase offset and speed tracking error curve in the parameter combination. For example, in a certain period of time, the stator current may fluctuate due to a small fluctuation in the power supply voltage or an instantaneous change in the load, forming a current fluctuation sequence. The extraction of the rotor flux phase offset is based on the data of the flux observer. When the motor is subject to external interference or its own performance changes, the phase of the rotor flux may shift relative to the ideal state. The speed tracking error curve reflects the deviation between the actual speed of the motor and the set speed. Calculate the time domain coupling coefficient of the stator current fluctuation sequence and the rotor flux phase offset, as well as the frequency domain correlation index of the speed tracking error curve and the load change characteristic parameter. Assuming that at a certain moment, the stator current fluctuation sequence shows a certain periodicity, and the rotor flux phase offset also has a corresponding change in the same period, the time domain coupling coefficient between them is calculated by a specific algorithm. For the speed tracking error curve and the load change characteristic parameter, when the load changes, it may cause the motor speed to fluctuate, thereby showing a certain correlation in the frequency domain, and the frequency domain correlation index is obtained by analysis and calculation. The parameter combination that meets the coupling coefficient threshold and the frequency domain correlation threshold is mapped as a candidate element of the initial parameter adjustment set. For example, if the calculated time domain coupling coefficient of the stator current fluctuation sequence and the rotor flux phase offset exceeds the preset coupling coefficient threshold, and the frequency domain correlation index of the speed tracking error curve and the load change characteristic parameter also meets the frequency domain correlation threshold, then this set of parameter combinations is considered to be an effective combination related to motor performance optimization, and is mapped as a candidate element of the initial parameter adjustment set. This initial parameter adjustment set contains multiple such screened parameter combinations, which provides a basis for the subsequent parameter adjustment subset division.

[0025] Step S130, dividing a plurality of parameter adjustment subsets according to the parameter correlation in the initial parameter adjustment set, configuring a dynamic error range for each parameter adjustment subset, and extracting at least one positive adjustment parameter from the first parameter adjustment subset corresponding to the first operating state, wherein the positive adjustment parameter satisfies a response convergence condition within the dynamic error range.

[0026] In the scenario of a conveyor belt permanent magnet synchronous motor, each parameter combination in the initial parameter adjustment set is processed as follows to divide the parameter adjustment subset. A three-dimensional parameter space is constructed based on the stator current response delay time, the rotor flux steady-state error rate, and the speed overshoot. For example, the stator current response delay time reflects how fast the stator current responds when the motor changes load or the command changes; the rotor flux steady-state error rate reflects the deviation ratio of the rotor flux from the ideal value under stable operation; the speed overshoot indicates the amplitude of the actual speed exceeding the set speed during the motor startup or speed adjustment process. The stator current response delay time, the rotor flux steady-state error rate, and the speed overshoot in the three-dimensional parameter space are normalized to generate a three-dimensional coordinate data set with a unified dimension standard. It is assumed that after normalization, each data point corresponds to the coordinates of a set of parameter combinations in the three-dimensional space. The three-dimensional coordinate data set is traversed based on a preset neighborhood radius threshold to calculate the local density value of each data point. For example, with a certain data point as the center, within the spherical area defined by the neighborhood radius threshold, the total number of data points contained is counted, and this total number of data points is the local density value of the data point. A density distribution map is generated according to the local density value, core data points whose local density values ​​are greater than the first density threshold are screened, and a spatial connectivity relationship between the core data points is established. In this process, if the Euclidean distance between two core data points is less than the neighborhood radius threshold, they are considered to have a spatial connectivity relationship. The core data points with direct or indirect spatial connectivity are aggregated into continuous high-density parameter areas, and isolated data points whose local density values ​​are less than the second density threshold are removed. For example, some isolated data points with weak correlation with other data points are removed, while core data points with strong correlation form different high-density parameter areas. Each high-density parameter area is extended outward to the boundary data points, and the judgment conditions of these boundary data points are that the distance from the core data point is less than the neighborhood radius threshold and is not classified into any high-density parameter area. A spatial partition mapping table is constructed based on the expanded high-density parameter area, which records the high-density parameter area number to which each data point belongs and the correlation strength coefficient with the core data point. According to the spatial division mapping table, the parameter combinations belonging to the same high-density parameter area number are clustered into parameter adjustment subsets, and a dynamic weight factor is added to each parameter adjustment subset. For example, if the average local density value of all data points in a parameter adjustment subset is high, and the correlation strength coefficient between the data points and the core data points is large, then the dynamic weight factor of this parameter adjustment subset will be large. Secondary clustering is performed on the overlapping boundary areas between adjacent high-density parameter areas. The judgment condition for secondary clustering is that there is a parameter combination in the boundary area and the difference in the correlation strength coefficients of the two high-density parameter areas is less than a preset overlap threshold. The parameter combination of the boundary area is allocated to the adjacent parameter adjustment subsets according to the correlation strength coefficient ratio.The corresponding parameter association calculation weight is adjusted according to the dynamic weight factor of each parameter adjustment subset to generate a priority sequence.

[0027] For the first parameter adjustment subset corresponding to the first operating state, positive adjustment parameters are extracted from it. For example, in the first operating state where the conveyor belt motor operates normally and the load is relatively stable, the parameter combination that meets the following conditions is screened in the first parameter adjustment subset: the amplitude change rate of the stator current fluctuation sequence is less than or equal to the first amplitude threshold. Assuming that the first amplitude threshold is set to a certain specific value, when the amplitude change rate of the stator current fluctuation sequence is within this threshold range, it means that the fluctuation of the stator current is in an acceptable small range. The cumulative error of the rotor flux phase offset is lower than the phase error threshold. For example, if the cumulative error of the rotor flux phase offset during long-term operation is lower than the set phase error threshold, it indicates that the operating state of the rotor flux is relatively stable. The target response index in the screened parameter combination is used as a positive adjustment parameter, and these positive adjustment parameters can make the motor response converge in the optimization direction within the dynamic error range.

[0028] Step S140: extracting at least one negative adjustment parameter from a second parameter adjustment subset, wherein a parameter correlation between the second parameter adjustment subset and the first parameter adjustment subset is lower than a preset correlation threshold.

[0029] Still taking the conveyor belt permanent magnet synchronous motor as an example, among the multiple parameter adjustment subsets divided, the second parameter adjustment subset has a low correlation with the parameters of the first parameter adjustment subset. Perform the following processing on the parameter combination in the second parameter adjustment subset to extract the negative adjustment parameter. Calculate the difference between the stator current fluctuation sequence of the parameter combination and the current fluctuation of the first parameter adjustment subset. For example, during the operation of the conveyor belt motor, due to some special circumstances (such as local failure of the motor or strong external interference), the stator current fluctuation sequence in the second parameter adjustment subset is greatly different from the relatively stable current fluctuation in the first parameter adjustment subset, and this difference is calculated by a specific algorithm. Detect whether the oscillation frequency of the speed tracking error curve of the parameter combination exceeds the preset frequency range. Assuming that the preset frequency range is set according to the speed fluctuation during normal operation of the motor, if the oscillation frequency of the speed tracking error curve of a certain parameter combination in the second parameter adjustment subset exceeds this range, it means that the speed control of the motor under this parameter combination is abnormal. If the current fluctuation difference is greater than the difference threshold and the oscillation frequency exceeds the preset frequency range, the target response index of the parameter combination is marked as a negative adjustment parameter. These negative adjustment parameters indicate that under this parameter combination, the motor's operating state develops in a direction that is not conducive to performance optimization, which may, for example, lead to reduced motor efficiency, unstable output torque, etc.

[0030] Step S150, using the plurality of the first operating states, the corresponding positive adjustment parameters and the negative adjustment parameters, to train a dynamic parameter adjustment model, and generate an optimized real-time parameter setting instruction set.

[0031] In the scenario of a conveyor belt permanent magnet synchronous motor, the first operating state is input into the first coding layer of the dynamic parameter adjustment model to generate a state feature vector. For example, information such as the stator current, rotor flux, speed fluctuation and load change characteristics in the first operating state are encoded into a specific state feature vector, which can fully describe the operating characteristics of the motor in the first operating state. The positive adjustment parameter is input into the second coding layer to generate a positive adjustment vector, and the negative adjustment parameter is input into the third coding layer to generate a negative adjustment vector. For example, the motor output torque optimization value and efficiency improvement value in the positive adjustment parameter are encoded as a positive adjustment vector, and the torque instability factor and efficiency reduction factor in the negative adjustment parameter are encoded as a negative adjustment vector. The similarity weight between the state feature vector and the positive adjustment vector, as well as the repulsion weight between the state feature vector and the negative adjustment vector are calculated by the contrast loss function. In this process, the contrast loss function calculates the weight according to factors such as the distance and correlation between the vectors. The parameter space distribution of the dynamic parameter adjustment model is updated according to the similarity weight and the repulsion weight to generate an optimized real-time parameter setting instruction set. For example, if the similarity weight between the state feature vector and the forward adjustment vector is high, it means that the operating state of the motor in this case is more consistent with the forward adjustment direction. The model will adjust the parameter space distribution according to this weight, so that the generated real-time parameter tuning instruction set is more inclined to optimize the performance of the motor.

[0032] After generating the optimized real-time parameter setting instruction set, the real-time parameter setting instruction set is loaded into the parameter adjustment module of the motor controller. For example, in the control system of the conveyor belt motor, the parameter adjustment module of the motor controller receives these instruction sets. The closed-loop parameters of the current loop, speed loop and flux loop of the permanent magnet synchronous motor are injected through the parameter adjustment module. For example, for the current loop, the proportional integral parameter of the current loop is adjusted according to the current loop proportional integral coefficient adjustment instruction in the real-time parameter setting instruction set; for the speed loop, the switching threshold of the sliding mode controller is updated according to the speed loop sliding mode controller switching threshold update instruction; for the flux loop, the gain of the flux observer is adjusted according to the flux observer gain adaptive adjustment instruction. The motor operation response data after parameter injection is collected in real time and fed back to the dynamic parameter adjustment model for incremental training. For example, after the parameters are injected, the operation data of the motor such as the stator current, rotor flux, speed fluctuation amount, etc. are continuously monitored, and these data are fed back to the dynamic parameter adjustment model. The model further optimizes its own parameters according to the new data to adapt to various changes in the motor operation process, thereby continuously improving the performance and operation efficiency of the motor.

[0033] Based on the above steps, the embodiment of the present application first monitors the permanent magnet synchronous motor's comprehensive dynamic operation data set in real time, covering stator current, rotor flux, speed fluctuation and load change characteristic parameters. Compared with the traditional monitoring method that only focuses on a few single parameters, it can capture the motor's operating status under various complex working conditions in an all-round and accurate manner, providing a rich and detailed data basis for subsequent parameter adjustment, greatly improving the ability to understand and grasp the motor's operating conditions, and helping to promptly discover potential problems and make targeted adjustments.

[0034] The method of generating the initial parameter adjustment set by mapping the feature correlation of multiple parameter combinations in the dynamic operation data set breaks through the limitations of the traditional fixed parameter mapping mode. The unique mapping relationship between the motor operating state and the target response index contained in each parameter combination enables the initial parameter adjustment set to flexibly adapt to the diverse needs of the motor under different operating states, innovatively constructing a parameter adjustment space that is more in line with the actual operating laws of the motor, and providing a data basis for the subsequent fine adjustment of parameters.

[0035] The strategy of dividing multiple parameter adjustment subsets according to the parameter correlation in the initial parameter adjustment set and configuring a dynamic error range for each subset is extremely creative. This method changes the previous setting mode of a unified and fixed error range, and can dynamically and accurately define the error range of each subset according to the intrinsic connection between the parameters and the actual operating characteristics, thereby achieving differentiated and personalized control of parameter adjustment in different operating scenarios. In particular, the positive adjustment parameters that meet the response convergence conditions within the dynamic error range are extracted from the first parameter adjustment subset corresponding to the first operating state, ensuring that the parameter adjustment under a specific operating state can be carried out in the direction of optimizing motor performance and fast convergence response, effectively improving the stability and response speed of the motor in this state.

[0036] Negative adjustment parameters are extracted from the second parameter adjustment subset whose correlation with the parameters of the first parameter adjustment subset is lower than a preset correlation threshold. This cross-subset parameter extraction method based on correlation difference can comprehensively consider the interaction between the motor operating parameters and adjust the parameters from different dimensions, avoiding the one-sidedness and limitations that may be caused by relying solely on a single subset parameter adjustment, further enriching the strategies and means of parameter adjustment, and achieving more comprehensive and detailed optimization of motor parameters.

[0037] Finally, the dynamic parameter adjustment model is trained using multiple first operating states, corresponding positive adjustment parameters and negative adjustment parameters to generate an optimized real-time parameter adjustment instruction set, giving the motor adaptive and intelligent parameter adjustment capabilities. The dynamic parameter adjustment model can continuously learn and optimize parameter adjustment strategies according to the changing operating states and parameter adjustment requirements, and generate accurate instruction sets in real time, so that the permanent magnet synchronous motor can maintain optimal operating performance under different working conditions, significantly improving the control accuracy, efficiency and reliability of the motor.

[0038] In a possible implementation, step S120 includes:

[0039] The following processing is performed for each parameter combination:

[0040] Step S121, extracting the stator current fluctuation sequence, the rotor flux phase offset and the speed tracking error curve in the parameter combination.

[0041] In the scenario of a conveyor belt permanent magnet synchronous motor in an industrial production workshop, for each set of parameter combinations, the stator current fluctuation sequence, rotor flux phase offset, and speed tracking error curve must first be extracted. For example, during the operation of the conveyor belt, the loading and unloading of materials is not completed instantly and is a continuously changing process, which will cause the motor load to change continuously. The extraction of the stator current fluctuation sequence is based on the data collected by the stator current sensor. For example, when the conveyor belt starts to load materials, the load gradually increases, the motor needs to provide a larger torque, and the stator current will rise accordingly; when the material is unloaded, the load is reduced and the current will drop again. The change of these current values ​​over time forms a stator current fluctuation sequence. The acquisition of the rotor flux phase offset depends on a high-precision flux observer. As the motor load changes, the magnetic field distribution inside the motor will also change, and the phase of the rotor flux may deviate from the ideal state. For example, after the motor has been running for a long time, the temperature increase may affect the electromagnetic characteristics of the motor, resulting in a rotor flux phase offset. The flux observer can accurately measure this offset. The speed tracking error curve reflects the deviation between the actual motor speed and the set speed. In the conveyor belt system, a specific speed value is set to ensure that the material is transported at the predetermined speed. However, due to factors such as changes in motor load and power supply voltage fluctuations, the actual speed will fluctuate around the set speed, and these fluctuations constitute the speed tracking error curve.

[0042] Step S122, calculating the time domain coupling coefficient between the stator current fluctuation sequence and the rotor flux phase offset, and the frequency domain correlation index between the speed tracking error curve and the load variation characteristic parameter.

[0043] When calculating the time domain coupling coefficient of the stator current fluctuation sequence and the rotor flux phase offset, the electromagnetic coupling relationship of the motor needs to be taken into account. For example, the fluctuation of the stator current will cause the change of the magnetic field, which in turn affects the state of the rotor flux. The stator current fluctuation sequence and the rotor flux phase offset within a period of time are analyzed and calculated through a specific mathematical algorithm. When a peak value in the stator current fluctuation sequence is associated with a specific change point of the rotor flux phase offset in time, this association is reflected in the time domain coupling coefficient. For the calculation of the frequency domain correlation index of the speed tracking error curve and the load change characteristic parameter, since the load change will affect the speed of the motor, a specific frequency characteristic may be shown in the frequency domain. For example, when the load of the conveyor belt changes at a certain frequency period, the speed tracking error curve of the motor will also be shown at the corresponding frequency. Through the frequency domain analysis technology, the frequency correlation between the speed tracking error curve and the load change characteristic parameter is analyzed, and the frequency domain correlation index is calculated.

[0044] Step S123: Mapping parameter combinations that satisfy the coupling coefficient threshold and the frequency domain correlation threshold as candidate elements of the initial parameter adjustment set.

[0045] For example, in this conveyor belt permanent magnet synchronous motor system, the coupling coefficient threshold and frequency domain correlation threshold are pre-set. For example, these thresholds are determined based on the motor design specifications, operating requirements, and previous empirical data. If the time domain coupling coefficient of the stator current fluctuation sequence and the rotor flux phase offset calculated by a certain set of parameter combinations exceeds the set coupling coefficient threshold, and the frequency domain correlation index of the speed tracking error curve and the load change characteristic parameter also meets the frequency domain correlation threshold, this indicates that this set of parameter combinations is of great significance in terms of the motor's operating status and performance optimization. This set of parameter combinations will be mapped as candidate elements of the initial parameter adjustment set, and these candidate elements will be further used for subsequent parameter adjustment subset division and other operations, providing basic data support for the efficient and stable operation of permanent magnet synchronous motors in conveyor belt systems.

[0046] In a possible implementation, step S130 includes:

[0047] The following processing is performed on each parameter combination in the initial parameter adjustment set:

[0048] Step S131, constructing a three-dimensional parameter space based on the stator current response delay time, the rotor flux steady-state error rate and the speed overshoot.

[0049] In this embodiment, during the operation of the aforementioned conveyor belt, the stator current response delay time is a key parameter. When the load of the conveyor belt changes, for example, a batch of new materials is suddenly loaded onto the conveyor belt, the motor needs to adjust the output torque accordingly. At this time, the stator current should respond in time to meet the new load requirements. If the stator current response delay time is too long, the speed of the conveyor belt may fluctuate, affecting production efficiency. The rotor flux steady-state error rate reflects the deviation of the rotor flux from the ideal value under stable operation. Due to the complex electromagnetic relationship inside the motor, the rotor flux may not be maintained in an ideal state due to factors such as temperature and magnetic field interference. For example, as the motor runs for a long time, the temperature increase may cause the electromagnetic characteristics of the motor to change, thereby causing an error between the actual value and the ideal value of the rotor flux. The speed overshoot indicates the amplitude by which the actual speed exceeds the set speed during the motor startup or speed adjustment process. When the conveyor belt starts, the motor accelerates from a stationary state to a predetermined speed. If the speed overshoot is too large, it may cause an impact on the mechanical structure of the conveyor belt and affect its service life. Using these three parameters as coordinate axes, a three-dimensional parameter space is constructed, and each parameter combination has a corresponding position in this space.

[0050] Step S132, normalizing the stator current response delay time, the rotor flux steady-state error rate and the speed overshoot in the three-dimensional parameter space to generate a three-dimensional coordinate data set with a unified dimension standard, wherein each data point corresponds to the coordinates of a set of parameter combinations in the three-dimensional space.

[0051] For example, in actual monitoring data, the stator current response delay time may be in seconds, the rotor flux steady-state error rate is a unitless proportional value, and the speed overshoot may be in revolutions per minute (rpm). In order to be able to analyze and process under the same mathematical model, these parameters of different dimensions need to be normalized. For example, the stator current response delay time is mapped to a specific interval, such as [0,1], through a specific mathematical formula; for the rotor flux steady-state error rate, since it is in the [0,1] interval (indicating the deviation ratio from the ideal value), it can be used directly; for the speed overshoot, it is also converted to the [0,1] interval through the corresponding algorithm. In this way, each data point corresponds to a set of coordinates of a parameter combination in three-dimensional space, and these coordinates have a unified dimension standard, which is convenient for subsequent calculation and analysis.

[0052] Step S133, traverse the three-dimensional coordinate data set based on a preset neighborhood radius threshold, and calculate the local density value of each data point, where the local density value is the total number of data points contained in a spherical area defined by the neighborhood radius threshold with the data point as the center.

[0053] In this conveyor belt permanent magnet synchronous motor system, a neighborhood radius threshold is preset. This threshold is determined based on the operating characteristics of the motor and the distribution of the data. With a certain data point as the center, the total number of data points contained in the spherical area defined by this neighborhood radius threshold is counted, and this total number is the local density value of the data point. For example, in a three-dimensional coordinate data set, a certain data point represents a set of specific parameters of stator current response delay time, rotor flux steady-state error rate, and speed overshoot. With this data point as the center, a spherical area is determined according to the preset neighborhood radius threshold, and then the number of other data points contained in this area is counted, and this number is the local density value of the data point. By performing such traversal calculations on the entire three-dimensional coordinate data set, the local density value of each data point can be obtained.

[0054] Step S134, generating a density distribution map according to the local density value, screening core data points whose local density values ​​are greater than a first density threshold, and establishing a spatial connectivity relationship between the core data points, wherein the determination condition of the spatial connectivity relationship is that the Euclidean distance between two core data points is less than the neighborhood radius threshold.

[0055] In this embodiment, the first density threshold is set according to the overall distribution characteristics of the motor operation data. In the density distribution map, data points whose local density values ​​are greater than the first density threshold are regarded as core data points. These core data points have a higher density in the entire data set, representing a data combination with a certain correlation in the three parameters of stator current response delay time, rotor flux steady-state error rate and speed overshoot. For two core data points, if their Euclidean distance is less than the neighborhood radius threshold, it is considered that there is a spatial connectivity relationship between them. For example, in three-dimensional space, the Euclidean distance between two core data points is calculated (calculated based on the coordinate values). If this distance is less than the preset neighborhood radius threshold, it means that the two core data points are connected in space. This connectivity relationship reflects that the parameter combinations they represent may have some kind of intrinsic connection in the motor operation characteristics.

[0056] Step S135, aggregating core data points with direct or indirect spatial connectivity into continuous high-density parameter areas, and simultaneously eliminating isolated data points whose local density values ​​are less than a second density threshold, where the second density threshold is less than the first density threshold.

[0057] In this embodiment, the second density threshold is less than the first density threshold, and is used to distinguish the boundary between core data points and isolated data points. In this process, the core data points that are interconnected through spatial connectivity are combined together to form a continuous high-density parameter area. The data points in these areas have a strong correlation in terms of motor operating parameters. For isolated data points whose local density values ​​are less than the second density threshold, since they have a weak correlation with other data points, they may not be representative when analyzing motor operating characteristics and parameter adjustments, so they are eliminated. For example, an isolated data point may be caused by a measurement error or a special one-time interference, and does not reflect the law of normal operation of the motor. Eliminating it helps to improve the accuracy of subsequent analysis.

[0058] Step S136, each high-density parameter area is expanded outward to the boundary data points, and the judgment condition of the boundary data points is that the distance to the core data points is less than the neighborhood radius threshold and is not classified into any high-density parameter area.

[0059] The judgment condition of the boundary data points is that the distance from the core data points is less than the neighborhood radius threshold and they are not classified into any high-density parameter area. At the edge of each high-density parameter area, there are some data points that are not core data points, but the distance from the core data points is less than the neighborhood radius threshold. These boundary data points are similar to the data points in the high-density parameter area to a certain extent. Including them in the corresponding high-density parameter area can more comprehensively describe the parameter relationship of the motor under different operating conditions. For example, a high-density parameter area represents a typical parameter combination of the motor under a specific load and operating condition, and its boundary data points may represent the parameter conditions of the transition from this typical state to other states. Including boundary data points can make this area more completely cover the relevant parameter variation range.

[0060] Step S137, constructing a space division mapping table based on the expanded high-density parameter region, wherein the mapping table records the number of the high-density parameter region to which each data point belongs and the correlation strength coefficient with the core data point.

[0061] In this spatial partitioning mapping table, each data point is clearly marked with the number of the high-density parameter area to which it belongs, so that the data points can be easily classified. At the same time, the correlation strength coefficient reflects the degree of correlation between each data point and the core data point. For example, for a data point, if it is close to the core data point and is similar to the typical situation represented by the core data point in terms of motor operating parameters, then its correlation strength coefficient with the core data point will be higher; conversely, if the distance is far or the parameter difference is large, the correlation strength coefficient will be lower.

[0062] Step S138, clustering the parameter combinations belonging to the same high-density parameter area number into parameter adjustment subsets according to the spatial division mapping table, and adding a dynamic weight factor to each parameter adjustment subset, wherein the dynamic weight factor is calculated by multiplying the average local density value of all data points in the parameter adjustment subset by the maximum correlation strength coefficient.

[0063] For example, during the operation of a conveyor belt permanent magnet synchronous motor, a parameter adjustment subset may contain multiple parameter combinations under a specific load range and operating conditions. These parameter combinations have similar characteristics in terms of stator current response delay time, rotor flux steady-state error rate, and speed overshoot. The dynamic weight factor is calculated by multiplying the average local density value of all data points in the parameter adjustment subset by the maximum correlation intensity coefficient. If the average local density value of the data points in a parameter adjustment subset is high and the maximum correlation intensity coefficient is also large, then the dynamic weight factor of this subset will be large, which means that this subset may have a higher importance in the subsequent parameter adjustment process.

[0064] Step S139, performing secondary clustering on the overlapping boundary areas between adjacent high-density parameter areas, the judgment condition for the secondary clustering is that there is a parameter combination in the boundary area and the difference in the correlation strength coefficients of the two high-density parameter areas is less than a preset overlap threshold, and the parameter combination of the boundary area is allocated to the adjacent parameter adjustment subset according to the proportion of the correlation strength coefficient.

[0065] Step S1310, adjusting the corresponding parameter association calculation weight according to the dynamic weight factor of each parameter adjustment subset to generate a priority sequence, wherein the priority sequence is used to control the calling order of subsequent parameter adjustment subsets.

[0066] In this embodiment, the judgment condition of secondary clustering is that there is a parameter combination in the boundary area and the difference of the correlation strength coefficient of the two high-density parameter areas is less than the preset overlap threshold. In the actual operation of the conveyor belt permanent magnet synchronous motor, there may be some overlapping boundary areas between adjacent high-density parameter areas, and the parameter combination in these areas may have a certain correlation with the two high-density parameter areas at the same time. For example, when the load of the conveyor belt is in the transition stage between two typical load states, the operating parameters of the motor may have similarities with the typical parameter combination in one high-density parameter area, and there is an association with the parameter combination in another adjacent high-density parameter area. Through secondary clustering, the parameter combination of the boundary area is allocated to the adjacent parameter adjustment subset according to the correlation strength coefficient ratio. According to the dynamic weight factor of each parameter adjustment subset, the corresponding parameter correlation calculation weight is adjusted to generate a priority sequence. This priority sequence is used to control the calling order of the subsequent parameter adjustment subset. During the operation of the motor, if parameter adjustment is required, the parameter adjustment subset with higher priority will be considered first, because these subsets may have more important significance for the optimization of motor performance under the current operating state.

[0067] In a possible implementation manner, extracting at least one positive adjustment parameter from the first parameter adjustment subset corresponding to the first operating state includes:

[0068] The parameter combinations that meet the following conditions are screened in the first parameter adjustment subset:

[0069] The amplitude change rate of the stator current fluctuation sequence is less than or equal to the first amplitude threshold.

[0070] The accumulated error of the rotor flux phase offset is lower than the phase error threshold.

[0071] The target response index in the screened parameter combination is used as a positive adjustment parameter.

[0072] And, extracting at least one negative adjustment parameter from the second parameter adjustment subset comprises:

[0073] The following processing is performed on the parameter combinations in the second parameter adjustment subset:

[0074] The difference between the stator current fluctuation sequence of the parameter combination and the current fluctuation of the first parameter adjustment subset is calculated.

[0075] It is detected whether the oscillation frequency of the rotation speed tracking error curve of the parameter combination exceeds a preset frequency range.

[0076] If the current fluctuation difference is greater than a difference threshold and the oscillation frequency exceeds a preset frequency range, the target response indicator of the parameter combination is marked as a negative adjustment parameter.

[0077] In this embodiment, in the scenario of a conveyor belt permanent magnet synchronous motor in an industrial production workshop, the first operating state can be set as the motor operating state when the conveyor belt is in a stable load transportation process. In this state, the first parameter adjustment subset includes a series of parameter combinations related to the motor operation.

[0078] First, a parameter combination that meets specific conditions is screened in the first parameter adjustment subset. One of the conditions is that the amplitude change rate of the stator current fluctuation sequence is less than or equal to the first amplitude threshold. When the conveyor belt is running stably, the stator current of the motor should remain relatively stable to ensure stable torque output to drive the conveyor belt. For example, the conveyor belt transports materials at a uniform speed, at which time the load of the motor is relatively constant and the stator current fluctuation should be small. Assume that the first amplitude threshold is set to a specific value, which is determined based on the rated current of the motor, the normal operating load of the conveyor belt, and previous operating experience data. If the amplitude change rate of the stator current fluctuation sequence is within this threshold range, it means that the fluctuation of the stator current is at an acceptably low level, which means that the motor is operating relatively stably in this regard.

[0079] Another condition is that the cumulative error of the rotor flux phase offset is lower than the phase error threshold. During the operation of the motor, the phase of the rotor flux has an important influence on the torque generation and energy conversion of the motor. Due to the complex electromagnetic environment inside the motor, the phase of the rotor flux may be offset by factors such as temperature and magnetic field interference. During the stable operation of the conveyor belt, the phase of the rotor flux should ideally remain in a relatively stable state. For example, as the motor runs for a long time, some slight phase offsets may occur, but if the cumulative error of these offsets is lower than the set phase error threshold, it means that the phase control of the rotor flux is within an acceptable range, which helps to ensure efficient and stable operation of the motor.

[0080] When the parameter combination meets the above two conditions, the target response index in the screened parameter combination is used as a positive adjustment parameter. These target response indicators may include the output torque, efficiency, speed stability, etc. of the motor. For example, under the parameter combination that meets the conditions of stator current fluctuation and rotor flux phase offset, the output torque of the motor can stably maintain the operating speed of the conveyor belt, and the efficiency of the motor is at a high level. In this case, the output torque, efficiency and other target response indicators are regarded as positive adjustment parameters. These positive adjustment parameters are of great significance for optimizing the performance of the motor in the first operating state, and can be used as a reference for subsequent adjustment of the motor parameters to further improve the operating efficiency and stability of the motor and ensure the stable operation of the conveyor belt.

[0081] Regarding the process of extracting at least one negative adjustment parameter from the second parameter adjustment subset, the second parameter adjustment subset has a lower correlation with the parameters of the first parameter adjustment subset, and represents a parameter combination of the motor under different operating characteristics.

[0082] Specific processing is performed on the parameter combinations in the second parameter adjustment subset. First, the difference between the stator current fluctuation sequence of the parameter combination and the current fluctuation of the first parameter adjustment subset is calculated. During the operation of the conveyor belt, the first parameter adjustment subset represents the parameter combination under a relatively ideal or stable operating state, while the parameter combination in the second parameter adjustment subset may reflect some abnormal or undesirable operating conditions. For example, when a component of the conveyor belt has a minor fault, such as increased friction of the conveyor belt roller or problems with the heat dissipation of the motor, the operating state of the motor will change, which may cause the stator current fluctuation sequence to differ from the normal fluctuation in the first parameter adjustment subset. This difference is calculated by a specific algorithm, which may involve comparative analysis of the amplitude, frequency and other characteristics of the two current fluctuation sequences in the same time interval.

[0083] Then, it is detected whether the oscillation frequency of the speed tracking error curve of the parameter combination exceeds the preset frequency range. In the conveyor belt system, in order to ensure the stable transportation of materials, the speed of the motor needs to run at the set speed, and the fluctuation of the speed should be within a certain range. The preset frequency range is set according to the normal operation requirements of the conveyor belt and the performance of the motor. For example, when the control system of the motor fails or is strongly disturbed by the outside, the oscillation frequency of the speed tracking error curve may exceed the normal range. If the oscillation frequency of the speed tracking error curve of a certain parameter combination in the second parameter adjustment subset exceeds this preset frequency range, it means that the speed control of the motor under this parameter combination has an abnormality.

[0084] If the current fluctuation difference is greater than the difference threshold and the oscillation frequency exceeds the preset frequency range, the target response index of the parameter combination is marked as a negative adjustment parameter. These target response indicators may also include the output torque, efficiency, speed stability, etc. of the motor. For example, under this abnormal parameter combination, the output torque of the motor may be unstable, resulting in large fluctuations in the speed of the conveyor belt, or a significant decrease in the efficiency of the motor. In this case, the target response indicators such as output torque and efficiency are marked as negative adjustment parameters. Identifying these negative adjustment parameters helps to detect problems in a timely manner and take corresponding measures when the motor has abnormal operation, such as fault diagnosis of the motor, adjustment of the control strategy, etc., to avoid greater impact on the normal operation of the conveyor belt.

[0085] In a possible implementation, step S150 includes:

[0086] Step S151: input the first operating state into the first encoding layer of the dynamic parameter adjustment model to generate a state feature vector.

[0087] In this embodiment, the first operating state includes a lot of information about the conveyor belt permanent magnet synchronous motor during stable operation, such as the value of the stator current under stable load and its fluctuation law, the amplitude and phase of the rotor flux in a stable state, the value of the speed fluctuation within an acceptable range, and the value of the load change characteristic parameter when it is relatively stable. This rich information is input into the first encoding layer of the dynamic parameter adjustment model. The first encoding layer uses a specific encoding algorithm to process the input information, which may be based on the quantization, mapping and feature extraction operations of different parameters. For example, for the stator current data, it may be quantified according to its value range and change trend, mapped to a specific value interval, and then extract the features related to the motor operation state, such as the stability of the current, the periodicity of the fluctuation, etc. For the rotor flux, in addition to quantifying its amplitude, the stability of its phase and other features will also be analyzed. Related features of the speed fluctuation, such as the frequency and amplitude of the fluctuation, will also be extracted. The load change characteristic parameters may be converted into features related to the motor operating efficiency and torque demand. Through these operations, the first encoding layer integrates all the information of the first operating state into a state feature vector, which comprehensively describes the operating characteristics of the motor in the first operating state and is the basis for subsequent model calculation and adjustment.

[0088] Step S152: input the positive adjustment parameter into the second coding layer to generate a positive adjustment vector, and input the negative adjustment parameter into the third coding layer to generate a negative adjustment vector.

[0089] In this embodiment, the positive adjustment parameters are selected from the first parameter adjustment subset. In the scenario of stable operation of the conveyor belt, these parameters reflect the factors that contribute to the optimized operation of the motor, such as the motor output torque, efficiency and other target response indicators when the stator current fluctuation amplitude change rate meets the requirements and the rotor flux phase offset cumulative error is low. After receiving these positive adjustment parameters, the second coding layer also uses a specific coding algorithm for processing. For example, for the positive adjustment parameter of the motor output torque, it may be quantified and feature extracted based on its proportional relationship with the rated torque, torque stability and other factors, and these features are combined into part of the positive adjustment vector. For the efficiency parameter, the encoding is performed considering factors such as its closeness to the ideal efficiency and the efficiency change trend under different loads. The negative adjustment parameters are selected from the second parameter adjustment subset and represent the relevant indicators of the motor under abnormal or undesirable operation conditions. When these negative adjustment parameters are input into the third coding layer, the third coding layer will perform similar encoding operations on them. For example, when the oscillation frequency of the speed tracking error curve exceeds the preset range and the stator current fluctuation is greatly different from the first parameter adjustment subset, the target response indicators such as the motor output torque and efficiency are input as negative adjustment parameters. The encoding process will highlight the abnormal characteristics reflected by these parameters, such as the instability of the torque, a significant drop in efficiency, etc., thereby generating a negative adjustment vector.

[0090] Step S153, calculating the similarity weight between the state feature vector and the positive adjustment vector, and the repulsion weight between the state feature vector and the negative adjustment vector by using a contrast loss function.

[0091] In this embodiment, the contrast loss function is constructed based on a specific mathematical model, which takes into account many factors such as the distance between vectors and the correlation of elements. For the calculation of the similarity weight between the state feature vector and the positive adjustment vector, since the positive adjustment vector represents the direction of the motor's optimized operation, the contrast loss function analyzes the degree of matching between the elements in the state feature vector and the elements in the positive adjustment vector. For example, if the stator current stability feature in the state feature vector is highly matched with the ideal stator current feature in the positive adjustment vector, and the stable phase feature of the rotor flux is also matched, then when calculating the similarity weight, these matching elements will have a positive impact on the weight. For the calculation of the repulsion weight between the state feature vector and the negative adjustment vector, since the negative adjustment vector represents a situation that is not conducive to the operation of the motor, the contrast loss function will focus on the difference elements between the two. For example, when the speed stability feature in the state feature vector is significantly different from the speed abnormal fluctuation feature in the negative adjustment vector, this will increase the repulsion weight between the two. In this way, the contrast loss function accurately calculates the similarity weight and the repulsion weight, which reflect the relationship between different vectors and provide a basis for parameter adjustment of the dynamic parameter adjustment model.

[0092] Step S154, updating the parameter space distribution of the dynamic parameter adjustment model according to the similarity weight and the rejection weight, and generating an optimized real-time parameter setting instruction set.

[0093] In this process, the dynamic parameter adjustment model adjusts its internal parameter space according to the calculated similarity weight and exclusion weight. For example, if the similarity weight is high, it means that the current first operating state is more consistent with the positive adjustment direction, and the model tends to adjust the parameter space to get closer to this optimization direction. This may involve the optimization of the adjustment strategy for the parameters related to the current loop, speed loop and flux loop. For the current loop, the parameters of the proportional integral controller may be adjusted according to the weight, such as increasing the proportional coefficient according to the similarity weight to improve the control accuracy of the stator current, or reducing the integral coefficient according to the exclusion weight to avoid integral saturation. For the speed loop, the parameters of the speed controller may be adjusted, such as optimizing the speed feedback gain according to the weight. For the flux loop, the relevant parameters of the flux observer may be adjusted, such as improving the accuracy of the flux observation according to the weight. Through these adjustments, an optimized real-time parameter tuning instruction set is generated, which contains the optimal parameter adjustment strategy for the current operating state of the motor, which can improve the operating efficiency, stability and performance of the motor.

[0094] Wherein, after step S154, the method further includes:

[0095] Step S160, loading the real-time parameter setting instruction set into the parameter adjustment module of the motor controller.

[0096] In the control system of the conveyor belt permanent magnet synchronous motor, the parameter adjustment module of the motor controller is responsible for receiving and executing these instruction sets. For example, the parameter adjustment module has a special interface for receiving real-time parameter setting instruction sets, and these real-time parameter setting instruction sets are transmitted to the parameter adjustment module in a specific data format.

[0097] Step S170: performing closed-loop parameter injection on the current loop, speed loop and flux loop of the permanent magnet synchronous motor through the parameter adjustment module.

[0098] For the current loop, the current loop proportional-integral coefficient adjustment instruction in the real-time parameter tuning instruction set is applied to the control algorithm of the current loop. For example, if the instruction requires an increase in the proportional coefficient of the current loop, the parameter adjustment module will inject the new proportional coefficient value into the proportional-integral controller of the current loop, thereby changing the control effect of the current loop on the stator current and making the stator current closer to the ideal state. For the speed loop, the switching threshold of the sliding mode controller is updated according to the switching threshold update instruction of the speed loop sliding mode controller. During the operation of the conveyor belt, this helps to improve the precise control of the motor speed by the speed loop, so that the motor speed can track the set speed more stably. For the flux loop, the gain of the flux observer is adjusted according to the flux observer gain adaptive adjustment instruction. This can improve the accuracy of flux observation, thereby optimizing the torque control and overall performance of the motor.

[0099] Step S180, collecting the motor operation response data after parameter injection in real time, and feeding it back to the dynamic parameter adjustment model for incremental training.

[0100] During the operation of the motor, various sensors installed on the motor continuously collect the operating data of the motor, including parameters such as stator current, rotor flux, and speed. These collected data reflect the actual operation of the motor after parameter injection. For example, the collected stator current data can show whether the current is more stably controlled as expected, the rotor flux data can reflect whether the accuracy of flux observation has improved, and the speed data indicates whether the motor speed is more stable. After feeding these data back to the dynamic parameter adjustment model, the model can be incrementally trained according to the new data. During the incremental training process, the model will further optimize its internal parameters according to the new operation response data to adapt to various changes in the operation of the motor, such as small fluctuations in the conveyor belt load, the aging of the motor itself, and other factors, thereby continuously improving the performance and operating efficiency of the motor and ensuring that the conveyor belt permanent magnet synchronous motor is always in the best operating state.

[0101] In a possible implementation, before step S110, the method further includes:

[0102] Step A110, initializing a reference parameter template of the permanent magnet synchronous motor, wherein the reference parameter template includes a nominal value of a stator resistance, a nominal value of a rotor inductance, and a nominal range of a flux linkage amplitude.

[0103] Step A120, generating a parameter disturbance data set based on the motor load step change test, and injecting the parameter disturbance data set into the pre-training stage of the dynamic parameter adjustment model.

[0104] Wherein, step A110 includes:

[0105] Step A111, injecting multiple sets of sinusoidal swept-frequency voltage signals in a no-load state, and collecting the frequency response curve of the stator current.

[0106] Step A112, adjusting the nominal value of the rotor inductance according to the resonant peak value of the frequency response curve, and correcting the nominal value of the stator resistance based on the current decay rate.

[0107] Step A113, estimating the initial amplitude of the rotor flux through a flux observer, and constraining the initial amplitude of the rotor flux to be within the nominal range of the flux amplitude.

[0108] In this embodiment, the reference parameter template includes important parameters such as the nominal value of the stator resistance, the nominal value of the rotor inductance, and the nominal range of the flux amplitude. During the initialization process, the motor is operated in a no-load state. At this time, multiple sets of sinusoidal swept-frequency voltage signals are injected into the motor, which is a voltage signal with a continuously changing frequency. While injecting these signals, the frequency response curve of the stator current is collected. For a conveyor belt permanent magnet synchronous motor, the no-load state refers to the situation where the conveyor belt does not carry any material. The injected sinusoidal swept-frequency voltage signal will cause the stator current to respond, and this response presents a specific curve shape as the frequency of the voltage signal changes, that is, the frequency response curve of the stator current. For example, when the frequency of the swept-frequency voltage signal gradually increases from a lower value, the stator current will change accordingly according to the electromagnetic characteristics inside the motor. In this process, a resonant peak may appear in the frequency response curve. The nominal value of the rotor inductance is adjusted according to this resonant peak, because the rotor inductance affects the magnetic field coupling relationship inside the motor, thereby affecting the response of the stator current to voltage signals of different frequencies. When the resonance peak occurs, it has a specific mathematical relationship with the rotor inductance, through which the nominal value of the rotor inductance can be accurately adjusted.

[0109] At the same time, the nominal value of the stator resistance is corrected based on the current decay rate. After the sinusoidal swept frequency voltage signal is injected, the stator current will decay during the change of the voltage signal, and this decay rate is closely related to the stator resistance. For example, when the voltage signal frequency changes, the stator current will decay according to a certain rule. By accurately measuring this decay rate and using the known mathematical relationship in the electromagnetic theory of the motor, the nominal value of the stator resistance can be corrected. This is because the stator resistance plays a key role in the energy loss and current transmission of the motor. The accurate nominal value of the stator resistance is very important for the performance analysis and control of the motor.

[0110] In addition, the initial amplitude of the rotor flux is estimated by the flux observer, and this initial amplitude is constrained to the nominal range of the flux amplitude. The flux observer is a device specifically used to measure and estimate the rotor flux of the motor. In the no-load state, although the motor has no load, the rotor flux still exists and has a certain initial amplitude. The flux observer uses the measurable parameters of the motor, such as the stator current and voltage, to estimate the amplitude of the rotor flux through a specific algorithm. This estimated initial amplitude needs to be constrained to the nominal range of the flux amplitude. The nominal range of the flux amplitude is a reasonable range determined in advance based on the design specifications and performance requirements of the motor. For example, if the estimated initial amplitude of the rotor flux exceeds this nominal range, it may indicate that there is an abnormality in the magnetic field inside the motor, such as magnetic circuit saturation or a large error of the flux observer. Constraining the initial amplitude of the rotor flux within the nominal range can ensure the accuracy of subsequent motor operation status analysis and control.

[0111] After completing the initialization of the benchmark parameter template of the permanent magnet synchronous motor, a parameter perturbation data set is generated based on the motor load step change test, and this parameter perturbation data set is injected into the pre-training stage of the dynamic parameter adjustment model. In the operation scenario of the conveyor belt permanent magnet synchronous motor, the motor load step change test is a test method that artificially creates a load mutation. For example, during the normal operation of the conveyor belt, a certain amount of material is suddenly loaded on the conveyor belt, which is equivalent to applying a step-change load to the motor. In this process, various operating parameters of the motor will change accordingly. While the load step changes, the motor's stator current, rotor flux, speed fluctuation, and load change characteristic parameters are continuously monitored, and the changes in these parameters are recorded to form a parameter perturbation data set. This data set contains the dynamic change information of various parameters of the motor when the load changes suddenly, which is very critical for understanding the performance changes of the motor under non-steady-state conditions.

[0112] This parameter perturbation data set is injected into the pre-training stage of the dynamic parameter adjustment model. The dynamic parameter adjustment model is an important tool for optimizing the motor operating parameters. In the pre-training stage, the model needs to learn the parameter change rules of the motor under various conditions, and the parameter perturbation data set just provides rich information under such load step changes. For example, when the model receives the parameter perturbation data set, it will analyze the sudden change of stator current, the dynamic response of rotor flux, the change trend of speed fluctuation, and the correlation between load change characteristic parameters and other parameters when the load step changes. By learning from these data, the model can establish a preliminary parameter adjustment strategy, laying the foundation for more accurate parameter adjustment based on real-time monitoring data in the actual operation process. This pre-training process can improve the model's adaptability and optimization ability to complex motor operation conditions, thereby better ensuring the stable and efficient operation of the conveyor belt permanent magnet synchronous motor under various load change conditions.

[0113] In a possible implementation, the method further includes:

[0114] When a sudden change in the motor load or a jump in the speed command of the permanent magnet synchronous motor is detected, a parameter self-tuning emergency response mechanism is triggered.

[0115] Under the parameter self-tuning emergency response mechanism, the historical optimal parameter combination is preferentially used to cover the current parameter adjustment subset, and the response cycle of the dynamic parameter adjustment model is shortened.

[0116] In this embodiment, first, when a sudden change in the motor load or a jump in the speed command of the permanent magnet synchronous motor is detected, the parameter self-tuning emergency response mechanism is triggered. During the operation of the conveyor belt, the sudden change in the motor load may occur due to a variety of reasons. For example, when a large amount of material is suddenly piled up on the conveyor belt or a large object is placed on the conveyor belt, the motor load will suddenly increase. Or during the material unloading process, if a large amount of material is unloaded quickly, the motor load will suddenly decrease. In addition, the speed command jump may also occur, such as a temporary adjustment of the production process, requiring the conveyor belt to speed up or slow down, resulting in a sudden change in the speed command of the motor.

[0117] In this case, the parameter self-tuning emergency response mechanism is triggered. Under this parameter self-tuning emergency response mechanism, the historical optimal parameter combination is preferentially used to cover the current parameter adjustment subset. The historical optimal parameter combination is a parameter combination that has been obtained through multiple tests and optimizations during the previous operation of the motor and has shown good performance in dealing with similar load mutations or speed command jumps. For example, in the case of a sudden increase in load, a set of parameter combinations that can restore the motor to stable operation in a short time and maintain high efficiency was found by adjusting and optimizing various parameters of the motor. This set of parameter combinations is recorded as the historical optimal parameter combination. When a new load mutation or speed command jump occurs, this set of historical optimal parameter combinations is directly used to cover the current parameter adjustment subset, which can quickly adjust the motor's operating parameters to a relatively stable and optimized state. At the same time, the response cycle of the dynamic parameter adjustment model is shortened. Under normal circumstances, the dynamic parameter adjustment model requires a certain amount of time to analyze data, calculate adjustment parameters, and other operations, but in this emergency situation, shortening the response cycle can enable the model to respond to changes in the motor more quickly. For example, reducing the number of iterations of some complex algorithms in the model during data processing, or speeding up data acquisition and transmission, so that the motor can adapt more quickly to the impact of sudden load changes or speed command jumps, ensuring the normal operation of the conveyor belt.

[0118] In a possible implementation, the method further includes:

[0119] The nominal value of the stator resistance is dynamically compensated according to the change in the motor's operating environment temperature.

[0120] The input feature dimension of the dynamic parameter adjustment model is updated based on the compensated stator resistance nominal value.

[0121] In industrial production workshops, the ambient temperature is a factor that is prone to change. For example, in different seasons or when the cooling equipment in the workshop is operating in different states, the ambient temperature of the motor will change significantly. As the ambient temperature rises, the resistance inside the motor will change. According to the principles of physics, the resistance of metal increases with the increase in temperature. For the stator resistance of a permanent magnet synchronous motor, temperature changes will affect the actual value of the stator resistance. The change in the ambient temperature of the motor is measured by a temperature sensor and other equipment, and then the nominal value of the stator resistance is dynamically compensated according to a predetermined temperature-resistance relationship model. For example, if the ambient temperature is measured to have increased by a certain value, the value by which the stator resistance should increase is calculated according to the temperature-resistance relationship model, and then the original nominal value of the stator resistance is compensated accordingly.

[0122] Update the input feature dimension of the dynamic parameter adjustment model based on the compensated stator resistance nominal value. The input feature dimension of the dynamic parameter adjustment model is related to various operating parameters of the motor. When the nominal value of the stator resistance changes, this parameter becomes a new input feature dimension or affects the original input feature dimension. For example, in the model, the nominal value of the stator resistance may have a certain mathematical relationship with other parameters such as stator current, rotor flux, etc., and the compensated nominal value of the stator resistance changes the coefficient or weight of this relationship. Therefore, it is necessary to update the input feature dimension of the model so that the model can accurately adjust and optimize parameters according to the new nominal value of the stator resistance. This may involve rebuilding the input layer structure of the model, adjusting the calculation method or weight distribution of the input parameters related to the stator resistance, and other operations.

[0123] In a possible implementation, the method further includes:

[0124] During the motor startup phase, fuzzy control rules are used to generate initial parameter setting instructions.

[0125] When the motor enters steady-state operation, it gradually switches to the real-time parameter setting instructions output by the dynamic parameter adjustment model.

[0126] In this embodiment, the motor start is a special operation process, at which the various parameters of the motor are not yet stable, and the motor needs to reach a predetermined operating state as soon as possible. Fuzzy control rules are a control method based on fuzzy logic. It does not require an accurate mathematical model, but is controlled based on experience and some fuzzy rules. When the conveyor belt permanent magnet synchronous motor is started, for example, the initial parameter setting instructions are generated by fuzzy control rules based on factors such as the type of motor, rated power, and initial load of the conveyor belt (such as whether the conveyor belt is started without load). If the motor is of high power and the conveyor belt is started without load, the fuzzy control rules may set a larger initial stator current value to quickly start the motor, but at the same time, the upper limit of this current value will be limited according to fuzzy logic to avoid excessive current shocks from causing damage to the motor. At the same time, for the initial parameters of the speed loop, a faster speed rise rate may be set, but the stability of the motor will also be considered according to the fuzzy rules to avoid excessive speed overshoot.

[0127] When the motor enters steady-state operation, it gradually switches to the real-time parameter tuning instructions output by the dynamic parameter adjustment model. After the motor starts, as time goes by, the various parameters of the motor gradually stabilize and enter the steady-state operation stage. In this process, it gradually switches from the initial parameter tuning instructions generated by the fuzzy control rules to the real-time parameter tuning instructions output by the dynamic parameter adjustment model. For example, at the beginning, it may still mainly rely on the parameter tuning instructions under the fuzzy control rules, but as the motor operation state gets closer and closer to the steady state, the weight of the dynamic parameter adjustment model output instructions is gradually increased. This switching process is a gradual process to ensure that the motor can run smoothly during the transition from startup to steady state. During the switching process, the operating parameters of the motor, such as stator current, rotor flux, speed fluctuation, etc., are monitored at the same time, and the switching speed and proportion are adjusted according to the changes in these parameters.

[0128] In a possible implementation manner, the real-time parameter setting instruction set includes at least one of the following instruction types:

[0129] Current loop proportional integral coefficient adjustment instruction.

[0130] Speed ​​loop sliding mode controller switching threshold update command.

[0131] Flux observer gain adaptive adjustment command.

[0132] The step of generating the current loop proportional integral coefficient adjustment instruction includes:

[0133] The rising edge time and the steady-state error rate of the stator current fluctuation sequence in the parameter combination are extracted.

[0134] The proportional coefficient step size is adjusted according to the ratio of the rising edge time to the preset time window.

[0135] The integral coefficient range is dynamically reduced based on the integral saturation trend of the steady-state error rate.

[0136] In this embodiment, for the real-time parameter setting instruction set, multiple instruction types are included. Taking the current loop proportional integral coefficient adjustment instruction as an example, its generation step involves extracting the rising edge time and steady-state error rate of the stator current fluctuation sequence in the parameter combination. During the operation of the conveyor belt permanent magnet synchronous motor, the stator current fluctuation sequence reflects the impact of factors such as motor load changes and power supply fluctuations on the current. The rising edge time of the stator current fluctuation sequence indicates the time required for the current to rise from a lower value to a higher value. For example, when the load on the conveyor belt increases slightly, the stator current will rise accordingly, and the time spent in this rising process is the rising edge time. The steady-state error rate indicates the deviation ratio between the actual value and the ideal value of the stator current when the motor is in a stable operating state.

[0137] The proportional coefficient step size is adjusted based on the ratio of the rise time to the preset time window. The preset time window is a fixed time interval set based on the performance requirements of the motor and previous operating experience. For example, if the rise time is short, that is, the current rises faster, it may mean that the motor responds more sensitively to load changes. In this case, the proportional coefficient step size may be appropriately increased based on the ratio of the rise time to the preset time window. This is because a larger proportional coefficient step size allows the current loop to respond more quickly to current changes, thereby better controlling the stator current. On the contrary, if the rise time is long, it means that the motor responds slowly to load changes, and the proportional coefficient step size may be appropriately reduced to avoid over-adjustment that causes unstable motor operation.

[0138] Dynamically narrow the integral coefficient range based on the integral saturation trend of the steady-state error rate. During the operation of the motor, the integral link is used to eliminate the steady-state error. However, if the steady-state error rate continues to be large or shows a trend of integral saturation, for example, the steady-state error of the stator current cannot converge to a smaller value for a long time, it may cause the integral term to accumulate continuously, making the integral coefficient too large, thereby causing system instability. By monitoring the integral saturation trend of the steady-state error rate, when such a trend is found, the integral coefficient range is dynamically narrowed. For example, if it is found that the steady-state error rate has not improved significantly over a period of time, and the integral term has a trend of increasing continuously, the value range of the integral coefficient will be narrowed to a smaller value according to the pre-set rules, thereby avoiding the occurrence of integral saturation, ensuring the stable control of the stator current by the current loop, and then ensuring the stable operation of the conveyor belt permanent magnet synchronous motor.

[0139] For the speed loop sliding mode controller switching threshold update instruction, during the operation of the conveyor belt permanent magnet synchronous motor, the speed loop sliding mode controller is used to control the motor speed. The sliding mode controller switching threshold is a key parameter that determines the switching conditions of the controller between different states. When the operating state of the motor changes, such as load changes or speed command adjustments, this switching threshold needs to be updated according to the real-time operating parameters of the motor. For example, when the load on the conveyor belt increases, the motor needs to output a larger torque to maintain the speed. At this time, it may be necessary to adjust the switching threshold of the sliding mode controller according to factors such as the magnitude of the load increase and the current speed deviation of the motor, so that the speed loop can more effectively control the motor speed and maintain the stable operating speed of the conveyor belt.

[0140] For the flux observer gain adaptive adjustment instruction, the flux observer is used to accurately measure and estimate the rotor flux of the motor. During the operation of the conveyor belt permanent magnet synchronous motor, the observation accuracy of the flux will be affected by factors such as the load change and temperature change of the motor. The flux observer gain adaptive adjustment instruction is to dynamically adjust the gain of the flux observer according to the operating state of the motor. For example, when the motor load increases, the distribution and intensity of the magnetic field will change, which may affect the accuracy of the flux observation. At this time, according to the load change and other related parameters, such as stator current, speed, etc., the gain of the flux observer is adaptively adjusted to improve the accuracy of the flux observation, ensure the accurate torque control of the motor, and then ensure the normal operation of the conveyor belt.

[0141] Figure 2 The hardware structure of the control system 100 for self-tuning parameters of a permanent magnet synchronous motor provided by an embodiment of the present invention for implementing the control method for self-tuning parameters of a permanent magnet synchronous motor is shown. Figure 2 As shown, the control system 100 for self-tuning parameters of a permanent magnet synchronous motor may include a processor 110 , a machine-readable storage medium 120 , a bus 130 , and a communication unit 140 .

[0142] The machine-readable storage medium 120 may store data and / or instructions. In some embodiments, the machine-readable storage medium 120 may store data acquired from an external terminal. In some embodiments, the machine-readable storage medium 120 may store data and / or instructions that the control system 100 for self-tuning parameters of a permanent magnet synchronous motor uses to execute or use to complete the exemplary method described in the present invention.

[0143] During the specific implementation process, one or more processors 110 execute computer executable instructions stored in the machine-readable storage medium 120, so that the processor 110 can execute the control method of self-tuning parameters of the permanent magnet synchronous motor in the above method embodiment. The processor 110, the machine-readable storage medium 120 and the communication unit 140 are connected through the bus 130, and the processor 110 can be used to control the sending and receiving actions of the communication unit 140.

[0144] The specific implementation process of the processor 110 can refer to the various method embodiments executed by the control system 100 for self-tuning parameters of the permanent magnet synchronous motor mentioned above. The implementation principles and technical effects are similar and will not be repeated in this embodiment.

[0145] In addition, an embodiment of the present invention further provides a readable storage medium, in which computer executable instructions are preset. When a processor executes the computer executable instructions, the above control method for self-tuning parameters of a permanent magnet synchronous motor is implemented.

[0146] It should be noted that in order to simplify the description of the present invention and thus help understand one or more embodiments of the invention, in the foregoing description of the embodiments of the present invention, various features are sometimes combined into one embodiment, drawing or description thereof.

Claims

1. A control method for self-tuning parameters of a permanent magnet synchronous motor, characterized in that: The method comprises: Real-time monitoring of the dynamic operation data set of the permanent magnet synchronous motor, wherein the dynamic operation data set includes stator current, rotor flux, speed fluctuation and load change characteristic parameters; Performing feature correlation mapping based on multiple parameter combinations in the dynamic operation data set to generate an initial parameter adjustment set, wherein each parameter combination includes a mapping relationship between the motor operation state and the target response index; Divide a plurality of parameter adjustment subsets according to the parameter correlation in the initial parameter adjustment set, configure a dynamic error range for each parameter adjustment subset, and extract at least one positive adjustment parameter from the first parameter adjustment subset corresponding to the first operating state, wherein the positive adjustment parameter satisfies a response convergence condition within the dynamic error range; Extracting at least one negative adjustment parameter from a second parameter adjustment subset, wherein a parameter correlation between the second parameter adjustment subset and the first parameter adjustment subset is lower than a preset correlation threshold; The dynamic parameter adjustment model is trained using the plurality of the first operating states and the corresponding positive adjustment parameters and negative adjustment parameters to generate an optimized real-time parameter setting instruction set.

2. The control method for self-tuning parameters of a permanent magnet synchronous motor according to claim 1, characterized in that: The step of performing feature correlation mapping based on multiple sets of parameter combinations in the dynamic operation data set to generate an initial parameter adjustment set includes: The following processing is performed for each parameter combination: Extracting the stator current fluctuation sequence, rotor flux phase offset and speed tracking error curve in the parameter combination; Calculating the time domain coupling coefficient between the stator current fluctuation sequence and the rotor flux phase offset, and the frequency domain correlation index between the speed tracking error curve and the load variation characteristic parameter; A parameter combination that satisfies a coupling coefficient threshold and a frequency domain correlation threshold is mapped as a candidate element of the initial parameter adjustment set.

3. The control method for self-tuning parameters of a permanent magnet synchronous motor according to claim 1, characterized in that: The dividing into a plurality of parameter adjustment subsets according to the parameter association degree in the initial parameter adjustment set comprises: The following processing is performed on each parameter combination in the initial parameter adjustment set: A three-dimensional parameter space is constructed based on the stator current response delay time, rotor flux steady-state error rate and speed overshoot; Normalizing the stator current response delay time, the rotor flux steady-state error rate and the speed overshoot in the three-dimensional parameter space to generate a three-dimensional coordinate data set with a unified dimension standard, wherein each data point corresponds to the coordinates of a set of parameter combinations in the three-dimensional space; Traversing the three-dimensional coordinate data set based on a preset neighborhood radius threshold, calculating a local density value of each data point, wherein the local density value is the total number of data points contained in a spherical area defined by the neighborhood radius threshold with the data point as the center; Generate a density distribution map according to the local density value, select core data points whose local density value is greater than a first density threshold, and establish a spatial connectivity relationship between the core data points, wherein the determination condition of the spatial connectivity relationship is that the Euclidean distance between two core data points is less than the neighborhood radius threshold; Aggregate core data points with direct or indirect spatial connectivity into continuous high-density parameter regions, and simultaneously remove isolated data points whose local density values ​​are less than a second density threshold, where the second density threshold is less than the first density threshold; Expand each high-density parameter region outward to a boundary data point, where the boundary data point is determined to be less than the neighborhood radius threshold from the core data point and is not classified into any high-density parameter region; Building a space partition mapping table based on the expanded high-density parameter region, the mapping table records the high-density parameter region number to which each data point belongs and the correlation strength coefficient with the core data point; Clustering parameter combinations belonging to the same high-density parameter region number into parameter adjustment subsets according to the spatial division mapping table, and adding a dynamic weight factor to each parameter adjustment subset, wherein the dynamic weight factor is calculated by multiplying the average local density value of all data points in the parameter adjustment subset by the maximum correlation strength coefficient; Perform secondary clustering on the overlapping boundary areas between adjacent high-density parameter areas, the determination condition of the secondary clustering is that there is a parameter combination in the boundary area and the difference in the correlation strength coefficients of the two high-density parameter areas is less than a preset overlap threshold, and the parameter combination in the boundary area is allocated to the adjacent parameter adjustment subset according to the correlation strength coefficient ratio; The corresponding parameter association calculation weight is adjusted according to the dynamic weight factor of each parameter adjustment subset to generate a priority sequence, and the priority sequence is used to control the calling order of subsequent parameter adjustment subsets.

4. The control method for self-tuning parameters of a permanent magnet synchronous motor according to claim 2, characterized in that: The extracting at least one positive adjustment parameter from the first parameter adjustment subset corresponding to the first operating state includes: The parameter combinations that meet the following conditions are screened in the first parameter adjustment subset: The amplitude change rate of the stator current fluctuation sequence is less than or equal to the first amplitude threshold; The cumulative error of the rotor flux phase offset is lower than the phase error threshold; The target response index in the screened parameter combination is used as a positive adjustment parameter; And, extracting at least one negative adjustment parameter from the second parameter adjustment subset comprises: The following processing is performed on the parameter combinations in the second parameter adjustment subset: Calculating the difference between the stator current fluctuation sequence of the parameter combination and the current fluctuation of the first parameter adjustment subset; Detecting whether the oscillation frequency of the speed tracking error curve of the parameter combination exceeds a preset frequency range; If the current fluctuation difference is greater than a difference threshold and the oscillation frequency exceeds a preset frequency range, the target response indicator of the parameter combination is marked as a negative adjustment parameter.

5. The control method for self-tuning parameters of a permanent magnet synchronous motor according to any one of claims 1 to 4, characterized in that: The method of using the plurality of the first operating states, the corresponding positive adjustment parameters and the negative adjustment parameters to train a dynamic parameter adjustment model and generate an optimized real-time parameter setting instruction set includes: Inputting the first operating state into a first encoding layer of the dynamic parameter adjustment model to generate a state feature vector; Inputting the positive adjustment parameter into the second coding layer to generate a positive adjustment vector, and inputting the negative adjustment parameter into the third coding layer to generate a negative adjustment vector; Calculating the similarity weight between the state feature vector and the positive adjustment vector, and the repulsion weight between the state feature vector and the negative adjustment vector by using a contrast loss function; updating the parameter space distribution of the dynamic parameter adjustment model according to the similarity weight and the exclusion weight, and generating an optimized real-time parameter setting instruction set; Wherein, after updating the parameter space distribution of the dynamic parameter adjustment model according to the similarity weight and the exclusion weight to generate an optimized real-time parameter setting instruction set, the method further comprises: Loading the real-time parameter setting instruction set into a parameter adjustment module of a motor controller; Performing closed-loop parameter injection on the current loop, speed loop and flux loop of the permanent magnet synchronous motor through the parameter adjustment module; The motor operation response data after parameter injection is collected in real time and fed back to the dynamic parameter adjustment model for incremental training.

6. The control method for self-tuning parameters of a permanent magnet synchronous motor according to any one of claims 1 to 4, characterized in that: Before the step of real-time monitoring of the dynamic operation data set of the permanent magnet synchronous motor, the method further includes: Initializing a reference parameter template of a permanent magnet synchronous motor, wherein the reference parameter template includes a nominal value of a stator resistance, a nominal value of a rotor inductance, and a nominal range of a flux linkage amplitude; Generating a parameter disturbance data set based on a motor load step change test, and injecting the parameter disturbance data set into a pre-training phase of the dynamic parameter adjustment model; The initialization of the reference parameter template of the permanent magnet synchronous motor includes: Inject multiple sets of sinusoidal swept-frequency voltage signals under no-load condition and collect the frequency response curve of stator current; adjusting the nominal value of the rotor inductance according to the resonant peak of the frequency response curve, and correcting the nominal value of the stator resistance based on the current decay rate; The initial amplitude of the rotor flux is estimated by a flux observer, and the initial amplitude of the rotor flux is constrained to be within the nominal range of the flux amplitude.

7. The control method for self-tuning parameters of a permanent magnet synchronous motor according to any one of claims 1 to 4, characterized in that: The method further comprises: When a sudden change in the motor load or a jump in the speed command of the permanent magnet synchronous motor is detected, a parameter self-tuning emergency response mechanism is triggered; Under the parameter self-tuning emergency response mechanism, the historical optimal parameter combination is preferentially used to cover the current parameter adjustment subset, and the response cycle of the dynamic parameter adjustment model is shortened.

8. The control method for self-tuning parameters of a permanent magnet synchronous motor according to any one of claims 1 to 4, characterized in that: The method further comprises: Dynamically compensate the nominal value of stator resistance according to the change in the motor operating environment temperature; Updating the input feature dimension of the dynamic parameter adjustment model based on the compensated stator resistance nominal value; And, the method further comprises: During the motor startup phase, fuzzy control rules are used to generate initial parameter setting instructions; When the motor enters steady-state operation, it gradually switches to the real-time parameter setting instructions output by the dynamic parameter adjustment model.

9. The control method for self-tuning parameters of a permanent magnet synchronous motor according to any one of claims 1 to 4, characterized in that: The real-time parameter setting instruction set includes at least one of the following instruction types: Current loop proportional integral coefficient adjustment instruction; Speed ​​loop sliding mode controller switching threshold update instruction; Flux observer gain adaptive adjustment command; The step of generating the current loop proportional integral coefficient adjustment instruction includes: Extracting the rising edge time and steady-state error rate of the stator current fluctuation sequence in the parameter combination; Adjusting the proportional coefficient step size according to the ratio of the rising edge time to the preset time window; The integral coefficient range is dynamically reduced based on the integral saturation trend of the steady-state error rate.

Images