A method and system for real-time optimization of high dynamic motor control parameters with reduced loss

Abstract

The application relates to a loss-reducing real-time optimization method and system for high-dynamic motor control parameters, and relates to the technical field of motor control parameter optimization.The method comprises the following steps: obtaining current working condition parameters of a target motor, and collecting control parameters of the target motor; performing clustering analysis based on historical data to obtain a historical working condition clustering set; obtaining a predicted loss value according to control parameter prediction, obtaining an optimization target weight, and obtaining an initial control parameter weight; obtaining a dynamic optimization target based on the optimization target weight, obtaining a control parameter influence degree, and adjusting calculation to obtain a control parameter weight; optimizing motor control parameters with the dynamic optimization target as a target and the control parameter weight as a constraint, obtaining optimized control parameters, and using the optimized control parameters to optimize control parameters of a high-dynamic motor.The application solves the problems that traditional motor control parameter optimization easily ignores loss, is difficult to adapt to dynamic working conditions, and causes low motor efficiency and poor running stability.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of motor control parameter optimization, in particular to a high-dynamic motor control parameter real-time optimization method and system for reducing loss. BACKGROUND

[0002] With the improvement of performance requirements of industrial equipment in high-efficiency operation, precise control and other fields, the operation efficiency and stability of high-dynamic motor have become a key technical requirement to ensure the reliable operation of equipment.

[0003] At present, the traditional motor control parameter optimization method does not fully consider the loss factor and current harmonic distortion problem, and it is difficult to adapt to the dynamic changes of motor working conditions, and it is difficult to adjust the control strategy to balance the loss and current harmonic distortion, which not only easily leads to low motor operation efficiency, but also reduces the stability of motor operation, and increases the risk of subsequent equipment failure and energy consumption exceeding the standard based on motor drive. SUMMARY

[0004] The present application provides a high-dynamic motor control parameter real-time optimization method and system for reducing loss, which improves the consideration degree of loss and the adaptation ability to dynamic working conditions in traditional motor control parameter optimization, and improves the operation efficiency and stability of high-dynamic motor.

[0005] The embodiments of the present application disclose the following technical solutions:

[0006] In a first aspect, the embodiments of the present application provide a high-dynamic motor control parameter real-time optimization method for reducing loss, the method comprising:

[0007] Obtaining the current working condition parameters of a target motor and collecting the control parameters of the target motor, wherein the control parameters include speed loop PI parameters and current loop PI parameters, and the working condition parameters include motor speed and torque;

[0008] Based on historical data, obtaining a plurality of historical control parameters, a plurality of corresponding historical working condition parameters, a plurality of corresponding historical loss values and a plurality of corresponding historical current harmonic distortions, and performing clustering analysis to obtain a historical working condition clustering set;

[0009] According to the control parameters, a predicted loss value is obtained, an optimization target weight is obtained in combination with the corresponding historical working condition clustering set, and an initial control parameter weight is obtained according to the control parameters and the corresponding historical working condition clustering set;

[0010] Based on the optimization target weight, a dynamic optimization target is obtained, a control parameter influence degree is obtained based on the dynamic optimization target, the initial control parameter weight is adjusted and calculated to obtain a control parameter weight;

[0011] With the stated dynamic optimization objective as the goal and the control parameter weights as constraints, the motor control parameters are optimized to obtain optimized control parameters. These optimized control parameters are then used to optimize the control parameters of the high-dynamic motor.

[0012] Secondly, embodiments of this application provide a real-time optimization system for high dynamic motor control parameters to reduce losses, the system comprising:

[0013] The operating condition and parameter acquisition module acquires the current operating condition parameters of the target motor and acquires the control parameters of the target motor. The control parameters include speed loop PI parameters and current loop PI parameters, and the operating condition parameters include motor speed and torque.

[0014] The historical operating condition clustering module is used to obtain multiple historical control parameters, corresponding multiple historical operating condition parameters, corresponding multiple historical loss values, and corresponding multiple historical current harmonic distortions based on historical data, and to perform cluster analysis to obtain a historical operating condition cluster set.

[0015] The loss prediction and weight calculation module is used to predict the loss value based on the control parameters, obtain the optimization target weight by combining the corresponding historical operating condition cluster set, and obtain the initial control parameter weight based on the control parameters and the corresponding historical operating condition cluster set.

[0016] The dynamic target and weight adjustment module is used to obtain a dynamic optimization target based on the optimization target weight, and to obtain the influence degree of the control parameters based on the dynamic optimization target, adjust and calculate the initial control parameter weight, and obtain the control parameter weight.

[0017] The control parameter optimization module is used to optimize the motor control parameters with the dynamic optimization target as the objective and the control parameter weights as constraints, to obtain optimized control parameters, and to use the optimized control parameters to optimize the control parameters of the high-dynamic motor.

[0018] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0019] This application proposes a real-time optimization method and system for control parameters of high-dynamic motors to reduce losses. By step-by-step methods—determining historical operating condition clusters, calculating control parameter weights, determining dynamic optimization objectives, obtaining control parameter influence, calculating optimization adjustment step size, iteratively optimizing control parameters, and applying optimized parameters—precise real-time optimization of control parameters for high-dynamic motors under different operating conditions is achieved. First, target motor control parameters and operating condition parameters are collected. Based on historical control parameters, operating condition parameters, loss values, and current harmonic distortion data of the same motor model, historical operating condition clusters are formed, and typical values ​​of control, loss, and current harmonic distortion are extracted from the cluster centers of each cluster. Then, a loss predictor constructed using a multi-branch machine learning model obtains predicted loss values, and the deviations from these predicted values ​​and from typical loss values, as well as the deviations of control parameters from typical control values, are calculated. The weights of the loss and current harmonic distortion optimization objectives are determined based on these deviations, resulting in dynamic optimization objectives. Subsequently, based on the historical operating condition cluster data, the correlation coefficients of control parameters relative to the two optimization objectives are calculated, and the control parameter influence is obtained by weighting. The initial control parameter weights are adjusted accordingly, and the optimization adjustment step size is calculated based on the typical values ​​of historical operating conditions. Finally, iterative optimization is carried out with the dynamic optimization target as the guide. The difference between the control results is collected and compared. If the error exceeds the allowable error, the optimized parameters are used as the new control parameters to continue the iteration. After the target is met, the optimized control parameters are applied to the real-time control of the motor.

[0020] The technical solution of this application solves the problems in traditional high dynamic motor control, such as the difficulty of adapting fixed parameters to operating condition fluctuations, the lack of quantitative basis for parameter adjustment, and the difficulty of coordinating the optimization of losses and current harmonic distortion. It avoids excessive losses, excessive current harmonic distortion, or performance fluctuations caused by blind adjustment of control parameters, and significantly improves the efficiency, stability, and adaptability of high dynamic motor operation, providing strong support for the long-term reliable operation of high dynamic motors under complex operating conditions. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 A flowchart illustrating a method for real-time optimization of high dynamic motor control parameters to reduce losses, provided in an embodiment of this application;

[0023] Figure 2 This is a schematic diagram of a high dynamic motor control parameter real-time optimization system for reducing losses, provided in an embodiment of this application.

[0024] The components represented by each number in the attached diagram are explained below:

[0025] Operating condition and parameter acquisition module 01, historical operating condition clustering module 02, loss prediction and weight initial calculation module 03, dynamic target and weight adjustment module 04, control parameter optimization module 05. Detailed Implementation

[0026] This application provides a method and system for real-time optimization of high dynamic motor control parameters to reduce losses, which solves the technical problems in the prior art where motor control parameter optimization easily ignores losses and is difficult to adapt to dynamic operating conditions, resulting in low motor operating efficiency and poor operating stability.

[0027] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0028] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0029] In the description of this application, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0030] Example 1, as shown in the appendix Figure 1 As shown, this application provides a method for real-time optimization of high dynamic motor control parameters to reduce losses. The method includes the following steps:

[0031] S110: Obtain the current operating parameters of the target motor and collect the control parameters of the target motor, wherein the control parameters include speed loop PI parameters and current loop PI parameters, and the operating parameters include motor speed and torque;

[0032] In this embodiment of the application, in high dynamic operating scenarios such as drive motors for new energy vehicles and servo motors for industrial robots, in order to accurately capture the real-time operating status of the motor and avoid subsequent parameter optimization from deviating from actual needs due to distortion of basic data, it is necessary to simultaneously acquire operating parameters and control parameters to construct real and reliable optimization benchmark data.

[0033] First, multi-dimensional monitoring is used to obtain the operating parameters of the target motor (including motor speed and torque) to accurately reflect the real-time operating status of the motor and provide a benchmark for subsequent operating condition matching and parameter optimization.

[0034] Specifically, the load torque value is collected in real time by a torque sensor, with the sampling frequency set to 1kHz to adapt to the rapid changes in high dynamic operating conditions, ensuring that the torque data can follow the fluctuations of the operating conditions in real time; at the same time, the motor output speed is recorded by a speed sensor. The two together constitute complete operating condition parameters, which directly correspond to the operating condition characteristics of the motor in real time.

[0035] Simultaneously, control parameters of the target motor (including speed loop PI parameters and current loop PI parameters) are collected to capture the core adjustment parameters of motor control, providing a control dimension basis for loss prediction, weight calculation and parameter optimization.

[0036] Specifically, the speed loop PI parameters (including proportional and integral coefficients) and current loop PI parameters (including proportional and integral coefficients) are directly acquired through the parameter reading interface of the motor controller. During the acquisition process, it is necessary to ensure the real-time and accuracy of parameter reading to avoid the parameters from becoming out of sync with the actual operating status due to transmission delay. The acquisition frequency should be synchronized with the monitoring of operating parameters to achieve dynamic correspondence between control parameters and operating parameters.

[0037] In addition, a spatiotemporal synchronization mechanism needs to be established during the data acquisition process. Hardware trigger signals are used to align the timestamps of operating parameters (speed, torque) with those of control parameters (speed loop PI, current loop PI), keeping the time deviation within 1ms to ensure that each set of data accurately corresponds to the motor state at the same operating moment. For the acquired data, a sliding window filtering algorithm, a technique already in use, is employed for preprocessing. The window size is set to 50ms to reduce interference noise and improve data stability while ensuring data real-time performance.

[0038] This step enables precise perception of the motor's operating and control status, laying a solid data foundation for subsequent adjustments to control strategies based on historical data comparison, loss prediction, and dynamic optimization.

[0039] S120: Based on historical data, obtain multiple historical control parameters, corresponding multiple historical operating condition parameters, corresponding multiple historical loss values ​​and corresponding multiple historical current harmonic distortions, and perform cluster analysis to obtain a historical operating condition cluster set.

[0040] In this embodiment of the application, in the scenario of real-time optimization of control parameters of high dynamic motors, in order to avoid blind optimization of control parameters due to lack of operating condition reference, it is necessary to first extract complete historical data from the operation records of motors of the same model, and form a structured historical operating condition cluster set through clustering processing, so as to build a parameter optimization reference benchmark covering multiple operating scenarios.

[0041] Specifically, the first step is to identify motors with the same model as the target motor and extract data from their historical operating database. The extracted data must include multiple historical control parameters, as well as multiple historical operating condition parameters, multiple historical loss values, and multiple historical current harmonic distortion values ​​corresponding to each historical control parameter, to ensure that each set of data can fully reflect the motor's control input and performance output under a certain operating state.

[0042] Furthermore, after acquiring sufficient historical data, cluster analysis is performed on the extracted historical control parameters, corresponding historical loss values, and corresponding historical current harmonic distortions. During the clustering process, using the above three types of data as joint features, the K-means clustering algorithm is applied to group historical data with similar operating conditions into one category, forming multiple historical operating condition clusters.

[0043] Each historical operating condition cluster represents a specific type of motor operating condition. The cluster contains the corresponding historical control parameters, historical operating condition parameters, historical loss values, and historical current harmonic distortion data for that operating condition, which can clearly show the correspondence between motor control input, operating conditions, and performance output under that type of operating condition.

[0044] Finally, all generated historical operating condition clusters are integrated to form a unified historical operating condition cluster set. This set systematically organizes the historical operating data of motors of the same model, providing structured historical data support for subsequent matching of similar operating conditions with the real-time control parameters of the target motor and calculating and optimizing the target weight and control parameter weight.

[0045] Step S120 in the method provided in this application embodiment includes:

[0046] Collect multiple historical control parameters of motors of the same model as the target motor, and obtain multiple historical operating condition parameters, multiple historical loss values, and multiple historical current harmonic distortions corresponding to the multiple historical control parameters;

[0047] Based on the multiple historical operating condition parameters, cluster analysis is performed on the multiple historical control parameters, the corresponding multiple historical loss values ​​and the corresponding multiple historical current harmonic distortions to obtain multiple historical operating condition clusters.

[0048] The multiple historical operating condition clusters are integrated to obtain a historical operating condition cluster set.

[0049] In the method provided in this application embodiment, each historical operating condition cluster in the historical operating condition cluster set includes a cluster center. The control parameters corresponding to the cluster center are used as the typical control values ​​of the historical operating condition cluster, the historical loss values ​​corresponding to the cluster center are used as the typical loss values ​​of the historical operating condition cluster, and the historical current harmonic distortion corresponding to the cluster center is used as the typical current harmonic distortion values ​​of the historical operating condition cluster.

[0050] In this embodiment of the application, in order to provide historical data references that fit the actual operating scenario for the optimization of the control parameters of the target motor and avoid deviation in the optimization direction due to the lack of data support for similar operating conditions, it is necessary to first extract complete control and performance data from the historical operating records of motors of the same model, and then form a structured historical operating condition cluster set through clustering processing to build an accurate and reliable parameter optimization benchmark.

[0051] Specifically, the first step is to collect historical data on motors of the same model. The data collection focuses on motors with the exact same model as the target motor to ensure the adaptability of the historical data to the operating conditions and avoid the reduction in the reference value of the data due to differences in motor models.

[0052] Meanwhile, the collected data includes multiple historical control parameters, as well as multiple historical operating condition parameters, multiple historical loss values, and multiple historical current harmonic distortions that correspond one-to-one with each historical control parameter.

[0053] Historical loss values ​​include comprehensive loss data such as copper loss and iron loss, while historical operating parameters include motor speed and torque. Each set of data must be labeled with basic information such as the collection time and operating scenario to ensure the integrity and traceability of the data.

[0054] During the data collection process, the data sample size needs to be controlled. It should typically cover different operating states of the motor, such as low load, medium load, and high load, with a sample size of no less than 1,000 groups, in order to avoid the subsequent clustering results being biased due to insufficient data.

[0055] After data acquisition, based on the extracted historical operating condition parameters, cluster analysis was performed on the extracted historical control parameters, corresponding historical loss values, and corresponding historical current harmonic distortion to clearly present the correlation between control parameters and loss and current harmonic distortion under various operating conditions.

[0056] Specifically, cluster analysis uses historical operating parameters (motor speed and torque) as the core classification basis, and uses historical control parameters, historical loss values, and historical current harmonic distortion as correlation features. The K-means clustering algorithm is used to measure the similarity by calculating the Euclidean distance of historical operating parameters between different data groups, and historical data with similar operating conditions are grouped into one category to form multiple historical operating condition clusters.

[0057] Each historical operating condition cluster represents a specific type of motor operating condition, such as "low speed-low torque-low loss-low harmonics" operating condition, "medium speed-medium torque-medium loss-medium harmonics" operating condition, and "high speed-high torque-high loss-high harmonics" operating condition.

[0058] At the same time, each historical operating condition cluster will generate a cluster center. The control parameters corresponding to the cluster center will be used as the typical control values ​​for that cluster, the historical loss values ​​corresponding to the cluster center will be used as the typical loss values, and the historical current harmonic distortion corresponding to the cluster center will be used as the typical current harmonic distortion values, so as to quickly match similar operating conditions in the future.

[0059] Finally, the generated historical operating condition clusters are integrated to form a unified historical operating condition cluster set. During the integration process, the typical control values, typical loss values, and typical current harmonic distortion values ​​of each cluster need to be labeled, and the clusters are sorted according to the gradient order of operating condition characteristics to facilitate quick retrieval of the corresponding operating condition data in subsequent calls.

[0060] Through the above steps, the compatibility between historical data and target motor is ensured, and the structured classification of operating conditions is achieved through clustering, laying a data foundation for subsequent optimization target weight calculation and control parameter weight adjustment based on historical operating conditions.

[0061] S130: Based on the control parameters, predict the loss value, combine it with the corresponding historical operating condition cluster set, obtain the optimization target weight, and based on the control parameters and the corresponding historical operating condition cluster set, obtain the initial control parameter weight.

[0062] In this embodiment of the application, in order to clarify the priority order of reducing losses and ensuring torque under the current working conditions, and at the same time determine the initial adjustment priority of each control parameter, it is necessary to first obtain the real-time loss of the motor by prediction, then combine historical working condition data to match similar scenarios, and then calculate the optimization target weight and the initial control parameter weight to ensure that the subsequent parameter optimization direction is in line with the actual needs.

[0063] Specifically, a loss predictor is first built and trained based on machine learning. This loss predictor takes motor control parameters as input and loss values ​​as output, and is trained with a large amount of historical data to achieve accurate loss prediction capabilities. The collected real-time control parameters of the target motor (speed loop PI parameters and current loop PI parameters) are input into the loss predictor. The predictor calculates and outputs the corresponding predicted loss values ​​based on the parameter characteristics and the internal training model, thereby understanding the loss level of the motor under the current operating state.

[0064] Furthermore, based on the acquired real-time control parameters and predicted loss values, a search is conducted in the constructed historical operating condition cluster set to locate the historical operating condition cluster with the highest similarity. This cluster represents the historical scenario most similar to the current motor operating state, providing a reference for subsequent weight calculation.

[0065] After matching similar historical operating condition clusters, the typical control value and typical loss value of the cluster are obtained first, and the deviation between the predicted loss value and the typical loss value, and the deviation between the real-time control parameter and the typical control value are calculated respectively.

[0066] Among them, the deviation between the predicted loss value and the typical loss value directly reflects the degree of deviation of the current loss state from similar historical operating conditions. The larger the deviation, the more significant the current loss optimization space is. Based on this, the target weight for loss optimization can be determined, and then the target weight for current harmonic distortion optimization can be determined based on the target weight for loss optimization.

[0067] Specifically, if the predicted loss value deviates significantly from the typical loss value, the weight of the loss optimization target should be higher, meaning that reducing loss is of higher priority under the current operating conditions. Conversely, if the deviation is small, the weight of the loss optimization target should be appropriately reduced, and the weight of the current harmonic distortion optimization target should be adjusted accordingly to clarify the order of importance between optimizing loss and optimizing current harmonic distortion under the current operating conditions.

[0068] Simultaneously, the initial control parameter weights are obtained based on the deviation between the real-time control parameters and the typical control values. The larger the deviation of the control parameter, the more significant the deviation between its current value and the optimal control parameter under similar historical operating conditions, and the more prominent the impact on motor performance may be. Therefore, it is given a higher initial control parameter weight and is given priority in subsequent parameter adjustment. Control parameters with smaller deviations have relatively lower initial control parameter weights and are given lower priority in subsequent adjustments.

[0069] This step, by first predicting the loss value, then matching historical operating conditions, and finally calculating the optimization target weight and the initial control parameter weight, clarifies the optimization direction of the current operating condition and determines the initial adjustment priority of each control parameter. This provides a key basis for the subsequent construction of dynamic optimization targets and adjustment of control parameter weights, ensuring that the parameter optimization process is more targeted and reasonable.

[0070] Step S130 in the method provided in this application embodiment includes:

[0071] Based on machine learning, obtain a loss predictor;

[0072] The control parameters are input into the loss predictor to predict the loss and obtain the predicted loss value.

[0073] Based on the control parameters and predicted loss values, a search is performed in the historical operating condition cluster set to obtain the historical operating condition cluster with the highest similarity.

[0074] Obtain the typical control value and typical loss value of the historical operating condition cluster, calculate the deviation between the predicted loss value and the typical loss value, and calculate the deviation between the control parameter and the typical control value;

[0075] Based on the deviation between the predicted loss value and the typical loss value, the target weight for loss optimization is obtained, and based on the target weight for loss optimization, the target weight for current harmonic distortion optimization is obtained.

[0076] The initial control parameter weights are obtained based on the control parameters and their deviations from the typical control values.

[0077] In this embodiment of the application, in order to accurately grasp the current loss status of the high dynamic motor, clarify the optimization priority of reducing loss and reducing current harmonic distortion under the current operating conditions, and determine the importance of the initial adjustment of each control parameter, it is necessary to build a loss predictor through machine learning, match similar scenarios with historical operating data, and then calculate the target weight of loss optimization and the target weight of current harmonic distortion optimization, so as to ensure that the subsequent parameter optimization direction is in line with the actual operating needs of the motor and avoid blind optimization.

[0078] Specifically, a loss predictor is first obtained based on machine learning to accurately capture the mapping relationship between control parameters and loss values ​​under different operating conditions.

[0079] The method provided in this application embodiment, "obtaining a loss predictor based on machine learning", includes:

[0080] Based on machine learning, multiple loss prediction branches are constructed;

[0081] Multiple historical operating condition clusters are used to train the multiple loss prediction branches respectively, thereby obtaining multiple loss prediction branches.

[0082] The multiple loss prediction branches are integrated to obtain a loss predictor.

[0083] In this embodiment of the application, in order to adapt the loss prediction to the characteristics of different operating conditions of the motor and avoid the prediction deviation of a single prediction model under multiple operating conditions, it is necessary to construct and train loss prediction branches for different historical operating condition clusters, and then integrate them to form a loss predictor covering all operating conditions, so as to ensure the accuracy of the prediction of real-time motor loss.

[0084] First, multiple loss prediction branches are constructed based on machine learning. Each branch uses a BP neural network model as its basic framework, and will be adapted to the loss patterns of each historical operating condition cluster.

[0085] In the specific construction, the input and output layers of the model need to be determined according to the correlation characteristics between the motor control parameters (speed loop PI parameters, current loop PI parameters) and the loss values. The input layer is set as the speed loop PI parameters (proportional coefficient, integral coefficient) and the current loop PI parameters (proportional coefficient, integral coefficient), and the output layer is set as the corresponding loss values. At the same time, a reasonable number of hidden layer nodes should be set to ensure that the model can capture the nonlinear relationship between the control parameters and the loss values.

[0086] Furthermore, multiple historical operating condition clusters are used to train multiple loss prediction branches separately. Before training, each loss prediction branch needs to be assigned corresponding historical operating condition cluster data to ensure that each loss prediction branch learns the loss characteristics under only one type of operating condition.

[0087] During training, the model parameters are iteratively optimized using the gradient descent method, with the loss function set to the mean square error between the predicted loss value and the actual historical loss value. For example, the learning rate is set to 0.01 and the number of iterations is set to 1000, until the loss function value is less than the preset threshold of 0.001. At this point, the loss prediction branch can accurately fit the mapping relationship between the control parameters and the loss value under the corresponding working condition.

[0088] After training each branch, the obtained loss prediction branches are integrated to obtain a loss predictor. During integration, a working condition matching module needs to be added to the loss predictor. This module can calculate the similarity between the input real-time control parameters and the typical control values ​​of each historical working condition cluster, and automatically select the prediction branch corresponding to the cluster with the highest similarity for loss calculation.

[0089] At the same time, during the branch training and integration process, it is necessary to ensure the validity and consistency of the data in order to avoid the model training effect being affected by the difference in the magnitude of parameters.

[0090] After integration, the predictor needs to be tested as a whole. Historical data from each cluster that was not used in training should be selected as the test set. If the loss prediction error of each loss prediction branch on the test set is less than 5%, the loss predictor is deemed qualified and can be put into use.

[0091] If the error of a certain loss prediction branch exceeds the preset threshold, the model parameters of that branch need to be readjusted, such as increasing the number of iterations or adjusting the number of hidden layer nodes, and then retrained until the error requirement is met.

[0092] Through the above steps, the constructed loss predictor can call the appropriate prediction branch for different operating conditions, effectively avoiding the prediction deviation of a single model under multiple operating conditions and ensuring the accuracy of real-time motor loss prediction.

[0093] Furthermore, the real-time control parameters are input into the constructed loss predictor to predict the loss value.

[0094] Specifically, the preprocessed control parameters for the real-time operation of the motor are first input into the loss predictor. Its internal operating condition matching module analyzes the parameter characteristics, compares them with the typical control values ​​of each historical operating condition cluster, selects the prediction branch corresponding to the cluster with the closest characteristics to carry out the calculation, and finally outputs the predicted loss value under the current control parameters to clarify the real-time loss level of the motor.

[0095] Furthermore, based on the control parameters and predicted loss values, a search is performed in the historical operating condition cluster set to obtain the historical operating condition cluster with the highest similarity. The search process uses the control parameters and predicted loss values ​​as joint search criteria, and determines the matching objects by calculating the comprehensive similarity between these data and the typical control values ​​and typical loss values ​​of each historical operating condition cluster.

[0096] Furthermore, the typical control values ​​and typical loss values ​​of the historical operating condition cluster are obtained, and then the deviation between the predicted loss value and the typical loss value is calculated. At the same time, the deviation between the control parameters and the typical control values ​​is also calculated.

[0097] Specifically, the deviation is calculated using the absolute difference method. The deviation between the predicted loss value and the typical loss value is the absolute value of the difference between the two values. For example, the deviation between the predicted loss value of 172W and the typical loss value of 180W is 8W (|172-180|=8).

[0098] In addition, the deviations between the control parameters and the typical control values ​​are calculated separately according to the parameter type. For example, the deviation of the PI proportional coefficient of the speed loop is |2.7-2.8|=0.1, and the deviation of the PI integral coefficient of the current loop is |1.15-1.2|=0.05. All deviation results are retained to two decimal places to ensure calculation accuracy.

[0099] Furthermore, based on the deviation between the obtained predicted loss value and the typical loss value, the target weight for loss optimization is obtained, and based on the target weight for loss optimization, the target weight for current harmonic distortion optimization is obtained, so as to clarify the priority order of reducing loss and reducing current harmonic distortion under the current operating conditions.

[0100] The target weight for loss optimization is calculated using the formula "deviation / typical loss value". For example, the ratio of 8W to 180W is 0.044 (8 / 180=0.044), meaning the target weight for loss optimization is 0.044. The target weight for current harmonic distortion optimization is 1 minus the target weight for loss optimization, i.e., 1-0.044=0.956. The sum of the two is always 1, thus clarifying the priority of optimizing loss and optimizing current harmonic distortion under the current operating conditions.

[0101] If the predicted loss value deviates significantly from the typical loss value, such as a deviation of 30W and a typical loss value of 150W (30 / 150), with a ratio of 0.2, then the weight of the loss optimization target is 0.2, indicating that more attention needs to be paid to reducing loss. If the deviation is small, such as a deviation of 3W and a typical loss value of 200W, with a ratio of 0.015 (3 / 200), then the weight of the loss optimization target is 0.015, in which case reducing current harmonic distortion is given priority.

[0102] Furthermore, the initial control parameter weights are obtained based on the control parameters and their deviations from the typical control values.

[0103] Specifically, the weights of each control parameter are determined during the calculation based on the degree of deviation. The larger the deviation, the more significant the difference between the current parameter and the optimal parameter under similar historical conditions, and the more prominent the impact on motor performance. Therefore, a higher initial weight is assigned, and the parameter is included in the adjustment range first. The smaller the deviation, the lower the initial weight, and the lower the adjustment priority.

[0104] For example, if the deviation of the speed loop PI proportional coefficient is 0.1 and the deviation of the current loop PI integral coefficient is 0.05, and the deviation percentages are 66.7% and 33.3% respectively, then the corresponding initial control parameter weights are 0.667 and 0.333 respectively, and the sum of the initial weights of all control parameters is 1. If the deviation of a certain parameter is 0, its initial weight is set to 0 and it will not participate in this round of adjustment, thereby reducing invalid calculations and improving optimization efficiency.

[0105] Ultimately, through the above steps, the priority of the optimization objectives of loss and current harmonic distortion was quantified, and the initial adjustment sequence of control parameters was determined, providing a clear and reliable decision-making basis for the subsequent real-time optimization of motor control parameters.

[0106] S140: Based on the optimization target weight, obtain the dynamic optimization target, and based on the dynamic optimization target, obtain the influence degree of the control parameters, adjust and calculate the initial control parameter weight, and obtain the control parameter weight;

[0107] In this embodiment of the application, in order to adapt the optimization target to the changes in the real-time operating state of the motor and accurately measure the actual impact of each control parameter on the optimization target, it is necessary to first determine the dynamic optimization target by combining the optimization target weight, then calculate the influence degree of the control parameters and adjust the initial control parameter weights to form control parameter weights that fit the current working conditions, so as to provide a scientific adjustment basis for subsequent parameter optimization.

[0108] Specifically, firstly, based on the optimization target weights, dynamic optimization targets are calculated and obtained. These dynamic optimization targets include loss optimization targets and current harmonic distortion optimization targets. The optimization target weights include the previously determined loss optimization target weights and current harmonic distortion optimization target weights. The dynamic optimization targets combine these two weights with the performance benchmarks of motor operation, transforming them into specific and actionable optimization indicators.

[0109] Furthermore, the influence of control parameters is obtained based on the dynamic optimization objective. Specifically, the influence of control parameters is obtained by calculating the correlation coefficients of different control parameters relative to the dynamic optimization objective within historical operating condition clusters.

[0110] Finally, based on the influence of the control parameters, the initial control parameter weights are adjusted to obtain the control parameter weights. Note that the initial control parameter weights are calculated only based on the deviation between the control parameters and typical control values, without considering the actual impact of the parameters on the optimization objective; therefore, they need to be corrected in conjunction with the influence.

[0111] During the correction, the initial control parameter weights are multiplied by the influence of the control parameters, and the calculation results are normalized to ensure that the sum of the weights of each control parameter is 1, so that the corrected control parameter weights can better reflect the importance of each parameter to the current dynamic optimization objective.

[0112] This step transforms abstract weights into specific dynamic optimization objectives, then combines correlation analysis to quantify the influence of control parameters, and finally corrects the initial control parameter weights to ensure that the obtained control parameter weights not only fit the real-time operating state of the motor, but also accurately match the optimization requirements.

[0113] Step S140 in the method provided in this application embodiment includes:

[0114] Based on the aforementioned optimization target weights, dynamic optimization targets are calculated and obtained, wherein the dynamic optimization targets include loss optimization targets and current harmonic distortion optimization targets;

[0115] Based on the dynamic optimization objective, the influence degree of the control parameters is obtained, wherein the influence degree of the control parameters is obtained by calculating the correlation coefficient of different control parameters relative to the dynamic optimization objective in the historical operating condition cluster, wherein the correlation coefficient is obtained by weighting the loss correlation coefficient and the current harmonic distortion correlation coefficient.

[0116] Based on the influence of the control parameters, the initial control parameter weights are adjusted and calculated to obtain the control parameter weights.

[0117] In this embodiment of the application, in order to ensure that the optimization of motor control parameters can meet the performance requirements under real-time operating conditions and avoid the optimization direction from deviating from the actual goal, it is necessary to first convert the optimization target weight into a specific dynamic optimization target, then calculate the influence of control parameters and correct the initial control parameter weights, so as to form scientific control parameter weights and provide an accurate basis for subsequent parameter adjustments.

[0118] Specifically, the dynamic optimization objectives are first obtained by calculating the weights of the optimization objectives. These dynamic optimization objectives include loss optimization objectives and current harmonic distortion optimization objectives. During the calculation, the abstract weights are transformed into quantifiable indicators by combining the current operating performance benchmark of the motor with the typical values ​​of historical operating condition clusters (typical values ​​of loss and typical values ​​of current harmonic distortion).

[0119] For example, if the current predicted loss value is 140W, the typical loss value of the historical similar working condition cluster is 120W, and the weight of the loss optimization target is 0.167 ((140-20) / 120≈0.167), then the loss optimization target is set as "to reduce the current predicted loss value to within 137W (140-(140-120)×0.167≈137)".

[0120] Furthermore, if the current harmonic distortion is 5.2%, the typical value of the current harmonic distortion in the historical similar operating condition cluster is 3.8%, and the weight of the current harmonic distortion optimization target is 0.833 (1-0.167=0.833), then the current harmonic distortion optimization target is set as "to reduce the current harmonic distortion to within 4.03% (5.2%-(5.2%-3.8%)×0.833≈4.03%)".

[0121] This calculation method ensures that the dynamic optimization objective is both related to the historical best state and matches the current working conditions, thus avoiding optimization failure caused by setting the objective too high or too low.

[0122] Furthermore, the influence of control parameters is obtained based on the dynamic optimization objective, so as to accurately quantify the actual effect of each control parameter on the current optimization objective.

[0123] In the method provided in this application embodiment, "obtaining the influence degree of the control parameter based on the dynamic optimization objective" includes:

[0124] Based on the historical operating condition clusters, the loss correlation coefficient of the control parameters relative to the loss optimization target is obtained, and the current harmonic distortion correlation coefficient of the control parameters relative to the current harmonic distortion optimization target is obtained.

[0125] The influence of the control parameter is obtained by weighting the loss correlation coefficient and the current harmonic distortion correlation coefficient. The weights in the weighting calculation are obtained based on the loss optimization target weight and the current harmonic distortion optimization target weight.

[0126] In this embodiment of the application, in order to accurately quantify the actual effect of each control parameter on the dynamic optimization target and avoid optimization deviation caused by judging the importance of parameters solely through experience, it is necessary to analyze the intrinsic relationship between control parameters and loss and current harmonic distortion targets from historical operating data, and integrate them with the priority of the current optimization target to form a quantitative index that can truly reflect the impact of parameters, providing a scientific basis for subsequent weight adjustment.

[0127] Specifically, firstly, based on historical operating conditions, clusters are formed to obtain the loss correlation coefficient of the control parameters relative to the loss optimization target, and then the current harmonic distortion correlation coefficient of the control parameters relative to the current harmonic distortion optimization target is obtained.

[0128] The historical operating condition cluster contains a large amount of historical control parameters and corresponding loss values ​​and current harmonic distortion values ​​for this type of operating condition, providing a reliable sample for correlation analysis. During the calculation, the Pearson correlation coefficient analysis method is used. This method can effectively measure the degree of linear correlation between two variables, with results ranging from [-1, 1]. Positive values ​​indicate positive correlation, negative values ​​indicate negative correlation, and the larger the absolute value, the stronger the correlation.

[0129] Furthermore, after obtaining the correlation coefficients of loss and current harmonic distortion, a weighted calculation is performed on both to obtain the influence of the control parameters. The weights in the weighted calculation are obtained based on the target weights for loss optimization and current harmonic distortion optimization to ensure that the calculation results are consistent with the priority of the current optimization objectives.

[0130] Specifically, the formula for calculating the influence of control parameters can be expressed as "Influence of control parameters = (Loss correlation coefficient × Loss optimization target weight) + (Current harmonic distortion correlation coefficient × Current harmonic distortion optimization target weight)".

[0131] For example, if the current loss optimization target weight is 0.3, the current harmonic distortion optimization target weight is 0.7, the loss correlation coefficient of the speed loop PI parameter is -0.78, and the current harmonic distortion correlation coefficient is -0.65, then the influence of its control parameters is "-0.78×0.3+(-0.65)×0.7≈-0.689"; the loss correlation coefficient of the current loop PI parameter is -0.83, and the current harmonic distortion correlation coefficient is -0.72, then its influence is "-0.83×0.3+(-0.72)×0.7≈-0.777". In the calculation results, the absolute value of the influence of the current loop PI parameter is higher, indicating that its comprehensive effect on the current dynamic optimization target is more significant; the absolute value of the influence of the speed loop PI parameter is lower, indicating that its effect is slightly weaker under the current optimization priority.

[0132] Furthermore, during the weighted calculation process, it is necessary to ensure the matching between historical operating condition clusters and current operating conditions. That is, the historical operating condition clusters used to calculate the correlation coefficient must be the clusters with the highest similarity to the current control parameters and predicted loss values, so as to ensure that the correlation analysis is based on similar operating condition data and avoid the distortion of correlation coefficients due to differences in operating conditions.

[0133] At the same time, the rationality of the calculated influence of the control parameters needs to be verified. If the absolute value of the influence of a certain control parameter is too large (e.g., exceeding 0.9, which is outside the reasonable range of strong correlation) or too small (e.g., below 0.1, which is of no practical significance for optimization due to weak correlation), the completeness of the historical data and the accuracy of the correlation calculation need to be checked. If necessary, the sample size should be increased for supplementary verification to ensure the reliability of the influence.

[0134] Through the above steps, the inherent relationship between control parameters and optimization objectives is first extracted from historical data, and then weighted and integrated with the current optimization priority. The resulting control parameter influence can reflect the objective effect of the parameters on the objectives and match the real-time optimization needs, providing a precise quantitative basis for subsequent adjustment of the initial control parameter weights.

[0135] Furthermore, based on the obtained influence of the control parameters, the initial control parameter weights are readjusted to obtain the control parameter weights, so as to avoid the misalignment of the optimization direction caused by the initial control parameter weights relying only on the deviation and ignoring the influence of the parameters.

[0136] Specifically, the absolute value of the influence of the control parameters should be taken first, because the sign of the influence of the control parameters only represents the direction of the parameter's effect on the target (i.e., positive or negative correlation), while the weight adjustment should focus on the intensity of the effect. Taking the absolute value can eliminate directional interference and only retain the magnitude of the influence.

[0137] Furthermore, the initial control parameter weights are multiplied by the corresponding absolute values ​​of their influence to obtain the preliminary adjustment weights.

[0138] For example, if the initial control parameter weights are speed loop PI parameter 0.4 and current loop PI parameter 0.6, and the corresponding absolute values ​​of influence are 0.689 and 0.777 respectively, then the initial adjustment weights are speed loop PI parameter "0.4×0.689≈0.276" and current loop PI parameter "0.6×0.777≈0.466".

[0139] After obtaining the initial adjusted weights, they need to be normalized so that the sum of the weights of all control parameters is 1, in order to ensure that the weight allocation is comparable and reasonable.

[0140] Specifically, the normalization calculation formula can be expressed as "control parameter weight = initial adjustment weight / sum of all initial adjustment weights".

[0141] Taking the above data as an example, the initial weight sum is "0.276 + 0.466 ≈ 0.742". Therefore, the normalized control parameter weights are "0.276 / 0.742 ≈ 0.372" for the speed loop PI parameter and "0.466 / 0.742 ≈ 0.628" for the current loop PI parameter. Through normalization, the current loop PI parameter, due to its higher deviation and stronger effect on the target, has a significantly increased weight and becomes the core adjustment target for subsequent parameter optimization; the speed loop PI parameter, due to its slightly weaker effect on the target, has a slightly lower final weight and is adjusted at a relatively lower priority.

[0142] Similarly, during the adjustment calculation process, it is necessary to ensure the consistency of data sources. That is, the parameter deviation used to calculate the initial control parameter weights and the historical operating condition clusters used to calculate the influence of the control parameters must come from the same set of historical data with the highest similarity to the current operating conditions, in order to avoid adjustment errors caused by different data sources.

[0143] Through the above steps, the final control parameter weights take into account both the degree to which the parameters deviate from the optimal state and their ability to influence the optimization objective. This provides a clear priority basis for the subsequent precise optimization of motor control parameters, ensuring that the adjustment actions can efficiently target the dynamic optimization objective.

[0144] S150: With the dynamic optimization objective as the goal and the control parameter weights as constraints, optimize the motor control parameters, obtain optimized control parameters, and use the optimized control parameters to optimize the control parameters of the high-dynamic motor.

[0145] In this embodiment of the application, in order to ensure that the optimization of control parameters can accurately target the dynamic goals of reducing losses and current harmonic distortion, and to prioritize the adjustment of key control parameters that have a significant impact on the goals, it is necessary to combine the dynamic optimization goals and the weights of control parameters to form optimized control parameters that meet the needs of the current operating conditions, so as to achieve a precise improvement in motor performance.

[0146] Specifically, the step size for optimizing the control parameters is first calculated based on the weights of the control parameters and the typical values ​​of historical operating condition clusters.

[0147] Among them, the control parameter weights determine the adjustment priority of each parameter. Parameters with higher weights need to be allocated a larger adjustment range, while parameters with lower weights use a smaller adjustment step size to avoid excessive adjustment of minor parameters from interfering with stability.

[0148] Furthermore, with the goal of dynamic optimization, the motor control parameters are optimized by adjusting the step size, and the optimized control parameters are obtained and applied to the real-time control of the high-dynamic motor to make the motor operating state move closer to the dynamic optimization goal.

[0149] This step, guided by dynamic optimization objectives and constrained by control parameter weights, achieves targeted optimization of motor control parameters. It ensures the priority adjustment of key parameters and avoids performance fluctuations through step size control. The resulting optimized control parameters can accurately adapt to the current operating conditions, providing a reliable guarantee for the efficient and stable operation of high-dynamic motors.

[0150] Step S150 in the method provided in this application embodiment includes:

[0151] The optimization adjustment step size of the control parameters is calculated based on the weights of the control parameters and the typical values ​​of the historical operating condition clusters.

[0152] With the aforementioned dynamic optimization objective as the goal, the motor control parameters are optimized by adjusting the step size to obtain the optimized control parameters;

[0153] The optimized control parameters are used to optimize the control parameters of the high-dynamic motor.

[0154] In this embodiment of the application, in order to ensure that the optimization of motor control parameters not only conforms to the current dynamic optimization goal, but also prioritizes the adjustment of key parameters that have a significant impact on the goal, it is necessary to determine a reasonable adjustment range by combining the control parameter weights and typical values ​​of historical operating conditions, and then optimize the control parameters according to the goal orientation and apply them to motor control, so as to achieve precise improvement of motor performance and stable operation.

[0155] Specifically, the step size for optimizing the control parameters is first calculated based on the obtained control parameter weights and the typical values ​​of historical operating condition clusters.

[0156] Among them, the control parameter weights reflect the degree of influence of each parameter on the dynamic optimization objective. Parameters with higher weights need to be given greater adjustment flexibility in order to quickly approach the optimization objective. The typical value of the historical working condition cluster is the optimal parameter benchmark under this type of working condition. The step size calculation needs to be carried out around the typical value in order to avoid the adjustment range being too large and exceeding the reasonable operating range.

[0157] In the specific calculation, the control parameter weights are converted into adjustment coefficients proportionally, and then the step size is determined by combining the typical values ​​of historical operating condition clusters. That is, step size = typical value × (control parameter weight / highest control parameter weight) × basic adjustment ratio.

[0158] The base adjustment ratio is a fixed value (0.05) used to standardize the adjustment scale. The ratio of "control parameter weight / highest control parameter weight" ensures that parameters with higher weights receive relatively larger adjustment coefficients. For example, the current loop PI parameter weight is 0.628 (highest weight), and its adjustment coefficient is 0.628 / 0.628=1; the speed loop PI parameter weight is 0.372, and its adjustment coefficient is 0.372 / 0.628≈0.592.

[0159] For example, if the typical value of the current loop PI proportional coefficient in the historical operating condition cluster is 3.5, the typical value of the speed loop PI proportional coefficient is 2.8, and the basic adjustment ratio is 0.05, then the optimization adjustment step size of the current loop PI proportional coefficient is "3.5×1×0.05=0.175", and the speed loop PI proportional coefficient is "2.8×0.592×0.05≈0.083".

[0160] Ultimately, the optimized adjustment step size determined through the above method ensures that the adjustment space of the high-weight parameters (current loop PI parameters) is maximized, while the adjustment range of the low-weight parameters (speed loop PI parameters) is reduced proportionally, which is consistent with the difference in the degree of influence of control parameters on the dynamic optimization objective.

[0161] Furthermore, after determining the optimal adjustment step size, with the goal of dynamic optimization, the optimal adjustment step size is used to optimize the motor control parameters and obtain the optimal control parameters.

[0162] Specifically, the optimization process needs to be adjusted step by step according to priority. First, the current loop PI parameter with the highest weight is adjusted first, and the adjustment direction is determined according to the dynamic optimization target: if the current predicted loss is higher than the target value, the current loop PI ratio coefficient is gradually reduced by step size. Each adjustment is made while monitoring the changes in loss and current harmonic distortion. If the loss decreases and the current harmonic distortion does not exceed the standard, the adjustment continues.

[0163] If the current harmonic distortion approaches the threshold (1.1 times the typical value of current harmonic distortion in the historical operating condition cluster), pause the adjustment of the current loop PI parameters and instead adjust the speed loop PI parameters, which have the next highest weight. Balance loss and harmonic control by optimizing the speed loop PI integral coefficients. After each adjustment, verify whether the parameters conform to the dynamic optimization objective until the parameter adjustments meet the requirements and obtain the final optimized control parameters.

[0164] Furthermore, the obtained optimized control parameters are used to optimize the control parameters of the high-dynamic motor.

[0165] The method provided in this application embodiment, "optimizing the control parameters of the high-dynamic motor using the optimized control parameters", includes:

[0166] Obtain the control results corresponding to the optimized control parameters;

[0167] Calculate the difference between the control result corresponding to the optimized control parameters and the dynamic target. When the difference is less than or equal to the allowable error, the optimized control ends.

[0168] When the difference is greater than the allowable error, the optimized control parameter is used as the control parameter, and the control parameter optimization is performed iteratively until the difference is less than or equal to the allowable error.

[0169] In this embodiment of the application, in order to ensure that the optimized control parameters can continuously adapt to the operating requirements of the high dynamic motor and avoid the performance deviation from the dynamic target due to the fluctuation of the operating conditions after a single optimization, it is necessary to realize the closed-loop control of parameter optimization by real-time monitoring of the control results, comparison of the target difference and iterative optimization, so as to ensure that the motor operates stably in the high-efficiency range for a long time.

[0170] Specifically, the first step is to obtain the control results corresponding to the optimized control parameters. That is, after applying the optimized control parameters to the high-dynamic motor, key operating data needs to be collected in real time through the sensors of the motor control device. This data includes the actual loss value of the motor, the actual current harmonic distortion rate, etc. The acquisition frequency is set to 100kHz to ensure that the performance changes after parameter adjustment can be captured in a timely manner.

[0171] Furthermore, the difference between the control result corresponding to the optimized control parameters and the dynamic target is calculated, and it is determined whether the difference is less than or equal to the allowable error. The dynamic target includes both loss optimization and current harmonic distortion optimization targets; the difference calculation needs to be performed separately for each target.

[0172] Specifically, the loss difference is "actual loss value - loss optimization target value", and the current harmonic distortion difference is "actual current harmonic distortion rate - current harmonic distortion optimization target value" (if the actual value is lower than the target value, the difference is 0, and only the performance exceeding the standard is considered).

[0173] In addition, the allowable error is set according to the performance requirements of motor operation. The allowable error for loss is usually 5% of the target value for loss optimization, and the allowable error for current harmonic distortion is 10% of the target value for current harmonic distortion optimization. These two error ranges are determined based on a large number of engineering practices, which can ensure that the motor performance meets the requirements and avoid frequent iterations caused by excessive pursuit of precision.

[0174] Conversely, if the calculated difference is greater than the allowable error, the current optimized control parameter is used as the new control parameter, and the control parameter optimization process is iteratively executed until the difference is less than or equal to the allowable error.

[0175] During specific iterations, the current optimized control parameters are used as the initial control parameters. The control parameter weights are recalculated, the adjustment step size is optimized, and the parameters are optimized again. After reapplication, the control results are collected, and the difference is calculated. If the condition is still not met, the iteration continues until the difference meets the allowable error requirements.

[0176] During iterative optimization, the upper limit of the number of iterations needs to be controlled (usually set to 5). If, after 5 iterations, the difference between the control result and the dynamic target is still greater than the allowable error, it is necessary to check whether the dynamic target setting is reasonable, whether the control parameter weight calculation is accurate, or whether there are any abnormalities in the sensor data. After troubleshooting, the optimization process should be restarted to avoid meaningless repeated iterations.

[0177] Finally, after completing iterative optimization and confirming that the control results meet the dynamic target and allowable error requirements, the final optimized control parameters are applied to the real-time control of the high-dynamic motor to achieve precise dynamic optimization of the control parameters and ensure that the motor maintains a high-efficiency operating state with low loss and low harmonics under all operating conditions.

[0178] The embodiments of this application, through the specific implementation methods described above, achieve the following technical effects:

[0179] This application proposes a real-time optimization method for high-dynamic motor control parameters to reduce losses. First, the current operating parameters of the target motor are obtained, and the control parameters of the target motor are collected. During data collection, a spatiotemporal alignment mechanism is established simultaneously, and data reliability is ensured through filtering preprocessing and validity verification. Then, based on historical data, multiple historical control parameters, corresponding historical operating parameters, corresponding historical loss values, and corresponding historical current harmonic distortions are obtained, and cluster analysis is performed to obtain a historical operating condition cluster set. Subsequently, a multi-branch loss predictor is used to match the corresponding operating condition branch according to the real-time parameters to obtain the predicted loss value, which is then calculated in conjunction with similar historical operating conditions. The system first identifies losses and parameter deviations, then determines the target weights for loss optimization, current harmonic distortion optimization, and initial control parameter weights. Next, it calculates the correlation coefficient between control parameters and dynamic optimization targets based on historical operating data, weights the control parameter influence, corrects and normalizes the initial control parameter weights, forming the final control parameter weights. Finally, guided by the dynamic optimization targets, it calculates the optimization adjustment step size for each parameter according to its weight, optimizes it step by step according to priority, and iterates by collecting control results in real time and comparing them with the dynamic targets until the results meet the allowable error. The optimized control parameters are then used for the real-time control of high-dynamic motors.

[0180] The method provided in this application, through the technical solution of "operating condition and control parameter acquisition - historical operating condition clustering - loss prediction and initial weight calculation - dynamic target and weight adjustment - control parameter optimization and iteration", solves the problems of traditional motor control parameter optimization relying on experience, high blindness, and difficulty in adapting fixed parameters to high dynamic operating condition fluctuations. It significantly improves the accuracy and real-time performance of control parameter optimization, effectively reduces motor losses and current harmonic distortion, and provides reliable technical support for high dynamic motor application fields such as new energy vehicle drive motors and industrial robot servo motors.

[0181] Example 2, as shown in the appendix Figure 2 As shown, based on the inventive concept of a real-time optimization method for high dynamic motor control parameters to reduce losses provided in Embodiment 1, this application also provides a real-time optimization system for high dynamic motor control parameters to reduce losses, specifically including:

[0182] The operating condition and parameter acquisition module 01 is used to acquire the current operating condition parameters of the target motor and to acquire the control parameters of the target motor. The control parameters include speed loop PI parameters and current loop PI parameters, and the operating condition parameters include motor speed and torque.

[0183] The historical operating condition clustering module 02 is used to obtain multiple historical control parameters, corresponding multiple historical operating condition parameters, corresponding multiple historical loss values ​​and corresponding multiple historical current harmonic distortions based on historical data, and to perform cluster analysis to obtain a historical operating condition cluster set.

[0184] The loss prediction and weight calculation module 03 is used to predict the loss value based on the control parameters, obtain the optimization target weight by combining the corresponding historical operating condition cluster set, and obtain the initial control parameter weight based on the control parameters and the corresponding historical operating condition cluster set.

[0185] The dynamic target and weight adjustment module 04 is used to obtain a dynamic optimization target based on the optimization target weight, and to obtain the influence degree of the control parameters based on the dynamic optimization target, adjust and calculate the initial control parameter weight, and obtain the control parameter weight.

[0186] The control parameter optimization module 05 is used to optimize the motor control parameters with the dynamic optimization target as the objective and the control parameter weight as the constraint, to obtain optimized control parameters, and to use the optimized control parameters to optimize the control parameters of the high dynamic motor.

[0187] In one embodiment, the historical operating condition clustering module 02 is also used for:

[0188] Multiple historical control parameters of motors of the same model as the target motor are collected, and multiple historical operating condition parameters, multiple historical loss values, and multiple historical current harmonic distortions corresponding to the multiple historical control parameters are obtained. Based on the multiple historical operating condition parameters, cluster analysis is performed on the multiple historical control parameters, the multiple historical loss values, and the multiple historical current harmonic distortions to obtain multiple historical operating condition clusters. The multiple historical operating condition clusters are integrated to obtain a historical operating condition cluster set.

[0189] Furthermore, the historical operating condition clustering module 02 also includes:

[0190] In the historical operating condition cluster set, each historical operating condition cluster includes a cluster center. The control parameters corresponding to the cluster center are used as the typical control values ​​of the historical operating condition cluster, the historical loss values ​​corresponding to the cluster center are used as the typical loss values ​​of the historical operating condition cluster, and the historical current harmonic distortion corresponding to the cluster center is used as the typical current harmonic distortion values ​​of the historical operating condition cluster.

[0191] In one embodiment, the loss prediction and weight calculation module 03 is also used for:

[0192] Based on machine learning, a loss predictor is obtained; the control parameters are input into the loss predictor to predict the loss and obtain the predicted loss value; based on the control parameters and the predicted loss value, a search is performed in the historical operating condition cluster set to obtain the historical operating condition cluster with the highest similarity; the typical control value and typical loss value of the historical operating condition cluster are obtained, the deviation between the predicted loss value and the typical loss value are calculated, and the deviation between the control parameters and the typical control value are also calculated; based on the deviation between the predicted loss value and the typical loss value, the loss optimization target weight is obtained, and based on the loss optimization target weight, the current harmonic distortion optimization target weight is obtained; based on the control parameters and the deviation from the typical control value, the initial control parameter weight is obtained.

[0193] Furthermore, the loss prediction and weight calculation module 03 also includes:

[0194] Based on machine learning, multiple loss prediction branches are constructed; the multiple historical operating condition clusters are used to train the multiple loss prediction branches respectively to obtain multiple loss prediction branches; the multiple loss prediction branches are integrated to obtain a loss predictor.

[0195] In one embodiment, the dynamic target and weight adjustment module 04 is further configured to:

[0196] Based on the optimization target weights, a dynamic optimization target is calculated and obtained, wherein the dynamic optimization target includes a loss optimization target and a current harmonic distortion optimization target; based on the dynamic optimization target, the influence degree of the control parameters is obtained, wherein the influence degree of the control parameters is obtained by calculating the correlation coefficients of different control parameters relative to the dynamic optimization target in the historical operating condition cluster, wherein the correlation coefficients are obtained by weighting the loss correlation coefficient and the current harmonic distortion correlation coefficient; based on the influence degree of the control parameters, the initial control parameter weights are adjusted and calculated to obtain the control parameter weights.

[0197] Furthermore, the dynamic target and weight adjustment module 04 also includes:

[0198] Based on the historical operating condition clusters, the loss correlation coefficient of the control parameters relative to the loss optimization target is obtained, and the current harmonic distortion correlation coefficient of the control parameters relative to the current harmonic distortion optimization target is obtained. The loss correlation coefficient and the current harmonic distortion correlation coefficient are weighted and calculated to obtain the influence degree of the control parameters. The weights in the weighting calculation are obtained based on the weights of the loss optimization target and the current harmonic distortion optimization target.

[0199] In one embodiment, the control parameter optimization module 05 is further configured to:

[0200] Based on the weights of the control parameters and the typical values ​​of historical operating condition clusters, the optimization adjustment step size of the control parameters is calculated; with the goal of dynamic optimization, the motor control parameters are optimized using the optimization adjustment step size to obtain optimized control parameters; the optimized control parameters are then used to optimize the control parameters of the high-dynamic motor.

[0201] Furthermore, the control parameter optimization module 05 also includes:

[0202] Obtain the control result corresponding to the optimized control parameter; calculate the difference between the control result corresponding to the optimized control parameter and the dynamic target. When the difference is less than or equal to the allowable error, end the optimized control; when the difference is greater than the allowable error, use the optimized control parameter as the control parameter and iteratively execute the control parameter optimization until the difference is less than or equal to the allowable error.

[0203] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0204] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0205] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

1. A method for reducing loss of high dynamic motor control parameter real-time optimization, characterized in that, The method comprises the following steps: acquiring current working condition parameters of a target motor and collecting control parameters of the target motor, wherein the control parameters include speed loop PI parameters and current loop PI parameters, and the working condition parameters include motor speed and torque; based on historical data, acquiring a plurality of historical control parameters, corresponding a plurality of historical working condition parameters, corresponding a plurality of historical loss values and corresponding a plurality of historical current harmonic distortions, and performing clustering analysis to obtain a historical working condition clustering set; obtaining a predicted loss value according to the control parameters, obtaining an optimization target weight in combination with the corresponding historical working condition clustering set, and obtaining an initial control parameter weight according to the control parameters and the corresponding historical working condition clustering set, comprising: based on machine learning, acquiring a loss predictor, comprising: based on machine learning, constructing a plurality of loss prediction branches; training the plurality of loss prediction branches respectively using a plurality of historical working condition clustering clusters to obtain a plurality of loss prediction branches; integrating the plurality of loss prediction branches to obtain a loss predictor; inputting the control parameters into the loss predictor to predict the loss, and obtaining a predicted loss value; based on the control parameters and the predicted loss value, searching in the historical working condition clustering set to obtain a historical working condition clustering cluster with the highest similarity; obtaining control typical values and loss typical values of the historical working condition clustering cluster, calculating the deviation of the predicted loss value and the loss typical value, and calculating the deviation of the control parameters and the control typical values; obtaining a loss optimization target weight according to the deviation of the predicted loss value and the loss typical value, and obtaining a current harmonic distortion optimization target weight according to the loss optimization target weight; obtaining an initial control parameter weight according to the deviation of the control parameters and the control typical values; based on the optimization target weight, obtaining a dynamic optimization target, and based on the dynamic optimization target, obtaining a control parameter influence degree, adjusting the calculation of the initial control parameter weight to obtain a control parameter weight; optimizing the motor control parameters with the dynamic optimization target as the goal and the control parameter weight as the constraint to obtain optimized control parameters, and using the optimized control parameters to optimize the control parameters of the high dynamic motor.

2. The method of claim 1, wherein, Based on historical data, a plurality of historical control parameters, corresponding a plurality of historical working condition parameters, corresponding a plurality of historical loss values and corresponding a plurality of historical current harmonic distortions are acquired, and clustering analysis is performed to obtain a historical working condition clustering set, comprising: collecting a plurality of historical control parameters of the same type of motor as the target motor, and acquiring corresponding a plurality of historical working condition parameters, corresponding a plurality of historical loss values and corresponding a plurality of historical current harmonic distortions of the plurality of historical control parameters; based on the plurality of historical working condition parameters, performing clustering analysis on the plurality of historical control parameters, corresponding a plurality of historical loss values and corresponding a plurality of historical current harmonic distortions to obtain a plurality of historical working condition clustering clusters; integrating the plurality of historical working condition clustering clusters to obtain a historical working condition clustering set.

3. The method of claim 2, wherein, The historical working condition clustering set includes a clustering center for each historical working condition clustering cluster, the control parameter corresponding to the clustering center is taken as a control typical value of the historical working condition clustering cluster, the historical loss value corresponding to the clustering center is taken as a loss typical value of the historical working condition clustering cluster, and the historical current harmonic distortion corresponding to the clustering center is taken as a current harmonic distortion typical value of the historical working condition clustering cluster.

4. The method of claim 1, wherein, Based on the optimization target weight, a dynamic optimization target is obtained, and based on the dynamic optimization target, a control parameter influence degree is obtained, the initial control parameter weight is adjusted and calculated to obtain a control parameter weight. Based on the optimization target weight, a dynamic optimization target is calculated and obtained, wherein the dynamic optimization target includes a loss optimization target and a current harmonic distortion optimization target. Based on the dynamic optimization target, the control parameter influence degree is obtained, wherein the control parameter influence degree is obtained by calculating the correlation coefficients of different control parameters with respect to the dynamic optimization target in the historical working condition clustering cluster, wherein the correlation coefficients are obtained by weighted calculation of loss correlation coefficients and current harmonic distortion correlation coefficients. Based on the control parameter influence degree, the initial control parameter weight is adjusted and calculated to obtain a control parameter weight.

5. The method of claim 4, wherein, Based on the dynamic optimization target, the control parameter influence degree is obtained, including: According to the historical working condition clustering cluster, a loss correlation coefficient of a control parameter with respect to a loss optimization target is obtained, and a current harmonic distortion correlation coefficient of a control parameter with respect to a current harmonic distortion optimization target is obtained. The loss correlation coefficient and the current harmonic distortion correlation coefficient are weighted and calculated to obtain the control parameter influence degree, wherein the weight in the weighted calculation is obtained based on the loss optimization target weight and the current harmonic distortion optimization target weight.

6. The method of claim 1, wherein, Based on the dynamic optimization target, the control parameter weight is taken as a constraint to optimize the motor control parameter to obtain an optimized control parameter, and the optimized control parameter is used to optimize the control parameter of the high-dynamic motor, including: According to the control parameter weight and the typical value of the historical working condition clustering cluster, an optimization adjustment step of the control parameter is calculated; Based on the dynamic optimization target, the optimization adjustment step is used to optimize the motor control parameter to obtain an optimized control parameter; The optimized control parameter is used to optimize the control parameter of the high-dynamic motor.

7. The method of claim 6, wherein the method further comprises: Using the optimized control parameter to optimize the control parameter of the high-dynamic motor further includes: The optimized control parameter is used to optimize the control parameter of the high-dynamic motor, further including: An adjustment result corresponding to the optimized control parameter is obtained; A difference between the adjustment result corresponding to the optimized control parameter and the dynamic target is calculated, and when the difference is less than or equal to an allowable error, the optimization adjustment is ended; 8. A system for real-time optimization of high dynamic motor control parameters with reduced loss, characterized in that, When the difference is greater than the allowable error, the optimized control parameter is taken as a control parameter, and the control parameter optimization is iteratively executed until the difference is less than or equal to the allowable error. The system is used to execute the real-time optimization method of the high-dynamic motor control parameter with reduced loss according to any one of claims 1-7, and the system includes: The working condition and parameter acquisition module acquires current working condition parameters of the target motor and acquires control parameters of the target motor, wherein the control parameters include speed loop PI parameters and current loop PI parameters, and the working condition parameters include motor speed and torque; The historical working condition clustering module is configured to acquire a plurality of historical control parameters, corresponding historical working condition parameters, corresponding historical loss values, and corresponding historical current harmonic distortions based on historical data, and perform clustering analysis to obtain a historical working condition clustering set; The loss prediction and weight preliminary calculation module is configured to predict a predicted loss value based on the control parameters, acquire an optimization target weight based on the corresponding historical working condition clustering set, and acquire an initial control parameter weight based on the control parameters and the corresponding historical working condition clustering set, including: Based on machine learning, a loss predictor is acquired, including: Based on machine learning, a plurality of loss prediction branches are constructed; The plurality of historical working condition clustering clusters are used to train the plurality of loss prediction branches respectively to obtain a plurality of loss prediction branches; The plurality of loss prediction branches are integrated to obtain a loss predictor; The control parameters are input into the loss predictor to predict a loss value; Based on the control parameters and the predicted loss value, a historical working condition clustering cluster with the highest similarity is acquired by searching in the historical working condition clustering set; Control typical values and loss typical values of the historical working condition clustering cluster are acquired, a deviation between the predicted loss value and the loss typical value is calculated, and a deviation between the control parameters and the control typical values is calculated; Based on the deviation between the predicted loss value and the loss typical value, a loss optimization target weight is acquired, and based on the loss optimization target weight, a current harmonic distortion optimization target weight is acquired; Based on the control parameters and the deviation from the control typical values, an initial control parameter weight is acquired; The dynamic target and weight adjustment module is configured to acquire a dynamic optimization target based on the optimization target weight, acquire a control parameter influence degree based on the dynamic optimization target, adjust the calculation of the initial control parameter weight, and obtain a control parameter weight; The control parameter optimization module is configured to optimize motor control parameters based on the dynamic optimization target and the control parameter weight as a constraint, acquire optimized control parameters, and use the optimized control parameters to optimize control parameters of a high-dynamic motor.

Images