A high-precision control method and system for a robot joint motor and a storage medium

Abstract

The application discloses a high-precision control method and system of a robot joint motor and a storage medium, which can significantly improve the motor control precision and reduce the influence on the robot operation. The application comprises the following steps: acquiring motor rotation angle position information, motor vibration data and motor speed data; constructing a motor operation state data set based on the motor rotation angle position information, the motor vibration data and the motor speed data; constructing a digital twin model according to the working data information of the motor; inputting the data output by the motor operation state data set and the digital twin model into an artificial intelligence model to obtain preliminary control parameters; inputting real-time operation data of the motor into a neural network model, and the neural network model is used for outputting nonlinear compensation parameters; acquiring an adaptive adjustment factor based on the scene characteristics under the motor operation state; and performing synergistic fusion optimization on the preliminary control parameters according to the nonlinear compensation parameters and the adaptive adjustment factor to obtain target parameters.

Description

Technical Field

[0001] This application relates to the field of motor control technology, and in particular to a high-precision control method, system and storage medium for robot joint motors. Background Technology

[0002] In the field of robotics, the motor is the core power actuator, and its operating performance and stability are directly related to the efficiency and reliability of the entire robot. Therefore, the control requirements for motors are becoming increasingly stringent. Not only is it necessary to achieve precise speed, position and torque control, but it is also required that the motor can operate efficiently and stably under complex and changing working conditions.

[0003] Currently, traditional motor control methods are mainly based on PID control. While these methods can achieve good control results under simple, linear operating conditions, they struggle to meet high-precision control requirements when faced with nonlinear, time-varying, and uncertain factors inherent in motor operation. For example, the dynamic characteristics and parameters of a motor change under different loads, temperatures, and speeds. Traditional control methods cannot adapt to these changes accurately and in real time, leading to decreased motor control accuracy, increased energy consumption, and even operational instability, thus impacting robot operation. Summary of the Invention

[0004] To address the aforementioned technical problems, this application provides a high-precision control method, system, and storage medium for robot joint motors.

[0005] The technical solution provided in this application is described below:

[0006] The first aspect of this application provides a high-precision control method for a robot joint motor, the method comprising:

[0007] Acquire motor rotation angle position information, motor vibration data, and motor speed data;

[0008] A motor operating status dataset is constructed based on the motor rotation angle position information, the motor vibration data, and the motor speed data;

[0009] A digital twin model is constructed based on the motor's operating data.

[0010] The motor operating status dataset and the data output from the digital twin model are input into the artificial intelligence model to obtain preliminary control parameters;

[0011] The real-time operating data of the motor is input into a neural network model, which is used to identify the nonlinear characteristics of the motor under different operating conditions and output nonlinear compensation parameters.

[0012] An adaptive adjustment factor is obtained based on the scene characteristics of the motor's operating state;

[0013] The preliminary control parameters are synergistically fused and optimized based on the nonlinear compensation parameters and the adaptive adjustment factor to obtain the target parameters;

[0014] Control commands are generated based on the target parameters, and adjustment parameters for the motor are generated according to the control commands. The adjustment parameters are then sent to the motor so that the motor can be adjusted and controlled according to the adjustment parameters.

[0015] Optionally, a motor operating status dataset is constructed based on the motor rotation angle position information, the motor vibration data, and the motor speed data, including:

[0016] The motor rotation angle position information, the motor vibration data, and the motor speed data are preprocessed to obtain preprocessed data.

[0017] Multi-dimensional features are extracted from the preprocessed data, including motor acceleration, vibration data, and angular acceleration.

[0018] The multi-dimensional features are integrated into a vector of a unified format according to the time series.

[0019] A mapping relationship between features and operating states is established based on the vectors in the unified format described above;

[0020] The motor operating status dataset is constructed based on the mapping relationship.

[0021] Optionally, a digital twin model is constructed based on the motor's operating data information, including:

[0022] Obtain the operating data information of the motor;

[0023] The working data information is filtered to obtain the target working data information;

[0024] The target working data information is sorted according to the target weight value to obtain the sorting result;

[0025] A digital twin model is constructed based on the sorting results.

[0026] Optionally, an adaptive adjustment factor is obtained based on scene characteristics of the motor's operating state, including:

[0027] Obtain historical scene data of the motor under different scenarios;

[0028] Extract historical scene features based on the historical scene data;

[0029] A mapping relationship between scene features and adaptive adjustment factors is established based on the aforementioned historical scene features;

[0030] Obtain scene information under the motor's operating state and extract target scene features;

[0031] Based on the mapping relationship, and according to the target scene features, an adaptive adjustment factor is obtained.

[0032] Optionally, the initial control parameters are synergistically fused and optimized based on the nonlinear compensation parameters and the adaptive adjustment factor to obtain the target parameters, including:

[0033] The nonlinear compensation parameters and the adaptive adjustment factor are preprocessed;

[0034] The nonlinear compensation parameter and the adaptive adjustment factor are used to perform synergistic optimization of the preliminary control parameter using a weighted summation formula to obtain the target parameter;

[0035] The weighted summation formula is as follows:

[0036] S = W× (1 + R × T + Y × P), where S is the target parameter, W is the initial control parameter, R is the nonlinear compensation parameter, T is the compensation weight, Y is the adaptive adjustment factor, and P is the adjustment weight.

[0037] Optionally, generating control commands based on the target parameters includes:

[0038] The target parameters are analyzed;

[0039] The parsed target parameters are matched with preset rules, and the matching results are obtained.

[0040] Decision information is generated based on the matching results;

[0041] The decision information is encoded and converted into control commands.

[0042] Optionally, the motor is adjusted and controlled according to the adjustment parameters, including:

[0043] Obtain the adjustment type and adjustment data of the adjustment parameter;

[0044] Generate adjustment operation instructions based on the adjustment type and the adjustment data;

[0045] Adjustment control is performed according to the adjustment operation instructions.

[0046] A second aspect of this application provides a high-precision control system for a robot joint motor, the system comprising:

[0047] The first acquisition unit is used to acquire motor rotation angle position information, motor vibration data, and motor speed data;

[0048] The first construction unit constructs a motor operating status dataset based on the motor rotation angle position information, the motor vibration data, and the motor speed data;

[0049] The second construction unit is used to construct a digital twin model based on the motor's operating data information;

[0050] The second acquisition unit is used to input the motor operating status dataset and the data output by the digital twin model into the artificial intelligence model to obtain preliminary control parameters;

[0051] The output unit is used to input the real-time operating data of the motor into the neural network model, which is used to identify the nonlinear characteristics of the motor under different operating conditions and output nonlinear compensation parameters.

[0052] The third acquisition unit acquires an adaptive adjustment factor based on the scene characteristics of the motor's operating state;

[0053] The optimization unit is used to perform collaborative fusion optimization of the preliminary control parameters based on the nonlinear compensation parameters and the adaptive adjustment factor to obtain the target parameters;

[0054] The adjustment unit generates control commands based on the target parameters, generates adjustment parameters for the motor according to the control commands, and sends the adjustment parameters to the motor so that the motor can be adjusted and controlled according to the adjustment parameters.

[0055] A third aspect of this application provides a high-precision control system for a robot joint motor, the system comprising:

[0056] Processor, memory, input / output units, and bus;

[0057] The processor is connected to the memory, the input / output unit, and the bus;

[0058] The memory stores a program, which the processor invokes to perform the method as described in the first aspect and any one of the first aspects.

[0059] A fourth aspect of this application provides a computer-readable storage medium on which a program is stored, which, when executed on a computer, performs the method as described in the first aspect and any one of the first aspects.

[0060] As can be seen from the above technical solutions, this application has the following beneficial effects:

[0061] 1. This application constructs an operating status dataset by acquiring multi-dimensional information such as motor rotation angle position information, vibration data, and speed data, which can comprehensively reflect the motor operating status. Compared with relying on single or small amounts of data, multi-source data fusion can provide a more accurate information basis for control decisions, effectively reduce control errors caused by data partiality, thereby significantly improving motor control accuracy and reducing the impact on robot operation.

[0062] 2. This application constructs a digital twin model based on motor operating data. This digital twin model can simulate the motor operation process. Then, the motor operating status dataset and the output data of the digital twin model are input into the artificial intelligence model to obtain preliminary control parameters. This helps to adjust the control strategy in advance, avoid motor instability due to abnormal operation, and enhance the stability of motor operation.

[0063] 3. The initial control parameters can be optimized by synergistic fusion based on nonlinear compensation parameters and adaptive adjustment factors. The adaptive adjustment mechanism can adjust the control parameters in real time according to the motor operating status and environmental changes, so that the motor is always in the best operating state, effectively cope with various complex working conditions, reduce the operating fluctuations caused by external interference or internal parameter changes, and improve the stability and reliability of motor operation. Attached Figure Description

[0064] To more clearly illustrate the technical solutions in this application, 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 this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0065] Figure 1 This is a schematic diagram of an embodiment of the high-precision control method for robot joint motors according to this application;

[0066] Figure 2 This is a schematic diagram of another embodiment of the high-precision control method for robot joint motors according to this application;

[0067] Figure 3 This is a schematic diagram of another embodiment of the high-precision control method for robot joint motors according to this application;

[0068] Figure 4 This is a schematic diagram of another embodiment of the high-precision control method for robot joint motors according to this application;

[0069] Figure 5 This is a schematic diagram of another embodiment of the high-precision control method for robot joint motors according to this application;

[0070] Figure 6This is a schematic diagram of another embodiment of the high-precision control method for robot joint motors according to this application;

[0071] Figure 7 This is a schematic diagram of another embodiment of the high-precision control method for robot joint motors according to this application;

[0072] Figure 8 This is a schematic diagram of an embodiment of the high-precision control system for the robot joint motor of this application;

[0073] Figure 9 This is a schematic diagram of another embodiment of the high-precision control system for the robot joint motor of this application. Detailed Implementation

[0074] Currently, traditional motor control methods are mainly based on PID control. While these methods can achieve good control results under simple, linear operating conditions, they struggle to meet high-precision control requirements when faced with nonlinear, time-varying, and uncertain factors inherent in motor operation. For example, the dynamic characteristics and parameters of a motor change under different loads, temperatures, and speeds. Traditional control methods cannot adapt to these changes accurately and in real time, leading to decreased motor control accuracy, increased energy consumption, and even operational instability, thus impacting robot operation.

[0075] Based on this, this application provides a high-precision control method, system and storage medium for robot joint motors, which can significantly improve motor control accuracy and reduce the impact on robot operation.

[0076] Please see Figure 1 The first aspect of this application provides a high-precision control method for robot joint motors, the method comprising:

[0077] 101. Obtain motor rotation angle position information, motor vibration data, and motor speed data;

[0078] 102. Construct a motor operating status dataset based on the motor rotation angle position information, the motor vibration data, and the motor speed data;

[0079] 103. Construct a digital twin model based on the motor's operating data information;

[0080] 104. Input the motor operating status dataset and the data output from the digital twin model into the artificial intelligence model to obtain preliminary control parameters;

[0081] 105. Input the real-time operating data of the motor into the neural network model. The neural network model is used to identify the nonlinear characteristics of the motor under different operating conditions and output nonlinear compensation parameters.

[0082] 106. Obtain an adaptive adjustment factor based on the scene characteristics of the motor's operating state;

[0083] 107. Perform synergistic fusion optimization on the preliminary control parameters based on the nonlinear compensation parameters and the adaptive adjustment factor to obtain the target parameters;

[0084] 108. Generate control commands based on the target parameters, generate adjustment parameters for the motor according to the control commands, and send the adjustment parameters to the motor so that the motor can be adjusted and controlled according to the adjustment parameters.

[0085] In this embodiment, the following steps are taken: First, the motor rotation angle position information, motor vibration data, and motor speed data are acquired. Then, a motor operating state dataset is constructed based on the motor rotation angle position information, motor vibration data, and motor speed data. Next, a digital twin model is constructed based on the motor's operating data information. The data output from the motor operating state dataset and the digital twin model are input into an artificial intelligence model to obtain preliminary control parameters. The real-time motor operating data is input into a neural network model, which is used to identify the nonlinear characteristics of the motor under different operating conditions and output nonlinear compensation parameters. Furthermore, an adaptive adjustment factor is obtained based on the scene characteristics of the motor operating state. The preliminary control parameters are then synergistically fused and optimized based on the nonlinear compensation parameters and the adaptive adjustment factor to obtain target parameters. Finally, control commands are generated based on the target parameters, and motor adjustment parameters are generated based on the control commands. The adjustment parameters are then sent to the motor so that the motor can adjust and control according to the adjustment parameters.

[0086] In step 101, the motor rotation angle position information, motor vibration data, and motor speed data are acquired. Specifically, high-precision sensors are installed on the motor to achieve real-time data acquisition. The encoder captures the motor rotation angle position information. The encoder converts the mechanical rotation angle of the motor shaft into an electrical signal output through photoelectric or electromagnetic induction principles. A specific pulse signal or digital code is generated for each rotation of a certain angle, thereby recording the rotation position of the motor.

[0087] An accelerometer is used to collect motor vibration data. The accelerometer can detect changes in vibration acceleration generated by the motor during operation and convert them into electrical signals. A Hall sensor is used to acquire motor speed data. The Hall sensor utilizes the Hall effect; when the motor rotates, it causes a change in the magnetic field. After sensing the change in the magnetic field, the Hall sensor outputs a pulse signal. The motor speed can be calculated based on the number of pulses per unit time. The data collected by these sensors is transmitted to the data acquisition module in the form of electrical signals. After signal filtering and amplification, it is converted into a digital signal suitable for subsequent processing, and step 102 is executed.

[0088] In step 102, a motor operating status dataset is constructed based on the motor rotation angle position information, the motor vibration data, and the motor speed data. Specifically, in practical applications, the collected motor rotation angle position information, vibration data, and speed data are integrated, and the three types of data at the same moment are combined into a record in a time series manner. Each record represents the operating status of the motor at a certain instant.

[0089] Meanwhile, after acquiring the records, corresponding labels are added to each record to indicate the current operating condition of the motor, such as normal operation, overload, fault, etc., in order to form a structured motor operating status dataset. This dataset reflects the operating status information of the motor at different times and under different operating conditions, providing basic data for subsequent model building and analysis.

[0090] In step 103, a digital twin model is constructed based on the motor's operating data information. Specifically, based on the motor's physical structure, working principle, rated power, rated speed, and winding parameters, a digital twin model of the motor is constructed using computer simulation technology.

[0091] During the model building process, mathematical modeling is used to establish the relationship equations between various physical fields. For example, an electromagnetic model of the motor is established based on the law of electromagnetic induction to describe the distribution of the magnetic field and the generation of electromagnetic force inside the motor. Specifically, a mechanical model of the motor is established based on Newton's laws of motion to analyze the rotor motion and mechanical vibration of the motor; and a thermal model of the motor is established based on the principle of heat transfer to study the temperature distribution and heat dissipation of the motor during operation.

[0092] These models are coupled to form a digital twin model that can accurately simulate the actual operating state of a motor. This model can run in real time on a computer. By inputting real-time operating data of the motor, it can simulate the operating state of the motor and provide a virtual simulation environment for motor control and optimization.

[0093] In step 104, the motor operating status dataset and the data output by the digital twin model are input into the artificial intelligence model to obtain preliminary control parameters. Specifically, the motor operating status dataset is used as the training data for the artificial intelligence model, and the data simulated by the digital twin model under different operating conditions is also input into the artificial intelligence model.

[0094] The artificial intelligence model is a deep learning model. Through learning and training on a large amount of data, the deep learning model establishes a mapping relationship between the motor's operating state and control parameters. During training, the model continuously adjusts its parameters to minimize the error between the predicted output and the actual expected output. After sufficient training, when new motor operating state data and digital twin model output data are input, the deep learning model can output a set of preliminary control parameters based on the learned patterns. These preliminary control parameters are derived from learning from simulated data and aim to make preliminary adjustments and optimizations to the motor's operating state. After obtaining the preliminary control parameters, step 105 is executed.

[0095] In step 105, the real-time operating data of the motor is input into the neural network model. This model identifies the nonlinear characteristics of the motor under different operating conditions and outputs nonlinear compensation parameters. Specifically, a neural network model is constructed to address the nonlinear characteristics exhibited by the motor during actual operation, such as the nonlinear changes in torque and speed under different loads and speeds. The real-time operating data of the motor is input into this neural network model. Through the complex computation and learning capabilities of its multi-layered neurons, the neural network model extracts and analyzes features from the input data to identify the nonlinear characteristics of the motor under the current operating condition.

[0096] For example, by learning from a large amount of operating data under different operating conditions, the neural network model can discover the nonlinear relationship between torque and current of the motor under low-speed and heavy-load conditions. Based on the identification of nonlinear characteristics, the neural network model outputs corresponding nonlinear compensation parameters. These parameters are used to correct the initial control parameters to compensate for the influence of the motor's nonlinear characteristics on the control effect and improve the accuracy of control.

[0097] In step 106, an adaptive adjustment factor is obtained based on the scene characteristics of the motor operating state. Specifically, by analyzing the motor operating state data, feature information that can reflect the current operating scene of the motor is extracted, such as operating environment temperature, load type, and operating time.

[0098] Based on these scenario characteristics, a model is built to calculate the adaptive adjustment factor. For example, when a high ambient temperature is detected in the motor's operating environment, the adaptive adjustment factor is appropriately increased according to pre-set rules to strengthen the adjustment of motor heat dissipation-related control parameters. When the motor is in a frequent start-stop operation scenario, the adaptive adjustment factor calculated based on the characteristics of this scenario is used to optimize and adjust the motor's start and stop control parameters. In other words, the adaptive adjustment factor can dynamically adjust according to the different operating scenario characteristics of the motor, enabling the motor to better adapt to complex and changing operating environments.

[0099] In step 107, the preliminary control parameters are synergistically fused and optimized based on the nonlinear compensation parameters and the adaptive adjustment factor to obtain the target parameters. Specifically, the nonlinear compensation parameters and adaptive adjustment factor obtained in the preceding steps are fused and optimized with the preliminary control parameters. A weighted summation algorithm is used to incorporate the nonlinear compensation parameters and adaptive adjustment factor into the preliminary control parameters according to certain weights.

[0100] For example, by analyzing the impact of nonlinear characteristics and scenario features on motor operation under different operating conditions, different weights are assigned to the nonlinear compensation parameters and adaptive adjustment factors. These weighted sums are then calculated with the preliminary control parameters to obtain the final target parameters. The target parameters fully consider the nonlinear characteristics of the motor and the characteristics of the current operating scenario, making them more accurate and reasonable than the preliminary control parameters, and enabling more effective control of the motor's operating state.

[0101] In step 108, control commands are generated based on the target parameters, and adjustment parameters for the motor are generated according to the control commands. These adjustment parameters are then sent to the motor so that the motor can be adjusted and controlled according to the adjustment parameters. Specifically, based on the calculated target parameters, corresponding control commands are generated according to the communication protocol and command format of the motor control system.

[0102] The control commands contain specific adjustment requirements for various motor control parameters, such as voltage, current, and frequency. These commands are then translated into adjustment parameters that the motor can recognize and execute, and sent to the motor controller via a data communication interface. Upon receiving the adjustment parameters, the motor controller controls the motor's drive circuit accordingly. For example, adjusting the inverter's output frequency changes the motor's speed, and adjusting the motor's input voltage and current controls its torque. This allows for precise adjustment and control of the motor's operating state, ensuring optimal motor operation, improving efficiency and reliability, and minimizing impact on robot operation.

[0103] Please refer to Figure 2 According to some embodiments of the present invention, step 103, which involves constructing a motor operating state dataset based on the motor rotation angle position information, the motor vibration data, and the motor speed data, may specifically include, but is not limited to, the following:

[0104] 201. Preprocess the motor rotation angle position information, the motor vibration data, and the motor speed data to obtain preprocessed data;

[0105] 202. Extract multi-dimensional features from the preprocessed data, including motor acceleration, vibration data, and angular acceleration;

[0106] 203. Integrate the multi-dimensional features into a vector of a unified format according to the time series;

[0107] 204. Establish a mapping relationship between features and operating states based on the vectors in the unified format;

[0108] 205. Construct the motor operating status dataset based on the mapping relationship.

[0109] In this embodiment of the application, after collecting the motor rotation angle position information, vibration data and speed data, the data is preprocessed to obtain preprocessed data. Then, based on the preprocessed data, mathematical calculations are used to extract multi-dimensional features.

[0110] Specifically, for motor acceleration, differential calculation is performed on the motor speed data to obtain the speed change at adjacent time points. Dividing this by the time interval yields the motor acceleration within that time period. For vibration data, frequency domain analysis is performed on the collected vibration signals. A fast Fourier transform is used to convert the time-domain vibration signal to the frequency domain, extracting features such as frequency components and amplitude to analyze the motor's vibration characteristics. In terms of angular acceleration extraction, second-order differential calculation is performed on the angle data based on the motor's rotation angle position information to obtain the acceleration due to angle changes at adjacent time points, i.e., angular acceleration. The multi-dimensional features obtained above can more comprehensively reflect the motor's operating status.

[0111] The extracted multi-dimensional features, such as motor acceleration, vibration data, and angular acceleration, are then arranged and integrated in chronological order. Using time as an index, all feature data corresponding to each time point are combined into a vector, with each element in the vector corresponding to a specific feature value.

[0112] For example, at a certain moment, the first element of the vector is the motor acceleration at that moment, the second element is the amplitude of the vibration signal at a certain frequency, the third element is the angular acceleration, and so on, eventually forming a series of vectors in a unified format arranged in time sequence.

[0113] Then, statistical analysis methods are used to analyze the integrated unified format vectors to obtain the intrinsic relationship between the features in the vectors and the actual operating state of the motor. Specifically, a training dataset is constructed by collecting a large amount of vector data on motor operating states. Then, a decision tree is used to train the training dataset, allowing the algorithm to learn the mapping pattern between vector features and operating states.

[0114] For example, when the motor acceleration in the vector exceeds a certain threshold and the amplitude of the vibration data increases abnormally at certain specific frequencies, the algorithm learns that this combination of features corresponds to the motor's possible unbalanced operating state. After sufficient training, an accurate mapping relationship model between features and operating state is established.

[0115] Based on the established mapping relationship between features and operating states, vector data in a unified format is associated with the corresponding motor operating states. For each vector in the unified format, its corresponding motor operating state is determined through the mapping relationship, and a corresponding label is added, such as "normal," "overload," or "fault." The labeled vector data is then aggregated and integrated to finally construct a complete motor operating state dataset. This dataset not only contains multi-dimensional feature information during motor operation but also clearly labels the operating state of the motor at each moment.

[0116] Please refer to Figure 3 According to some embodiments of the present invention, the construction of a digital twin model based on the motor's operating data information in step 103 may specifically include, but is not limited to, the following:

[0117] 301. Obtain the operating data information of the motor;

[0118] 302. Filter the working data information to obtain the target working data information;

[0119] 303. Sort the target working data information according to the target weight value to obtain the sorting result;

[0120] 304. Construct a digital twin model based on the sorting results.

[0121] In this embodiment of the application, multiple sensors are deployed on the motor to acquire motor operating data information. Since the raw operating data information collected by the sensors is inevitably affected by external electromagnetic interference, sensor noise and other factors, there will be a large amount of useless noise signals, which need to be filtered.

[0122] For different types of data and noise characteristics, appropriate filtering algorithms should be selected. Specifically, for periodic noise, notch filters can be used to accurately suppress noise signals of specific frequencies. For example, during motor operation, if interference generated by the power supply frequency and its harmonic frequencies is detected, a notch filter can effectively eliminate noise components of these frequencies. For random noise, median filtering algorithms are used. This algorithm replaces the value of each point in the data sequence with the median of the data in its neighborhood, effectively removing random noise interference while preserving the true trend of data change.

[0123] For complex noise signals containing multiple frequency components, wavelet filtering, based on wavelet transform theory, can analyze the signal at different scales, decompose the signal into components of different frequency bands, selectively remove the components in the frequency band where the noise is located, and then reconstruct the useful signal components to obtain the target working data information.

[0124] Based on the importance of each motor's operating data to its running status and the construction of the digital twin model, a target weight value is determined for each data item. For example, the motor's current and speed data are directly related to its output power and operating efficiency, and have a significant impact on its operating status, so they are assigned higher weight values. On the other hand, some auxiliary environmental monitoring data, such as ambient temperature, although they also have some impact on motor operation, are relatively minor and are assigned lower weight values.

[0125] Therefore, a weighted sorting algorithm is adopted, which multiplies each target work data information by its corresponding weight value to obtain a weighted value, and then sorts the target work data information according to the size of the weighted value.

[0126] During the sorting process, a merge sort algorithm can be used. The final sorting result clearly shows the order of importance of each piece of data, with the most important data listed first. This provides a basis for the rational use of data when building the digital twin model, ensuring that key data information is given priority during the model building process, thereby improving the accuracy and reliability of the model.

[0127] Guided by the sorting results, and combined with the physical structure and working principle of the motor, a digital twin model is constructed using computer modeling technology. At the beginning of the construction, the most important data information is started and integrated into the corresponding parts of the model.

[0128] For example, based on the top-ranked current and voltage data, an electromagnetic model of the motor is established to simulate the distribution of electromagnetic fields and the generation process of electromagnetic forces inside the motor; based on the speed and torque data, a mechanical dynamics model of the motor is constructed, using Newton's laws of motion and rotation to describe the motion state of the motor rotor and the interaction between the mechanical load; for temperature data, a heat conduction model of the motor is established, and based on the principle of heat transfer, the heat transfer and temperature changes of each component of the motor during operation are analyzed.

[0129] Then, these sub-models, built based on different data information, are organically integrated, and data interaction and collaborative simulation between the sub-models are realized through data interfaces. Simultaneously, computer graphics technology is further utilized to create a 3D model of the motor's appearance and internal structure, enabling the digital twin model to visually and intuitively present the motor's true form. This allows the digital twin model to accurately simulate the motor's operating state under different working conditions, achieving a digital mapping of the motor's actual operating process and providing an effective virtual simulation platform for motor fault diagnosis, performance optimization, and operation control.

[0130] Please refer to Figure 4 According to some embodiments of the present invention, the step 103 of obtaining the adaptive adjustment factor based on the scene characteristics of the motor operating state may specifically include, but is not limited to, the following:

[0131] 401. Obtain historical scene data of the motor under different scenarios;

[0132] 402. Extract historical scene features based on the historical scene data;

[0133] 403. Establish a mapping relationship between scene features and adaptive adjustment factors based on the aforementioned historical scene features;

[0134] 404. Obtain scene information under the motor's operating state and extract target scene features;

[0135] 405. Based on the mapping relationship, and according to the target scene features, obtain an adaptive adjustment factor.

[0136] In practical applications, to obtain motor operating information under various working conditions, it is necessary to retrieve historical operating records from the motor operation database. These records contain the motor's operating parameters at different times and under different workloads, such as speed and current variations when the motor drives the conveyor belt in different production batches, as well as voltage and temperature data under long-term continuous operation and intermittent start-stop conditions. On the other hand, historical data on the motor's operating environment is also collected, including temperature and humidity changes in different seasons and electromagnetic interference intensity at different times. These data are then summarized to form historical scenario data containing information such as motor operating parameters, environmental conditions, and maintenance status.

[0137] After obtaining historical scene data, historical scene features are extracted from it. Specifically, for numerical data, such as motor speed, current, and temperature, mean, variance, and extreme value statistics are calculated to reflect the central tendency and degree of fluctuation of the data. For example, the stability of the current can be measured by calculating the standard deviation of the motor current over a certain period. For categorical data, such as load type and environmental category, one-hot encoding or label encoding methods are used for feature processing.

[0138] For time-series data, Fourier transform or wavelet transform are used to extract frequency domain features, analyze the frequency components in motor vibration data, and determine whether there are abnormal vibration frequencies. Furthermore, while retaining the main information of the data, the feature dimensions are reduced and redundant features are removed, ultimately obtaining a set of feature vectors that accurately characterize the historical operating scenarios of the motor. These feature vectors will serve as the foundational data for establishing mapping relationships.

[0139] Then, a mapping model was built using machine learning algorithms. Historical scene features were used as input variables, and an adaptive adjustment factor was pre-set as the output variable. A linear regression method was used to establish the mathematical relationship between the two. When training the model, the historical scene feature dataset was divided into a training set and a test set. The training set was used to adjust and optimize the model parameters. By minimizing the error between the predicted adaptive adjustment factor and the actual value, the model was able to accurately capture the intrinsic relationship between scene features and the adaptive adjustment factor.

[0140] After training, the model's generalization ability is evaluated using a test set. If the model performs well on the test set, it indicates strong adaptability and reliability. The final mapping model can accurately output the corresponding adaptive adjustment factor based on the input scene characteristics, providing a decision-making basis for adaptive adjustment during motor operation.

[0141] Furthermore, during the real-time operation of the motor, scene information is collected in real time through a sensor network and monitoring module. The collected scene information is integrated, and the data is preprocessed and features extracted using the same methods and processes as those used to extract historical scene features. A sliding window technique is employed for real-time data to dynamically analyze data over a short period, extracting target scene features that reflect the immediate state of the motor's current operating scene. These target scene features are input into a mapping relationship to obtain the corresponding adaptive adjustment factor.

[0142] The extracted target scene features are input into a previously established mapping model between scene features and adaptive adjustment factors. Based on the learned mapping rules, the model analyzes and calculates the input target scene features and outputs the corresponding adaptive adjustment factor. For example, when the model detects that the target scene features indicate the motor is operating under high temperature and heavy load, it outputs a large adaptive adjustment factor based on the trained mapping relationship. This factor will subsequently be used to strengthen the motor's heat dissipation control and torque regulation, ensuring stable and efficient operation of the motor under complex conditions. Therefore, dynamically acquiring the adaptive adjustment factor based on the motor's real-time operating scenario provides strong support for precise motor control.

[0143] Please refer to Figure 5According to some embodiments of the present invention, step 103, which involves synergistically fusing and optimizing the preliminary control parameters based on the nonlinear compensation parameters and the adaptive adjustment factor to obtain the target parameters, may specifically include, but is not limited to, the following:

[0144] 501. Preprocess the nonlinear compensation parameters and the adaptive adjustment factor;

[0145] 502. The nonlinear compensation parameters and the adaptive adjustment factor are used to perform synergistic fusion optimization on the preliminary control parameters using a weighted summation formula to obtain the target parameters;

[0146] The weighted summation formula is as follows:

[0147] S = W× (1 + R × T + Y × P), where S is the target parameter, W is the initial control parameter, R is the nonlinear compensation parameter, T is the compensation weight, Y is the adaptive adjustment factor, and P is the adjustment weight.

[0148] In this embodiment, the nonlinear compensation parameters and adaptive adjustment factors are preprocessed to ensure the validity and consistency of the data and improve the accuracy and stability of collaborative fusion optimization.

[0149] After preprocessing, a weighted summation formula is used to synergistically fuse and optimize the initial control parameters with the nonlinear compensation parameters and the adaptive adjustment factor to obtain the target parameters. Specifically, the weighted summation formula is: S = W × (1 + R × T + Y × P), where S is the target parameter, W is the initial control parameter, R is the nonlinear compensation parameter, T is the compensation weight, Y is the adaptive adjustment factor, and P is the adjustment weight.

[0150] The compensation weight T and adjustment weight P are predetermined based on the motor characteristics and operating conditions, reflecting the importance of the nonlinear compensation parameter and the adaptive adjustment factor in the target parameter calculation. In the calculation process, the nonlinear compensation parameter R is first multiplied by the compensation weight T to obtain the correction amount of the nonlinear characteristics to the initial control parameters. Then, the adaptive adjustment factor Y is multiplied by the adjustment weight P to obtain the adjustment amount of the operating scenario characteristics to the initial control parameters. These two adjustment amounts are then added together, and 1 is added to obtain a comprehensive adjustment coefficient. Finally, this comprehensive adjustment coefficient is multiplied by the initial control parameter W to obtain the final target parameter S.

[0151] Please refer to Figure 6 According to some embodiments of the present invention, the generation of control instructions based on the target parameters in step 103 may specifically include, but is not limited to, the following:

[0152] 601. Analyze the target parameters;

[0153] 602. Match the parsed target parameters with the preset rules and obtain the matching results;

[0154] 603. Generate decision information based on the matching results;

[0155] 604. After encoding and converting the decision information, control instructions are generated.

[0156] In this embodiment, after obtaining the target parameters, since the target parameters are comprehensive data containing multiple control information, they need to be parsed to extract specific control parameters. The target parameters store information on multiple control quantities such as motor speed, torque, and voltage in a specific encoded form. Through pre-set parsing rules, the target parameters are split and parsed according to the data structure. The validity of each parsed control parameter is checked to determine whether it is within the acceptable operating range of the motor. If it exceeds the range, it is marked as abnormal and not used, thus ensuring that the parsed control parameters are accurate and usable.

[0157] The preset rules are a set of rules formulated based on the motor's safety operation requirements, performance indicators, and operating parameters under different working conditions. For example, it may stipulate that under normal load, the motor speed should be maintained within a certain error range of the rated speed, and the torque should not exceed the maximum allowable torque.

[0158] Each parsed target control parameter is compared one by one with its corresponding preset rule. Taking motor speed as an example, the parsed speed value is compared with the preset normal speed range to determine whether the speed is within the normal range. For torque parameters, it is checked whether they exceed the maximum allowable torque value, etc. Based on the comparison results, the matching status of each control parameter with the preset rule is determined, and the matching result is obtained. The matching result is represented by a simple logical judgment, such as "match successful" or "match failed".

[0159] Based on the matching results obtained in the previous step, corresponding decision information is generated. If all parsed target parameters match the preset rules successfully, it indicates that the current control parameters of the motor are set reasonably, and the decision information can be "maintain the current control state".

[0160] If some parameters fail to match, such as the motor speed exceeding the normal range, specific decision information is generated based on the degree of deviation and the preset decision strategy. For example, if the speed is slightly out of range, the decision information could be "fine-tune the speed, reduce it by X revolutions per minute"; if the speed is significantly out of range, the decision information would be "immediately reduce the speed and check the motor's operating status." For other control parameters such as torque and voltage, corresponding decision information is generated in a similar manner based on the matching results and preset strategies. The decision information not only clearly indicates the control actions to be taken but also includes an assessment and suggestions regarding the motor's operating status, providing guidance for subsequent control.

[0161] Furthermore, in order for the motor controller to recognize and execute decision information, this information needs to be encoded and converted. First, according to the communication protocol and command format requirements of the motor control system, the decision information is converted into a specific data format. For example, if the motor controller uses the Modbus protocol, the decision information needs to be organized according to the Modbus protocol's data frame format, including adding device address, function code, data segment, and other information.

[0162] For textual descriptions in decision-making information, such as "fine-tune the speed, reduce by X revolutions per minute," this is converted into specific numerical codes and control operation instruction codes. Then, the generated data undergoes encoding processing, such as using cyclic redundancy check (CRC) to verify the data and ensure its accuracy and integrity during transmission. Finally, the encoded, converted, and verified decision information is encapsulated into complete control instructions and sent to the motor controller via the data communication interface. This enables the motor to adjust and control accordingly, achieving precise control of the motor's operating status.

[0163] Please refer to Figure 7 According to some embodiments of the present invention, the generation of control instructions based on the target parameters in step 103 may specifically include, but is not limited to, the following:

[0164] 701. Obtain the adjustment type and adjustment data of the adjustment parameter;

[0165] 702. Generate adjustment operation instructions based on the adjustment type and the adjustment data;

[0166] 703. Adjust and control according to the adjustment operation instructions.

[0167] In this embodiment of the application, after receiving the adjustment parameters generated by the target parameters, it is necessary to delve into the adjustment parameters to obtain the specific adjustment direction and value. The adjustment parameters are usually stored and transmitted in a specific data structure, which contains key information about the adjustment type and adjustment data.

[0168] The adjustment parameters are formatted as "parameter type identifier + specific data value". The adjustment type is determined by parsing the parameter type identifier. Common adjustment types include speed adjustment, torque adjustment, voltage adjustment, and current adjustment. If the identifier indicates speed adjustment, the subsequent data value is the speed-related data to be adjusted, such as the speed change or target speed value. If it is voltage adjustment, the data value represents the voltage adjustment range or target voltage value. Simultaneously, to ensure data accuracy and validity, the acquired adjustment data is verified to check if it conforms to the preset data format specifications and is within a reasonable numerical range. If abnormal data is found, the corresponding error handling mechanism is triggered, such as re-acquiring the adjustment parameters or issuing a warning to the operator.

[0169] After clarifying the adjustment type and adjustment data, the system transforms this information into adjustment operation commands that the motor can understand and execute, based on the operating logic and communication protocol of the motor control system. Different adjustment types have corresponding command generation rules. Taking speed adjustment as an example, if the adjustment data requires increasing the motor speed by 100 rpm, the system will generate the corresponding command according to a pre-set command format. During command generation, the system further optimizes and refines the command based on the motor's operating status and safety requirements, checking whether the generated adjustment operation command will cause the motor's operating parameters to exceed its safe operating range. If a risk exists, the command is corrected or supplemented with corresponding protective measures to ensure that the command achieves the adjustment target without damaging the motor.

[0170] After the adjustment operation command is generated, it is sent to the motor controller through the data communication interface. After receiving the command, the motor controller calls the corresponding control algorithm and drive circuit to execute the adjustment operation according to the adjustment type and adjustment data in the command, thereby realizing precise control of the motor.

[0171] Please see Figure 8 The second aspect of this application provides a high-precision control system for robot joint motors, comprising:

[0172] The first acquisition unit 801 is used to acquire motor rotation angle position information, motor vibration data and motor speed data;

[0173] The first construction unit 802 constructs a motor operating status dataset based on the motor rotation angle position information, the motor vibration data, and the motor speed data;

[0174] The second construction unit 803 is used to construct a digital twin model based on the working data information of the motor;

[0175] The second acquisition unit 804 is used to input the motor operating status dataset and the data output by the digital twin model into the artificial intelligence model to obtain preliminary control parameters;

[0176] The output unit 805 is used to input the real-time operating data of the motor into the neural network model, which is used to identify the nonlinear characteristics of the motor under different operating conditions and output nonlinear compensation parameters.

[0177] The third acquisition unit 806 acquires an adaptive adjustment factor based on the scene characteristics of the motor's operating state;

[0178] Optimization unit 807 is used to perform collaborative fusion optimization of the preliminary control parameters based on the nonlinear compensation parameters and the adaptive adjustment factor to obtain target parameters;

[0179] The adjustment unit 808 generates control commands based on the target parameters, generates adjustment parameters for the motor according to the control commands, and sends the adjustment parameters to the motor so that the motor can be adjusted and controlled according to the adjustment parameters.

[0180] Please see Figure 9 This application also provides a high-precision control system for robot joint motors, comprising:

[0181] Processor 901, memory 902, input / output unit 903, bus 904;

[0182] The processor 901 is connected to the memory 902, the input / output unit 903, and the bus 904;

[0183] The memory 902 contains a program, and the processor 901 calls the program to execute any of the methods described above.

[0184] This application also relates to a computer-readable storage medium on which a program is stored, which, when run on a computer, causes the computer to perform any of the methods described above.

[0185] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0186] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0187] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0188] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0189] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A high-precision control method for robot joint motors, characterized in that, The method includes: Acquire motor rotation angle position information, motor vibration data, and motor speed data; A motor operating status dataset is constructed based on the motor rotation angle position information, the motor vibration data, and the motor speed data; A digital twin model is constructed based on the motor's operating data. The motor operating status dataset and the data output from the digital twin model are input into the artificial intelligence model to obtain preliminary control parameters; The real-time operating data of the motor is input into a neural network model, which is used to identify the nonlinear characteristics of the motor under different operating conditions and output nonlinear compensation parameters. An adaptive adjustment factor is obtained based on the scene characteristics of the motor's operating state; The preliminary control parameters are synergistically fused and optimized based on the nonlinear compensation parameters and the adaptive adjustment factor to obtain the target parameters; Control commands are generated based on the target parameters, and adjustment parameters for the motor are generated according to the control commands. The adjustment parameters are then sent to the motor so that the motor can be adjusted and controlled according to the adjustment parameters.

2. The high-precision control method for robot joint motors according to claim 1, characterized in that, A motor operating status dataset is constructed based on the motor rotation angle position information, the motor vibration data, and the motor speed data, including: The motor rotation angle position information, the motor vibration data, and the motor speed data are preprocessed to obtain preprocessed data. Multi-dimensional features are extracted from the preprocessed data, including motor acceleration, vibration data, and angular acceleration. The multi-dimensional features are integrated into a vector of a unified format according to the time series. A mapping relationship between features and operating states is established based on the vectors in the unified format described above; The motor operating status dataset is constructed based on the mapping relationship.

3. The high-precision control method for robot joint motors according to claim 1, characterized in that, A digital twin model is constructed based on the operating data information of the motor, including: Obtain the operating data information of the motor; The working data information is filtered to obtain the target working data information; The target working data information is sorted according to the target weight value to obtain the sorting result; A digital twin model is constructed based on the sorting results.

4. The high-precision control method for robot joint motors according to claim 1, characterized in that, The adaptive adjustment factor is obtained based on the scene characteristics of the motor's operating state, including: Obtain historical scene data of the motor under different scenarios; Extract historical scene features based on the historical scene data; A mapping relationship between scene features and adaptive adjustment factors is established based on the aforementioned historical scene features; Obtain scene information under the motor's operating state and extract target scene features; Based on the mapping relationship, and according to the target scene features, an adaptive adjustment factor is obtained.

5. The high-precision control method for robot joint motors according to claim 1, characterized in that, The initial control parameters are synergistically fused and optimized based on the nonlinear compensation parameters and the adaptive adjustment factor to obtain target parameters, including: The nonlinear compensation parameters and the adaptive adjustment factor are preprocessed; The nonlinear compensation parameter and the adaptive adjustment factor are used to perform synergistic optimization of the preliminary control parameter using a weighted summation formula to obtain the target parameter; The weighted summation formula is as follows: S = W× (1 + R × T + Y × P), where S is the target parameter, W is the initial control parameter, R is the nonlinear compensation parameter, T is the compensation weight, Y is the adaptive adjustment factor, and P is the adjustment weight.

6. The high-precision control method for robot joint motors according to claim 1, characterized in that, Based on the target parameters, control commands are generated, including: The target parameters are analyzed; The parsed target parameters are matched with preset rules, and the matching results are obtained. Decision information is generated based on the matching results; The decision information is encoded and converted into control commands.

7. The high-precision control method for robot joint motors according to claim 1, characterized in that, The motor is adjusted and controlled according to the adjustment parameters, including: Obtain the adjustment type and adjustment data of the adjustment parameter; Generate adjustment operation instructions based on the adjustment type and the adjustment data; Adjustment control is performed according to the adjustment operation instructions.

8. A high-precision control system for a robot joint motor, characterized in that, The system includes: The first acquisition unit is used to acquire motor rotation angle position information, motor vibration data, and motor speed data; The first construction unit constructs a motor operating status dataset based on the motor rotation angle position information, the motor vibration data, and the motor speed data; The second construction unit is used to construct a digital twin model based on the motor's operating data information; The second acquisition unit is used to input the motor operating status dataset and the data output by the digital twin model into the artificial intelligence model to obtain preliminary control parameters; The output unit is used to input the real-time operating data of the motor into the neural network model, which is used to identify the nonlinear characteristics of the motor under different operating conditions and output nonlinear compensation parameters. The third acquisition unit acquires an adaptive adjustment factor based on the scene characteristics of the motor's operating state; The optimization unit is used to perform collaborative fusion optimization of the preliminary control parameters based on the nonlinear compensation parameters and the adaptive adjustment factor to obtain the target parameters; The adjustment unit generates control commands based on the target parameters, generates adjustment parameters for the motor according to the control commands, and sends the adjustment parameters to the motor so that the motor can be adjusted and controlled according to the adjustment parameters.

9. A high-precision control system for a robot joint motor, characterized in that, The system includes: Processor, memory, input / output units, and bus; The processor is connected to the memory, the input / output unit, and the bus; The memory stores a program, which the processor invokes to perform the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium contains a program that, when executed on a computer, performs the method as described in any one of claims 1 to 7.

Images