CN-120915196-B - High-precision position maintaining method, device and storage medium for multi-dimensional motor cooperative control

CN120915196BCN 120915196 BCN120915196 BCN 120915196BCN-120915196-B

Abstract

The application discloses a high-precision position maintaining method, a device and a storage medium for cooperative control of a multi-dimensional motor, which are used for maintaining the high-precision position of the multi-dimensional motor. The method comprises the steps of constructing a multi-dimensional data set, carrying out standardization processing on the multi-dimensional data set to obtain standard data, calculating a covariance matrix based on the standard data and carrying out eigenvalue decomposition, determining a plurality of principal component eigenvectors according to the accumulated variance contribution rate to form a principal component space, projecting the standard data into the principal component space to form a dimension-reduction data set, inputting the dimension-reduction data set into a pre-trained deep belief network, outputting comprehensive eigenvectors through hierarchical features of a limited Boltzmann machine, inputting the comprehensive eigenvectors into a depth deterministic strategy gradient algorithm model, constructing a reward function, outputting a real-time adjustment strategy based on the reward function to obtain optimal control parameters, and realizing high-precision position maintenance of a plurality of motors according to the optimal control parameters.

Inventors

ZHU ZHONGLEI
SUN MENG
ZHOU YANQIONG
WANG YUANSHU

Assignees

雷文斯（深圳）科技有限公司

Dates

Publication Date: 20260512
Application Date: 20250804

Claims (8)

1. The high-precision position maintaining method for the coordinated control of the multi-dimensional motor is characterized by comprising the following steps of: Collecting position data, rotating speed data, torque data, environment data, vibration data and current data of a plurality of motors in real time; aligning the position data, the rotational speed data, the torque data, the environment data, the vibration data and the current data based on a time stamp to construct a multi-dimensional dataset; carrying out standardization processing on the multidimensional data set to obtain standard data; calculating a covariance matrix based on the standard data and decomposing the eigenvalue; Determining a plurality of principal component feature vectors according to the accumulated variance contribution rate to form a principal component space; projecting the standard data into the principal component space to form a reduced-dimension data set; Inputting the dimension reduction data set into a pre-trained deep belief network, and outputting a comprehensive feature vector through the hierarchical features of the limited Boltzmann machine, wherein the comprehensive feature vector is used for representing the position error dynamic characteristics, the load torque coupling relation, the environment disturbance response characteristics of a plurality of motors and dynamic compensation parameters required by the cooperative control of the plurality of motors; inputting the comprehensive feature vector into a depth deterministic strategy gradient algorithm model to construct a reward function; Outputting current loop parameters, speed loop compensation coefficients and position loop parameters of a plurality of motors based on the reward function, and obtaining optimized control parameters; According to the optimized control parameters, the current loop parameters, the speed loop compensation coefficients and the position loop parameters of the motors are adjusted in real time so as to realize high-precision position maintenance of the motors; inputting the comprehensive feature vector into a depth deterministic strategy gradient algorithm model to construct a reward function, wherein the method comprises the following steps of: Inputting the comprehensive feature vector into a depth deterministic strategy gradient algorithm model to generate initial control parameters, wherein the initial control parameters comprise current loop parameters, speed loop compensation coefficients and position loop parameters of the motor; Controlling a plurality of motors to run according to the initial control parameters, and collecting actual position data of the plurality of motors, synchronous data among the plurality of motors and energy consumption data of the plurality of motors in real time to generate actual control parameters; Based on the actual control parameters, calculating position errors, synchronization errors and energy consumption respectively, and constructing a reward function comprising the position errors, the synchronization errors and energy consumption optimization; Based on the actual control parameters, respectively calculating a position error, a synchronization error and energy consumption, and constructing a reward function comprising the position error, the synchronization error and energy consumption optimization, wherein the method comprises the following steps: and designing a punishment term of the position error by adopting an exponential decay function, wherein the formula is as follows: Wherein, the As a coefficient of sensitivity to a position error, As an actual position of the motor, Is the target position of the motor; and designing a penalty term of the synchronous error by adopting a Gaussian kernel function, wherein the formula is as follows: wherein M is the number of motors, In order to synchronize the error-sensitive coefficients, And The actual positions of motors i and j; The optimization term of the energy consumption is designed by adopting a current level integral form, and the formula is as follows: Wherein, the As the weight coefficient of the energy consumption, Is the motor current.
2. The high-precision position maintaining method according to claim 1, wherein after obtaining an optimal control parameter based on a real-time adjustment strategy of current loop parameters, speed loop compensation coefficients, and position loop parameters of the plurality of motors output by the bonus function, before adjusting the current loop parameters, the speed loop compensation coefficients, and the position loop parameters of the plurality of motors in real time according to the optimal control parameter to achieve high-precision position maintaining of the plurality of motors, the method further comprises: calculating deviation values of actual positions and target positions of a plurality of motors to form a position deviation sequence; Calculating a comprehensive error index based on the position deviation sequence; triggering a reward function correction mechanism if the comprehensive error index exceeds a preset threshold; Inputting the current multidimensional data set into the deep belief network, outputting the current comprehensive feature vector, and analyzing an error source by combining a preset fault feature library; Dynamically correcting a reward function in the depth deterministic strategy gradient algorithm model according to the error source; and inputting the corrected reward function into a depth deterministic strategy gradient algorithm model, and regenerating the adjustment strategies of a plurality of motors to obtain new optimized control parameters.
3. The high-precision position maintaining method according to claim 1, wherein the outputting the current loop parameters, the speed loop compensation coefficients, and the position loop parameters of the plurality of motors based on the bonus function in real time to obtain the optimized control parameters includes: evaluating the running states of a plurality of motors under the actual control parameters based on the reward function, and outputting an evaluation result; and generating an optimized control parameter based on an adjustment strategy of a depth deterministic strategy gradient algorithm according to the evaluation result.
4. The high-precision position maintaining method according to claim 1, characterized in that, after aligning the position data, the rotational speed data, the torque data, the environment data, and the current data based on time stamps, a multidimensional dataset is constructed, the multidimensional dataset is subjected to normalization processing, and before standard data is obtained, the method further comprises: and carrying out multi-scale decomposition on the multi-dimensional data set by adopting wavelet transformation to remove high-frequency noise.
5. The high-precision position maintaining method according to claim 4, wherein the multi-scale decomposition of the multi-dimensional dataset using wavelet transform to remove high-frequency noise comprises: Performing multi-scale decomposition on the multi-dimensional data through a wavelet basis function to obtain different decomposition layers; Adopting an adaptive threshold to reduce noise of the different decomposition layers; Processing the high-frequency coefficients in the different decomposition layers through a soft threshold function to obtain noise-reduced coefficients; And reconstructing the noise-reduced coefficient and the low-frequency approximate coefficient to finish high-frequency noise removal.
6. A high precision position maintaining apparatus for coordinated control of a multi-dimensional motor, characterized by performing the method of any one of claims 1 to 5, comprising: The acquisition unit is used for acquiring position data, rotation speed data, torque data, environment data, vibration data and current data of the motors in real time; An alignment unit configured to align the position data, the rotation speed data, the torque data, the environment data, the vibration data, and the current data based on a time stamp, and construct a multi-dimensional dataset; the processing unit is used for carrying out standardized processing on the multidimensional data set to obtain standard data; the decomposition unit is used for calculating a covariance matrix based on the standard data and decomposing the eigenvalue; a forming unit for determining a plurality of principal component feature vectors according to the cumulative variance contribution ratio to form a principal component space; The projection unit is used for projecting the standard data into the principal component space to form a reduced-dimension data set; The output unit is used for inputting the reduced data set into a pre-trained deep belief network, and outputting a comprehensive feature vector through the hierarchical features of the limited boltzmann machine, wherein the comprehensive feature vector is used for representing the position error dynamic characteristics, the load torque coupling relation, the environment disturbance response characteristics of a plurality of motors and dynamic compensation parameters required by the cooperative control of the plurality of motors; the construction unit is used for inputting the comprehensive feature vector into a depth deterministic strategy gradient algorithm model to construct a reward function; The acquisition unit is used for outputting current loop parameters, speed loop compensation coefficients and position loop parameters of a plurality of motors based on the reward function and acquiring optimal control parameters; And the holding unit is used for adjusting the current loop parameters, the speed loop compensation coefficients and the position loop parameters of the motors in real time according to the optimized control parameters so as to realize high-precision position holding of the motors.
7. A high-precision position maintaining device cooperatively controlled by a multi-dimensional motor, comprising: A processor, a memory, an input-output unit, and a bus, the processor being connected to the memory, the input-output unit, and the bus, the memory holding a program, the processor invoking the program to perform the high-precision position maintaining method according to any one of claims 1 to 5.
8. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon a program which, when executed on a computer, performs the high-precision position maintaining method according to any one of claims 1 to 5.

Description

High-precision position maintaining method, device and storage medium for multi-dimensional motor cooperative control Technical Field The embodiment of the application relates to the field of industrial automation control, in particular to a high-precision position maintaining method, device and storage medium for multi-dimensional motor cooperative control. Background In the fields of high-precision control of industrial automation, robots, aerospace and the like, the multi-dimensional motor cooperative system is widely applied to complex motion control scenes, such as multi-joint mechanical arms, precision machining platforms, satellite attitude adjustment systems and the like. The system generally requires that multiple motors still can maintain the position accuracy of micron level and even submicron level under the complex conditions of dynamic load, environmental interference and the like, and simultaneously satisfies multiple constraints such as energy consumption optimization, synchronization control and the like. The traditional single-motor control method cannot meet the coupling effect and the cooperative requirement among multiple motors, so that the multidimensional motor cooperative control technology becomes the core direction of current research. At present, in the multi-dimensional motor cooperative control, only data of limited dimensions such as position, rotating speed, current and the like are collected. The data are subjected to linear dimension reduction processing such as traditional low-pass filtering or principal component analysis and the like, and then are input into a preset motor mathematical model or a fixed parameter PID control algorithm to generate a control instruction. The control algorithm based on the preset model lacks self-adaptive capability, when facing complex working conditions such as dynamic load change, environmental parameter fluctuation and the like, parameters are required to be manually reset, parameters of a current loop, a speed loop and a position loop cannot be automatically optimized in a real-time data driving mode, and finally position errors of the system are gradually accumulated in a long-time operation process, so that severe requirements of high-end equipment on high-precision position maintenance are difficult to meet. In addition, neglecting the influence of temperature and humidity change on the resistance of the motor winding and the mechanical resonance risk reflected by the vibration sensor data, the control strategy is easy to sense the state of the system comprehensively, and the nonlinear characteristics in the motor operation data are difficult to effectively process by the linear dimension reduction method. Disclosure of Invention The application discloses a high-precision position maintaining method, a device and a storage medium for cooperative control of a multi-dimensional motor, which are used for maintaining the high-precision position of the multi-dimensional motor. The first aspect of the application discloses a high-precision position maintaining method for cooperative control of a multi-dimensional motor, which comprises the following steps: Collecting position data, rotating speed data, torque data, environment data, vibration data and current data of a plurality of motors in real time; aligning the position data, the rotational speed data, the torque data, the environment data, the vibration data and the current data based on a time stamp to construct a multi-dimensional dataset; carrying out standardization processing on the multidimensional data set to obtain standard data; calculating a covariance matrix based on the standard data and decomposing the eigenvalue; Determining a plurality of principal component feature vectors according to the accumulated variance contribution rate to form a principal component space; projecting the standard data into the principal component space to form a reduced-dimension data set; Inputting the dimension reduction data set into a pre-trained deep belief network, and outputting a comprehensive feature vector through the hierarchical features of the limited Boltzmann machine, wherein the comprehensive feature vector is used for representing the position error dynamic characteristics, the load torque coupling relation, the environment disturbance response characteristics of a plurality of motors and dynamic compensation parameters required by the cooperative control of the plurality of motors; inputting the comprehensive feature vector into a depth deterministic strategy gradient algorithm model to construct a reward function; Outputting current loop parameters, speed loop compensation coefficients and position loop parameters of a plurality of motors based on the reward function, and obtaining optimized control parameters; and according to the optimized control parameters, the current loop parameters, the speed loop compensation coefficients and the position loop parameters of the motors