Research On Parameter Identification Of Wireless Power Transmission System Based On Magnetic Coupling Resonance

With the rapid development of modern industrial technology,higher requirements are put forward for the intelligence,safety,and convenience of the power supply system.Wireless Power Transfer(WPT)technology has the advantages of freedom,flexibility,safety,and reliability.Aroused people’s attention.Due to the existence of inevitable factors such as coil offset and transmission distance change in actual power supply occasions,the coupling coefficient between coils will fluctuate randomly.In addition,changes in the internal resistance of the battery during the charging process will also cause the WPT system to deviate from its optimal working condition.Therefore,it is very important to improve the robustness and efficiency of the system to carry out the research on the parameter identification of the wireless power transmission system.For the study of WPT system parameter identification,traditional methods need to obtain the impedance angle information of the original side of the WPT system or the evolutionary optimization algorithm to identify the parameters through an iterative process,which requires high-precision sampling equipment and a lengthy iterative process.For this reason,this article focuses on the coupling coefficient and load identification of the Series-Series(SS)type WPT system.By analyzing the relationship between the electrical information of the primary side and the parameters to be identified,a WPT system parameter identification method combining machine learning model and mechanism model is proposed.Finally,the effectiveness of the identification method is verified by simulation and experiment.The main research work of this paper is as follows:The circuit model of the SS-type WPT system is established based on the circuit theory.On this basis,the influence of the coil coupling coefficient and load resistance fluctuation on the output performance of the system is analyzed.A parameter identification method combining machine learning model and mechanism model is proposed.This method analyzes the basic principles of support vector machine(SVR),BP neural network and RBF neural network in machine learning algorithms,and builds the mathematical model of the algorithm on this basis.The above three methods all use the input voltage and current on the primary side as input factors,and the coupling coefficient as the label,and build learning models of three different algorithms.According to the identification value of the coupling coefficient,load identification is completed through the mechanism model.The accuracy and generalization ability of the above three parameter identification models are compared and analyzed.The research results show that the recognition accuracy and generalization ability of the SVR algorithm are better than the other two algorithms in the small training sample data.On this basis,a WPT system experimental platform was built to verify the effectiveness of the SVR-based parameter identification method under the conditions of varying transmission distance and horizontal misalignment.