LightGBM-TabTransformer-Based Hybrid Data-Driven Parameter Estimation Method for Under-Water WPT Systems

Wireless Power Transmission (WPT) is characterized by high efficiency and environmental friendliness, and accurate parameter estimation is a prerequisite to ensure normal and safe operation of the system. Complex and changeable environment often leads to the generation of uncontrollable factors, for example, underwater wireless transmission often leads to the position of the coil to produce offset caused by the wave point of the data, which has an impact on the output voltage. This paper proposes LightGBM-TabTransformer (LMTT) to estimate parameter. Thus, it can effectively solve the problem of large errors due to the traditional formula derivation algorithms easily receive the influence of the original delay, current distortion and other problems. The LMTT model collects and saves all the data that appear to change in the system uniformly to form a mixed data set, and collects, processes and learns the important features through LMTT, so as to generate perfect and accurate parameter estimation of the artificial intelligence model. And the feasibility of the method is verified by comparing and improving it with some common deep learning models. Finally, a simulation testbed is constructed to verify the feasibility and accuracy of the proposed hybrid data-driven parameter estimation method.