Dynamic Decoupling Control of PMa-BSynRM Based on Neural Network Inverse System Optimized by Enhanced Chaos PSO

To solve the problems of slow convergence, susceptibility to local optima, low fitting accuracy, and generalization ability caused by poor connection weights in the backpropagation neural network inverse system (BPNNIS) decoupling control for permanent magnet assisted bearingless synchronous reluctance motor (PMa-BSynRM), a novel dynamic decoupling control strategy in full-speed domain based on enhanced chaos particle swarm optimization (ECPSO)—adaptive switch Adam to SGD (ASATS)—BPNNIS is proposed in this study. First, the structure and principle of the PMa-BSynRM are introduced, and the reversibility analysis is also deduced based on the mathematical models. Second, the ECPSO algorithm and ASATS algorithm are expounded and verified to realize the optimization and real-time update of the connection weights. Finally, the experiments show that the dynamic decoupling performance among speed, <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">x</i>- and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">y</i>-direction radial displacement, is improved, and the static stability and dynamic responsiveness of this system verify the effectiveness of the scheme.