An improved Hopfield Lagrange network with application on motor efficiency optimization

In this paper, an improved Hopfield Lagrange network (IHLN) is proposed and applied to solve motor efficiency optimization in an electric vehicle (EV) with a permanent magnet synchronous motor (PMSM). In contrast to the Hopfield Lagrange network (HLN) incorporating deterministic gradient descent, a stochastic searching method, simulated annealing (SA), is introduced in IHLN to prevent the network from falling into local minima and improve the optimality. Some important issues regarding the performance of the IHLN are thoroughly investigated such as whether the optimization problem is nonconvex with local minima, the stability of the network's equilibria, and the convergence condition of the SA. Simulations are conducted to validate the effectiveness of IHLN, and results demonstrate the improvement brought by IHLN concerning different indices.