Modeling and simulation of switched reluctance machine based aircraft electric brake system by BP neural network

Based on the electromagnetic characteristics of switched reluctance machine (SRM) obtained by finite element method (FEM), two nonlinear mapping relations, namely i(ψ,θ) and T(i,θ), are modeled by BP neural network (BPNN) with Levenberg-Marquardt(LM) algorithm. On this basis, the dynamic simulation model of SRM based aircraft electric brake system (SRM-EBS) is built in Matlab. The performance of SRM-EBS is simulated with dry runway, and many results including brake torque and distance are presented. The simulation process shows that the BPNN model of SRM has advantages including fast learning speed, small convergent error, strong generalization ability and small network size. The simulation results indicate that SRM is suitable for application in aircraft electric brake system.