Chaos Optimizing BP-NNG Speed Recognition in DTC System

According to the non-linear relationship of direct torque control (DTC) system, multiorbit chaos optimizing algorithm is put forward, which resolve the slower converging problem of single-orbit and single-non-linearity-function chaos algorithm. In addition, the conception of neural network group (NNG) is proposed to resolve the bigger periodical errors problem. As the sub-network of NNG is intended to deal with different data group, the accuracy has been improved greatly. The result of DTC system simulation in MATLAB/SIMULINK shows that NNG speed recognition optimized by the multiorbit chaos optimizing algorithm has better tracking capability and fitness, as well as favorable static and dynamic properties.