基于改进型BP神经网络的PID控制器在惯性测量系统中的应用

o adapt slow time variation of the hardware parameters and nonlinearities in Magnetic Inertial Measurement System (MIMS), the improved Back Propagation Neural Network (BPNN) algorithm is effectively integrated to the traditional PID controller. BPNN has the ability to represent any nonlinear functions, which can achieve the best combination of PID three coefficients in real time online learning. Using BPNN can help to build the coefficient self-learning PID controller in MIMS. The simulation which was established in MATLAB indicates that the improved BPNN PID controller can improve the robustness of system and has high accuracy control. At last, experiments in MIMS convincingly verified the simulation.