Research of Control and Simulation of 1:5 Tilting Bogie Based on Neural Network Prediction

The mechanical power system model of 1:5 trial tilting bogie is established in the ADAMS, and it´s considered to predict the filtered acceleration (velocity or displacement) from the frame using neural network, and the tilting actuation systems is controlled in accordance with the signal and research will be proceeded on the effect of according control algorithms imposed on the performance of swing bolsters. The research results shows that: under a certain satisfaction, the reference input of 1:5 trial tilting bogie control system established by predictive signal can not only guarantee the accuracy of controlling the start and stop, but also ensure the real time of control, and realize the compensation of filter delay.