Prediction Method of Air Braking Force of Heavy Load Train Based on Fuzzy Logic-based Neural Network

Abstract

Due to the strong nonlinearity of air braking of heavy load train, the large error of feedback decompression amount, and the coupling relationship between charging and exhaust time and decompression process, it is difficult to guarantee the operation accuracy of circulating air braking of heavy load train, which affects its operation safety. In order to improve the control accuracy of circulating air braking of heavy load train, a prediction method of air braking force based on fuzzy logic-based neural network (FLNN) is proposed in this paper. Firstly, radial basis function neural network (RBF-NN) is used to train off-line data of air braking, and off-line prediction rules of air braking force in the form of fuzzy logic are obtained. Then, matching degree between current data and off-line prediction rules of air braking force is calculated, and corresponding prediction rules are obtained. Finally, according to the current data and the corresponding prediction rules, predicted value of air braking force is output. By processing the data, this method can get rid of the dependence on traditional air braking model, avoid coupling analysis between charging and discharging time and decompression process, and can get air braking force prediction value more accurately. Test results show that air braking force prediction method based on FLNN can improve the prediction accuracy of heavy load train air braking force to 99% within 100 kN, which verifies that the method can effectively realize air braking force prediction under different working conditions.

Keywords

• heavy load train
• prediction of air braking force
• fuzzy logic-based neural network (flnn)
• radial basis function neural network (rbf-nn)
• prediction rules