A Mind Evolutionary Algorithm Optimized Back-Propagation Neural Network Model for Tire-Road Friction Coefficient Prediction

The tire-road friction coefficient has a critical impact on the driving stability of vehicles, and it is the key parameter of vehicle dynamics control systems. This paper aims to improve the prediction accuracy and efficiency of the tire-road friction coefficient. Therefore, a Mind Evolutionary Algorithm optimized Back-Propagation (MEA-BP) neural network model for the prediction of the tire-road friction coefficient is proposed for tire-road friction coefficient predicting; and compared with the extreme learning machine (ELM) and BP neural network algorithms. The results show that the prediction accuracy rate of MEA-BP neural network algorithm is 8.8% higher than ELM algorithm, and 5.6% higher than BP neural network algorithm. In addition, different types and numbers of input variables are selected to study the efficiency and accuracy of tire-road friction coefficient prediction. The study found that the slip angle, tire longitudinal force, tire lateral force and tire vertical force have a significant impact on the prediction accuracy of the tire-road friction coefficient. As the number of input variables increases, the prediction accuracy gradually improves. When the number of input variables reaches 12, the growth rate of prediction accuracy slows down.