Research On The Decoupling Algorithm Of Tire-Road Contact Three-Way Force Sensor Based On ReLU-BPNN

Tire-road contact three-way force is one of the important indicators for studying the early causes of road damage,improving pavement structure design and extending road service life,and also one of the important parameters for studying vehicle and component safety performance.However,due to factors such as mechanical machining accuracy and structural design principle of tire-road contact three-way force sensor,three-way force dimension coupling is caused,so reducing dimension coupling is essential for improving measurement accuracy of tire-road contact three-way force sensor.In recent years,domestic and foreign scholars have carried out a lot of research on reducing three-way force dimension coupling,such as optimizing sensor structure and numerical decoupling methods.These methods reduce three-way force dimension coupling to a certain extent,but there are still problems of incomplete decoupling.Based on reducing three-way force dimension coupling,this paper deeply analyzes the existing decoupling methods,and proposes a tire-road three-way force decoupling method based on Re LU-BPNN.Aiming at the problems that existing decoupling methods are not completely decoupled and have unsatisfactory decoupling effect,this paper improves the accuracy of three-way force decoupling by increasing neural network depth.At the same time,it introduces Dropout regularization strategy to randomly delete neurons and prevent network overfitting from leading to poor model prediction effect.Aiming at the problem of gradient disappearance caused by soft saturation activation functions Sigmoid and Tanh in backward propagation n process,this paper proposes a tire-road three-way force decoupling method based on Re LU-BPNN by using sparse characteristics of Re LU function.Experimental results show that this method can further reduce tire-road contact three-way coupling force and improve measurement accuracy.