Research On Torque Test System Of Driving Wheel Based On RBF Neural Network

As an important component of the propulsion device of tracked vehicles,the measurement of the axial torque of the driving wheel provides important reference and effective technical support for optimizing the motion mechanism.The design of the test system is constrained by space conditions.The combination of neural network algorithm and indirect measurement technology provides a new idea for parameter measurement of difficult measurement points,which has practical significance.In this paper,in the test of the torque of the driving wheel of the tracked vehicle,the storage test technology was used to collect the strain signals at the four hubs of the driving wheel.The RBF neural network algorithm based on the conjugate gradient descent method,combined with virtual instrument technology,solved the problem of indirect torque testing of tracked vehicle driving wheels.First of all,the test environment and the design requirements of test system were analyzed in this paper,the corresponding solutions were proposed,and the overall design scheme was formed.Then,in order to ensure the reliability of the tester,the mechanical structure of the tester was designed.According to the characteristics of the measured signal,combined with the storage test technology,the analog circuit and the digital circuit modules in the hardware circuit were designed respectively,and the strain signals of four hub test points were obtained.Then,based on the standard RBF neural network algorithm and the idea of exponential smoothing algorithm,the RBF neural network prediction algorithm based on the conjugate gradient descent method was proposed,and the validity of the algorithm was verified by the calibration data.Finally,in order to obtain the torque value of indirect dynamic measurement of wheel axle,a comprehensive multi-channel information processing software platform based on LabVIEW and neural network was designed.It had the functions of multi-channel data receiving,displaying and data processing for the data stored in the hardware circuit test.It showed that the software had advantages in data acquisition and analysis,could reduce the application threshold of intelligent algorithm,and was convenient for users.This test method could not only indirectly test the torque of active wheel,solve the problem of multi parameter fitting of complex nonlinear system,but also could be used in the test of stress field,temperature field and other field variables,to establish the relationship between the relevant parameters,or indirectly accept the parameters of the target.