Abstract:Aiming at the system composed of solenoid valve, wet clutch and pressure sensor in hydro-mechanical continuously variable transmission (HMCVT), in order to establish the dynamic response model of the pressure sensor system more accurately, and to improve the calibration accuracy of the pressure sensor under specific operating conditions, based on the improved simulated annealing (SA) algorithm and the radial basis function neural network (RBF-NNN), the methods to identify the system and calibrate the pressure sensor are proposed. The experimental results show that the dynamic response of the system is consistent with that of the second order system. The natural frequency is 1312 Hz, the damping ratio is 038, and the system sensitivity is 045. The identification accuracy of the system is high, and the average error is only 007. According to the measurement data at the zero point, only the samples under 2 working conditions are needed as the training sample, and the pressure sensor under the input step signal condition can be calibrated accurately with the RBF-NN.