Aiming at the problem that theoutput accuracy of pressure sensor is not high due to the temperature drift phenomenon in the actual use,a calibration method based on improved genetic algorithm(GA)to optimize back propagation(BP)neural network was proposed.This method mainly improves the initial population and genetic operator of genetic algorithm and introduces t distribution disturbance.Taking the output voltage and temperature voltage of the pressure sensor module as the input,a set of optimal initial weights and thresholds were calculated by genetic algorithm and a BP neural network model was established.Finally,the trained neural network model was transplanted to the conditioning chip of the sensor module for experimental verification.The results show that this method can effectively reduce the influence of temperature on the output of pressure sensor.The full-scale relative error of pressure sensor after algorithm calibration reaches 0.01%FS,the zero drift reaches 8.444×10-⁶,and the sensitivity drift reaches 2.055× 10-⁶.The calibration effect is better than other algorithms.