The error compensation principle of pressure transmitter is deep studied，BP neural network algorithm is used to compensate the output error caused by temperature change of the pressure sensor First the factors affecting the measurement of pressure transmitter is analysed , A neural network model is established, next part of the experimental data for training to get neural network structure, And the effectiveness is verified by the experimental data of residual neural network, The results show that the method can effectively restrain the influence of temperature on the output of the pressure sensor and improve the stability and accuracy of the sensor, the method of temperature compensation can be used in the transmitter design, production.