Considering the current temperature compensation method is used to establish the temperature compensation model using intelligent algorithm,and use swarm intelligent optimization algorithm to optimize and improve the compensation precision,has good compensation effect for nonlinear sensor temperature drift,but for the low efficiency of this method has good linearity,and the use of linear least squares fitting method the routine can get better compensation effect,so this will be the least squares fitting method and RBF neural network model integration,a model of temperature compensation of pressure sensor,using ant colony algorithm to optimize the conventional RBF neural network,improve the performance of compensation model. Through the MPX53 pressure piezoresistive pressure sensors were studied. The results showed that after using the temperature compensation method,sensor at various temperatures is basically the same temperature,with the use of the whole ant colony optimization RBF neural network method of temperature compensation effect is similar,but the intermediate temperature region due to the linear fitting method,the efficiency of the whole temperature compensation improved.