The accuracy of infrared body temperature measurement is affected by many factors, which is characterized by nonlinearity and high complexity. In order to improve the accuracy of infrared body temperaturemeasurement, the influence of ambient temperature, measurementdistance and emissivity on the accuracy of infrared body temperature measurement analyzed. The temperature field diffusion model of the hip surface temperature is transformed into the actual body temperature of the human body. The partial least squares method and the artificial neural network are used to optimize the temperature field model, which effectively solves the problem of the temperature field diffusion method. The problem of multiple correlation between influencing factors and the nonlinear problem of compensation modelare served effectively, and the reliability of the system is improved. The experimental results show that the temperaturemeasurement error of the proposed infrared temperature compensation model is -0.12~0.11 ℃, which has higher measurement accuracy and stronger adaptability.