Nowadays, the manual inspection is considered as a low efficiency way for substation inspection. The potential abnormal for equipments might not be warned since. The temperature measurement of electrical equipments is inaccurate. Therefore, an temperature early warning system for substation equipments is proposed based on the mobile infrared temperature measurement in this paper. This system takes the inspection robot as a mobile platform and new technologies such as infrared temperature measurement, WiFi communications and remote monitoring display, etc are incorporated. It is able to conduct temperature measurement early warning. In addition, the effects factors of infrared temperature measurement error are analyzed and the BP neural network is employed to modify the infrared temperature measurement errors. The proposed system combines the absolute and relative temperature measurements early warning method so as to achieve the ideal early warning effect. It has been run successfully in Deqing substation to verify its rationality and reliability.