According to the current technical specifications for infrared diagnostic technology for live equipment, positive temperature rise of 1 K is a criterion for low-value porcelain insulators. However, the 1 K temperature difference threshold is loose in many real cases, which may cause a high miss rate as to the infrared detection. This paper summarizes and statistically analyzes the infrared detection data of porcelain insulators in a provincial electric power research institute. Firstly, the distribution function of low-value temperature difference samples is fitted, and then the log-likelihood, Kolmogorov-Smirnov and AIC criteria are used to test the goodness of fit of the model, and then a low-value insulator temperature difference mathematics based on inverse Gaussian distribution is proposed. The model is used to obtain the model parameters by using the maximum likelihood estimation method, and the error analysis is carried out. Finally, the accurate temperature difference threshold model of the infrared detection low value porcelain insulator is obtained. The research shows that the inverse Gaussian distribution can well fit the low-value temperature difference data, and the resulting temperature difference threshold model can improve the infrared detection accuracy of low-value porcelain insulators, and can be used to reasonably optimize the temperature difference threshold for low-value porcelain insulators.