To solve the problem that it is difficult to accurately predict the degradation variation of smart meter under extreme temperature stress, a degradation prediction model based on the combination of fusion kernel support vector regression (FSVR) and genetic algorithm (GA) was proposed. Firstly, to improve the learning and generalization ability of prediction model, a new fusion kernel function based on RBF kernel and Sigmoid kernel was proposed, and the FSVR model based on fusion kernel function was further established to predict the degradation process of smart meter under extreme temperature stress. Then, in the kernel parameter adjustment stage, the kernel parameters of SVR are optimized by GA to improve the prediction accuracy of the model. Based on the actual smart meters test data for 14 months in high temperature (50℃) and low temperature (-40℃), the experimental results show that the proposed prediction model can accurately track the degradation trend of smart meters under different extreme temperatures, which can provide guidance for reliability analysis of smart meters in typical regions of China.