In order to improve the safety and stability of transformer operation, a transformer insulation aging state evaluation model based on improved whale optimization algorithm?back propagation (IWOA?BP) algorithm is proposed. An oil?paper insulation model of transformer for polarization and depolarization current (PDC) method experiment is established. In order to avoid the influence of moisture content on the evaluation results, the comprehensive conductivity is extracted as the aging characteristic quantity. An evaluate algorithm model for the aging state of oil?paper insulation basing on BP neural network is established, in which PDC current data is the input and comprehensive conductivity is the output. In order to promote the evaluation effect of the algorithm model, the whale optimization algorithm (WOA) is improved by adding nonlinear convergence factors and adaptive weights. And IWOA is used to optimize the BP neural network to obtain better BP weights and threshold initial values. Thus, the convergence speed and accuracy of the BP neural network are improved. The final result proves that the proposed IWOA?BP evaluation model has certain accuracy and practicability, and has some development space in the direction of oil?paper insulation life evaluation.