During China’s ‘14th Five-Year Plan’, the rapid expansion of AC and DC power grids has led to more complex air gap breakdown characteristics in transmission lines. This highlights the urgent need to predict the air gap discharge voltage accurately under various conditions, forming a crucial basis for insulation design. Existing prediction methods that rely on a single AI algorithm often fall short in accurately classifying complex data. To improve this, the paper used the SVM to reduce the feature dimension, and introduce the KNN to handle the voltage prediction. An air gap discharge voltage prediction algorithm based on KNN-SVM is developed and used for predicting the rod-plane air gaps. Its performance is compared with traditional SVM predictions and previous experimental results, showing a relative error of 0.013, which is lower than that of the SVM approach. This confirms the effectiveness of the proposed method and provides theoretical support for predicting air gap discharge voltage under various conditions.