<p>During the operation of digital temperature meter of main transformer, its stability will be reduced due to the change of ambient temperature. Therefore, a temperature compensation algorithm for digital temperature meter of main transformer based on Back Propagation Neural Network (BPNN) is proposed. Firstly, the algorithm uses BPNN as the compensation core, and constructs a Support Vector Machine (SVM) classification model based on BPNN. Considering the possible noise or redundant information, measures are taken to deal with the temperature data according to the model, and the communication connection between the compensation interface and the digital temperature instrument is established. Pre-set the temperature compensation warning value of the main transformer, and carry out real-time temperature rise warning to realize the temperature compensation measured by the digital temperature instrument of the main transformer. The experimental results show that the maximum temperature of the nodes in the upper layer, middle layer and lower layer appears at about 5.5&#xa0;h, which can reduce the temperature of each area of the digital temperature instrument, ensure the stability of its operation, and has obvious temperature compensation effect.</p>