Abstract: Aiming at the problems that BP neural network diagnosis accuracy is not high enough defect in transformer fault diagnosis, an algorithm combining the BP neural network with the beetle antennae search algorithm is proposed. The initial weights and thresholds of the optimization algorithm are taken as the initial weights and thresholds of the BP neural network, the model is presented and the simulation test is performed. The results showed that, compared with the traditional BP neural network, the new model overcomes the shortcoming of long training time and slow convergence. The presented method for transformer fault diagnosis is new