Abstract Online voltage transformer error prediction is an important research direction in the field of smart grids. This article mainly focuses on the online error prediction and the parallelization method of voltage transformers. First, an optimized multi-layer perceptron model based on the sparrow search algorithm (SSA) is proposed. The weight initialization process is optimized using the SSA to improve the prediction accuracy of the multi-layer perceptron. Considering the massive amount of data in real-world scenarios, a distributed sparrow search optimization algorithm for the multi-layer perceptron model was then developed, and the acceleration and scalability were tested on different data scales. In addition, transformer error prediction experiments were conducted to demonstrate the performance of the proposed algorithm.