Prediction of Dissolved Gas Content in Transformer Oil Based on SSA-BiGRU-Attention Model

The accurate prediction of the dissolved gas content in oil is of great significance to the safe and stable operation of a transformer. In order to overcome the limitations of traditional prediction methods that only consider a single variable of dissolved gas content in oil and rely on experience as the model selection parameter, an adaptive prediction method of dissolved gas content in transformer oil where multiple factors are taken into account is proposed. First, online monitoring data of dissolved gas in oil, temperature data (environment, oil in the top of transformer), and ambient wind speed data under normal and abnormal operating conditions of the transformer are collected. And the attention mechanism is used to calculate the specific weight of each data variable as the input of the prediction model. Then, the sparrow search algorithm (SSA) is used for intelligent iterative optimization of the four important parameters, such as batch processing, learning rate, number of hidden layers and number of neurons, in the bidirectional gated recurrent unit (BiGRU) model，and the optimal parameter values are obtained. On this basis, the SSA-BiGRU-Attention optimization model is constructed to realize the prediction of the dissolved gas content in oil. The results of the study show that the model proposed in this paper is used to predict the dissolved gas content in oil under normal operating conditions, and the prediction accuracy can reach 98.3%. For the prediction of the dissolved gas content in oil that reaches the attention value, the peak and inflection point of the gas data can be predicted better, and the prediction accuracy can reach 86.6%. The prediction method of gas content in transformer oil proposed in this paper can provide important technical supports for transformer state assessment and fault early warning. © 2022 Science Press. All rights reserved.