Prediction method of insulation degradation trend of power transformer windings based on iot sensing data

To solve the problem of the lack of prediction methods that depend on IoT sensing data for the insulation degradation of power transformer windings, an online transformer winding insulation degradation prediction method based on multi-source IoT sensing data is innovatively proposed, which can realize the online prediction of winding insulation deterioration under the condition of considering the influence of various performance deterioration factors such as electricity, heat and machinery. First, the GAT-LSTM algorithm is used to predict the future trend of power transformer voltage, current, temperature, and partial discharge IoT sensing data, which have periodic and fluctuating features. Then, based on the prediction results of transformer IoT sensing data, the electrical, thermal, and mechanical performance degradation damage indicators of the transformer winding insulation are constructed. Finally, based on the tensor fusion algorithm, the three degradation damage indicators are fused into a comprehensive degradation trend prediction model for winding insulation, using the minimum quantization error method to achieve accurate prediction of the winding insulation degradation trend. This method was validated using actual IoT sensing data from transformers, and the experimental results indicate that the proposed method can accurately predict the evolution trend of transformer IoT sensing data and the degradation trend of winding insulation. Relying on the data of the previous three years and five years to carry out the deterioration trend prediction experiment of the winding insulation, the mean absolute error accuracy of the proposed method are 0.0120 and 0.0121, respectively. Compared with other algorithms, the accuracy advantage exceeds 0.005. The significance of this research is to predict the performance degradation trend of winding insulation by constructing a degradation model, and to take operation and maintenance measures before serious failures occur.