Research on Temperature Situation Awareness and Auxiliary Decision-Making System Scheme of Substation Equipment

Abstract

ObjectivesTo enhance the intelligent management of substation equipment maintenance, timely identify and mitigate the risks of failures caused by device overheating, and ensure the safe and stable operation of the power grid, the temperature situation awareness and auxiliary decision-making scheme of substation equipment were proposed.MethodsThe research was carried out from four aspects: the perception layer, the understanding layer, the prediction layer, and the auxiliary decision-making layer. In the perception layer, the K-nearest neighbor (KNN) classification algorithm was used to analyze the correlation of multi-class temperature data. In the understanding layer, a BP neural network was employed to construct a historical data transmission model to address missing historical data issues. In the prediction layer, a temperature prediction model combining autoregressive integrated moving average (ARIMA) and support vector machine (SVM) was designed to handle nonlinear data and noise. In the auxiliary decision-making layer, the grey relational analysis was applied to analyze the relationship between equipment temperature changes and fault risks.ResultsThe verification results of numerical examples based on the proposed scheme show that the scheme realizes the effective perception of the future temperature variation trends of the equipment and provides a basis for the identification of equipment defects.ConclusionsThrough multi-dimensional and deep-level temperature data analysis, the proposed scheme reveals the potential correlation between equipment temperature and fault risk, realizes the prediction of the operational trend of substation equipment, and provides a reference for the optimization of operational mode and the formulation of equipment maintenance plan.

Keywords

• power system
• substation
• temperature state awareness
• auxiliary decision-making
• autoregressive integrated moving average (arima) model
• bp neural network
• support vector machine (svm)