Aiming at the problems of serious sensor drift, poor environmental adaptability and insufficient fault prediction accuracy in SF6 humidity monitoring of high-voltage equipment, an optimization scheme of sensor network with double closed-loop cooperation of abnormal drift compensation and fault dynamic prediction is proposed. This approach employs the Sparrow Search Algorithm (SSA) to optimize the Backpropagation Neural Network (BPNN)—SSA-BPNN—to construct a multi-parameter fusion compensation model integrating temperature, humidity, and pressure, dynamically correcting output drift in quartz resonator sensors. Integrating fuzzy reasoning with Long Short Term Memory (LSTM) networks, a dual layer fault prediction architecture consisting of short-term trend prediction and long-term health assessment is established to quantify fault risks and predict the remaining lifespan of equipment. The experimental results show that the root mean square error of the compensation model is reduced to 7.23 ppm(v) under the working conditions of −30 °C∼50 °C and 0.1∼0.5 MPa, and the average absolute percentage error is only 2.2%, which is significantly better than the traditional method. Based on the field data of GIS equipment for 18 months, the scheme accurately predicts the seal failure two months in advance, and the health index early warning mechanism effectively prolongs the maintenance response time. This research provides a full chain technology solution from sensor compensation to fault prediction for intelligent monitoring of high voltage equipment.
Yang et al. (Sun,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: