GIS PARTIAL DISCHARGE DIAGNOSING METHOD, MODEL TRAINING METHOD, DEVICE AND SYSTEM

A GIS partial discharge diagnosing method, a model training method, a device and a system are disclosed. Sensor modules are in communication with each other, so that sensor network position distribution data of each sensor module in a wireless transmission network can be determined. In a training process of a partial discharge diagnosing model, a spatial-temporal feature of the partial discharge is introduced, so that the trained partial discharge diagnosing model is adaptive to different GIS equipment and different sensors layout solutions, and has better model universality and applicability, thus greatly saving a training time of the model and expediting the deployment of the partial discharge diagnosing model. Moreover, the model trained in the present disclosure accounts for the relationship between the position where partial discharge occurs and the sensor network position distribution. This allows the trained partial discharge diagnosing model of the present disclosure to eliminate interference from discharge signals occurring outside the GIS, enhancing the accuracy of partial discharge type identification.