A method, device and product for evaluating the insulation state of a hydro-generator stator winding

Abstract

The application relates to the technical field of water-turbine-generator detection, and discloses a water-turbine-generator stator winding insulation state evaluation method, device and product, the method comprising the following steps: obtaining original multi-dimensional characteristic quantity data sets of the stator winding of a water-turbine generator at the current maintenance time, and obtaining a multi-dimensional characteristic set through data preprocessing and feature extraction; based on the multi-dimensional characteristic set, the actual insulation health index value of the stator winding is obtained through insulation health index model calculation; based on the multi-dimensional characteristic set, the fault type of the stator winding is obtained through a fault recognition model based on a graph convolutional neural network and a partial discharge mode recognition model based on a convolutional neural network; the insulation state of the stator winding is comprehensively evaluated according to the actual insulation health index value and the fault type, and the insulation state comprehensive evaluation result of the stator winding is generated, the multi-source data driven stator winding insulation state quantitative evaluation and fault intelligent recognition are realized, and the diagnosis comprehensiveness and objectivity are improved.

Images