ROA-DBN transformer fault diagnosis method based on data expansion

The invention belongs to the technical field of power equipment monitoring, and relates to a ROA-DBN transformer fault diagnosis method based on data expansion, and the method comprises the steps: taking fault gas data as a fault sample; standardizing the expanded data sample and dividing the data sample into a training set and a test set; building a DBN network according to the initialization parameters, and inputting the training set into the DBN network; by taking the fault diagnosis accuracy of the DBN neural network model as fitness, optimizing the DBN model by using a bucket jellyfish optimization algorithm, and returning an optimal parameter; and constructing an ROA-DNB fault diagnosis model according to the returned optimal parameters, inputting the test set, and outputting a transformer fault diagnosis result. According to the method, the accuracy of transformer fault diagnosis can be improved by optimizing the key parameters of the ROA-DBN model. 本发明属于电力设备监测技术领域，涉及基于数据扩充的ROA-DBN变压器故障诊断方法，将故障气体数据作为故障