Fault diagnosis method based on density-induced support vector data description

The invention discloses a fault diagnosis method based on density-induced support vector data description recursive feature elimination, and the method comprises the steps: collecting data, and carrying out the standardization preprocessing of the data, and obtaining normal data and fault data; training by using normal data to obtain a detection model; training by using the fault data to obtain a diagnosis model, and obtaining a fault feature sorting set through the diagnosis model; performing standardized preprocessing on the test data, inputting the test data into the detection model for detection, if the test data is predicted to be abnormal data, entering the next step, otherwise, outputting the test data as normal data: obtaining a feature sequence of the abnormal data, comparing the feature sequence with a fault feature sequence set, and confirming a fault type. According to the method, on the basis of density-induced support vector data description, a feature selection method of recursive feature elimi