CNN LSTM based depth learning method and multi attribute time sequence data fault diagnosis method

The invention relates to the technical field of system fault diagnosis, in particular to a CNN-LSTM-based depth learning method and a multi-attribute time sequence data fault diagnosis method. The method comprises the following steps of S1, collecting system historical operation data and performing data preprocessing, and then building a fault diagnosis model based on CNN and LSTM; S2, collectingsystem real-time operation data and performing data preprocessing, then, sending the data to the fault diagnosis model built in S1 to be processed, and outputting a diagnosis result. According to thefault diagnosis model built on the basis of CNN and LSTM, attribute dimension characteristic information and time dimension delay information can be well integrated, and therefore the fault diagnosisaccuracy and noise resistance can be well improved.