Abstract: Based on fuzzy clustering and LSTM network, a data driven anomaly analysis method of launch vehicle engine oxygen turbopump data is proposed. Firstly, fuzzy clustering is used to pre classify the data samples with complex working conditions and incomplete labels to obtain complete labels and analyze the feature contribution, which lays a foundation for feature screening and training of LSTM network. The LSTM network is used to predict the data of the oxygen turbopump, and the average error between the predicted results and the original data is calculated. Then the threshold criterion calculated by the non parametric threshold calculation method is used to determine whether the turbopump is abnormal. Finally, the fault detection and alarm driven by the oxygen turbopump data are realized, and the accuracy is improved by 7% compared with the red line threshold detection method.

Key words: Launch vehicle engine, Anomaly detection, Oxygen turbopump, Fuzzy clustering, LSTM networks