A New Method for Anomaly Detection and Diagnosis of Ocean Observation System based on Deep Learning

Due to the harsh conditions of the ocean observation environment and the influence of many uncertain factors, data anomalies often appear in the ocean observation system. Therefore, how to detect and diagnosis data anomalies and take corresponding measures has become an urgent problem to be solved. In this paper, a novel method based on deep learning is presented for detecting and diagnosing the data anomalies of ocean observation system, which combines convolution neural network (CNN) and bidirectional long short-term memory recurrent neural network (BiLSTM) with attention mechanism (CBLA for short). By constructing a prediction model based on the CBLA, the short-term trend of ocean observation data is predicted, and the error threshold is set to realize the detection of data anomalies. In addition, through the discussion of different types of data anomalies, the causes of data anomalies are analyzed, and the diagnosis of data anomalies is actively explored. The experimental results show the accuracy of the prediction model and the effectiveness of the overall data anomalies detection and diagnosis algorithm.