Marine sensor monitoring data drift detection method

The invention relates to the technical field of electric digital data processing, and discloses a marine sensor monitoring data drift detection method, which comprises the following steps of: smoothing original monitoring data of a marine sensor by adopting an S-TWIMA method to obtain smoothed data; a CEEMDAN method is adopted to decompose the smoothed data, all decomposition items after decomposition are obtained, and the decomposition items comprise an IMF item and a residual item; the method comprises the following steps: decomposing a plurality of decomposition items, carrying out feature extraction on all the decomposition items after decomposition through TimesNet, mapping the extracted features to a feature vector with a fixed dimension, carrying out drift phenomenon classification on the feature vector by using a K-means clustering algorithm, finally detecting the drift phenomenon by using an improved ADF detection method, and marking the drift time. According to the method disclosed by the invention,