Objectives  The ship shafting monitoring system has many characteristics such as many monitoring objects, large amount of test data, and large storage space requirements. If there is no reasonable storage scheme and the data are not pre-counted and classified, it will cause inconvenience in data retrieval, calculation and analysis. In this end, a method for batch statistics and storage in different tables of characteristic data is proposed in this paper.

  Methods  Firstly, based on the existing database, the time series characteristic data table and the summary table were designed, and a statistical method for extracting the value of the mean, extreme variance and skewness of the test data was designed. Then, through simulation comparison, the mean and skewness of the test data were selected as characteristic vectors, and the sensor anomaly recognition method was designed by DBSCAN(Density-Based Spatial Clustering of Application with Noise) clustering algorithm so as to validate the effectiveness of recognition.

  Results  The results show that the proposed method has a good recognition effect on the abnormal data of the sensor system.

  Conclusions  The method can be used to extractcharacteristic data in practical application.