Research on data anomaly discrimination method for multi-station fusion measurement system

Under the condition of multi station integration, the anomaly detection of data streams with different characteristics has high requirements for accuracy and real-time, and the traditional threshold detection method can not meet the application. Based on the data communication architecture of multi station fusion, a data flow anomaly detection method under the condition of multi station fusion based on improved density clustering algorithm and single class support vector machine algorithm is proposed. The density clustering algorithm is used to preprocess the normal samples, and then the single class support vector machine algorithm is used to establish the data flow detection model. Finally, a simulation example is used and compared with the traditional algorithm to verify the effectiveness of this method.