A Novel Measurement Information Anomaly Detection Method for Cooperative Localization

Abstract

To deal with the influence of abnormal underwater acoustic ranging errors in cooperative localization of autonomous underwater vehicles, a novel measurement information anomaly detection method based on adaptive neuro-fuzzy inference system (ANFIS) is proposed. The method can accurately identify and isolate acoustic distance information with errors exceeding the threshold range even when multiple distance information is alternately used as the measurement data for the filtering algorithm. The adaptive cubature Kalman filter is used to extract the characteristic information, which can better reflect the change of measurement information. According to the preset state threshold, the flag bit obeying Bernoulli distribution is obtained, and the hybrid database is established. This method combines the online data training mechanism with the ANFIS-based detection system to update the ANFIS rules online, which can effectively improve the reliability and accuracy of anomaly detection method, especially when the sample data are insufficient. Experimental results based on data obtained from the actual lake-water trial show that the method can accurately identify the abnormal acoustic distance information and retain the accurate distance information to ensure the stable operation of cooperative localization system.