Abnormal Data Detection Method based on Fuzzy Autoregressive Hidden Markov Model
Fang Liu, Weixing Su, Jianjun Zhao, Xiandan Liang · Journal of Residuals Science and Technology · 2016
Abstract：In various fields of research, the use of data-based intelligent method to establish the research object model has received more and more attention. Accuracy of the data model is critical to the accuracy of subsequent studies, and accurate modeling data is necessary to establish an accurate model. In this paper, an anomaly detection method based on ARHMM is proposed, which is suitable for modeling data anomaly detection. The method divides the data into the correct data set and the abnormal data set by three-step detection respectively, and determines the membership function of the correct data set. Finally, the membership function of the three sets is used to judge whether the data is abnormal. Fuzzy ARHMM adopts the method of updating parameters online, which ensures that the method can be applied to the process modeling and data modeling, which is suitable for the process modeling data quantity and real-time property. The simulation results and practical application show that the method of anomaly data detection based on fuzzy ARHMM has good detection effect and practical application value.