Method of Generalization and Inference of the Results of the Anomalies Analysis and Patterns of Changes in the Characteristics of Storage Systems

The paper describes a method for generalizing and logically inferring the results of anomaly analysis and patterns of changes in the characteristics of the storage system. This method includes three stages: detecting anomalies using effective machine learning methods (using an LSTM network as an example). The second stage is filtering the results of time series analysis by analyzing the patterns of the subject area, enclosed in the form of a subject OWL ontology. The third stage is the logical inference of recommendations and results of the analysis of time series of characteristics of the state of the storage system based on the results of the first two stages. The paper also presents a model of the ontology of data storage system state analysis. In conclusion, the results of the experiments on the data of the storage system are presented, which showed the effectiveness of the proposed approach subject to the development of a correct ontology of the subject area and its constant updating.