Gas data anomaly detection based on time-series ARIMA model

Abstract

With the rise of artificial intelligence technology in recent years, combining artificial intelligence technology with industry has become a popular research direction, and safety issues in industry have gradually become important. Gas data can directly show the health condition of gas, and the research of anomaly detection on gas data is one of the important means to improve gas safety. To address the gas data problem, this paper proposes a gas data anomaly detection method based on a time-series ARIMA model. Firstly, the data in the SCADA system is pre-processed, then the features of the gas data are extracted, then a time-series model is established using ARIMA, and finally the anomaly is determined by the difference between the model prediction data and the real data. And the method proposed in this paper is verified in real data, and the results show that the average values of the indexes of true positive rate, false positive rate and accuracy rate of the method in this paper are 0.18%, 0% and 96.6% respectively, which verifies the effectiveness of the method in this paper.