Online Forecasting and Anomaly Detection Based on the ARIMA Model

Abstract

Real-time diagnostics of complex technical systems such as power plants are critical to keep the system in its working state. An ideal diagnostic system must detect any fault in advance and predict the future state of the technical system, so predictive algorithms are used in the diagnostics. This paper proposes a novel, computationally simple algorithm based on the Auto-Regressive Integrated Moving Average model to solve anomaly detection and forecasting problems. The good performance of the proposed algorithm was confirmed in numerous numerical experiments for both anomaly detection and forecasting problems. Moreover, a description of the Autoregressive Integrated Moving Average Fault Detection (ARIMAFD) library, which includes the proposed algorithms, is provided in this paper. The developed algorithm proves to be an efficient algorithm and can be applied to problems related to anomaly detection and technological parameter forecasting in real diagnostic systems.

Autoencoders for Anomaly Detection in an Industrial Multivariate Time Series Dataset

“…where e th is a predetermined threshold value. The optimal threshold is properly selected to maximize the usual metrics, namely the Accuracy (A), the Precision (P) and the Recall (R) [5].…”

“…Conventional AD methods include clustering-based and statistical-based [1,2] methods; the k-NN [3] and the K-means [4] methods are probably the most popular clustering methods, while the autoregressive-moving-average models are typical statistical choices [5]. However, conventional methods demonstrate poor performance and face challenges such as low anomaly recall rate, noise-resilient anomaly detection, difficulty to deal with complex anomalies and in high-dimensional date-especial in case of interdependencies [2], etc.…”

“…Magnitude. Both positive and negative deviations were induced, with addition or subtraction of random numbers with magnitude: [1.6 bar, 2.6 bar] for pressure, [1 dB, 4 dB] for noise and [5,20]% of the maximum speed value of the curve for speed.…”

Unsupervised Offline Changepoint Detection Ensembles

“…Moreover, existing algorithms and approaches in the CPDE field are not systematized, and it is questionable to consider any of them as SOTA. Quite noteworthy is that all of the added-to-compare algorithms except for unsupervised ARIMAFD [46] are semi-supervised since they need a training set with the healthy operation mode, while all algorithms tested in our work are unsupervised. Therefore, compared algorithms are applicable in different situations and are not competing.…”

A New False Data Injection Attack Detection Model for Cyberattack Resilient Energy Forecasting

“…In [38], an adversarial ML approach is suggested for cyberattack-resistant load forecasting that is capable of detecting a broad spectrum of attacks without detecting outliers. Kozitsin et al [39] introduced a new computationally simple technique based on the auto-regressive integrated moving average model for both anomaly detection and forecasting systems.…”