LSTM-based Encoder-Decoder for Multi-sensor Anomaly Detection

TL;DR

The paper addresses anomaly detection in multi-sensor time-series where external factors can render data unpredictable. It introduces EncDec-AD, an LSTM-based encoder-decoder trained only on normal sequences to learn typical behavior and detect anomalies via reconstruction error modeled with a Gaussian, enabling robust detection across predictable and unpredictable time-series. Extensive experiments on power demand, space shuttle, ECG, and engine data show EncDec-AD achieving high anomaly-likelihood and superior performance on unpredictable sequences, while prediction-based LSTM-AD can outperform on strictly predictable data. The approach is effective for both short and long sequences, offering a practical, data-efficient solution for real-world industrial monitoring with limited anomalous data.

Abstract

Mechanical devices such as engines, vehicles, aircrafts, etc., are typically instrumented with numerous sensors to capture the behavior and health of the machine. However, there are often external factors or variables which are not captured by sensors leading to time-series which are inherently unpredictable. For instance, manual controls and/or unmonitored environmental conditions or load may lead to inherently unpredictable time-series. Detecting anomalies in such scenarios becomes challenging using standard approaches based on mathematical models that rely on stationarity, or prediction models that utilize prediction errors to detect anomalies. We propose a Long Short Term Memory Networks based Encoder-Decoder scheme for Anomaly Detection (EncDec-AD) that learns to reconstruct 'normal' time-series behavior, and thereafter uses reconstruction error to detect anomalies. We experiment with three publicly available quasi predictable time-series datasets: power demand, space shuttle, and ECG, and two real-world engine datasets with both predictive and unpredictable behavior. We show that EncDec-AD is robust and can detect anomalies from predictable, unpredictable, periodic, aperiodic, and quasi-periodic time-series. Further, we show that EncDec-AD is able to detect anomalies from short time-series (length as small as 30) as well as long time-series (length as large as 500).

Figures (3)

• Figure 1: Readings for a manual control sensor.
• Figure 2: LSTM Encoder-Decoder inference steps for input $\{\mathbf{x}^{(1)},\mathbf{x}^{(2)}, \mathbf{x}^{(3)}\}$ to predict $\{\mathbf{x'}^{(1)},\mathbf{x'}^{(2)}, \mathbf{x'}^{(3)}\}$
• Figure 3: Sample original normal (first column) and anomalous (second column) sequences (first row, blue color) with corresponding reconstructed sequences (second row, green color) and anomaly scores (third row, red color). The red regions in the original time-series for anomalous sequences correspond to the exact location of the anomaly in the sequence (whenever available). Plots in same row have same y-axis scale. The anomaly scores are on log-scale.