Research on abnormal detection of gas load based on LSTM-WGAN

Abstract

Traditional anomaly detection methods cannot take into account the influence of external factors. Therefore, it is difficult to obtain an accurate detection effect when performing gas anomaly detection. In this paper, we propose an anomaly detection model based on LSTM-WGAN. The multi-layer LSTM network captures temporal dependencies and introduces a self-attention layer to integrate embedded weather, holidays and other influencing factor data. Embed different features into the WGAN framework and make the detection results more accurate by fused models. This paper conducts experiments on the real gas load data set. The experimental results show that the model in this paper has obvious performance advantages in precision, recall and F1 value compared with traditional time series anomaly detection methods.

A Data-Distillation-Enhanced Autoencoder for Detecting Anomalous Gas Consumption

“…Based on the anomaly types identified in the previous stage, the second stage trains a Bayesian maximum likelihood classifier to identify the anomalies. Recently, researchers investigated data-driven methods for gas theft detection [2], [3], [6], [7], which are closely related to this study. In [2], Yang et al proposed a method based on normal user modeling and RankNet for detecting gas theft suspects among restaurant users.…”

“…Similarly, this method relies on labelled abnormal data for model training, and is limited to the detection of gas-theft suspects among boiler room users. In [7], Xu et al proposed an anomaly detection model based on LSTM-WGAN for gas load. Although this method can be used for different types of gas users, it requires that the training data does not contain anomalies.…”