Anomaly Detection for Water Treatment System based on Neural Network with Automatic Architecture Optimization

TL;DR

This paper presents a neural network-based anomaly detection method for water treatment systems, utilizing genetic algorithms for architecture optimization and various techniques to enhance detection accuracy and interpretability.

Contribution

It introduces a genetic algorithm approach for automatic neural network architecture optimization tailored for water treatment anomaly detection.

Findings

Optimized neural network architectures improve anomaly detection accuracy.

Proposed techniques enhance detection quality and interpretability.

The approach effectively identifies anomalies in the SWaT dataset.

Abstract

We continue to develop our neural network (NN) based forecasting approach to anomaly detection (AD) using the Secure Water Treatment (SWaT) industrial control system (ICS) testbed dataset. We propose genetic algorithms (GA) to find the best NN architecture for a given dataset, using the NAB metric to assess the quality of different architectures. The drawbacks of the F1-metric are analyzed. Several techniques are proposed to improve the quality of AD: exponentially weighted smoothing, mean p-powered error measure, individual error weight for each variable, disjoint prediction windows. Based on the techniques used, an approach to anomaly interpretation is introduced.