Audio-based Anomaly Detection in Industrial Machines Using Deep One-Class Support Vector Data Description

TL;DR

This work addresses aerially detecting anomalies in industrial machines using audio signals, proposing a comparison between a baseline dense autoencoder and a one-class deep SVDD approach applied to Log-Mel spectrogram features from the MIMII dataset. The deep SVDD method, especially with a subspace dimension of 2, yields the strongest anomaly detection performance across SNRs (average AUC of 0.84, 0.80, 0.69) while using far fewer trainable parameters than the baseline. Valve-specific preprocessing further enhances performance for the baseline, yet the deep SVDD approach remains superior and robust to dimensionality choices. The findings support the viability of low-cost acoustic sensing for real-time industrial monitoring, with potential extensions to multimodal or graph-based representations to further improve accuracy and efficiency.

Abstract

The frequent breakdowns and malfunctions of industrial equipment have driven increasing interest in utilizing cost-effective and easy-to-deploy sensors, such as microphones, for effective condition monitoring of machinery. Microphones offer a low-cost alternative to widely used condition monitoring sensors with their high bandwidth and capability to detect subtle anomalies that other sensors might have less sensitivity. In this study, we investigate malfunctioning industrial machines to evaluate and compare anomaly detection performance across different machine types and fault conditions. Log-Mel spectrograms of machinery sound are used as input, and the performance is evaluated using the area under the curve (AUC) score for two different methods: baseline dense autoencoder (AE) and one-class deep Support Vector Data Description (deep SVDD) with different subspace dimensions. Our results over the MIMII sound dataset demonstrate that the deep SVDD method with a subspace dimension of 2 provides superior anomaly detection performance, achieving average AUC scores of 0.84, 0.80, and 0.69 for 6 dB, 0 dB, and -6 dB signal-to-noise ratios (SNRs), respectively, compared to 0.82, 0.72, and 0.64 for the baseline model. Moreover, deep SVDD requires 7.4 times fewer trainable parameters than the baseline dense AE, emphasizing its advantage in both effectiveness and computational efficiency.

Figures (2)

• Figure 1: Deep SVDD model is illustrated with its preprocessing steps.
• Figure 2: Examples of initial and preprocessed time-domain waveforms with corresponding log-Mel spectrograms for valve data.