MIMII Dataset: Sound Dataset for Malfunctioning Industrial Machine Investigation and Inspection

TL;DR

The paper introduces the MIMII dataset, a comprehensive collection of normal and anomalous industrial machine sounds recorded in real factory environments, aimed at advancing machine fault detection and maintenance automation.

Contribution

It provides the first publicly available dataset of industrial machine sounds under real operating conditions, including various fault types, to support research in acoustic anomaly detection.

Findings

Dataset includes sounds from valves, pumps, fans, and slide rails.

Contains recordings of contamination, leakage, unbalance, and rail damage.

Aims to facilitate development of automated maintenance systems.

Abstract

Factory machinery is prone to failure or breakdown, resulting in significant expenses for companies. Hence, there is a rising interest in machine monitoring using different sensors including microphones. In the scientific community, the emergence of public datasets has led to advancements in acoustic detection and classification of scenes and events, but there are no public datasets that focus on the sound of industrial machines under normal and anomalous operating conditions in real factory environments. In this paper, we present a new dataset of industrial machine sounds that we call a sound dataset for malfunctioning industrial machine investigation and inspection (MIMII dataset). Normal sounds were recorded for different types of industrial machines (i.e., valves, pumps, fans, and slide rails), and to resemble a real-life scenario, various anomalous sounds were recorded (e.g., contamination, leakage, rotating unbalance, and rail damage). The purpose of releasing the MIMII dataset is to assist the machine-learning and signal-processing community with their development of automated facility maintenance. The MIMII dataset is freely available for download at: https://zenodo.org/record/3384388