Feasibility Study of Magnetism-based Indoor Positioning Methods in an Incineration Plant

TL;DR

This study explores magnetism-based indoor positioning in incineration plants, proposing three fingerprinting methods and demonstrating their effectiveness with varying accuracy levels for real-time worker location monitoring.

Contribution

The paper introduces a novel magnetism-based indoor positioning system tailored for incineration plants, addressing cost and infrastructure limitations of traditional radio wave methods.

Findings

Point matching error: 6.89 meters

Path matching error: 0.05 meters

DTW matching error: 0.06 meters

Abstract

In an incineration plant, remote operation from a centralized control room is now possible, but inspection and cleaning of equipment still require a worker to visit the site. When the plant owner reduces the number of workers due to operation costs, it will be standard for a single worker to visit the site. Therefore, it is necessary to monitor the location of workers in real-time to detect unexpected human accidents quickly. Conventional methods use radio waves, such as Wi-Fi and Bluetooth, but there is little demand for communication equipment in the incineration plant. However, there is not enough demand for communication facilities in the incineration plant. It is too large to bear the cost of installing wireless access points, and Bluetooth Low Energy (BLE) beacons just for positioning. Therefore, we are focusing on magnetism using for indoor positioning method. In addition, the incineration plant has a lot of types of equipment that contains a wide range of magnetized metals, large motors, and generators. We could observe the magnetic peculiarity at each point. Based on these assumptions, we have developed a new indoor positioning method at the incineration plant. This paper describes the development of an indoor positioning system for an incineration plant. And we propose three methods for fingerprinting matching: Point matching, Path matching, and DTW matching. The average positioning errors of these methods are 6.89 m, 0.05 m, and 0.06 m, respectively.