Review of smoothing methods for enhancement of noisy data from heavy-duty LHD mining machines

Jacek Wodecki,Paweł Stefaniak,Anna Michalak,T.A Przylibski,D Kasza

doi:10.1051/e3sconf/20182900011

Jacek Wodecki, Paweł Stefaniak + Show 3 more

Open Access

https://doi.org/10.1051/e3sconf/20182900011

Copy DOI

Abstract

Appropriate analysis of data measured on heavy-duty mining machines is essential for processes monitoring, management and optimization. Some particular classes of machines, for example LHD (load-haul-dump) machines, hauling trucks, drilling/bolting machines etc. are characterized with cyclicity of operations. In those cases, identification of cycles and their segments or in other words – simply data segmentation is a key to evaluate their performance, which may be very useful from the management point of view, for example leading to introducing optimization to the process. However, in many cases such raw signals are contaminated with various artifacts, and in general are expected to be very noisy, which makes the segmentation task very difficult or even impossible. To deal with that problem, there is a need for efficient smoothing methods that will allow to retain informative trends in the signals while disregarding noises and other undesired non-deterministic components. In this paper authors present a review of various approaches to diagnostic data smoothing. Described methods can be used in a fast and efficient way, effectively cleaning the signals while preserving informative deterministic behaviour, that is a crucial to precise segmentation and other approaches to industrial data analysis.

Highlights

Load-Haul-Dump machines (LHD) are common assets used for ore transportation in production area of a mine with room-and-pillar system
Signals measured in underground mine reality are susceptible to interference, which very often reduces the level of their informativeness, and this in turn might lead to wrong interpretation of the analysis results
The reasons for this are partly to be found in harsh mining conditions where machines operate in a cyclic way and their work consists of many sub-processes

Summary

Introduction

Load-Haul-Dump machines (LHD) are common assets used for ore transportation in production area of a mine with room-and-pillar system Their character of the work seems very simple – machine loads blasted ore in mining face using bucket, transports it to dumping point where dumps material onto conveyor belt, and returns to the mining face with empty bucket. It is even possible to notice the underlying tasks of loading and unloading Such signal is highly contaminated with noise as well as fluctuations coming from the fact that the bucket, either full or empty, is prone to wobbling when the machine is moving, very often while driving on uneven surfaces. In this paper authors compare the performance of six basic smoothing methods for real-life signal denoising and presents the segmentation based of them

Methods

Results

Conclusion