Predictive Maintenance of Mining Machines Applying Advanced Data Analysis

Abstract

The production of mines depends on how the mining machines work. It is, therefore, essential to maintain them. For an extensive mining transport system, the maintenance process is extremely demanding because it consists of many components. Maintenance techniques in mine sites exist in different forms. Prevention, failure, and predictive groups may identify mining maintenance strategies. Since the machines’ reliability depends on several variables, it is not possible to fix the repair time for each component beforehand. Therefore, predictive maintenance is the most appropriate method. This approach provides continuous information on the state of the analyzed unit, thus monitoring the deterioration process and allowing the most appropriate duration of repairs to be scheduled. In the industry, development in online and standard acquisition systems is currently popular. Predictive maintenance today relies on the use of data fusion to continuously analyze data obtained from various machines in real time. It is necessary to suggest a set of time series indicators for management and maintenance purposes that allow for a full and objective evaluation of the artifacts in terms of technology, economics, and organization, as well as an estimation of the remaining life of the artifacts. This type of analysis is a Big Data solution on an industrial scale. Consequently, the appropriate techniques for data analysis must be applied. The amount of the data collected from machines and machinery at mines increases very rapidly through the production of sensing technologies for service and maintenance processes. By analyzing the data obtained by techniques of advanced methods such as Machine Learning (ML) and Artificial Intelligence (AI), valuable information and expertise can be extracted from mining operations and maintenance. The mining industry requires the continuous operation, efficiency, and productivity of its machinery and machine maintenance. This is why equipment defects must be identified and fixed before any failure in the production processes can be avoided. As a promising tool for Predictive Maintenance (PdM) products, ML and AI methods have been developed to avoid machinery/equipment failure. However, PdM applications efficiency depends on the correct ML and AI method range.