Driving deviation analysis for open sky mining vehicles

an experimental analysis based on pattern recognition algorithms

Abstract

One of the most relevant sectors of the Industry, namely open pit mining, has one of the greatest demands on the performance of processes optimized by technologies for intelligent processes. During the mining process, some risks are inherent to the activities, which can be consistent with a real apprehension about the care that human resources are exposed to. Thus, this research focuses on the vehicle driving process, with a proposal for predictive analysis, supported by pattern recognition algorithms, to predict possible driving deviations for their identification and enable preventive treatments, which represents an investment in the protection of life. of the drivers involved. Studies on the subject can be found, which focus on predictive analyzes of driver or vehicle behavior, but the proposed research contributes to an analysis of the production process. During the experiments, pattern recognition algorithms based on Multilayer Perceptron (MLP), with a variation of types and parameters based on the literature, were used to measure the ability to predict the Fatigue effect of the other effects demonstrated in a mining company in operation. The process had a statistical evaluation of both the provided database and the experiments via MLP, with predictive capabilities that reached an accuracy (based on AUC - Area Under the Curve, which is presented in values ​​between 0 and 1) of 0.9742.