Application of Data Mining Techniques for Fault Diagnosis and Prognosis of High Pressure Fuel Pump Failures in Mining Haul Trucks

  • Author / Creator
    Alla, Hemanth Reddy
  • Mining companies are investing in fewer but larger equipment, and downtime associated with larger equipment now represents a higher percentage of operational capacity loss. Thus, it is essential to frequently and accurately monitor the health of this equipment to avoid unscheduled breakdowns and expensive repairs. Modern mining is facilitated by the use of sensors for real-time monitoring of equipment operating parameters, external environmental conditions, and various key performance indicators. Although this data has existed within some companies for years, it is vastly underutilized in the mining industry. Thus, the problem statement for this research is: “The development of fault diagnostic and fault prognostic models using data from multiple sources and implementation of various data mining techniques to address critical failures in haul trucks”.
    In this research, the primary objective is to develop, implement and validate a robust engineering methodology to identify critical failures, diagnose and predict their remaining useful life in haul trucks using machine learning-based and deep learning-based data-driven approaches. To address a major shortcoming of the previous research works, this research does not use any fabricated data or data generated by simulations in the lab, but instead uses actual data originating from multiple haul trucks and various mines.
    In order to achieve the objectives of this research dissertation, a complete framework for developing data-driven fault diagnostic and prognostic models has been developed. These models were able to diagnose a critical failure in haul trucks at various mines and predict the remaining useful life of haul trucks diagnosed with the critical failure. This research demonstrated the use of several aspects of data-driven models such as data collection, data pre-processing, implementing supervised and unsupervised learning models, hyperparameter tuning, and evaluating model performance.
    The main contribution of this research is the development and implementation of an integrated methodology for diagnosing critical issues in haul trucks and estimating their remaining useful life using several data mining techniques. Based on the results obtained in this research, various data mining techniques can be confidently employed for fault diagnosis and prognosis in haul trucks. In addition, the performance of several data-driven fault diagnostic and fault prognostic models are compared to identify the best-performing model for each task. This provides a better understanding of the applicability of various machine learning-based and deep learning-based models on various types of data and facilitated a more reliable detection of failures and prediction of their remaining useful life.

  • Subjects / Keywords
  • Graduation date
    Fall 2021
  • Type of Item
  • Degree
    Doctor of Philosophy
  • DOI
  • License
    This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.