Development of a Data Analysis Process in the Context of Condition-based Maintenance

Unexpected machine breakdowns are one of the primary concerns in production companies today. Due to the drawbacks of reactive and preventive maintenance strategies with regard to machine availability and costs, current research focuses on the application of condition-based maintenance. Condition-based maintenance is characterized by the monitoring and analysis of machine conditions through Intelligent Maintenance Systems (IMS), which facilitate the surveillance of the machine status based on sensors installed at the machine. These sensor data provide information to identify and forecast device failures, resulting in more precise maintenance and spare parts requirements and thus higher machine reliability.

Condition-based maintenance is characterized by two main tasks, namely fault diagnostics and prognostics. The former refers to the analysis of real-time condition data in order to detect failures as soon as they become visible in the condition data, while the latter estimates the remaining useful life of a machine until breakdown occurs. The main goal of both tasks is the analysis of real-time condition data. In order to gain useful knowledge from available data, the usage of existing structural approaches for data analysis is encouraged. Those structural approaches present different steps that are necessary to perform data analysis including among others data preprocessing, algorithm selection and performance evaluation. By this, these approaches provide guidelines in extracting expedient information from data.

The objective of this thesis is to investigate existing structural approaches for the analysis of data and to assess them in the context of condition-based maintenance. In a first step, the thesis should cover a literature review of structural approaches for data analysis in general as well as approaches already applied to condition-based maintenance. This is followed by the identification of the main data analysis steps necessary to perform condition-based maintenance as well as requirements for a structural approach in condition-based maintenance. Based on these results, identified structural approaches for data analysis should be qualitatively assessed with regard to their suitability in the context of condition-based maintenance.

This thesis can be written in either English or German. Since scientific literature and domain-related terms are primary available in English, it is recommended to write the thesis in English.

Recommended reading:

  • Jardine, Andrew KS, Daming Lin, and Dragan Banjevic. "A review on machinery diagnostics and prognostics implementing condition-based maintenance." Mechanical systems and signal processing 20.7 (2006): 1483-1510.
  • Maimon, Oded and Rockach, Lior. "Introduction to Knowledge Discovery in Databases." Maimon and Rockach (eds.) Data Mining and Knowledge Discovery Handbook (2005). New York: Springer, pp. 1-15.
  • Wagner, Saalmann, and Hellingrath. "Machine Condition Monitoring and Fault Diagnostics with Imbalanced Data Sets based on the KDD Process." (2016)