Fault Detection Characterization, Design, and Reliability Analysis

• Author / Creator

  Yang, Shuonan

• This thesis develops fault detection characterization, design, and reliability analysis within a real-time, multilayered fault detection and diagnosis (FDD) framework proposed according to fault tolerant control systems (FTCS). With the development of sophisticated control and monitoring systems, it has become more challenging to carry out routine maintenance, tuning, troubleshooting, and thus keep the systems operate in their desired or optimal states. Nevertheless, the system is required to present competitive performance and timely response to abnormal conditions. Compared with pure data-driven FD techniques, integrated schemes with model-based FD approaches utilize the available inherent dynamic relationship among various signals hence can render more precise diagnosis results. Fitting in the FTCS scope, the real-time integrated FDD framework is thus highlighted as a strategic solution platform, where various specific FD approaches may be conceived and realized.

  Within the framework, research has been carried out in various aspects, including but not limited to fault detection (FD) design/characterization, real-time frequency estimation, and reliability of fault detection. Firstly, as the major part of the thesis, FD design/characterization takes up more than two chapters where analytical forms characterizing the (first) detection/hitting time (FHT) are developed based on the general-likelihood ratio (GLR) detection method. Both probability expressions and single-valued performance indices are proposed for both additive and multiplicative faults. Secondly, as a substantial technical solution promoted by FD techniques, online frequency estimation has been researched using the gradient estimator approach with leakage. In the last research topic, semi-Markov kernel modeling and real-time reliability are discussed with respect to long term fault and detector sequences. The characterization and design plans have been tested on practical data sequences and industrial system models, demonstrating the feasibility of implementation.

• Subjects / Keywords

  • Fault detection

• Graduation date

  Spring 2013

• Type of Item

  Thesis

• Degree

  Doctor of Philosophy

• DOI

  https://doi.org/10.7939/R3FK95

• 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.