Zhang W. (ed.) Fault Detection

InTech, 2010, -224 p.Fault diagnosis technology is a synthetic technology, which relates to several subjects, such as modern control theory, reliability theory, mathematical statistics, fussy set theory, information handling, pattern recognition and artificial intelligence.
The United States is the first study to carry out fault diagnosis countries. Since 1961, the United States at the beginning of the implementation of the Apollo program, has witnessed a series by equipment failure led to the tragedy, therefore, in April 1967, at NASA’s the idea of, by the Office of Naval Research (ONR), opened the Society of Mechanical Failure Prevention Group (MFPG) the establishment of the General Assembly, began to systematically fault diagnosis sub-topic for research. In addition to MFPG, the Society of Mechanical Engineers (ASME), Johns Mitchel company, Spire Corporation are all carried out relevant research.
In Europe, the United Kingdom machine health centers in the late’60s began to study the initial diagnosis. In addition, the Norwegian ship diagnostic techniques, sound detection system in Denmark, Sweden, SPM’s bearing detection technology are all more advanced. Japan’s fault diagnosis technology in the steel petrochemical sectors such as railways developed rapidly, and in the international market certain advantages.
In 1971, Massachusetts Institute of Technology Beard in his Ph.D. thesis was first put forward the concept of fault detection filter, which is to use analytical redundancy instead of hardware redundancy approach and make the system self-organization through the system closed-loop stability, through the observer output to be systems.
Fault diagnosis method based on analytical redundancy is to be diagnosed by an object exists in the analytical redundancy and other priori knowledge analysis and processing, enabling detection of fault diagnosis, isolation, identification. In the same year, Mehra and Peschon published relevant papers in Automatica, which marked the beginning of fault diagnosis. In 1976, Willsky published the first articles on fault detection and diagnosis technology, an overview of the article.
In 1978, Himmelblau published the first book on the international level fault detection and diagnosis (FDD) technology in academic works. Since then a lot of academic institutions, government departments, universities and enterprises are involved in or the fault diagnosis technology research, and made a large number of results.
Fault diagnosis tasks, from low level to high, can be divided into the following aspects:
Failure Modeling: In accordance with a priori information and input-output relations, build a mathematical model of system failure, as a basis for fault diagnosis.
Failure Detection: From the measured or unmeasured variables estimation, to determine whether there was a fault diagnosis system. The main task of Fault Detection system is to determine whether there are failures. In general, any fault detection subsystem can not detect a variety of failures correctly 100 percent. Therefore, improving the correct fault detection rate and reduce the failure rate of omission (which occurs without the fault is detected) and false alarm rate (not failure but police) has been the interesting topic in the area of fault detection and diagnosis.
Fault Isolation: After the fault has been detected, the location of the fault source should be given. Fault isolation is also known as fault identification or fault location.
Fault identification: After a fault has been isolated, determine the time of fault occurred and time-varying characteristics of the fault.
Failure evaluation and decision-making: Determine the severity of fault and its impact on the diagnosis of the object and trends in the different conditions for different measures.
After several decades of development, the formation of fault diagnosis technology generally three types of methods, analytical model-based fault diagnosis method, signal processing based fault diagnosis method and knowledge-based fault diagnosis method.
Dynamic system model-based fault diagnosis method developed the earliest and most in depth. It needs to establish accurate mathematical model of the object. The advantage of this method is that it can fully use of the deep knowledge within the system, which will help the system fault diagnosis.
But in fact a complex engineering system is very difficult to obtain accurate mathematical model, and the system modeling errors and uncertainties disturbance and measurement noise is always inevitable.
When diagnosed analytical mathematical model of the object is difficult to be established, signal processing-based method is very effective. The method uses the signal model directly, such as the correlation function, higher-order statistics, spectrum, autoregressive moving average, wavelet techniques to extract the amplitude, phase, spectral characteristics of value, and analyses these characteristics in order to achieve fault detection. This approach avoided the difficulties of building an object model. Obviously, this method is not only suitable for linear system is also suitable for nonlinear systems. This method of mining the system information contained in the signal and the structure of the system is not concerned.
Knowledge-based fault diagnosis method and signal processing-based fault diagnosis method is similar. It does not require quantitative mathematical model. The difference is that it introduces a lot of information of the diagnosis object, in particular, can take advantage of expert diagnostic knowledge. It is a promising method of fault diagnosis, particularly in the field of nonlinear systems.
In this book, a number of much innovative fault diagnosis algorithms in recently years are introduced. These methods can detect failures of various types of system effectively, and with a relatively high significance.The use of the GLRT for revealing faults in atomic frequency standards
Qualitative Fault Detection and Hazard Analysis Based on Signed Directed Graphs for Large-Scale Complex Systems
Customized Fault Management System for Low Voltage (LV) Distribution Automation System
Faults Detection for Power Systems
Fault Detection and Isolation Scheme Based on Parity Space Method for Discrete Time-Delay System
Model-Based FDI Schemes For Robot Manipulators Using Soft Computing Techniques
Condition Monitoring and Fault Detection of Electric Drives
Fault Detection in Crypto-Devices
Sensor and Actuator/Surface Failure Detection Based on the Spectral Norm of an Innovation Matrix
Fault Detection, Isolation, and Control of Drive by Wire Systems
Wavelet Based Diagnosis and Protection of Electric Motors
Fault detection and diagnosis for in-vehicle networks
Fault Detection of Actuator with Digital Positioner Based on Trend Analysis Method
Rotor Fault Detection in Line-fed Induction Machines Using Complex Wavelet Analysis of Startup Transients
Recent Advances in Localization of Winding Deformation in a Transformer
Sensor fault detection and isolation by robust principal component analysis
Single Line-to-Ground Fault Detection in Face of Cable Proliferation in Compensated Systems
Condition Monitoring of Power Plant Milling Process using Intelligent Optimisation and Model Based Techniques
Robust and Reduced Order H-Infinity Filtering via LMI Approach and its Application to Fault Detection
Mechanical fault detection in induction motor drives through stator current monitoring - Theory and application examples
Fault Detection Method based on Extended Locally Linear Embedding