Online Sensor Fault Diagnosis And The Application In Industry Process

With the development of industrial processes, modern control systems continue to expand the scale and rapidly grow the complexity, which leading to increasing maintenance cost, tremendous shutdown losses or serious accidents. Therefore fault detection and diagnosis technology are attracting more attention. Sensor and actuator fault has become the main cause of control system failure. Quickly and accurately detecting sensor and actuator fault is the key to guarantee control system operating stability and safety. This thesis reviewed the previous methods of fault detection and diagnosis, presented a rapid sensor fault detection algorithm and applied to an industrial process. The main contributions of this thesis are as follows:1. Systematically reviewed the development of fault diagnosis methods and application. This chapter discussed fault diagnosis systems' structure, fault diagnosis category, as well as the various methods' features and their applications in process industry.2. Studied industrial sensor signal and found that the variance of process data could be a good characterization of the sensor performance. This chapter presented a rapid sensor fault detection algorithm. The test results on actual industrial data showed that this algorithm could rapidly detect early fault with clear physical meaning.3. Developed an online sensor fault detection system based on the above rapid fault detection algorithm. The system consists of fault database, process data sampling, fault diagnosis module, man-machine interface, etc.4. Based on the demand analysis of a large oil refinery, applied the online fault detection system for a fluidized catalytic cracking unit (FCCU). Industrial application results showed its availability to improve operational safety and it could be applied to large-scale industrial processes.