Development and research on fault diagnosis system of solar power tower plants

According to the system configuration and operating characteristic of a constructing Solar Power Tower (SPT) plant in China in this paper, the Fault Diagnosis System (FDS) was researched and developed. Furthermore, evaluation system of fault grade was established by the method of fuzzy comprehensive evaluation. In this FDS, the fault diagnosis structure was designed to adopt the expert system for priority and the Radial Basis Function (RBF) neural network for assistant. The monitoring index of diagnosis object was built in expert system to set the fault symptom threshold and represent the fault symptom in quantification with the comprehensive methods of expert knowledge, fuzzy mathematics, and low probability identification and so on. The model of neural networks is established based on the structure of RBF multi-neural subnets. According to the analysis and verification results of a fault case, the structure design is reasonable and diagnosis methods are feasible in this FDS. Moreover, the fault could be accurately diagnosed and the evaluation of the fault grade could be made reliably with the great practical value.