| As thermal power station is developing in the direction of higher capacity and parameters, the dynamic and static clearance is getting smaller and smaller, resulting in a greater possibility of occurrence of rub-impact fault. Meanwhile, the inducement of rub-impact fault is complicated, and different kinds of rub-impact faults may have different kinds of effect, which make rub-impact fault have been the key and difficult points of research. This paper takes some 1000MW unit as the study object, and uses both FMEA and FTA method to finish rub-impact fault analysis. Specified symptoms of rub-impact faults were obtained and a thorough rub-impact fault knowledge base was established including fault modes, fault causes, fault symptoms, fault effect and treatment measures.The aim of this paper is to establish an intelligent fault diagnosis scheme. Starting with fault analysis and big data analysis, the scheme combines both fault knowledge and data analysis result to achieve automatic diagnosis of rub-impact faults. Firstly, multivariate state estimate technique was adopted to build steam turbine generator set abnormal detection model. By reconstructing operating data and testing residuals, abnormal information can be detected, so that the subsequent fault diagnosis procedure can be triggered. Then, residual threshold analysis, vibration frequencies statistics, trend test, correlation test and invariant moment theory are adopted to quantify symptoms, realizing automatic symptom acquisition. At last, evidence theory is taken to fuse symptoms so as to identify the fault mode. Combining the fault mode and the fault knowledge base, the further fault information can be searched out and accomplish the whole fault diagnosis task. By applying this scheme to the analysis of fault of some unit, a good result is gained, proving the effectiveness of the scheme. Finally, this paper designed the frame and function of this intelligent fault diagnosis system using modularization idea. |
PDF Full Text Request