| Fault diagnosis of Heating boiler’s operation process is an important part o f the safe operation of the Heating boiler, because the Heating boiler system has the characteristics of high complex degree,variables and strong coupling,so it is hard to set up the precise mechanism model for fault detection and diagnosis. This paper introduces the basic theory of principal component analysis, and for the operating characteristics of heating boilers, the use of fault diagnosis method based on principal component analysis, historical data in order to establish normal conditions PCA model, realized the historical data dimensionality reduction and decoupling, and performs fault monitoring by calculating T2 and SPE statistics and then complete fault isolation and identification of the contribution according to the conventional view and a contribution to the reconstruction method of figure. By fault simulation experiments show that the method for fault diagnosis of industrial boilers to be effective, it has a certain reference value.In this paper, using powerful MATLAB matrix computing power and easy function call function to achieve fault diagnosis algorithm PCA-based, and through OPC technology and boiler distributed control system(DCS) for real-time data exchange, complete the online fault monitoring and diagnosis. In the industrial field, through the process variable data fault simulation, the validity of on-line monitoring and fault diagnosis system.However, based on the traditional method of heating boilers PCA fault diagnosis, there will be some false positives. Based on the characteristi cs of heating boiler operation, the introduction of exponentially weighted dynamic kernel principal component analysis(EWDKPCA) concept, the establishment of a PCA model for fault diagnosis of nonlinear dynamic real-time fault diagnosis. Through analysis of simulation results proved that this method reduces false positives for troubleshooting heating boiler with good results. |
PDF Full Text Request