Fault Detection And Diagnosis For Thermal Boiler Based On Principal Component Analysis

The whole control level has been improved for the commonly use of DCS (Distribution Control System) and EIS (Enterprise information System) in thermal boiler process recently. At the same time lots of spot monitoring data was gathered by the monitoring system. To discover the tiny fault information of system in advance for fault detection and diagnosis has been one of the urgent problems to be solved. It is difficult to found precise math model in many industry processes, the fault detection method base on math model has much more difficulty in actual application. The method of fault detection and diagnosis based on MSPC (Multivariate Statistics Process Control) doesn't depend on the math model of system, this method has much more applied function and it has been a study hotspot.In this paper, the development history of SPC (statistics process control) theory was summarized firstly, then lots of current research results of scholars in this realm was introduced ,at last some existence problem needed more thorough research was introduced. The method of fault detection based on PCA making full use of PCA algorithm well and it has the function of declining the dimension while handling line related data, It can make the monitor process carrying out from multivariate space into the low dimension. Rule of the PC's number influence to the SPE and T2 index and their control limit was explored, This offered a new principle to choose appropriate PC number. Fault detection and diagnosis sensitive capacity of PCA model was analyzed and calculated in detail.Aim at the limitation of fault diagnosis qualitative way based on the SPE and T2 index's contribution diagram, this method was improved to diagnose fault in quantitative way.According to the condition of General PCA method can't apply directly on frequently changing working conditions of the thermal boiler, A new detection method based on dynamic multi-principal comment models is being proposed for this purpose. Process data for various steady-states are obtained by data classification based on the shortest distance method. To detect and diagnosis the process fault with multi-model of PCA and to improve the sensitivity. The method of PCA got a better application in the thermal boiler process.Take the actual movement data of the power station boiler as foundation in the research process, to thesis the introductive contents method made use of a calculator to write procedure to carry on a great deal of contrast analysis experiment, and the result show PCA model can more accurate detect the boiler process fault...