| Coal pulverizing system as an important auxiliary system of the thermal power plant, its operating status would will directly affect the safety and economy of boiler operation. The accurate measurement of coal powder amount in the coal pulverizing system is benefit of saving energy and economical operation. The first step of soft sensing technology is to select some process parameters that can be accurately measured as the secondary variables, and then use the mathematical model based on the secondary variables and the primary variable to get the online prediction of primary variable which could not be measured by using the traditional sensors. In this paper, the soft sensing method based on support vector machine is proposed for the measurement of coal powder amount.In this paper studies the soft sensing technique of coal powder amount in the coal pulverizing system, including the selection of secondary variables, data preprocessing, soft sensor modeling and model calibration. Firstly, the corresponding secondary variables are selected through mechanism analysis, and the rationality of the secondary variables selection is verified by correlation analysis; secondly, using Pauta criterion and sliding average filtering method to eliminate the gross error and random error in the data. In order to reduce the correlation among the secondary variables and solve the problem of high dimension of the secondary variables, the data with the error processed is processed by the principal component analysis;thirdly, using support vector machine theory to establish the soft measurement model.The selection of the kernel function and the initial parameters are analyzed and studied; finally, the model of soft measurement is corrected.It is proved by the actual operation data of the power plant, and the soft sensor model can obtain satisfactory results, which can be used to provide a strong theoretical basis for the future practical application. |
PDF Full Text Request