The Research On Soft Sensing Technology Of Flue Gas Flow In Coal-fired Power Units

At present,energy saving,environmental protection is the problem that each Power Generation Enterprises must face.Flue gas flux is a very important variable in the process of energy saving and environmental control.But the measurement of gas flow is facing a variety of problems,on the one hand,to expand the capacity of the unit so that the traditional hardware sensor measurement errors,on the other hand,affected by the sensor,such as the impact of the current sensor in high temperature,high dust,high corrosion hardware,often sensor failure,maintenance difficulties and so on,these factors restrict automation degree of Power Enterprise operation on the energy saving and environmental protection,seriously restricting the efficiency of Power Enterprises.In this paper,the soft measurement technology of flue gas flow is studied.Firstly,analysis of various factors affecting gas flow rate,such as the impact of the amount of smoke generated and the impact of gas flow,etc.To solve the problem of the relationship between the auxiliary variables and the influence of the auxiliary variables on the dominant variables.Through the analysis of the importance variable projection of the PLS and the forward search algorithm,the variables were selected.Secondly,in the aspect of modeling,according to the unit operation condition,typical process data is selected for static modeling;In the aspect of data preprocessing,using Pauta criterion and normalized processing method for singular point,isolated point,eliminate the influence on the process of modeling data;The improved least square support vector machine algorithm is used for modeling.About the loss of sparsity problem for LSSVM,the similarity function method is used to deal with the data redundancy in this paper.Pruning algorithm is used to solve data redundancy problem in the modeling process in order to increase the generalization ability.In the static modeling process,unable to select the whole conditions of modeling,online correction is essential for soft sensor modeling step.Here,we use the adaptive algorithm of the sliding window recursive algorithm to modify the model parameters and increase the capability of online prediction.In this paper,based on the research of the theoretical method,Collect data and the use of MATLAB programming to complete the modeling of data processing and selection of auxiliary variables,static and dynamic modeling of flue gas flow.The simulation results show that the model prediction of the conditions of the results to achieve the desired effect,provide the basis for optimization of coal-fired units completed energy saving and environmental protection.