Prediction Of Reduction Products At Exit Of 40MW Grade Controlled High Temperature Preheating Decompression Unit

In China’s power generation system,thermal power almost always occupies half of the country,while the huge environmental pollution caused by thermal power generation is increasing day by day.In the face of the increasingly strict requirements of green development,the traditional combustion organization is optimized by adding a controllable high-temperature preheating device outside the boiler to achieve ultra-low nitrogen combustion.The WIPM-LSSVM model constructed in this paper can effectively predict the reduction products CH₄,H₂ and CO at the outlet of the controllable high temperature preheating decompression device when the coal type is changed and the working condition is changed.In high temperature solution devices,realize the control of coal powder fluidization since maintain preheated combustion,realize the directional transformation and reduction of fuel nitrogen product precision control,realize the H₂,hydrocarbons such as directional transformation,reduction of active pyrolysis components to achieve compound containing different active reduction component of pyrolysis gas,so as to get the best product reducing gas composition has important significance.The accurate control technology of the separator is beneficial to the preheating solution system to realize the effective control of the proportion of reduction products in different combustion areas of the boiler.The NOx concentration at the boiler outlet can be controlled stably below 50mg/Nm~3.This paper takes the experimental device of the 40MW controllable high temperature preheating solution test platform as the research object,and uses the algorithm to predict the experimental data of the reduction products at the outlet of the controllable high temperature preheating solution device,so as to provide the basis and guarantee for the subsequent realization of efficient reduction of NOx.First of all,referring to relevant literature,combined with the basic idea of algorithm modeling,determine the idea of this paper from data denoising to data outlier processing,and then to extract important variables of data and reduce weight,and finally to use the algorithm for prediction modeling.And according to the characteristics of the experimental data to determine the algorithm model used in each stage.Secondly,in order to remove the adverse effect that bad data points may have on later modeling,we try to use wavelet threshold denoising in Matlab,and study the influence of different wavelet,different decomposition layers and different wavelet order on denoising.The scheme of removing bad data by using sym6 wavelet denoising and calculating the score of each data point by isolated mori algorithm is determined,which lays a foundation for the realization of effective prediction of restored products.Then,in order to reduce the dimension of the input data,the partial least square method is used to calculate the significance index of the input variables and the cumulative contribution to the input data.In order to guarantee the prediction accuracy of the model,we try to improve the traditional gray wolf algorithm with multiple strategies.In order to improve the generalization ability and learning ability of the algorithm,the penalty parameter C and the kernel parameterσof the least square support vector machine are optimized by the improved gray wolf algorithm.Finally,combined with the optimization of the above parameters,the least squares support vector machine algorithm is optimized by using the multi-strategy improved gray Wolf algorithm,and its characteristics are compared with other prediction algorithms by predicting the H₂ content of Shenhua coal.Subsequently,the WIPM-LSSVM model was used to predict the other two reduction products CH₄ and CO from the pyrolysis of Shenhua coal.On this basis,in order to explore the predictive ability of the WIPM-LSSVM model for the reduction products produced by different coal types,the reduction products such as CH₄,H₂ and CO from the pyrolysis of Shenhua mixed coal were also predicted.The prediction of reduction products under various working conditions by WIPM-LSSVM model is realized,which provides support for the device to achieve NOx reduction through the precise separation technology of the separator.