Research On Algorithms Of Blending Thermal Coal In Shangan Power Plant

At present,the diversification of coal sources has led to the unstable quality of coal.Coal blending has become one of the main fuel management methods for power enterprises.However,due to their limitations,power enterprises do not have scientific methods to analyze the element composition,features,and combustion characteristics of blended coal.According to the situation,the paper chooses thermal coal as research object,which is commonly used in Shangan Power Plant.First,the paper analyzes related factors of thermal coal affecting boiler indexes.Then,the objective function and the constraint function of blended coal are established.After that,main factors are used to train the prediction model so as to predict the indexes of blending two kinds of coal.Last,the blending ratio of the two kinds of coal is calculated based on these indexes.The main work of this paper includes the following aspects:Firstly,based on the key constraints of coal blending in Shangan Power Plant,an optimization model of coal blending is established,which takes moisture,volatile matter,ash and calorific value as constraints;and burnout index,ignition index and sulfur content as objective functions.The paper analyzes and studies multiple linear regression and neural network.According to the recognition of the change rule of coal quality after blending,two proportion models of dynamic coal blending is established using multiple linear regression and neural network respectively.After comparing the fitting effect,error analysis and prediction results of the two models,the paper finds the neural network model is more feasible and superior.Secondly,in order to improve the training efficiency of the network and avoid the problem of falling into local minimum,the genetic algorithm and its process are analyzed and studied.The original population is coded,initialized,selected,crossed and mutated.Using the global optimization characteristics of the genetic algorithm,the optimal neural network is obtained with the most suitable initial weights and thresholds.Based on the neural network optimized by genetic algorithm,the paper selected different inputs of neural network,establishing models for coal element composition analysis,combustion ignition and burnout index.Genetic algorithm is used to solve the model,and the validity in solving the coal blending optimization model is verified.Finally,through the analysis of coal blending process in Shangan Power Plant,the paper uses C# language,B/S software design framework to design and realize each functional module of the system.The system is consisted mainly of four functional modules.The coal blending scheme module is designed to facilitate users to formulate different coal blending schemes for different units.The information module is designed for users to select specific constraint indexes according to working conditions of the unit,realizing overall supervision of the coal blending indexes.The blending scheme management module is mainly responsible for the prediction and result analysis of thermal coal blending.The evaluation module evaluates the prediction results of coal blending,collects data according to the actual working condition of the unit,and realizes data visualization.