Research On Combustion Optimization Of Circulating Fluidized Bed Boiler Based On Improved Optimal Foraging Algorithm

Since the beginning of the Industrial Revolution,coal has been widely used as fuel,which has caused serious environmental pollution,and it is also one of the main causes of global warming.With the rapid economic development,the demand for electrical energy is also increasing,and people’s increased environmental awareness requires the power plant to increase the amount of power generation while reducing pollutant emissions as much as possible.In response to this problem,this article uses artificial neural network combined with group intelligent optimization algorithm to optimize the combustion process of the circulating fluidized bed boiler,so as to achieve the purpose of reducing the boiler NOx emissions and improving the boiler’s thermal efficiency.Aiming at the difficulty of modeling circulating fluidized bed boilers,this paper uses Window-Two-Hide Extreme Learning Machine(WTELM)to model the combustion process of the boiler.The WTELM network is an improvement based on the Two-Hide Extreme Learning Machine(TELM).The WTELM model adds a window mechanism before the input layer of the TELM network,uses a weight sharing strategy to reduce the number of random weights in the input layer,and introduces the idea of ??mean value in the second hidden layer to reduce errors and improve the stability of the network.Combine the WTELM model with Extreme Learning Machine(ELM),Fast Learning Network(FLN),Fast Parallel Learning Network with Parallel Layer Perceptron(PLP-FLN),and Parallel Extreme Learning Machine(PELM)and TELM are used for comparative simulation experiments.Experimental results show that WTELM has good prediction accuracy and stability on the benchmark test set.The model of NOx emission and boiler thermal efficiency of circulating fluidized bed boiler was established by WTELM network,and compared with the models established by ELM,FLN,PLP-FLN,PELM and TELM.The simulation experiment results show that WTELM established CFBB combustion characteristic model has higher accuracy and stronger generalization ability.In order to improve the Optimal Foraging Algorithm(OFA),it is easy to fall into the local best.Introducing adaptive inertia weights and global optimal solutions into theoptimal foraging algorithm to improve the heuristic search method of the algorithm,while combining phase space search with real number space search,an improved optimal foraging algorithm Phase-Space Optimal Foraging algorithm(POFA).The POFA algorithm,Artificial Bee Colony(ABC),Gbest-Guided Artificial Bee Colony(GABC),Particle Swarm Optimization(PSO),Grey Wolf Optimization(GWO)and OFA algorithm for comparison.The simulation experiment results show that,compared with OFA algorithm,POFA algorithm has faster convergence speed and higher optimization accuracy.Based on the combustion characteristics model of the circulating fluidized bed boiler based on the WTELM network,the POFA algorithm is used to optimize the adjustable parameters of the boiler under certain constraints,so as to reduce the pollutant emissions and improve the boiler as much as possible The goal of thermal efficiency.