Study On BP Neural Network Optimization In Power Station Boiler Application Based On Particle Swarm Optimization

Electric energy is one of the most commonly used energy in people’s daily life. With the development of technology and human’s civilization,no matter personal living environment or large factory enterprises can not separable from the electric energy to bring great convenience and benefits.Electric energy production mainly include:thermal power,wind power,hydro power and nuclear reactor power,etc.Power generation enterprises in China,coal-fired power plants account for about 80% of the total output,so the coal-fired power plants is the main source of electric energy in our country. Coal is the mainly fuel used in the power generation,when coal combustion in high temperature,will produces a large number of pollutants,including gas pollutants and solid contaminants,seriously affect the ecological environment.State promulgated a large number of documents for constraints the power plant pollutants discharge. At the same time,because of the power plant boiler’s internal and external factors,can affect the efficiency of boiler operation,so it will increase the power generation cost,and also aggravating the economic burden of the coal-fired power plant.In power generation enterprises increasingly competitive situation,coal-fired power plants want to improve their competitive power must be well solution two problems which are power boiler efficiency and pollutant emission when in the power generation process.For the high cost and high pollutant discharge of power generation,this design using BP neural network to established the boiler model,then using particle swarm optimization to optimize the BP neural network’s method to study the operation of the coal-fired power plant boiler efficiency and low pollution’s characteristics.First of all,detailed analysis the coal-fired power burning coal which produced the gas pollutants NOx’s formation and failure mechanism,as well as the factors that affect NOx generation,this analysis well provide effective basis for the parameters adjustment when in the operation process of the power plant boiler.Then,this paper respectively introduced the principle and process of BP neural network and particle swarm optimization,improved the lack of BP neural network and brief analysis of the reason that why selection the particle swarm optimization to optimize the BP neural network.The specific process of particle swarm optimization optimize the BP neural network is:input boiler operation parameters to the BP neural network for training and learning,which can let neural network better predict the boiler efficiency and NOx emission under different working conditions.Then,using particle swarm optimization to optimize the BP neural network’s weights and threshold that can further improved neural network’s generalization ability on the basis of the original.According to the above methods,using the data of Luohe power plant boiler operation which applied to this design and preprocessing those data,following established three different neural network model are respectively:BP neural network,GA-BP neural network,PSO-BP neural network.Using simulation software input those after the processing’s data to the three kinds of network model to obtain the predict output of network model.Analysis the error relationship between the network predict output and actual output by using graph and table.Achieve the conclusion is:particle swarm optimization optimize the BP neural network can better improve the level of network prediction,make the predict output of network more close to the actual output which can help staff obtain more favorable parameters guidance,at the same time provided a new thinking way to achieve high efficiency and low pollution of coal-fired power plant boiler operation characteristics.