Structure Optimization Design Of Axial Flow Fan

In recent years, the gradual deterioration of air quality, haze and other extreme weather events gradually increased, which case a serious threat to the health of the residents. As one of the main sources of haze, Industrial dust has attracted more and more attention of scholars. Effectively control the generation and diffusion of industrial dust, become a hot topic of the relevant scholars. Due to its high efficiency, flexibility, low cost and other advantages, spray has become an important means to control the dust in construction site. Axial fan, which performance directly affects the dust elimination effect is the core component of the spraying device. Therefore, it is very necessary to optimize the structure parameters to improve the performance of axial flow fan.At present, the method of optimizing the structure of axial flow fan, which is based on CFD technology, is mainly focused on the method of one-variable method, orthogonal experimental design, Uniform Design and so on. Because of the coupling effect of the structural parameters on the performance axial flow fans, so one-variable method can’t give the optimum solution. The optimal solution given by orthogonal experimental design or uniform design is the combination of discrete test point of the structural parameters, so these methods have poor approximation ability to the optimal solution. In this paper, BP Neural Network and Genetic Algorithm are used as optimization method to optimize the structure of the axial fans to obtain the optimal solution.First, according to the basic theory of axial flow fan, certainty the hub ratio, tip radial clearance, and collector form reasonably, then fix these structural parameters. After that, the influence of blade number, blade angle and other structural parameters to be optimized on the performance of axial flow fan was analyzed by CFD simulation, and the range of configuration parameters to be optimized were determined.Secondly, construct BP neural network by using C language in the MATLAB environment, and determine the number of hidden layer, hidden layer nodes, transfer function, etc. Construct BP neural network training sample database, through orthogonal experiment table and CFD simulation. Using the database to train the BP Neural Network, and the fitting of the mapping from the structural parameters to the axial flow fan performance is made. Then, test its predictive effect.Finally, according to the principle of genetic algorithm in combination optimization problem, and based on MATLAB, the programming language for the structure parameters and its combination optimization of axial flow fan is complied. Combined with the actual problem, determine the genetic algorithm initialization parameters, selection, crossover, mutation and other genetic operators. Eventually, through genetic algorithm, combined with the BP neural network, the optimization of the structure parameters and its combination of axial flow fan is realized, and the optimal structure is simulated.