Aerodynamic Noise Prediction Of Tube-axial Fan Based On BP Artificial Neural Network

Tube-axial fans are widely used in dustry.As people pay more and more attention to noise problems,the aerodynamic noise problems caused by tube-axial fans gradually becomes important.The traditonal experiment methods and numerical simulation methods both consume a lot of manpower,material resources and computing resources.Under this background,this paper proposed a method of using artificial neural network technology to predict the aerodynamic noise caused by tube-axial fans.By combining neural network and numerical simulation methods,the neural network was trained using the data obtained by numerical simulation.The loss of this method on the test set was only 2.55×10^-3,showing high accuracy.This method can greatly cut down the calculation time and has a good application prospect in industry.This article includes the follwings:In Chapter I,this paper introduced the current mainstream methods for obtaining the aerodynamicn noise field of tube-axial fans,including experimental mersurement methods and numerical simulation methods,as well as some problems with these methods,showing the good application prospects of neural network.This chapter also introduced some preliminary research of neural network in this aspect.In Chapter II,this paper studied the noise generation mechanism of tube-axial fan and the characteristics of the noise source,also deduced and analyzed the aeroacoustic equations which are applicable in different situations.In Chapter III,this paper first introduced the numerical calculation methods,and then used ANSYS Fluent simulation software to simulate the tube-axial fan,obtained sound pressure data of aerodynamic data under different working conditions.Via the cpmparing of sound pressure level data which was acquired by FFT with experimental data,this simulation method was proved to be effective.In Chapter IV,this paper proposed a method of using artificial neural network technology to predict the aerodynamic noise caused by tube-axial fans under the bachground of the complexity of experimentle methods and simulation methods.First,this paper compared the fluid mechanics theory with the aeroacoustic theory,discussing the selection of input and output of neural network.Then established a BP neural network model,extracted the corresponding data set from the simulation data and substituted it into the neural network for training.The loss of the completed neural network on the test set was only 2.55×10^-3,which proved the high accuracy of this method.This method can greatly reduce the amount of calculation and save calculation resources and time.In Chapter V,we summarized the work of this paper,and put forward the future work propests.