| As a kind of low-pressure fan,multi-blade centrifugal fan is widely used in household appliances such as range hoods and air conditioners because of its unique advantages such as compact structure,high pressure coefficient and high flow coefficient.However,due to the influence of the pipeline system and other components,the fan often deviates from the design working condition during the actual operation,resulting in the impeller stalling during the small flow operation and the blockage of the large flow running channel,which is difficult to meet the industrial requirements of environmental protection and energy conservation.Therefore,in this paper,a multi-blade centrifugal fan for a lampblack engine is taken as the research object.Based on the performance calculation program of a multi-blade centrifugal fan for a lampblack engine,taking the total pressure and total pressure efficiency of the fan as the optimization objective,and combining with the optimal Latin square test method to design training samples,the approximate model of BP neural network for predicting the performance of the fan is established,and GA-PSO combined algorithm is used to optimize the design of multi-conditions and multi-objectives,which finally achieves the purpose of improving the performance of the fan and expanding the stable operation area.The specific research contents and results are as follows:(1)The three-dimensional internal flow analysis of the fan is carried out by numerical simulation.The flow mechanism and the influence of the structural parameters on the performance of the fan are deeply studied.The geometric parameters of impeller(blade inlet angle,blade outlet angle,blade number,impeller width)which have a great influence on the aerodynamic performance of the fan are obtained.The range of variables to be optimized is determined according to the calculation results and theoretical analysis.(2)In order to shorten the period of structural optimization design,the performance calculation model of multi-blade centrifugal fan is compiled by Visual C++ based on the formula of total pressure and total pressure efficiency of fan.Considering the influence of actual gas viscosity effect,a loss coefficient correction method is adopted to optimize the calculation program with the aid of Pointer algorithm.The prediction error is less than 5%.This provides a theoretical basis for the multi-objective optimization design of multi-blade centrifugal fan under multi-conditions.(3)Using the Latin hypercube design method to construct 120 sets of sample values,with different working conditions point weighting efficiency and total pressure as the optimization goal,BP neural network is used to construct the multi-blade centrifugal fan performance parameter prediction model,in which the full pressure and efficiency prediction output The maximum error relative error is 1.5% and 1.2%.Combined with the GA-PSO algorithm,the original fan is multi-objective optimized,and finally two sets of new impeller structural parameter combinations are obtained.(4)The flow field distribution in the fan before and after optimization under different working conditions was compared and analyzed,and the noise was predicted by LES / FW-H method.The results show that the total noise of the fan in the high frequency band around 2 kHz is effectively controlled.Aerodynamic performance and noise tests were carried out to verify the accuracy of the numerical simulation results.At the same time,the following conclusions were drawn: after optimization,the total pressure and static pressure of the fan were increased by about 32 Pa and 30 Pa respectively,the maximum total pressure efficiency was increased by about 3.2% compared with the original fan,and the weighted sound pressure level of A was reduced by 1.5~3 dB at various flow conditions.In view of the shortcomings of multi-blade centrifugal fan,such as narrow operating range and low efficiency,a new structure optimization method is proposed in this paper.The variables to be optimized are determined by numerical simulation,and the BP neural network combined with GA-PSO algorithm is used for multi-objective and multi-condition optimization design.At the same time,aiming at the practical problems of traditional optimization design methods,the optimized Visual C++ self-programmed fan performance calculation program is adopted in this paper,and the training samples are designed by combining the optimal Latin hypercube test method,which can effectively shorten the optimization design cycle and improve the design efficiency. |
PDF Full Text Request