Evaluation Research On Sound Quality Of Cooling Fan Based On Binaural Effect

Noise pollution is becoming a serious problem as the industry developing rapidly.Cooling fan is the main noise source because its common usage,and its noise can disturb hearing,sleep,even nervous system of people around cooling fan.Thus,weakening the cooling fan noise is necessary.This paper has done objective and subjective analyses on the cooling fan sound samples collected from the signal stations.The result of analysis is used for establishing a predictive model.The model aims at providing reference for sound design of cooling fan.First,the paper collects the sound samples with HEAD Acoustics and edits them.This paper analyzes Hearing Area,Binaural Effect and Critical Band to see advantage of using HEAD Acoustics.The samples are collected in signal areas.Sound samples are edited with HEAD Monitor to remove the useless parts in all samples.The edited samples are used in the following objective and subjective evaluations.Second,objective evaluation is done for the cooling fan sound samples.This paper take advantage of psychoacoustic parameters and standard parameters to analyze sound samples.As for psychoacoustics,parameters including sharpness,loudness,fluctuation strength,roughness and so on are selected.Also,sound pressure level and A-weighted sound level are cited to carry on objective evaluation as standard parameters.Third,this paper has done subjective evaluation for the sound samples.The subjects are invited to grade sound samples based on irritability degree.The subjective evaluation experiment is carried on with Paired Comparison Method.After calculating marks of each samples,regression analysis is done for screening parameters,and the parameters are used in the evaluation model.Finally,a BP neural network predictive model is established based on the previous objective and subjective evaluation after standard on evaluation of sound quality is established.Loudness,sharpness and A-weighted sound level work as input,while the subjective evaluation result is taken as output.The model is trained repeatedly after neural network structure is settled and the model is tested.Besides,several suggestions are made for improving the cooling fan sound quality.