Research On Key Active Noise Reduction Technologies For Computer Fans

With the rapid development of internet technology, computers have become an integral part of daily life and work. In the mean time, computer fan noise has also drawn more and more attention. Traditional passive noise reduction method can only eliminate a certain amount of noise within a short period of time. Changes in ambient temperature, aging of the fan and accumulation of dust will gradually weaken its performance. Active Noise Control (ANC), through adaptive algorithm, can respond to fan noise in real time, and has very good noise reduction performance for low-frequency narrow-band noise. ANC technology therefore will have a greater role in computer fan noise control.This article studies how to design an ANC system, establishing its general design process and the key issues that must be resolved in the process. Research has shown the general design process of an ANC system:Firstly, analyze source noise properties, then select the appropriate channel features and control features based on analysis results. Channel features include pipe dimensions, the number of secondary channels and distribution of electro-acoustic equipment. Control features include control structures, control algorithms, selection of algorithm parameters and error channel identification. Finally, choose the appropriate hardware according to channel and control features.This article has mainly studied pipe design, adaptive control algorithm, spatial arrangement of adaptive filters, sensors and speakers and adaptive identification of sound channel. Pipe design is based on sonic catheter theory and optimization of speaker and error sensor locations; sound channel identification uses controllable offline adaptive identification. This article has studied three control algorithms. Considering computer fan noise characteristics and pipe size, FbLMS algorithm is used as the control algorithm of the ANC system. Then, through Matlab simulation, the effect of algorithm parameters on noise reduction effect is determined and following conclusion is drawn:1. the larger the convergence coefficient, the quicker the system converges. However, a convergence coefficient too large will cause system divergence;2. a larger filter length will lead to quicker convergence, but will also increase data throughput;3. algorithm can provide satisfactory reduction only for narrowband noise;4. certain error is allowed for estimating the error channel transfer function. A error too large, however, will lead to system divergence. Finally, in order to verify the active noise reduction theory and control algorithm, a hardware test platform is built. Noise reduction is performed for simulative sinusoidal noise, which has delivered certain effect.