Research On Noise Active Control System In Cab Based On Self-adapting Neural Network

With the development of automobile industry, enhancement of environmentalprotection consciousness and the improvement of ride comfort, The control ofautomobile noise, especially noise in cab, is increasing drawn the attention of peopleand the relevant departments. The traditional control technology which mainly includessound insulation, noise elimination and noise absorption and so on, have good effect onhigh-frequency, but the method is unable to achieve effective control for low-frequencynoise that is radiated by vehicle vibration of pavement incentive or engine incentive.Relative to passive, Active Noise Control can effectively reduce low-frequency noise incab. On the basis of research method at home and aboard, this paper establishes noiseactive control model and obtains good effect of noise reduction. It is used to provide thereference for noise active control technology.According to the character of the noise with high noise pressure level and mainlylow-frequency in cab, relying on the active control theory, using the method ofcombining finite element analysis with mathematical modeling, the paper respectivelyconstructs neural network active control system and adaptive neural network activecontrol system, and then compares the effect with two systems. The paper can divideinto two sections, the first section is the prediction of noise in cab and another section isnoise active control in cab.The noise prediction is refers to predicting cab’s noise by structure-sound fieldcoupling finite analysis. The implementation steps are as follow: Firstly, structuremodal analysis and sound model analysis, which obtain acoustic oscillation frequency,are respectively simulated for the body structure of ceiling and floor and acoustic cavity,and then provide the basis for structure improvement. The dynamic model of1/2car sixdegrees of freedom was establish and the simulation model was built by Matlab andSimulink. The simulation model was operated and the vibration acceleration of the bodystructure of ceiling and floor and acoustic cavity was obtained. Finally, the vibrationacceleration was introduced into LMS Virtual LAB acoustic modules forstructure-sound field coupling analysis and the frequency response function curve andpower spectral density curve were realized, which means achieving the prediction ofnoise in cab.The cab’s noise active control is refers to achieving effective control for low frequency noise, combining the theory of signal control with the theory of noise controltheory. Its specific content as follows: First of all, the schematic diagram of noise activecontrol system about neural network and self-adapting neural network were establishedon the basis of the theory of neural network and self-adapting filter. Both most suitablefilter order number, convergence factor and the weighs, thresholds about neural networkare obtained by Matlab and the modules of DSP Blockset/Neural Network Blockset.Then, according to the schematic diagram of noise active control system, the model ofnoise active control system about neural network and self-adapting neural network areestablished, operating the model to get the time domain-sound pressure curves,frequency response function curves and power spectral density curves with and withoutcontrol, though the comparison of the curves between the two kinds of active noisecontrol system of noise reduction effect. The results show that the self-adapting neuralnetwork noise active control system has better effect than neural network noise activecontrol system.