神经网络方法在车内噪声信号预测中的应用

针对以往车内噪声有源控制过程中存在的次级声源在初级声场采样传声器上的声反馈问题，提出了车内噪声信号识别和预测的神经网络方法。对被试面包车在稳态和非稳态两种工况下的试验研究表明，利用车身悬置点和发动机的振动信号，通过BP神经网络来识别和预测车内驾驶员耳旁噪声是可行的，可以解决次级声源在初级声场采样传声器上的声反馈问题。

Application of NN Method to Signal Forecast of Vehicle Interior Noise

A common problem on the active noise control in vehicle is the sound feedback from the secondary sound source to sampling microphone of the primary sound field. In order to solve this problem, a neural network(NN) method used for identifying and forecasting the noise signal in vehicle is proposed. A comparison between predicted and measured results in the minibus under the stable and unstable state indicates that it is feasible to identify and forecast the noises at the driver's ear position by the BP NN according to the signals of the suspension vibrations and the engine vibrations. Sound feedback problem of secondary sound source on sampling microphone of primary sound field can be solved by this method.