Research On In-vehicle Active Noise Control Technology Based On BP Neural Network

With the development of high-speed digital signal processing technology,the active noise reduction technology in the car has been realized from theory,and the Active Noise Control(ANC)system has been gradually applied to automobiles.At present,most mainstream in-vehicle ANC systems use adaptive algorithms to suppress engine noise.However,when the speed of vehicle is higher than 70km/h,the tire noise and wind noise begin to dominate in the interior noise,which leads to the deterioration of the noise suppression capability of the ANC system in the vehicle when the vehicle is driving at a high speed.In addition,consumers generally report that when there is sudden noise(whistling,sudden braking,etc.)in the surrounding environment,there will be a short-term noise increase in the car.To this end,this thesis mainly studies and improves the control algorithm in the vehicle ANC system for the above two problems,aiming at ensuring good noise suppression effect and system stability under different vehicle speeds.Firstly,the audio data acquisition system is built to collect data on the engine noise signal,wind noise signal,tire noise signal and noise signal of the desired noise reduction position of a brand car in different working conditions,and analyze the characteristics of various noise signals.According to the BP neural network model,the engine noise signal,wind noise signal and tire noise signal are used as the input of the neural network.The noise signal at the noise reduction position is expected to be used as the output.The BP neural network model proposed in this thesis is established to establish the nonlinearity of input and output.The model describes the comprehensive influence of various types of noise on the noise at the desired noise reduction position under different working conditions,and ensures the accuracy of the input signal of the active noise reduction adaptive algorithm.Then,the part of the adaptive control algorithm in the ANC system is improved: the traditional normalized least mean square algorithm(LMS)is used as the basic control algorithm.On this basis,the Sigmoid function is introduced to control the step factor of the normalized LMS algorithm,and the sigmoid function is fully utilized for the insensitivity of the larger input to improve the whole adaptive algorithm.Anti-interference ability to sudden noise in the environment.Then,the hardware experiment platform based on DSP chip is designed and built,and the improved algorithm is experimentally verified.Finally,the BP neural network model is combined with the improved adaptive algorithm.The engine noise signal,wind noise signal and tire noise signal are the input of the BP network.The output of the BP network is used as the reference input signal of the adaptive control algorithm.The comprehensive simulation and experimental results show that the anti-interference ability of the improved active noise reduction algorithm is improved.Moreover,in the case of high-speed(70km/h)driving,the noise reduction effect of the improved ANC algorithm is improved by 5dB compared with the current mainstream ANC algorithm.