Research on noise prediction of urban road traffic based on parameter learning

摘要

Noise prediction is an important part for urban road traffic noise study, at present, there is no predictive model widely recognized in our country, due to the larger differences in urban road traffic environment, it easily leads to low accuracy when foreign predictive model is used directly. So many factors affecting the urban road traffic noise are analyzed, relevant foreign predictive models are studied, noise predictive model of urban road traffic based on parameter learning is proposed, the process of predictive model using gradient descend method to proceed parameter learning based on that the initial value of empirical parameters are given is described, the optimal parameters are determined and used in the predictive model based on parameter learning. In the end, the feasibility of the predictive model is verified by the prediction examples of test sample collection and the comparison with the prediction results of other predictive models. Experiment results show that noise predictive model of urban road traffic based on parameter learning can adapt to different traffic conditions in our country and effectively improve the accuracy of noise prediction of urban road traffic.