Study On The Frequency Characteristics And Prediction Models Of Highway Traffic Noise

As the fast development of domestic automobile manufacturing industry and the rapid growth in automobile amount, the levels and frequency characteristics of the highway traffic noise changed simultaneously. Since the 1980 s, traffic noise prediction models have been developed by many countries. However, most of them use the average energy radiation level as the prediction of the basic vehicle noise; the energy distribution characteristics at different frequencies were usually ignored. Meanwhile, the source of the vehicle noise is assumed to be mono-pole source. The physical structure of the vehicles is ignored, which might also affect the prediction of the vehicles with long length or running in low speeds. Considering the problems existing in traffic noise prediction, the energy distribution of the source noise at different frequencies and their attenuation characteristics because of the construction of sound shields and green belts were investigated in this study. Based on above experimental results, a frequency related source intensity prediction model and a modified Leq(20 s) prediction model of the traffic acoustic noise were developed, and the precision of the FHWA model with average energy radiation level as their prediction target was assessed. The major results obtained in the paper are listed as following:(1) Spectral characteristics of the acoustic noise caused by different types of vehicles were clarified. It was determined that for the vehicles with small and middle size, the acoustic noises emitted from them were mainly with middle or high frequencies. For the vehicles with large size, however, the acoustic noises emitted from them mainly showed middle or low frequency characteristics. Accordingly, the effects of traffic noise on the sensitive points could not be accurately attained using 500 Hz as the equivalent frequency in predicton. For the sound shield design of the roads mainly with small vehicles and the green belt construction of the roads with a mixed large, middle, and small vehcles, the application of different equivalent frequency are necessary. Results on the acoustic noise adsorption caused by the roads with green belts showed that in the range of middle and low frequencies, the amount of the attenuated energy increased gradually with the increase of the octave center frequency. The maximum attenuation was obtained at approximately 500 Hz. In high frequencies, the maximum attenuation was obtained at around 4000 Hz, which might be mainly caused by the barrier effects of the branches.(2) A frequency related source intensity prediction model of the traffic acoustic noise was developed. For a specific vehicle type and running speed, the emitted acoustic noised at different frequencies could be predicted effectively. By comparing the determined data with the predicted value, the precision and applicability of the frequency related source intensity prediction model was verified.(3) The quantitative relationship between Leq(20 s) and 0 E()iL was developed based on the FHWA model. The errors and restrictions of the present Leq(20 s) model were analyzed, and modified by adding an infinite road correction term and increasing the constant term from-22.55 to-22. This modification promotes the application and improves the simulation precision of Leq(20 s).(4) The precision of the average energy radiation level and the rationality of the mono-pole source assumption of the FHWA model were assessed by comparing the determined data with calculated values of Leq(20 s),Leq(10 s),and Leq(5 s) from the average energy level of the traffic noise. It was determined that for small vehicles, both the present prediction models used in China and the FHWA model assumed the traffic noise as the monopole source were acceptable. For the medium-size and large vehicles, however, the variation of the average energy level of the traffic noise with time showed mixed source type characteristics at both the positions of 7.5 m and 15 m from the center of the road. Meanwhile, as the increase of the average energy levels of the medium-size and large vehicles, the errors caused by the assumption of the traffic noise as the monopole source decreased gradually.By substituting the average energy level term of 0 E()iL in the FHWA model with Leq(20 s) as the basic noise, the errors caused by the assumption of the source type could be eliminate effectively, especially for the middle and large vehicles with low speed. Meanwhile, the mixed mono-pole and di-pole characteristics of the noise source was also verified applying the relationship between the 20 s sound pressure level and the time that the vehicles from a certain distance passing by the observing point. Finally, the model parameters were calibrated using a non-linear fitting method.