| Consumers are often deeply affected by their subjective impression of vehicle interior noise when choosing a car, while A weight sound pressure level, as one evaluating indicator in the traditional research on automotive noise, can't fully show the annoyance grade of a driver or passenger to the vehicle interior noise. So the concept Sound Quality to reflect personal subjective feeling of noise arises, and has been one of the important indexes used to evaluate all kinds of sound. And the research on evaluation, analysis and control of sound quality has come to more and more attention.The key to improve the sound quality of vehicle interior noise lies in reducing the specific noise components to put listeners to annoyance. In this aspect, active noise control (ANC) technique, based on the principle of superposition whereby an unwanted primary noise can be canceled by the additional secondary noise of equal amplitude and opposite phase, has a good application prospect because of the obvious effect on low-frequency noise, the strong especial purpose and the selective noise elimination.The evaluation and control for the sound quality of vehicle interior noise was studied, and improving sound quality of the interior noise in a car with ANC method was attempted in this paper. First, an equation between the subjective evaluation and the objective psycho-acoustical parameters was gained. Then an active sound-quality control (ASQC) adaptive algorithm and the relevant control strategy were brought forward. After the ASQC system of binaural channels had been established, the active control experiments were conducted in a car at last. The main works completed in the paper are as follows:The sound quality of automobile product can be described with psycho-acoustical objective parameters as lots of foreign research indicates, so the principle of choosing proper psycho-acoustical parameters suitable to interior noise characteristic was given, and by this means four parameters loudness, sharpness, roughness and fluctuation were selected for the subsequent analysis. Then a set of software to assess sound quality objectively with the aforementioned four parameters was developed by our group according to ISO532B and other common models. A group of steady noise signals from an automotive cabin were calculated with this software and ArtemiS, one of sound-quality software developed by German company Head Acoustics, and the contrast between the two results of four parameters shows a small difference. It is obvious that our software can be used to analysis and appraise the sound quality of automobile because of enough credibility and accuracy.The 32 steady samples of vehicle interior noise near the ears of assistant driver from 8 types of cars at different gears and velocities in a flat asphalt road were selected as evaluating objects, and the subjective evaluation tests were carried out with the magnitude estimation method by a jury composed of 30 appraisers. After the evaluating data strictly verified, there are 6 persons' results were not accepted due to the significant discrepancy with others, and the left 24 ones' evaluating scores were averaged as annoyance grade of every sample. Then by means of correlation and regression analysis between the subjective appraisal results and four psycho-acoustical objective parameters of 32 sound samples, a sound-quality annoyance model with which the subjective evaluation be quantitatively expressed with objective parameters was established. The model shows that sound quality of the interior noise mainly depends on loudness and sharpness, especially the former, under the stable conditions that vehicles running at various uniform speeds. And it gives a direction for how to improve sound quality.On the basis of the above-mentioned work, theoretical research and simulation analysis on an ANC system, its algorithm and the relevant control strategy for improving sound quality were done. At first, according to the close correlation between dominant noise components and vibrating acceleration signals of engine-body and auto-body suspension, the reference signals of interior noise were distinguished and predicted by Elman dynamic neural network (NN) that the vibrating acceleration signals were used as its input, and after operational factors of the Elman NN had been chose through the simulation compare, the acoustic feedback disturbance of feed-forward ANC system was successfully avoided.The next work was finishing an active sound-quality control algorithm of the vehicle interior noise with the active noise equalization (ANE) and filtered-error least mean square (FELMS) technique as its core. The ASQC algorithm can make up the insufficiency of the traditional ANC system that the noise cancelled degree can not be controlled depending upon different frequency bands and be easily disturbed by random sound components, as a result of two prominent characteristics: First, gain factor, one new parameter, is applied to control the acoustic attenuation, and the noise of different frequency will be active cancelled to different degrees, if they have respective different gain factors. Second, passband disturbance caused by some random noise components uncorrelated to reference signals can be reduced through two-layer adaptive filters. Simulations under various parameters show remarkable performance of this ASQC algorithm.After convergent rate of the ASQC algorithm and computing speed of psycho-acoustical parameters had been considered together, the active sound-quality control strategy of the vehicle interior noise was put forward: The primary noise was divided into some simple harmonic waves with different frequencies and amplitudes after being digitally filtered and frequency-divided, and these simple harmonic signals were active controlled at the respective gain factors corresponding to different critical bands. At the same time, the aforementioned sound-quality model was adopted as an optimizing aim, when ASQC system worked and every gain factors was cyclically changed, the sound-quality annoyance grades of residual noise were continuously calculated and compared, and the best gain vector was determined as the lowest annoyance grade being obtained. Thereout, a simulation model of adaptive active sound-quality control system was developed and built with Matlab/Simulink, two real noise samples were active controlled and their sound quality improved obviously. This result proves that both the ASQC algorithm and the relevant control strategy are of good effect.Based on the previous research, an adaptive active controller with the digital signal processor (DSP) as hardware key was developed, simultaneously the matching software of the system programmed with the standard C and assembly language was developed too. Thus the active sound-quality control of the interior noise can be tested in an actual automobile since both the hardware and software of the ASQC system with the binaural channels have been finished.Finally, active sound-quality control of the vehicle interior noise was carried out in a local space near the two ears of assistant driver by means of the system developed. The experimental results indicate that the ASQC system is of the good stability and satisfactory convergence when the car running at four different velocities in the flat asphalt road. And both the loudness and sharpness, especially the former, of the interior noise evidently are better than before control, thereinto the loudness is reduced by 22%～32% and sharpness by 8%～13%. Consequently the sound quality is improved remarkably and the annoyance grade of the interior noise is dropped by more than 25%, which is verified by both objective compare according to the sound-quality model and subjective evaluation of the same jury.
|
PDF Full Text Request