Research On Intelligent Evaluation System For Sound Quality Of Nonstationary Vehicle Interior Noise

With the rapid development of social economy and the continuous improvement of people’s living standards, to ascertain the characteristics of non-stationary vehicle noise and establish a vehicle noise evaluation system, and to make a quick and accurate evaluation of the noise in the vehicle, has important theoretical and practical significance. A sound quality evaluation system aiming at both the stationary and non-stationary vehicle interior noises is developed in this thesis.Firstly, on the basis of previous studies of parameters of objective psychoacoustics, some objective evaluation parameters, such as loudness, sharpness, roughness of the stationary and non-stationary sample data are calculated and used as system inputs. A jury test with semantic differential method is conducted to the measured noise samples and the results are used as outputs. A sound quality evaluation model based on generalized regression neural(GRNN) aiming at both the stationary and non-stationary vehicle interior noises is developed. The unused noise samples are imported into the GRNN quality evaluation algorithm model to verify its validity.Secondly, based on the demand of non-stationary vehicle interior noises, the microphone, amplifier, data acquisition card, etc are been selected and a hardware system is been made, which provides reliable support for the normal operation of the hardware environment to achieve stationary and non-stationary noise acquisition.Thirdly, based on the hardware system of intelligent sound quality evaluation of non stationary vehicle noise, the application program interface function of the data acquisition card is encapsulated by the object oriented method, and the data acquisition module of the software system is designed by using the method of multithreading. Calling the GRNN evaluation algorithm of sound quality of nonstationary vehicle interior noise is realized in VC++ by using the mixed programing method in which Matlab engine is called in VC++. The data exchange between VC++ and Matlab is also implemented.Finally, according to the relevant national standard, the noise samples in four kinds of vehicle condition(respectively are idle, uniform, acceleration, braking) are collected using the developed intelligent evaluation system for sound quality of nonstationary vehicle interior noise. With this system, a sound quality evaluation model is established and the unknown noise samples are introduced into the intelligent evaluation system to verify its effectiveness by comparing the predicted and the real values.