This paper discussed the shortcomings of traditional sound pressures level, such as the lack of subjectivity consideration for noise and so on. A virtual method of prediction loudness characteristic and identifying noise sources which based on human auditory properties were applied to analyze vehicle interior noise. In order to reduce the commercial vehicle interior low booming , it was the first found a characteristic loudness calculation model of diffuse sound field in matlab, which was based on Zwicker model. Secondly, the excitation singles at the suspension mountings and responding singles of the interior noise were measured with road testing. And to predict and calculation the loudness mapping and structure contribution of the cab, an acoustic-structure coupled finite element model was established. In both applications, they were used for analysis relationship of vehicle interior noise source from different plates and sound quality frequency spectrum design. The results of noise source identification in both applications revealed that the loudness mapping of noise source identification method enabled the detection rattling sources from cab plates vibration to be more effective compared to the use of SPL. Thus, according this method, the acoustic optimization of the cab was designed and displayed excellent performance.