Abstract: A automobile buzz, squeak and rattle (BSR) automatic recognition system is proposed to address the issues of unstable recognition and low efficiency in the current subjective evaluation methods for diagnosing automobile BSR. Through experiments, the recognition success rates of two detection methods, namely the abnormal noise recognition system and manual subjective evaluation, are compared. The experimental results show that the comprehensive recognition success rate of manual subjective evaluation is only 66.2%, while the designed automobile BSR recognition system had a comprehensive recognition rate of up to 93.6%, and its stability is also better than manual recognition. The designed automobile BSR recognition system eliminates the problem of low manual recognition accuracy caused by experience differences and has certain market promotion value.

Key words: Automobile BSR recognition; Subjective evaluation; Characteristic parameters; Recognition rate

摘要： 针对目前汽车异响（BSR）诊断的人工主观评价方法的识别不稳定、效率低等问题， 提出了一种汽车异响自动识别系统。通过实验对比了异响识别系统和人工主观评价两种检测 方法的识别成功率，实验结果表明，人工主观评价的综合识别成功率只有 66.2%，而所设计 的异响识别系统综合识别率高达 93.6%，并且稳定性也优于人工识别。所设计的汽车异响识 别系统消除了经验差异所造成的人工识别准确率低问题，具备一定的市场推广价值。

关键词: 汽车异响识别；主观评价；特征参数；识别率