利用模式识别技术提高燃烧激励响应信号信噪比

针对实测缸盖表面振动信号中的耦合低频干扰信号,提出了在内燃机曲轴转角域建立往复惯性力激励响应信号模型的方法,利用模式识别技术辨识模型参数,剔除往复惯性力激励的影响,以提高燃烧激励响应信号的信噪比,并利用195型单缸柴油机的测试结果对所提出的方法进行验证.结果表明:选用压缩过程初期的数据进行模型参数辨识、采用4阶模型描述往复惯性力激励响应信号可以获得较为理想的效果;而瞬时转速波动对模型参数辨识的影响不显著;采用模式识别技术剔除往复惯性力激励响应信号后,实测的振动速度信号与压力升高率信号在峰值压力前后角度域内的相关性明显提高,从而验证了所提出方法的有效性.

Improvement of SNR of Vibration Response Signal Excited by In-Cylinder Pressure Using Pattern Recognition Method

Abstract： A model was established to describe the response signal excited by reciprocating inertia force. The parameters of the model were identified by the pattern recognition method. The processing results of the experiment data from a 195 diesel engine indicate that the influence of the instantaneous speed could be ignored. A fourth order model is appropriate to describe the response signal and the data of early stage in compression stroke is suitable for identifying the model parameters. The trend of the processed internal combustion engine vibration velocity remains consistent with that of the pressure rise rate and the correlation between them is improved, which verifies the proposed method.