中文摘要：

研究了气体分子碰撞能量转移模型和弛豫声衰减理论．分析了混合气体中某些成分的浓度低于0．01％时，分子复合弛豫声衰减系数与气体浓度的模型变为崎岖不平的原因．利用信号处理中的滤波方法，使用平滑窗处理弛豫声衰减结果，改善了弛豫声衰减理论．得到了检测三元混合气体中微弱气体浓度的算法，检测浓度达到0．001％．设计了3种不同权重的平滑窗，均可降低由微弱浓度引起的检测误差扩大，其仿真结果一致证明弛豫平滑处理算法的合理性．

英文摘要：

By analyzing the energy transfer models in molecular collision processes and the acoustic relaxation attenuation in gas, we find the reason of ruggedness in the dependence of molecular multi-relaxation attenuation coefficient on the mixture concentrations when the gas constituents have concentrations less than 0. 01%. We use a typical smooth window to process acoustic relaxation attenuation data, and develop an algorithm to detect weak gas concentration of three-component gas mixtures. The measuring accuracy of this method achieves 0. 001%. Three designed matrices with different weighted averages can all reduce errors resulting from weak gas concentration. The simulation results of consistent error-reduction using all the matrices show that the relaxation smooth processing algorithm is reasonable.