基于误差估计的发动机排气声源特性间接识别模型求解方法研究

发布时间：2018-10-05 10:24

【摘要】：用于发动机声源特性识别的线性时不变频域模型仍然是较为实用的方法,但尚存在求解误差大、对输入误差敏感、识别及求解过程复杂等问题。为了提高声源识别的准确性,建立了间接法声源特性识别的误差估计方法,即声源声压离散度估计和声源阻抗值偏差估计。将识别模型中超定非线性方程组中的方程式进行两两组合,通过解析几何的方法直接求解方程组,获得声源特性参数的多个解,再利用声源阻抗值偏差估计选取最优解。经分析计算,获取的声源特性参数的误差估计值要明显低于传统四负载法。最后,采用三维声流耦合仿真方法对获取的声源特性参数进行验证,仿真预测的远场响应点声压值与试验结果吻合良好,表明基于误差估计的声源特性参数最优选取法可以获取更加准确的发动机声源特性参数,减少识别误差和识别的复杂程度。
[Abstract]:The linear time-invariant domain model used to identify the characteristics of engine sound source is still a practical method, but there are still some problems such as large error, sensitive to input error, complex identification and solution process, etc. In order to improve the accuracy of acoustic source identification, an error estimation method for indirect acoustic source identification is established, that is, sound pressure dispersion estimation and source impedance deviation estimation. The equations of overdetermined nonlinear equations in the identification model are combined and solved directly by analytic geometry to obtain multiple solutions of acoustic source characteristic parameters. Then the optimal solution is selected by estimating the deviation of acoustic source impedance. Through analysis and calculation, the error estimate of the characteristic parameters of sound source is obviously lower than that of the traditional four-load method. Finally, the acoustic source characteristic parameters obtained are verified by using the three-dimensional acousto-flow coupling simulation method, and the predicted sound pressure values of the far-field response point are in good agreement with the experimental results. It is shown that the optimal selection method of acoustic source characteristic parameters based on error estimation can obtain more accurate characteristic parameters of engine sound source and reduce the recognition error and the complexity of identification.
【作者单位】：华东交通大学机电与车辆工程学院;清华大学汽车安全与节能国家重点实验室;
【基金】：国家自然科学基金(51268014) 华东交通大学科研基础经费(26541022)
【分类号】：U464.134.4

【相似文献】