基于区域分解的快速多极基本解法预测发动机表面辐射声场

摘要

Due to the complex structure of engine surface and a lot of long narrow borders,standard approaches to solve the distribution of source points cannot gain enough computational precision in using the method of fundamental solution(MFS).Thus,an improved algorithm for solving the distribution of source points was presented and used to deal with long narrow borders.Combining with the fast multipole algorithm,a domain decomposition based fast mutipole method of fundamental solution (FMMFS)was proposed.The improved algorithm was used to solve the distribution of source points,while the FMMFS was used to predict the acoustic radiation field on engine surface.The results show that the optimal distance between the collocation points and source points is 34.5 mm and the computational precision is improved.When the number of truncation terms is 15,the uncertainty in low frequency due to too few number of truncation terms can be avoided and the computational efficiency can be guaranteed.To improve the computational precision,the method uses domain decomposition to solve the distribution of source points,it differs from the fast multipole BEM which adopts mesh refinement.The experimental results are in good agreement with the computational results,which indicate that the domain decomposition based FMMFS can improve substanlially the computational precision of the prediction of acoustic radiation field on engine surface.%针对发动机表面结构复杂，多狭长边界，用通常算法求解源点分布无法保障基本解法计算精度问题，提出改进的计算源点分布算法，利用区域分解处理影响基本解计算精度的狭长边界；结合快速多极算法，形成适于三维复杂表面辐射声场预测基于区域分解的快速多极基本解法。利用区域分解求解发动机计算模型的源点分布，在源点分布已知基础上利用快速多极基本解法预测发动机表面辐射声场。结果表明，利用区域分解求得源点与配置点最优距离为34．5 mm，可提高发动机辐射声场预报精度；截断项数为15时可避免截断项数过小引起的低频不稳定，并保证计算效率；不同于快速多极边界元用网格加密提高计算精度，基于区域分解的快速多极基本解法通过区域分解进行优化源点分布可提高精度，且不增加计算负担。通过发动机表面辐射噪声实验，计算值与实验值吻合较好，说明该解法可提高发动机表面辐射声场的预测精度。