ABSTRACT

Fuzzy inference method is applied to formulate an algorithm capable of estimating material elastic constants (ECs) of a specimen by solving an inverse problem with a group of measured resonance frequencies obtained via Resonant Ultrasound Spectroscopy (RUS). The algorithm is validated with RUS data from a specimen of polycrystalline aluminium alloy. Then the algorithm is found to be sensitive to the initial ECs by processing RUS data from a specimen of fine-grain polycrystalline Ti–6Al–4V, the same as the Levenberg–Marquardt (L–M) method popularly used in solving inverse problems. To overcome such a drawback, a hybrid method of Particle Swarm Optimization (PSO) and Density-Based Spatial Clustering of Applications with Noise (DBSCAN) is proposed. And it is used to generate several groups of initial ECs for the fuzzy inference method. There is a trade-off between computational time and accurately estimated ECs, since the hybrid method needs more time to directly find out accurate ECs.

摘要

应用模糊推理方法，通过求解一组通过共振超声光谱法（RUS）获得的共振频率反演问题，提出了一种算法，能够估计样品的材料弹性常数（EC）。用多晶铝合金样品的RUS数据验证了该算法。然后，通过处理来自细晶粒多晶Ti–6Al–4V样品的RUS数据，发现该算法对初始EC敏感，这与通常用于解决反问题的Levenberg-Marquardt（LM）方法相同。为了克服这种缺点，提出了一种混合粒子群优化（PSO）方法和基于密度的带有噪声的应用程序空间聚类（DBSCAN）方法。并用它来为模糊推理方法生成几组初始EC。