论文部分内容阅读
针对缺陷定量辨识的非线性、不适定性提出了采用混沌-LM(Levenberg-Marquardt)混合解法结合单面红外检测技术对缺陷进行定量辨识的新算法,并提出了相对敏感系数分析缺陷可检测性的新概念。通过相对敏感系数分析发现,缺陷尺寸、位置的不同描述参数的可检测性并不相同,而且采用检测面达到最大温差时刻点的瞬态检测比稳态检测更具优越性。文中采用数值实验对瞬态和稳态检测均进行了检验,实验结果与相对敏感系数分析结论一致。此外,数值实验还证明了混沌-LM混合解法的有效性,表明了辨识结果的稳定性和全局最优性。
Aiming at the non-linear and ill-posedness of quantitative identification of defects, a new algorithm for the quantitative identification of defects using the hybrid method of Levenberg-Marquardt and single-sided infrared detection is proposed. The relative sensitivity coefficient is analyzed to detect the defects new concept. Through the analysis of relative sensitivity coefficient, it is found that the detectability of the different description parameters of the defect size and position are not the same, and the transient detection using the detection surface at the point of maximum temperature difference has more superiority than the steady state detection. In this paper, both transient and steady-state tests are tested by numerical experiments. The experimental results are consistent with the results of relative sensitivity analysis. In addition, the numerical experiment also proves the effectiveness of the chaos-LM hybrid method, indicating the stability and global optimality of the identification results.