在对面红外检测条件下,利用混沌算法全局收敛和LM算法快速收敛的相对优势提出了识别试件内部矩形缺陷的混沌-LM混合算法,并引入了相对敏感系数的概念用于评估缺陷参数的定量可检测性,同时采用附加内置热电偶测温的方法弥补了红外测温信息的不足。通过数值实验证明了以上研究内容的有效性。 Based on the relative advantages of the global convergence of the chaos algorithm and the rapid convergence of the LM algorithm, the chaotic-LM hybrid algorithm for identifying rectangular defects in the specimen is proposed under the condition of opposite-side infrared detection. The concept of relative sensitivity coefficient is introduced to evaluate the quantitativeness of the defect parameters Detectability, while using an additional method of built-in thermocouple to make up for the lack of infrared temperature measurement information. The numerical experiments prove the validity of the above research.