改进多种群遗传算法在墙土系统损伤识别中的应用

针对墙土系统损伤识别方法进行研究，提出了一种基于改进多种群遗传算法的墙土系统损伤识别方法。首先，建立了墙土系统动测简化模型，同时对土体发生损伤时墙土系统的特征方程进行理论分析，基于系统的特征方程构造目标函数；其次，对多种群遗传算法进行改进，改进的内容主要包括采用实数编码、采用自适应交叉概率、采用自适应变异概率；最后，利用改进多种群遗传算法分别进行了无噪声条件和噪声条件下的墙土系统损伤定位和定量研究。通过分析结果表明：无论对单处损伤还是多处损伤、单一损伤程度还是多损伤程度，按所提出的方法都能较好的识别出损伤位置和损伤程度，具有较强的抗噪声能力。因此，所提出的方法为墙土系统的损伤识别提供一种简单有效的途径。

Application of Improved Multi-population Genetic Algorithm to Damage Identification of Soil-wall System

The method of damage identification in soil-wall system was studied; a new approach based on improved multi-population genetic algorithm (IMGA) was developed. First, the simplified dynamic-detection model of soil-wall system was established, meanwhile, the theoretical analysis of characteristic equations in soil-wall system was conducted when soil in damage status. The objective function based on characteristic equations was established. Then, the improvements of multi-population genetic algorithm, including the adoption of real-valued representation, adaptive cross operator and adaptive mutation operator, were conducted. Finally, the localization and quantification of the soil-wall system damage were performed by IMGA with and without the consideration of noise, respectively. The results indicate that damage location and damage extent can be detected efficiently, and anti-noise performance is better.