基于不完备损伤指标和遗传算法的特大桥损伤识别和传感器布点优化

DAMAGE DETECTION AND OPTIMUM SENSOR LOCALIZATION BASED ON INCOMPLETE DAMAGE INDEXES AND GENETIC ALGORITHMS FOR LONG-SPAN BRIDGES

特大桥健康监测系统不可能在所有自由度安放传感器,该文讨论了用由不完备振型建立的损伤指标的损伤识别和传感器布点优化方法。与过去用遗传算法优化传感器布点的适应度函数不同,该文用损伤指标最灵敏来建立适应度函数。对桥梁的单个损伤,该文用不完备模态柔度矩阵差和截断模态应变能变化率两个不完备损伤指标作为适应度函数来优化传感器布点,并与传统的COMAC指标对比,还改进了多种群遗传算法,以提高收敛速度和全局寻优能力。并以西堠门悬索桥有限元模型为例,识别不同部位的损伤。算例表明:该方法在损伤识别和传感器布点优化方面不仅可行而且有效。

The methods for damage detection and optimum sensor localization with the incomplete damage indexes based on the incomplete mode shapes are developed, and the multi-species genetic algorithms for the structural health monitoring system of long-span bridges are improved. Two incomplete damage indexes, the difference in the incomplete mode flexibility matrixes, and the change ratio in the truncated mode strain energies, are developed as the fitness parameters in comparison with traditional COMAC. The modified multi-species genetic algorithms combine the advantages of general genetic algorithms and multi-species genetic algorithms, and are better in convergence and global optimization. The FE model of Xihoumen suspension bridge is used as an example for damage detection and optimum sensor localization. The results show the methods are feasible and effective.