为解决结构多损伤情况下的位置识别和损伤程度判定问题,提出了一种基于改进粒子群优化算法和贝叶斯理论的两阶段损伤识别方法,该方法采用频率和模态应变能作为损伤定位源数据,分别用基于频率改变和基于应变能耗散率的识别方法进行损伤信息的初步提取,再利用贝叶斯融合理论对损伤位置进行较为精确的判定.然后,利用粒子群优化(PSO)算法对损伤位置和程度进行更为精确的二次识别.考虑到简单PSO算法易陷入局部最优解,提出了3种改进措施,即粒子位置突变、最优记忆粒子微搜索和双收敛措施.数值仿真结果表明:采用贝叶斯融合理论可以有效地识别出可能的损伤单元,在此基础上用改进的PSO算法可以更精确地识别损伤的位置和程度,同时采用3种改进措施的PSO算法的识别精度明显优于其他PSO算法和遗传算法.%In order to solve structural multi-damage identification problem, a two-stage identification method based on the particle swarm optimization(PSO) algorithm and the Bayesian theory was proposed. In this method, structural modal strain energy (MSE) and frequency are considered as two information sources, and the methods based on frequency change and MSE dissipation ratio are utilized to extract damage information. Then, the Bayesian theory is utilized to integrate the two information sources and preliminarily detect structural damage locations. Finally, the PSO algorithm is adopted to precisely identify structural damage locations and extents. In order to improve the identification results of a simple PSO algorithm, three improved strategies, particle position mutation, elitist micro-search and double convergence criterion, were presented. The simulation results for a two-dimensional truss structure show that the Bayesian theory can identify the suspected damage locations, the improved PSO algorithm can precisely detect the damage extent, and the identification precision of the PSO algorithm with the three improved strategies is obviously better than those of the other PSO algorithms and the genetic algorithm.
展开▼