基于混合PSO算法和损伤概率均值的两阶段梁式结构损伤识别

受算法随机参数和测量噪声等因素的影响,基于群智能算法的结构损伤识别法容易出现单次识别误差大和多次识别波动大的问题。针对该问题,本文基于混合PSO算法和损伤概率均值提出一种两阶段梁式结构损伤识别方法。定义了包括单元容许损伤值α_cr和临界概率p_c的损伤有效原则,以计算损伤概率均值。第一阶段,对混合PSO算法的多次识别结果分批计算一阶段损伤概率均值,第二阶段,在此基础计算两阶段损伤概率均值。本文方法能够有效地提高高噪声水平影响下的结构损伤识别精度。简支梁和两跨连续梁的损伤工况试验研究表明,该算法能够有效地应用于不同结构和不同工况的结构损伤识别,在减少损伤误判的同时保证损伤单元的识别精度。

A two-stage structural damage detection method for beam-like structures based on hybrid PSO and probabilistic mean of damage vector factor

Due to the random parameters of algorithms and measurement noise,structural damage detection methods based on swarm intelligence algorithms are prone to large errors in single identification and large fluctuation in multiple identification.Therefore,this paper proposes a two-stage a damage detection method for beam structures based on a hybrid PSO algorithm and the probabilistic mean of a damage vector factor.An effective damage principle is defined including an elemental allowable damage value α_cr and two critical probability parameters p_c1 and p_c2. In the first stage,multiple SDD results based on the hybrid PSO are divided into several batches and the corresponding probabilistic means of the damage vector factor are calculated as the first stage probabilistic means.Then,a new probabilistic means of the damage vector factor is determined as the final SDD results based on the results in the first stage.The proposed method can effectively improve the accuracy of structural damage detection under the influence of high noise.Numerical studies on damage cases of a simply-supported beam and a two-span continuous beam demonstrate the effectiveness and efficiency of the proposed method.