基于IPOA的巨型组合框架结构震损快速预测模型研究?

为实现巨型组合框架结构的地震损伤程度快速评估，提出了一种基于改进鹈鹕优化算法（IPOA）的多参数震损预测方法。设计了5 个不同参数的巨型组合框架结构模型，利用振动台试验和有限元软件进行非线性时程分析获取结构动态响应数据，并采用结构损伤指数量化评估结构的损伤程度。同时，引入K 均值聚类优化策略和惯性权重自适应优化策略改进传统的鹈鹕优化算法。基于振动台试验和有限元分析结果数据，比较了不同输入参数组合预测结构损伤的准确性，构建了能反映结构参数与结构损伤之间非线性关系的智能算法快速预测模型。最后，将模型预测结果与一缩尺比为1/15 的振动台试验模型结构损伤程度进行对比验证。结果表明：（1）改进鹈鹕优化算法模型的准确性和泛化能力均优于其他算法模型；（2）最大层间位移角与结构损伤相关性最高，增加影响结构损伤的输入参数可提高预测模型的准确度和泛化能力；（3）模型预测的结构损伤指数与试验结果相比误差均小于10%，预测结构损伤等级与试验结果一致，所提出的快速预测模型能高效准确地预测结构的损伤指标，为巨型组合框架结构震后损伤快速评估提供了一种新方法。

Rapid Prediction Model for Seismic Damage in Mega Composite Frame Structures Based on IPOA Methods

To rapidly assess the extent of seismic damage in mega composite frame structures, thisstudy introduced a multi-parameter seismic damage prediction method utilizing Improved Pelican Opti-mization Algorithm (IPOA). Five models with different parameters were developed, and dynamic re-sponse data for the structures were obtained through shaking table tests and nonlinear time-history analyses using finite element (FE) software. Structural damage indices were quantified to assess the ex-tent of damage. Additionally, the traditional Pelican Optimization Algorithm (POA) was enhanced byincorporating K-means clustering optimization and inertia weight adaptive optimization strategy. Basedon the data from shaking table tests and FE analyses, the accuracy of structural damage predictions us-ing different input parameter combinations was compared. A rapid prediction model using an intelli-gent algorithm was constructed to reflect the nonlinear relationship between structural parameters andits damage. Finally, the model''s predictions were compared and verified against the seismic damageextent from shaking table tests on a 1/15 scale model. The results indicated that: (1) The IPOA modelexhibited superior accuracy and generalization capability compared to other algorithm models; (2) Themaximum inter-story drift angle showed the highest correlation with structural damage. The introduc-tion of additional input parameters that affected structural damage could enhance the model''s predic-tion accuracy and its generalization capability; (3) The predicted structural damage indices exhibited anerror of less than 10% compared to the experimental results, and the predicted levels of structuraldamage aligned with the experimental results. The proposed rapid prediction model can effectively andaccurately predict structural damage indicators.