基于参数化模型的架空直立式码头损伤诱因反演

近年来架空直立式码头成为长江上游地区码头主要的结构形式之一,由于其构件数量多、外部损伤因素复杂,又缺乏有效的损伤获知手段,导致很多码头处于“亚健康”服役状态。桩基作为整个码头结构的关键构件,是“窥视”码头结构受力状态及外界损伤诱因作用的重要“窗口”。通过构建神经网络反演模型对采集的有限元计算数据进行分析,采用Python-MANSYS建立了简化计算模型,通过与实体单元模型进行对比,验证了简化计算模型的适用性。通过抽样的方式获取参数化分析的数据样本空间,并通过降维提取反演模型所需数据。在此基础上,对主要损伤诱因的相关参数进行反演,并对反演结果和模型性能进行分析。

Inversion of damage inducement of overhead vertical wharf based on parameterization model

Overhead vertical wharfs have become one of the main structural types of wharves in the upper reaches of the Yangtze River in recent years.Due to their large number of components,complex external damage factors and lack of effective means to know the damage,many wharves are in sub-health service status.As the key component of the whole wharf structure,the pile foundation is an important“window”to“peep”the stress state of wharf structure and the effects of external damage inducement.By constructing a neural network inversion model to analyze the collected finite element calculation data,a simplified calculation model is established using Python-MANSYS,and the applicability of the simplified calculation model is verified by comparison with the solid element model.The data sample space of parametric analysis is obtained by sampling,and the data needed by inversion model is extracted by dimension reduction.On this basis,the relevant parameters of the main damage inducement are inverted,and the inversion results and model performance are analyzed.