基于POD-ANNS的架空输电线路舞动响应预测  

Response Prediction of the Overhead Transmission Line Galloping Based on the POD-ANNS Method

作　　者：蔡萌琦 田博文 闵光云 杨曙光 包婉玉 CAI Mengqi;TIAN Bowen;MIN Guangyun;YANG Shuguang;BAO Wanyu(School of Architecture and Civil Engineering,Chengdu University,Chengdu 610106,Sichuan,China;Failure Mechanics and Engineering Disaster Prevention and Mitigation Key Laboratory of Sichuan Province,Sichuan University,Chengdu 610065,Sichuan,China;School of Mechanical Engineering,Chengdu University,Chengdu 610106,Sichuan,China;Sino-French Institute of Nuclear Engineering and Technology,Sun Yat-sen University,Zhuhai 519082,Guangdong,China;School of Civil Engineering,Central South University,Changsha 410075,Hunan,China)

机构地区：[1]成都大学建筑与土木工程学院,四川成都610106 [2]四川大学破坏力学与工程防灾减灾四川省重点实验室,四川成都610065 [3]成都大学机械工程学院,四川成都610106 [4]中山大学中法核工程与技术学院,广东珠海519082 [5]中南大学土木工程学院,湖南长沙410075

出　　处：《电网与清洁能源》2024年第5期10-18,共9页Power System and Clean Energy

基　　金：国家自然科学基金项目(51507106)。

摘　　要：针对塔线体系下覆冰八分裂导线的舞动特性,通常采用有限元法(FEM)获取舞动响应,但是FEM往往会花费大量时间,获取舞动响应的时间成本巨大,有限模型搭建难度大、类型繁多,同时也必须考虑动力学求解中计算不收敛等棘手问题。因此,获得不同参数、工况下的舞动响应比较困难。该文提出了一种混合代理模型,该模型能基于本征正交分解(POD)和人工神经网络(ANNS)实现舞动响应的快速预测。将FEM得到的舞动响应组合成快照矩阵,然后基于POD方法得到快照矩阵的POD模态基系数和POD模态;使用3种代理模型(BPNN、RBFNN、CNN)预测POD模态基系数,线性组合预测的POD模态基系数和POD模态,实现舞动响应的快速预测;还对比了3种混合模型下(POD-BPNN、POD-RBFNN、POD-CNN)的预测误差,预测时间。结果表明:3种混合模型均能快速、准确地得到目标响应,并且POD-RBFNN混合模型的误差更小;快照数量达到60个,基本可表征出全部样本的特征。The finite element method(FEM)is usually used to obtain the galloping response of the iced 8-buddle conductor under the tower line system.However,the FEM often takes a long time and the time cost of obtaining the galloping response is huge.The finite model is difficult to build and has many types,and it must consider thorny questions such as the calculation non-convergence in the kinetic solution,so it is difficult to obtain the galloping response under different parameters and working conditions.Based on this,this article proposes a hybrid proxy model that can achieve fast prediction of the galloping response based on proper orthogonal decomposition(POD)and artificial neural networks(ANNS).First,the galloping response obtained by the FEM is combined into a snapshot matrix,and then the POD modal basis coefficients and POD modes of the snapshot matrix are obtained based on the POD method.Second,three proxy models(BPNN,RBFNN and CNN)are used to predict the POD modal basis coefficients,and a linear combination of the predicted POD modal basis coefficients and POD modes is used to achieve rapid prediction of the galloping response.Finally,the prediction error and prediction time of three hybrid models(POD-BPNN,POD-RBFNN and POD-CNN)are compared.The results show that all hybrid models can quickly and accurately obtain the target response,and the error of the POD-RBFNN hybrid model is smaller;The number of snapshots reaching 60 can basically characterize the characteristics of all samples.

关 键 词：本征正交分解 神经网络 有限元分析 舞动特征 塔线体系 

分 类 号：TM752[电气工程—电力系统及自动化]