【作者】 林近山； 窦春红； 马汝建；

【Author】 Lin Jinshan (School of Electromechanical Engineering,Weifang University,Weifang City,Shandong Province),Dou Chunhong Ma Rujian.

【机构】 潍坊学院机电工程学院； 潍坊学院信息与控制工程学院； 济南大学机械工程学院；

【摘要】 利用 ANSYS 软件建立了 W12—1平台的有限元模型,通过在模型识别点上施加激励得到响应的方法获得了神经网络的训练数据。建立了关于 W12—1海洋平台的神经网络模型,利用前面获得的数据训练网络至收敛,把实测响应数据输入训练好的网络中,得到了 W12—1平台的识别载荷。识别结果表明,在对海洋平台这样的大型复杂结构进行载荷识别时,神经网络方法识别的平均相对误差为8.5%,相对于传统载荷识别方法有很大的优势。更多还原

【Abstract】 The finite-element model of W12-1 platform is established by using ANSYS software.The training data of neural network is obtained by applying stimulation to get response at the identification point of the model.The neural network model of W12-1 offshore platform is set up.The network is trained into the convergence state by using the above data and the measured response data are inputted to the trained network,obtaining the identification load of W12-1 platform.The identification result shows that in the load identification of such large complex structure as offshore platform,the average relative error of neural network identification is 8.5%,showing great advantage compared with traditional load identification method.更多还原