基于神经网络技术的南京长江大桥桥墩船舶撞击力识别

IDENTIFICATION OF THE IMPACT FORCE DURING SHIP-BRIDGE COLLISION ON THE PIER OFNYRB BASED ON TECHNIQUES OF ANN

主要研究基于神经网络技术的南京长江大桥桥墩船撞力识别。在建立了南京桥4#桥墩的空间有限元分析模型后,对4#桥墩进行了动力特性分析;基于MATLAB平台建立了“船撞力-响应”神经网络,由沃辛试验船撞力曲线计算得到了船对桥墩的撞击响应并作为神经网络训练的样本。仿真计算结果表明,由桥墩墩顶的响应可有效识别船舶撞击力的大小、方向和撞击位置,该神经网络可用于有、无噪声干扰的船撞力识别,并且具有良好的容错性和鲁棒性。

The identification of the impact force during ship-bridge collision on the pier of the Nanjing Yangtze River Bridge (NYRB) is focused based on techniques of the artificial neural network (ANN). The 3D finite element model of pier 4 of the NYRB is built, and the analysis of its dynamic characteristics is completed. The responses due to the impact of the ship on the pier are obtained by use of the Osin's experimental curve of ship collision, and samples for the training of ANN are constructed. Based on the platform of MATLAB, the impact force versus response ANN used for identification of impact forces is established. The results of simulation demonstrate that the magnitude, direction and location of impact force of ship collision can be effectively identified either with noise or without noise from the responses at the pier top. In addition, the ANN has good fault tolerance and robustness.