基于FEA和ANN的水火弯板表面变形预测方法

The Line-heating Deformation Prediction Method Based on FEA and ANN

水火弯板是船体曲面的主要加工方法。目前,水火弯板的表面变形质量主要由加工技师的技术和经验决定。为了科学合理地预测其表面变形量,确定水火弯板的加工参数,提高产品的质量和生产效率,本文提出了一种高效、准确的水火弯板表面变形预测方法。首先应用有限元分析法(FEA)对水火弯板的成型过程进行模拟仿真;然后建立水火弯板加工参数与板件表面变形之间的神经网(ANN)模型,并依据FEA的分析结果对网络进行训练;最后应用该网络对钢板的表面变形量进行预测。

Line-heating is the main method for forming compound-cured shells of hull. And it is mainly carried out by skilled workers, so the accuracy of final shape and the productivity solely depend on the experience of the workers. To predict the plate deformation, determine the processing parameters, and improve the productivity, a new method which includes three steps is developed. Firstly, the LH fabrication process is simulated by Finite Element Analysis （FEA） due to the complexity of the LH mechanism. Secondly an Artificial Neural Network （ANN） model is established. Finally, the calculation results of simulation by FEA, are used to train the network. In this way, a LH deformation prediction method is proposed for real time analysis.