基于推理与神经网络的造船火焰矫正工艺自动生成法

选择船体建造过程中常见的3类焊接构件的火焰矫正工艺进行研究，实现了工艺的自动生成．采用分析工艺规则、建立逻辑推理代码的方法实现火焰矫正方法的自动生成；通过分析矫正变形以及焊接变形的影响因素，建立了影响因素之间的映射关系式，使用人工神经网络拟合映射关系式，从而确定并优化了火焰矫正的关键工艺参数．研究结果经过实验验证获得了令人满意的效果，该方法有效地提高火焰矫正的效率与矫正质量的稳定性，并为实现火焰矫正自动化提供了基础．

Logic Reasoning and Neural Network Based Process Automation Method of Flame Rectification in Shipbuilding

Three kinds of typical welding components in the process of hull construction were chosen to analyze their flame rectification process, which, as a result, realizes the process automation. By means of analyzing process rule and establishing logic reasoning codes to generate the flame rectification method automatically; by analyzing the parameters that influence the distortion of flame rectification and welding process to build the mapping formula; and then use artificial neural network algorithm to resolve it so as to determine the critical process parameters of flame rectification. Satisfying results were achieved through the experiment in shipyard. It proves that the proposed method can improve the productivity and quality of flame rectification, and provide with the fundament for the automation of flame rectification.