基于神经网络算法的船体结构腐蚀分析及预测

Corrosion Analysis and Prediction of Ship Hull Structure Based on Neural Network Algorithm

针对船体测厚数据未能被有效利用的问题,提出一种基于神经网络算法的结构腐蚀分析、预测方法,为船舶的健康管理提供评估及决策支持。依据神经网络算法的基本原理和建模方法,研究测厚数据结构、神经网络输入参数的选择、腐蚀状况的分析及预测等关键问题的实现方案,并在此基础上开发船体可视化模型关联测厚结果的展示、分析功能,应用于实船测厚数据并形成有针对性的分析结论、决策建议。采用该方法对某船实际测厚数据进行分析,预测出该船型随船龄的增长腐蚀严重的区域,并在3D模型中展示。

Based on the situation that hull thickness measurement data hasn？ t been effective utilized, a kind of neural net-work based corrosion analysis and prediction method was put forward, providing assessment and decision support for ship health management. According to the basic principle and modeling method of neural network algorithm, the solutions to key issues were investigated, such as establishing the data structure based on the thickness measurement, the selection of neural network input parameters, and the analysis and prediction of corrosion condition. The display and analysis functions of hull visible model con-nected with the data of thickness measurement were programmed and applied in real ship thickness measurement successfully to get analysis conclusion and suggestions. Based on the above method, a data structure was established, providing a certain techni-cal basis for ship comprehensive analysis and risk assessment.