基于神经网络和遗传算法的系泊优化设计

Mooring optimization design based on neural network and genetic algorithm

  • 摘要:
      目的  为使船舶维持作业位置的稳定,需采用系泊系统来减小浮体的平动。
      方法  以一艘南海作业的铺管船为例,通过优化锚泊线的布置方式来最大限度地减小系泊状态下锚链的平动位移量,保障船舶作业安全。首先,以抛锚距离和锚链方位角作为正交试验的因素,得到不同的系泊布置方案,利用Moses软件计算不同布置方案在不同浪向下的时域运动位移和锚链受力情况。然后,将结果作为样本,对BP神经网络进行训练,实现BP网络对Moses时域计算的仿真。最后,将抛锚距离和锚链方位角作为优化变量,取不同浪向下的加权平动位移为优化目标,并以BP神经网络的泛化能力代替Moses的时域计算,采用遗传算法进行优化求解。
      结果  结果表明,该铺管船各个浪向下的平动位移均有了显著的减小,优化效果明显,
      结论  可为海上浮式结构物的系泊布置提供参考。

     

    Abstract:
      Objectives  In order to maintain the stability of the position of a ship, a mooring system is required to reduce the translational motion of floating structures.
      Methods  Taking a pipe-laying vessel in the South China Sea as an example, it is possible to minimize the translational displacement of the anchor chain in the mooring state by optimizing the arrangement of the anchor line to ensure the safe operation of the ship. First, we can obtain several different layouts through orthogonal testing after selecting the azimuth and distance of the anchor chain as the test factors. We then calculate the different movements and force in time domain value at different wave direction angles for each layout using Moses. With the calculation results as samples, the BP neural network method achieves time domain simulation in Moses. After choosing the azimuth and distance of the anchor chain as the optimization variables, and with each wave-weighted translational displacement probability as the optimization objective, we find that the generalization capability of the BP neural network method can replace the time domain calculation of Moses.
      Results  Using a genetic algorithm optimization solution, movement is significantly reduced at different wave direction angles.
      Conclusions  This conclusion can provide a reference for the mooring arrangements of floating structures.

     

/

返回文章
返回