深海潜水器载人舱腐蚀损伤识别方法研究

深海潜水器载人舱是海洋深潜结构物的重要组成部分,保障着工作人员的安全,所以掌握该结构在长期使用过程中的健康安全状况非常必要。本文首先对4500m深海潜水器载人舱6个监测点进行加速度监测,得到其加速度的结点能量；再通过广义回归神经网络和概率神经网络对这些加速度结点能量进行训练；最后用训练好的神经网络进行预测。研究结果表明,对深海潜水器载人舱结构损伤位置的预测正确率可以达到90%左右,可以达到健康监测的目的。

Research on the health monitoring method of deep-sea submersible manned cabin

The deep-sea submersible manned cabin is an important part of the deep-sea structure, which guarantees the safety of the staff, so it is necessary to master the health and safety of the structure in the long-term use. In this paper, the acceleration of six monitoring points of the 4500m deep-sea submersible manned cabin is monitored firstly, and the node energy of acceleration is obtained; then the node energy of acceleration is trained by the generalized regression neural network and probability neural network; finally, the trained neural network is used to predict. The results show that the prediction accuracy of damage location of deep-sea submersible manned cabin structure can reach more than 90%, and the purpose of health monitoring can be achieved.