深度堆栈自编码网络在船舶重量估算中的应用  

Application of Deep Stack Autoencoder Network in Ship Weight Estimation

作　　者：陈健[1] 唐俊遥 朱生光 周兆钊 CHEN Jian;TANG Junyao;ZHU Shengguang;ZHOU Zhaozhao(School of Electro-Mechanical Engineering,Guangdong University of Technology,Guangzhou 510006,China)

机构地区：[1]广东工业大学机电工程学院,广州510006

出　　处：《计算机工程》2019年第5期315-320,共6页Computer Engineering

基　　金：广东省科技计划项目(2016A010101025;2014A010103027);惠州市科技计划项目(2015B020005007)

摘　　要：传统的船舶重量估算方法多数存在误差大、成本高等问题。为此,提出一种基于深度学习的船舶重量估算算法。利用多层神经网络逐层无监督学习训练初始化参数,通过反向梯度下降的方式微调参数。运用深度堆栈自编码网络挖掘深层次的数据特征,并在ShipWE自建数据库上进行分析。实验结果表明,与传统吃水估算方法相比,该算法具有更强的稳定性和更高的准确性,与BP神经网络算法和径向基函数神经网络算法相比,该算法的精度更高,能有效解决船舶估算可信度低的问题。Aiming at the problems of large error and high cost of most traditional ship weight estimation methods,an algorithm based on deep learning is proposed.It trains initialization parameters according to layer by layer unsupervised learning algorithm,which is based on multi-layer neural network,and uses inverse gradient descent fine-tuning the parameters.The deep stack autoencoder network is used to mine the deep data features and do analysis on the ShipWE self-built database.Experimental results show that compared with traditional estimation methods,this algorithm has better stability and higher accuracy.Compared with Back Propagation Neural Network(BPNN) algorithm and Radial Basis Function Neural Network(RBFNN) algorithm,the proposed algorithm has higher accuracy and can effectively solve the problem of low reliability of ship estimation.

关 键 词：气囊船舶下水 深度学习 反向梯度下降 深度堆栈自编码 逐层无监督学习 参数微调 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]