摘要
由于耙吸挖泥船艏吹疏浚作业的过程复杂,以往国内外学者大多从利用耙吸挖泥船的关键设备的工作特性进行研究并建立相关模型,但从原理上建立耙吸挖泥船控制系统这样一个多参数非线性的输入时输出模型是不准而且不能满足实际控制系统的需要。文中采用数据驱动的方法解决耙吸挖泥船艏吹控制过程中控制变量与顺时产量之间的黑盒问题,通过大量的数据为基础,训练GRU神经网络,,使预测值和真实值更加接近。实验结果表明,GRU神经网络的预测效果要优于RNN神经网络,Nadam算法的优化效果好于Adam算法,可以有效地预测艏吹泥泵出口处产量,给施工提供有效的参考。
Because of the trailing suction hopper dredger stem blow dredging operations process complex,in the past most scholars both at home and abroad from the use of trailing suction hopper dredger operating characteristics of key equipment for research and to establish related model,but in principle to establish such a trailing suction hopper dredger control system of multi-parameter nonlinear input output model is not and can not meet the needs of practical control system.In this paper,a data-driven method is used to solve the black box problem between the control variable and the on-time output in the process of the bow blowing control of rake suction dredger.Based on a large number of data,GRU neural network is trained to make the predicted value closer to the real value.The experimental results show that the prediction effect of GRU neural network is better than that of RNN neural network,and the optimization effect of Nadam algorithm is better than that of Adam algorithm,which can effectively predict the output of bow blowing mud pump outlet,and provide effective reference for construction.
作者
章亮
俞孟蕻
袁伟
ZHANG Liang;YU Menghong;YUAN Wei(School of Electronic and Information,Jiangsu University of Science and Technology,Zhenjiang 212100)
出处
《计算机与数字工程》
2023年第10期2470-2473,共4页
Computer & Digital Engineering
