摘要
大型船舶监控是一个参数众多的非线性复杂系统,传统的专家系统和智能化系统面临着挑战。基于耦联式多层神经网络的大型船舶临控系统有别于其他传统的系统,基本思路是:首先将大型船舶分成数量众多的多阶子系统,每个底层的子系统由单元神经网络进行神经计算,给出各因素的表现程度,然后将各因素的评判程度作为该子系统的上一阶子系统的评价网络的输入值,经过神经计算,得出该子系统的上一阶子系统的评价结论,以此类推,直至给出大型船舶的评价结论。实验数据与结果符合神经网络系统给出的结果。系统知识库易于更新和维护,提高了系统的自学习能力,避免了传统专家系统中的"知识爆炸"问题。
Monitoring and management system of large-sized watercraft is a non-linear complex system with many parameter. Monitoring and management system of large-sized watercraft based on networks is different from other conventional system, its basic idea is：at first large-sized watercraft is divdied into many muhi-layer subsystem, computation of every rock-bottom system is finished by networks,and then give effect level of every factor,and then the effect level is taked as input value of the previous subsystem, after neural computation, and we can know evaluation of the previous subsystem of the subsystem, analogy according to this, until we give evaluation of the large-sized watercraft. Experiment data and result is the same as the reslult the system give. Knowledge library of the system is easy to renew and maintain, and enhance self-learning ability of the system, and avoid the problem of ＂knowlegye explosion＂ in conventional expert system.
出处
《舰船科学技术》
2009年第8期124-126,137,共4页
Ship Science and Technology
基金
安徽省自然科学基金资助项目(70416241)
关键词
神经网络
船舶
监控
管理
neural networks
watercraft
monitoring
management
