舰船实船训练系统中混合建模方法研究  

Study on Hybrid Modeling Method in Warship on Board Training System

作　　者：孔庆福[1] 曾凡明[1] 吴家明[1] 吴雄学[1] 

机构地区：[1]海军工程大学船舶与动力学院,武汉430033

出　　处：《船海工程》2010年第2期122-124,129,共4页Ship & Ocean Engineering

基　　金：海军工程大学科研基金项目(HGDJJ06009)

摘　　要：提出舰船实船训练系统中基于神经网络的混合建模方法,综合运用机理建模方法和辨识建模方法,既较好地保证了模型的全局特性,又有效提高了模型的精度。以舰船柴油机推进装置中常见部件——液力耦合器为例,研究液力耦合器混合模型的开发方法,验证混合建模方法的可行性。The hybrid modeling method based on neural network was proposed for the warship on-board training system（OBTS）.By means of integration of both reason-based modeling method and identification modeling method,the overall performance of the model can be assured better,also the accuracy of the model can be improved effectively.Taking hydraulic coupling,which can be frequently found in marine diesel propulsion plant,as an example,the developing method of hybrid model of hydraulic coupling was studied,and the feasibility of hybrid modeling method proposed was testified.

关 键 词：实船训练系统 混合建模 机理模型 辨识模型 神经网络 

分 类 号：U664[交通运输工程—船舶及航道工程]