摘要
为了快速计算船体分段装配的工时定额,提出了人工神经网络及工作研究的方法.将船体分段的总重量、分段装配总长度、分段类型,以及分段所含的小组立个数、中组立个数等作为神经网络的输入,分段的装配总工时作为输出训练网络,利用此网络估算装配工时.在工作研究的方法中,首先将分段装配的整个过程按加工阶段进行工种细分,建立各细分工种的工时定额计算模型.根据规则关联相应的计算模型,实现了船体分段装配工时的智能计算.实验证明,神经网络估算方法具有早期估计、快速计算的特点,而基于定额模型的装配工时智能计算结果准确度较高,但计算模型的建立较为困难.
In order to calculate the ship block assembly man-hour total quickly,the neural network and the work study methods were proposed in this paper.First,the total weight and assembly length of the ship blocks,the block type,the number of small assembly,and the subassembly were taken as the inputs of a neural network.Next,the total block assembly man-hour total was taken as an output,which was then used to train the neural network.Last,the assembly man-hour total was calculated using this trained network.When using the work study method to calculate the man-hour total,the whole process of the block assembly was divided into different sub-divisional sections first,and then the man-hour calculation models were built for each section.Lastly,the detailed rules were defined;using these rules the system was able to find the proper model and calculate the block assembly man-hour total intelligently.By actual measurements,it was proven that the neural network method has the characteristics of early estimation and quick calculation;on the other hand,the assembly man-hour total calculated by the intelligent system based on the models is more accurate,but those models are difficult to establish.
出处
《哈尔滨工程大学学报》
EI
CAS
CSCD
北大核心
2012年第5期550-555,共6页
Journal of Harbin Engineering University
基金
国家自然科学基金资助项目(70971085)
关键词
船体分段
装配工时定额
人工神经网络
ship block
block assembly
assembly man-hour
neural network
