摘要
利用BP神经网络训练数据,以深松机关键部件的三个结构参数—翼张角、刃角和翼倾角为输入参量,输出目标参量为牵引阻力,建立深松机牵引阻力的人工神经网络模型。结果表明:训练良好的BP网络输出数据与实测数据吻合较好,网络模型具有较高的精度,并具有收敛速度快等特点。同时,确定了一组最优结构参数,为深松铲的设计提供理论依据。
The paper establish artificial neural networks model of deep -tiller＇ s tractional resistance with three structure parameters （wingspan angel.edge angel.wingluff angel） of the key parts of deep-tiller as the input and tractional resistance output when training and learning datum utilizing BP neural networks. The results indicated that： better output datum trained by BP neural networks inosculated experiment datum, the networks model was possessed the characteristics of better precise and rapid constringency speed. At the same time, the optimal construct parameters were obtained which offered a reference basis of optimizing design for deep-shovel.
出处
《黑龙江八一农垦大学学报》
2008年第3期37-39,共3页
journal of heilongjiang bayi agricultural university
基金
黑龙江省大庆市课题(SGG04-063
SGG2005-027)。
关键词
BP神经网络
深松机
关键参数
BP neural networks
deep-tillage
critical structure parameter
