基于BP神经网络的焊接工时定额模型研究  

Study on model of welding man-hour quota based on BP neural network

作　　者：于付龙 仲梁维[1] 李阳[1] 

机构地区：[1]上海理工大学机械工程学院,上海200093

出　　处：《现代制造工程》2014年第4期17-21,共5页Modern Manufacturing Engineering

基　　金：国家自然科学基金资助项目(71171136);上海市(教委)重点学科建设资助项目(J50503)

摘　　要：为了提高焊接工时定额的计算效率和精确度,消除传统定额方法过度依赖线性公式进行计算而忽略工时定额与其加工工艺参数之间的非线性关系的不足,将BP神经网络技术应用到焊接工时定额中,建立多输入单输出的神经网络模型,并运用实际数据对建立的神经网络进行训练,通过测试样本验证了该网络模型输出的仿真值与实际值之间的误差在5%以内。验证结果表明,用BP神经网络方法建立的焊接工时定额模型能够对焊接工时进行准确的预测。In order to improve the calculation efficiency and accuracy of welding man-hour quota ,and to eliminate the traditional quota method relied on linear formula excessively but ignored the lack of a non -linear relationship between man-hour quota and their processing technical parameters .The BP neural network technology is applied to the welding man-hour quota ,the establish-ment of a neural network model of multi-input and single-output,and the training on the establishment of the neural network ap-plying actual data ,the error between the simulation output values of the network model and actual values is validated within 5%by testing samples .The verification results show that establishing welding man-hour quota model based on the BP neural network method can predict welding man-hour accurately .

关 键 词：神经网络 焊接 工时定额 

分 类 号：TP391.7[自动化与计算机技术—计算机应用技术]