基于机器学习的典型零件特征加工方案智能决策研究

Research on Intelligent Decision Making Technology of Typical Parts Feature Machining Scheme Based on Machine Learning

针对目前典型零件特征加工方案决策过程主要依赖工艺人员经验、设计效率低下、加工知识提取与梳理过程比较困难、智能化程度较低等问题,提出一种基于改进蝗虫优化(IGOA)算法改进BP神经网络的特征加工方案智能决策方法:使用改进蝗虫优化(IGOA)算法方法对BP神经网络初始权值和阈值、学习速率以及反馈过程等进行优化,提高了神经网络的训练效率与准确性;通过对神经网络的节点数量、网络层数、学习策略的研究,建立了适应特征加工方案智能决策的IGOA-BP神经网络模型;最后以某船用柴油机上典型零件的历史特征加工方案数据进行实例验证,快速决策并生成了船用柴油机关键部件的特征加工方案结构树,证明了该算法及特征加工方案智能决策方法的可行性和有效性。

Aiming at the characteristics of typical parts processing scheme decision-making process mainly depends on process engineering experience,the design efficiency is low,processing and knowledge extraction and carding process more difficult,intellectualized degree is low,this paper puts forward a kind of based on improved locusts optimization algorithm(IGOA)the characteristics of the improved BP neural network processing scheme of intelligent decision-making method:The improved Locust optimization(IGOA)algorithm was used to optimize the initial weights and thresholds,learning rate and feedback process of BP neural network,which improved the training efficiency and accuracy of neural network.The number of nodes,network layers and learning strategies of neural network were studied,and the IGOA-BP neural network model was established to adapt to the intelligent decision of feature processing scheme.Finally,the historical feature machining scheme data of a typical part of a Marine diesel engine is used to verify the fast decision and generate the feature machining scheme structure tree of the key parts of Marine diesel engine,which proves the feasibility and effectiveness of the algorithm and the intelligent decision method of feature machining scheme.