基于GA-SVM的多规格小批量生产的装配工时估算模型

An Assembly Man-Hour Estimation Model Based on GA-SVM for Multi-specification and Small-Batch Production

对于制造承包商来说,在正式接收订单之前,为了指导报价和预测交货日期,有必要对工时(Man-hours,MH)进行评估。装配工时作为工时的重要组成部分,具有重要的实际研究意义。针对多规格、小批量生产的特点,提出了一种基于支持向量机(Support vector machine,SVM)的装配工时估算模型。除了单部件属性、装配过程和历史工时数据外,还考虑了可量化装配复杂性的最短路径长度平均值(Average of shortest path length,ASPL)作为装配MH的影响因素,并提出了基于Creo JLink三维模型的这些因素的自动计算方法。通过对几种算法的比较,选择SVM作为装配体MH建模的最优算法。将遗传算法(Genetic algorithm,GA)应用于SVM中,有利于在SVM中搜索最优参数c和g时避免了局部求解,加快了收敛速度。最后,对所提出的GA-SVM模型进行训练,并应用于雷达装置仿生腿的装配工时预测。实验结果表明,GA-SVM具有比本文其他方法更高的预测精度,整个预测过程仅需3 min左右。

It is necessary to evaluate man⁃hour(MH)before receiving the order to guide the quotation and forecast the delivery date for a manufacturing contractor.As an important part of assembled MH,it has important practical significance.Aiming at the characteristics of multi-specification and small-batch production,an assembly MH estimation model based on support vector machine(SVM)is proposed.Apart from single component attributes,assembly process,and historical MH data,we also consider the average of shortest path length(ASPL),which quantifies the complexity of an assembly,as influencing factors of assembly MH.Furthermore,the auto calculating methods of these factors based on 3D models with Creo JLink are proposed.Through the comparison of several algorithms,SVM is chosen as the optimal algorithm for assembly MH modeling.Genetic algorithm(GA)is used to avoid the local solution and accelerate convergence when searching for the optimal parameters of SVM(c and g).Finally,the proposed GA-SVM model is trained and applied to predict the assembly MH of the bionic leg for the radar device.Experimental results show that GA-SVM has higher prediction accuracy than other methods in this paper and the whole predicting process only takes about 3 min.