Similar Instances Reuse Based Numerical Control Process Decision Method for Prismatic Parts
Objective With the rapid development of Intelligent Manufacturing,it has become a research hotspot in both industry and academia.Specifically,intelligent NC process decision-making is an essential supporting method for raising Intelligent Manufacturing,as it focuses on optimizing and deciding the NC process scheme for a part automatically and effectively.Since CAD/CAPP/CAM technologies have been widely applied in mod-ern manufacturing environments,abundant CAD models associated with NC process results have been generated and stored in repositories,impli-citly containing engineers'valuable knowledge and experience.As a result,if the existing NC process achievements can be fully reused,the effi-ciency and accuracy of NC process decision-making will improve significantly.In contrast,current NC process reuse methods focus on the fea-ture/subpart levels,indicating that research on NC process decision-making at the part level remains limited.Thus,the NC process reuse methods cannot be implemented directly on the shop floor,leading to significant time and labor waste.In addition,the existing NC process decision meth-ods,such as the genetic algorithm,ant colony algorithm,and particle swarm algorithm,have not been combined with the reuse approach.The NC process determination for features still depends on engineers'interactions.The diversity of NC processes is ignored,as the dynamic changes of manufacturing resources on the shop floor are not considered.Therefore,the flexibility and efficiency are relatively low and cannot meet the re-quirements for product innovation in manufacturing environments.A similar instance reuse-based NC process decision method for prismatic parts is proposed to overcome shortcomings.Methods Firstly,several concepts are introduced,including operation and process scheme digraph(PSD).The operation is defined as a basic unit consisting of the NC process and involves various characteristics of features.The PSD is a directed graph that describes the precedence of opera-tions for a part.Then,the framework of the proposed approach is illustrated,which explains four significant steps:similar subpart instances re-trieval,optimal target function construction,PSD establishment,and NC process scheme optimization.Then,the mathematical model of the eval-uation criterion,which considers manufacturing costs,is established.The usage cost of machines and tools and the alteration cost of machines,tools,and fixtures are considered,and the corresponding mathematical functions are provided.The total manufacturing cost is calculated as the weighted sum of the five types of cost.The adaptive generation of multiple NC processes for a new subpart is presented.The previous 3D model retrieval method retrieves and returns similar subpart instances with their NC processes.Considering the dynamic changes in shop floor resources,the detailed parameters of the NC process are modified adaptively,including operation type,processing stage,machine,tool,and tool approach direction.Specifically,flexible candidates for machines,tools,and tool approach directions are added to raise the flexibility of manufacturing.The similarity of each retrieval result is employed to determine the probability of being selected during the optimization iteration with the roulette wheel method.The flexible elements are chosen randomly except for the fixed operation type and processing stage.Once the NC process for each subpart of the new part is determined,the operations are mapped as the vertices of the PSD.The constraint rules of operation precedence are provided to demonstrate the execution priority of operations.Thus,the directed edges of the PSD are obtained,and the entire PSD is determined.Therefore,the optimization objective of the proposed approach is to search the PSD for the shortest path,which corresponds to the optimal NC process scheme with the minimum manufacturing cost.The ant colony and simulated annealing algorithms are combined to solve the problem due to their strong ability for graph searching and global convergence,respectively.Two concepts of graph theory are explained.Then,the basic ant colony parameters are calculated for the optimization objective.The search principle is illustrated:only the vertices with an in-degree equal to 0 can be regarded as the next destination to be visited,which means that the prior operations have been implemented and the precedence is satisfied.The PSD-based ant colony algorithm is proposed and described in detail to generate several locally optimal paths.Then,the paths are input into the simulated annealing algorithm to overcome local convergence and search for the globally optimal NC process scheme.Results and Discussions In the experiments,a subpart consisting of 3 features is utilized to describe the feasibility of adaptive generation of mul-tiple NC processes.A test part with the MBD model is adopted to verify the approach,which is made up of 8 subparts.The state of manufactur-ing resources is listed,and the precedence constraints of features'operations are analyzed.When the iteration times reach 100,the global optimal solution is obtained with the minimum manufacturing cost.The machining simulation of the global optimal NC process scheme is conducted with CATIA.The simulation result shows no overcut or interference occurs,and the dimensional accuracy satisfies the quality requirements of the part.Detailed analysis of tool paths for a subpart is implemented to confirm the approach's effectiveness.The proposed approach's decision efficiency and time consumption and the typical NC process decision method with CAPP software are compared.The proposed approach requires fewer than 150 human interactions,while the typical method needs more than 1 200 interactions to generate the NC process for subparts.This is because the NC process candidates are obtained adaptively through reusing similar instances.The NC process decision efficiency is improved by 84.6%.On the other hand,the manufacturing cost of the optimal NC process scheme is 16.6%lower.Conclusions The experimental results showed that the proposed approach can generate optimal NC process schemes for parts effectively and automatically,decrease production costs,and shorten the development cycle.It helps to raise the efficiency and intelligence level of NC process decisions for manufacturing environments.
numerical control process decisionreusesubpartprocess scheme digraph
万方数据
