Research On Intelligent Decision-making Methods For Discrete Operations And Dynamic Resources In Computer-aided Process Planning

Computer-aided process planning(CAPP)is an important tool for manufacturing industry to reduce machining cost,enhance process efficiency and improve production quality.The discrete decision-making for the processes happens in the configuration of all methods and resources involved in machining.However,CAPP is a decision-making system with complex procedures and numerous constraints,which cannot make process deicisons automatically.Aiming at promoting the decision-making capability of the CAPP system,this paper focuses on investigating the subtasks of the CAPP with decision-making difficulties.The research contents can be summarized as follows:Feature interactions may result in many process alternatives in part machining.This paper presents an optimization approach to handle the interacting feature recognition problem,by subdividing the removal volume into cells first and then combining the cells into features.A two-level feature generation method is developed.On the lower level,individual features are formed according to a given part face;on the upper level,the feature distributions are explored by rearranging the part face orders.An optimization model is proposed,which considers the factors of feature number,tool approaching direction,and cutting direction.The optimization model is solved with the simulated annealing algorithm.Setup planning and operation sequencing are two interrelated problems in CAPP.However,most existing methods treat them separately,which leads to the imbalance of the requirements between precision and cost.In this paper,a constrained optimization approach is proposed to address the setup planning and operation sequencing problems in a combined way.With an objective to minimize the manufacturing cost,the optimization model is generated with various constraints,especially the tolerance constraints on setup error,tool error,and tolerance accumulation.The evaluation algorithm for each kind error is then designed and a modified Particle Swarm Optimization algorithm is developed to search for the optimal solution.The manufacturing resource configurations in the process plans that are generated in advance often suffer from a revision,when they are subject to the last-minute production changes.Deep reinforcement learning is employed in this paper for process planning,aiming at promoting the response speed by exploiting the reusability and expandability of past decision-making experiences.The Monte Carlo sampling method is used to evaluate the decision-making policy,and the reinforcement learning algorithm based on Actor-Critic method is used to improve the policy.In addition,a policy network and an evaluation network are constructed for policy storage,and a gradient descent method is used for strategy training.Based on the above three tasks,this paper takes the UG software as a platform for the development of the CAPP intelligent system,which contains four modules,including feature recognition,setup planning,resources configuration,and the database.By implementing the process plan generated from the system,the efficiency of this system is validated.