Queuing Network Modeling And Buffer Allocation Of A Manufacturing Cell With An Automatic Transporting Robot

Buffer allocation optimizations in robotic manufacturing cells is an important research topic.When planning and designing manufacturing cells,not only investment costs but also system performance should be considered.However,there is no closed expression of the system performance for the buffer allocation optimization problems of this kind of stochastic systems,so an effective method for system performance evaluation is a basic premise to solve the optimization problems.Aiming at a manufacturing cell with an automatic transporting robot in a stochastic environment,this paper establishes the corresponding queueing network model.Due to restrictions of finite capacities and general distributions,this queueing network model has no product solution,so it is difficult to solve.On the other hand,existing methods have limitations and cannot solve this queueing network model effectively.Therefore,considering the characteristics of the queueing network model of the robotic manufacturing cell,this paper proposes an approximate method for efficient evaluation of system performance and applies this method in the buffer allocation optimization problem of the robotic manufacturing cell.Firstly,for a robotic manufacturing cell with general distributions,an open queueing network model with dual resource constraints,resource sharing and finite capacities was established to describe the problem.Two-moment parameters of general distributions were used to describe the input and output processes of each queueing node.For the dual resource constraint problem with general distributions,existing literature has proposed approximate solving algorithms.For the resource sharing problem with general distributions,this paper proposes a two-moment approximation method to estimate system performance.For a multi-stage system with finite capacities and general distributions,this paper proposes an aggregation method to evaluate system performance.Based on them,for the queueing network model of a robotic manufacturing cell,an aggregation method based on two-moment approximations was proposed to evaluate system performance.Next,a series of numerical experiments are carried out,which verify the feasibility and accuracy of the proposed queueing network solving algorithm.For the resource sharing synchronization station,a corresponding simulation model is established to verify the effectiveness of the queueing network approach.For the multi-stage system with finite capacities,the validity of the queueing network method is verified by comparing this method with other methods in the existing literature.For the robotic manufacturing cell,a corresponding simulation model is established.By comparing statistics of steady-state system performance in the simulation model with evaluations of steady-state system performance in the queueing network model,the validity and accuracy of the approximate queueing network approach are verified,and steady-state system performance of the robotic manufacturing cell is analyzed.Finally,for the buffer allocation optimization problem of the robotic manufacturing cell,a stochastic nonlinear integer programming model is established.This optimization model requires that the average throughput and average cycle time of the robot manufacturing cell meet constraints and minimizes the investment cost of buffer allocation.For this stochastic nonlinear integer programming model,since the system performance in constraints cannot be expressed in closed-form by system parameters,a simulated annealing algorithm embedding the queuing network model is proposed to solve the problem,and the feasibility of the solving algorithm is verified by variant optimization cases.Based on queuing network theories and basic principles of stochastic processes,this paper models a manufacturing cell with an automatic transporting robot,evaluates the system performance,and optimizes the buffer capacity allocation.The research results of this thesis can expand the application of queuing network theories in stochastic manufacturing systems,provide important theoretical support for resource configuration of robot manufacturing cells,and provide a scientific method to reduce investment cost for long-term factory planning of customized manufacturing enterprises.