Optimization Control Of CSPS System Under Production-Marketing Integration Model

In many real-world manufacturing enterprises, there is usually a type of production station situated along a constant-speed conveyor, by which some parts are randomly conveyed to the station for necessary processing. Generally, the station is equipped with a finite-capacity buffer to store temporarily the parts to improve flexibility. This production model is called a conveyor-serviced production station (CSPS), which originates from Ford’s assembly line. It can be viewed as an abstract model of present-day automated manufacturing processes such as robotic assembly lines. In fact, as an intellectual production model, CSPS has received wide use in real production. Therefore, the optimal control of such type of production model has practical significance and is an important issue in the field of industrial engineering.This study investigates the optimization control of a conveyor-serviced production station (CSPS) system under production-marketing integration model. Such system is viewed as a production center, which is connected to a sales center. The parts after processing will flow into the product bank of the sales center, and the whole two-center system is characterized by random part arrival, random customer demand, random processing time, and limited buffer or bank capacities. So, the decision-making on the look-ahead range of such CSPS under production-marketing integration model is subject to the constraints of production and sales levels, and in this paper we will focus on modeling the stochastic control problem and providing solutions on finding the optimal look-ahead control policy under either average-or discounted-cost criteria. In this part we first establish in detail a semi-Markov decision process for the look-ahead control of the CSPS under production-marketing integration model by combining the vacancies of both the buffer and the bank into one state, which can be solved by policy iteration or value iteration if the system parameters are known precisely. Then, to avoid the curse of dimensionality and the curse of modeling in the numerical optimization methods, we also propose a Q-learning algorithm combined with simulated annealing technique to derive the approximate solutions. Simulation results are finally presented to show that, by our established model and proposed optimization methods the system can achieve an optimal or suboptimal look-ahead control policy after that the capacities of both the buffer and the bank are designed appropriately.In real production, the service rate of production station is one of the key determinants of the system efficiency, thus to control the appropriate service rate can improve the productivity of the system, reduce waiting time and cycle time, which can make the system optimal under certain conditions. The optimization control of CSPS system under production-marketing integration model with changeable service rate is also concerned in this paper. We focus on modeling the stochastic control problem and providing solutions. First, the vacancies of the buffer and the bank are jointed to be viewed as the system state, and the look-ahead rang and service rate are viewed as the action variable. Then we establish in detail a semi-Markov decision process for the optimization control problem. So policy iteration and Q-learning algorithm combined with simulated annealing technique can be used to obtain the optimal look-ahead control range and service rate under either average-or discounted-cost criteria.