Manufacturing Supply Chain Planning Based on Extended State Task Network

Manufacturing supply chain planning is a key factor of manufacturing supply chain management. Assignment of the production tasks, inventory control and transportation between the plants or the enterprises needs to be considered in manufacturing supply chain planning. Extended state task network (ESTN) is proposed in order to abstractly describe the production tasks with complex product production processes (assembly process, distilling process, process with many kinds of input material and many kinds of output material), the store tasks, and the transportation tasks with different modes (one mode is that transfering only one kind of material and another several kinds of material). In the extended state task network, proportion transform task is used to abstractly describe the production task, the store task and the transportation task that transfers only one kind of material. The combination of virtual proportion transform task and the combination move task are applied to describe the transportation task that transfers several kinds of material. It is easier to encode and operate the solution of the manufacturing supply chain planning if metaheuristic methods are used to solve the problem based on ESTN than on other models. A path relinking algorithm is developed to solve manufacturing supply chain planning model based on ESTN. The algorithm maintains a reference set of solution with good quality in evolutionary process. A list of the new solution (path) is created by inserting properties of a guide solution into an initiate solution. The list is used to update reference set in order to evolute the solution. The solution update method of the reference set with diversification detection and decentralization mutation strategy is proposed in the path relinking algorithm. The results of computations demonstrate that the path relinking algorithm can obtain better solutions than the standard genetic algorithm, the standard Tabu search procedure and the general path relinking algorithm, and prove that the solution update method of the reference set with diversification detection is very important to path relinking. Moreover, the decentralization mutation strategy can enhance the capacity of searching a better solution.