Research And Application Of Improved Genetic Algorithm In The Multi-Echelon Inventory Optimization Problem Of Coal Enterprise

At present, the problems of the low communication level and the old supervisor mode are existed in coal enterprise. The inventory of each node enterprise in SCE (Supplier Chain Environment) mainly adopts the independent supervisor mode, so the phenomenon of "Each does things in his own way" appears, which leads to the unfeasibility of system consolidation of the inventory of supplier chain and high enterprise inventory cost. In order to control the inventory cost and reduce the stock, we established core enterprise, which coordinated the multi-inventory management between upriver and downriver enterprises under the supplier chain and made a decision of the inventory allocation scheme among each warehouse.Based on multi-echelon inventory system combined multi-warehouse of coal enterprises in SCE, the thesis analyzed the optimization problem of enterprise inventory allocation scheme, which made a further design and improvement on original inventory management system.First of all, the thesis established the inventory optimization model for the multi-inventory system of coal enterprise. Generally, original system only considered the effect of inventory cost. By practical research and analysis of coal enterprise, the effect of shortage time and shortage times to the enterprise credit was important and the effect of coal peculiar volatile factors to inventory cost was ubiquitous. For strengthening the practical value of the mathematical model, the thesis established the inventory optimization model, which considered the effects of four aspects factors of the inventory cost, the shortage time, the shortage times and the coal volatile factor.In the second place, the thesis developed the improved algorithm for the inventory optimization model and designed the experiment to prove the model and the algorithm. Generally, in the process of dealing with the large-scale problem, the slow convergence problem of GA (Genetic Algorithm) appeared. By simulating the practical inventory scheduling process of warehouse management personnel and combining with the characteristics of algorithm data structure, the thesis developed an ASHA (Artificial Simulation Heuristic Algorithm). The model was fixed by the method of ASHA embedded AGA (Adapt Genetic Algorithm). The simulation experiment results indicated that the ASHA greatly improved the feasibility and the efficiency of the generation of the initial population, which largely reduced the running time of algorithm, which made the improved GA have a qualitative breakthrough in dealing with the large-scale practical problem, thus fixed the problem of slow convergence on GA.Finally, based on the RSS (Reconfigurable System Solution) application development framework platform, the thesis developed the inventory management subsystem by using the technology with Flex RIA (Rich Internet Application), which researched on the feasibility of algorithm in practical application by an example of the application system. The thesis did some inventory optimization experiments under a certain scale by obtaining the real data of coal enterprise. The result indicated that the method of inventory optimization in this thesis had saved some inventory cost, which proved the correctness of the model and the availability of algorithm.