Research On Planning Problem With Different Inventory Strategy

Production planning problem is a very important problem in the process of production management, and it is the foundation of making the delivery efficient. This article discusses the solution of production planning problem in modern petrochemical industry. In modern petrochemical industry there are three main features:a) the inventory standard is high; b) the supply chain structure is complex; c) the customer's demands change frequently. The features mentioned above mean that new method has to be developed, and hence we remodeled this kind of problem and developed new efficient heuristic algorithm.In production planning process, materials and production capacities are resources which need to be planned. In this paper we discussed the capacitated problems with different inventory strategy using the Quantitative Methods, and the final purpose is providing the theory and method for production support.Firstly, we summarized the research on production planning and based on this, we presented models of capacitated planning with different inventory strategy. For the outsourcing problem and stockout problem, we provide problem-specific heuristic algorithm.Secondly, we improved the traditional genetic algorithm to make the algorithm more efficient in solving the problem considered in this article. For general capacitated outsourcing problem, we use real coding strategy to overcome the complexity of the cost function. In the paper the consolidated process of the algorithm is introduced and the improvement of some operations are detailed. For stockout problem, we illustrate an important feature of the optimal solution and according to this feature, a problem specific binary-coded genetic algorithm is developed, in which, we designed a smart decoding strategy.Finally, we designed the simulation experiments, in which we use statistic method to testify the searching efficiency, including the searching accuracy and the computing time, and the testing results show that the two genetic algorithms can satisfy actual industry requests well and have strong potentials.