Research Of Coal Distribution Optimization Of S Group

Coal has a very high ratio in our country’s primary energy production and consumption. The supply of coal resources is of great significance in our country’s economic construction. However, the distribution characteristics of coal resource and market lead to the "West-East coal transportation" and "North-South coal transportation" supply pattern, making railway transport capacity being the bottleneck of coal resource supply. S Group is a large energy enterprise, which has a dragon coal business including coal production, transportation and sale. It is the largest coal enterprise of China and the largest coal dealer in the world. The efficient operation of S Group’s coal distribution business is not only very important for itself in improving its management and risk-resist ability, but also of great importance to the normal operation of our country’s economy. After analyzing S Group’s three-stage distribution planning system and its coal distribution business operation rules, the thesis finds its current distribution mode has some problems, including extensive distribution planning, backward balance calculation method and lack of metric for business results. The above problems have not been solved. Therefore, it is necessary to do the research of coal distribution optimization for S Group.Aiming at the above problems, the thesis puts forward the idea for improvement, which is to improve technical means, study coal distribution optimization model, leading the making of coal distribution planning and decision to be more detailed and scientific. On the basis of study on current distribution model, the thesis focuses on S Group coal distribution’s specific situation, which is a multi-dimension mass transportation problem with multiple constraints, multiple targets and interval parameter, choosing the type of modeling, building a multi-objective nonlinear interval integer programming model, under the premise of determining the distribution network topological structure, input and decision factors, and modeling assumption. Then the thesis studies the methods of solving the model. After transforming the interval constraints into certain constraints by using the concept of interval inequality degree, the thesis selects an improved genetic algorithm based on integer coding, and makes creative improvement on population initialization, introducing a certain percentage of zero factors. By the introduction of zero factors, the proportion of zeros in initial group coding increases, which improves the quality of initial population, avoids the problem that randomly generated low-quality initial population increases difficulty of searching feasible solution. The improvement of the algorithm also decreases its iteration calculation steps and accelerates its convergence speed, making the algorithm more efficient and effective. Then the thesis eliminates the soft and hard constraints respectively by goal programming method and penalty function technique, making the model being suitable for using genetic algorithm. Finally, the thesis proves the validity of the model and algorithm through an example.