Optimized Supply Planning Problem With CVaR

Under undetermined demand, making supply plan is very important for every enter-prise of supply chain. A reasonable and feasible supply plan is necessary for the enterprise.Supply plan is a multi-objection problem. It has to make the cost, profit and riskto reach the best condition simultaneously. Meanwhile, since the problem bases on theundetermined demand of the market, it's undetermined too. This undetermined characteraffects each index which the enterprise cares. With setting up multi-objection randomplan model, we can describe the essence of supply plan in mathematics. It makes thesolution more scientific and effective.The first problem to solute is the random demands. In this paper, we introduce theMont Carlo Simulation which simulate the future demand of the market with the historicdata (expectation, variance, relate coefficient and so on). Before all of this, it is needto obtain the uniform random number on [0, 1]. The module 2~ωpseudorandom numbergenerator has developed a lot since 1960s when came up with. The MT algorithm isused widely due to its long-period, high decision and good random performance in thehigh-dimension. In this paper, it will be used.Traditionally, there are many classical methods to solve the multi-objection plan.However, the performance of them is not satisfying. Evenly, as this is a random problem,it is more difficult. Thus, in this paper, the genetic algorithm is induced in. The comparingstrategy introduced in the paper would help the decision makers to select the best solution.In this paper, the method of CVaR(Conditional Value-at-Risk) is introduced. Theopportunity loss in a stage multiply its probability, which obtained through historic data.In this way, the infeasible opportunity loss value will be controlled greatly.After all the discussion, a numerical experiment is showed to prove: To this multi-objection problem under undetermined demand, the method in this paper has made abetter effect. Applying CVaR onto the opportunity loss, we can raise the max profit,which means that this method is more closed to the truth and predict more accurately.