Prediction-driven Optimization Of Production-sales-stock Decision-making Of Refined Oil Enterprises

In recent years,with the increasing complexity of the political,economic and social environment,the price and demand of refined oil products are facing increasing volatility.For refined oil production enterprises,how to avoid the risk of price and demand fluctuations to the greatest extent,and make optimal production-sales-stock decision-making,has become a top priority.For this purpose,in view of the opportunities and challenges in the refined oil market at the current stage,this paper will investigate the optimization strategies of production-sales-stock decision making of refined oil production enterprises under the new background of the gradual opening up of the refined oil market from the perspective of price and demand forecasting.Among them,how enterprises should reasonably use the existing relevant data to accurately predict the price and market demand of refined oil products,and how to effectively optimize the production-sales-stock decision-making of refined oil products in terms of the prediction results,is the core issue of this paper.The main research contents of this paper mainly include the following five aspects:(1)For the refined oil price prediction,this paper constructs a memory-trait-driven decomposition-ensemble forecasting model.The model includes three main stages:memory-trait-driven data decomposition,memory-trait-driven component prediction,and memory-trait-driven ensemble output of component prediction results.Using this model,the market prices of two common refined oil products,93# gasoline and 0# diesel oil,were predicted and analyzed.Empirical results show that the memory-trait-driven decomposition-ensemble model can obtain better prediction results than the single models listed in this study.(2)In the demand forecasting of refined oil products,this paper constructs a trend-and periodicity-trait-driven decomposition-ensemble forecasting model.The model includes four main stages: trend-and periodicity-trait identification and testing,periodicity-trait-driven data decomposition,trend-and periodicity-trait-driven component prediction,and ensemble output of component prediction results.Using this model,the market demands of two common refined oil products,93# gasoline and 0# diesel oil,were predicted and analyzed.The empirical results show that the prediction accuracy of the decomposition ensemble model with different prediction techniques is higher than that of the single models and decomposition-ensemble models with the same prediction techniques for the component.This shows that the selection of the suitable prediction techniques in terms of specific data traits can effectively improve the prediction accuracy.(3)In the sequential optimization of production-sales-stock decision-making,this paper constructs a n-period sequential optimization model for production-sales-stock decision-making of refined oil enterprises based on the prediction results of refined oil price and demand.The analytical results have three main findings.First of all,in any period of decision-making,the optimal inventory replenishment time point is monotonically increasing in terms of production cost,inventory cost and the total duration of any period,but not related to sales incentives.As the price prediction value increases,the enterprise will advance the ordering point.If the price is predicted to be higher than the actual value,the enterprise will delay the ordering point,and the more higher than the actual value,the more obvious the delay effect will be.Second,the sales incentive and expected profit of the enterprise increase with the increase of the price prediction value,and the marginal increase of this increase effect.Moreover,the sales incentive of the enterprise weakens with the increase of the demand prediction value and the relative prediction error,and this weakening effect increases marginally.When price and demand predictions increase at the same time,the strengthening effect of the above sales incentives will mask their weakening effects,resulting in an overall upward trend in sales incentives.Finally,the output of enterprises increases with the increase of price and demand prediction values,but the growth in demand is relatively conservative and the growth in price is more aggressive.(4)In the simultaneous optimization of production-sales-stock decision-making,this paper constructs a n-period simultaneous optimization of production-sales-stock decision-making of refined oil production enterprises in terms of the price and demand prediction results of refined oil products,and uses the particle swarm optimization(PSO)algorithm to solve the approximate optimal solution of the simultaneous optimization of production-sales-stock decision-making of refined oil enterprises in each period.The analytical results reveal the following four main findings.First of all,in the solution results of any period,the sales incentive amount of the enterprise will increase with the increase of the price and demand prediction values,but the price prediction value has a stronger impact on the sales incentive decision,while the demand prediction value has a weaker impact on the sales incentive decision.Second,the production of the enterprise increases with the increase of the price and demand prediction values,but the demand prediction value has a greater impact on the production decision,and the price prediction value has a relatively small effect on the production decision.Third,the time point of inventory replenishment of enterprises will advance with the increase of price and demand prediciton values,and this advance effect is more obviously affected by the demand prediction value.Finally,the expected profit of the enterprise decreases with the decrease of price and demand prediction values,and this reduction effect is more affected by the price prediction value.(5)In the comparison of sequential optimization strategy and simultaneous optimization strategy of production-sales-stock decision-making,we can get the conclusion that the profit of refined oil enterprises is equivalent under the two optimization strategies.Moreover,there is no significant difference in production decisions between simultaneous optimization and sequential optimization,but the sales incentive is higher and the replenishment time point is earlier.For this purpose,this paper extends the basic model of production-sales-stock decision-making,and expands the generic model from the situation of one refinery facing one oil depot to the situation of one refinery facing two oil depots.The analytical results found that compared with the underlying model,sales incentives,production and profits of the extended model are higher,and inventory replenishment is earlier.However,on the whole,the enterprise profits of the extended model are very close to the sum of the profits of the two oil depots under the basic model.This shows that in the practical operation,sequential optimization and simultaneous optimization are feasible strategies for the optimization of production-sales-stock decision-making of refined oil enterprises.