电网物资需求预测中区间估计的应用研究
Research on Interval Estimate Application of PGM Demand Forecasting

在电网工程物资管理中，需求预测的有效性十分重要。有效的需求预测可以防止采购过剩和供应不足。统计分析是常用的需求预测方法，其中尤以回归分析应用较为成熟。简单回归分析仅给出一个结果数值，而不反映预测的可靠性。对于电网物资供应链决策而言，由于资金和库容条件限制，物资采购总量只能维持在安全区间。管理者在面临需求不确定性时，需要明确当前采购量的可靠性。本文基于区间估计建立回归预测模型，并给出一些应用案例。预测模型同时返回需求区间和可靠性，这样需求不确定性被量化为风险概率，为供应链管理提供二元决策模型。二元决策模型的优点在于管理者可以根据可靠性目标分配电网物资。基于显著性检验方法，二元决策模型也可以衡量某一段时间内协议库存满足率。
The effectiveness of demand forecasting is crucial for power grid material (PGM) management. It contributes to both surplus purchase and short supply prevention. Statistical analysis is the most common method for demand forecasting. Especially, regression is rather apply proven. Simple regression analysis only returns amount as a result which does not reflect reliability of demand forecasting. In terms of PGM supply chain decision, due to the limitation of budget and capacity the purchase amount must remain on the safe side. The managers need clarify the reliability of current amount when they face the uncertainty of demand. This paper establishes a regression forecasting model based on interval estimate and gives a few application cases. Forecasting model simultaneously returns the interval and reliability about demand forecasting. So that the uncertainty of demand being measured as risk probability affords binary model for supply chain management. The advantage of binary model is that managers could distribute PGM according to the reliability target. Binary model also could measure the probability of agreed stock fulfillment.