摘要
物资需求预测过程中忽略了季节性因素对物资需求量的影响,导致预测结果与物资实际用量产生偏差。为此,提出基于历史数据挖掘的电网建设物资需求预测系统。确定典型项目所需物资,按照影响因素权重,赋值历史数据,将电网物资训练数据预处理后分离;采取神经网络多次迭代输出局部最优解,通过浮动偏差,修正局部最优解,输出物资需求数量预测结果。预测110kV新建线路工程所需的14种物资,实验结果表明,在保证预测稳定性和预测效率的基础上,所提方法降低了物资预测值偏差度,预测需求量与物资实际用量更加接近。
In the process of material demand forecasting,the influence of seasonal factors on material demand is ignored,which leads to the deviation between forecasting results and actual material consumption.Therefore,a forecasting system of power grid construction material demand based on historical data mining is proposed.The result is done by determining the materials required for typical projects,assigning the historical data according to the weight of influencing factors,and separating the training data of power grid materials after preprocessing;then,neural network is adopted to output the local optimal solution for many iterations,and the local optimal solution is modified through floating deviation to output the material demand quantity prediction results.The experiment results show that the proposed method reduces the deviation degree of material forecast value,and the predicted demand is closer to the actual material consumption on the basis of ensuring the prediction stability and prediction efficiency.
作者
丁靖
林明晖
陈凌
DING Jing;LIN Ming-hui;CHEN Ling(State Grid Ningbo Electric Power Supply Company,Ningbo 315000,Zhejiang Province,China)
出处
《信息技术》
2021年第9期144-148,154,共6页
Information Technology
关键词
数据挖掘
电网建设
物资需求
历史数据
偏差度
data mining
power grid construction
material demand
historical data
accuracy
作者简介
丁靖(1986-),男,硕士,工程师,研究方向为物资管理、物资采购标准管理、物资供应链管理等。
