短期负荷预测的实用数据挖掘模型

基于数据挖掘决策树算法和通用的决策支持对象(DSO)建模工具，结合区域电网气象负荷数据库设计与实现了决策树形式的数据挖掘模型并运用于日负荷预测。首先描述了决策树分类方法，分析研究了日负荷预测数据挖掘模型的决策树构造过程，给出了基于DSO的程序化实现，并给出用决策树算法的日负荷预测过程以及实际的预测结果情况。统计分析结果表明该数据挖掘模型完全满足实用标准，具有智能自适应、自学习和全过程自动化、通用可靠以及准确率高等特性。

APPLIED DATA MINING MODELS FOR SHORT-TERM LOAD FORECASTING

Based on decision tree algorithm of data mining and general decision support objects (DSO), an efficient model-building tool, this paper designed and implemented the decision tree formed data mining models and used them to the daily load forecasting in terms of the weather-load database of district power network. It first described decision tree classification theory, analyzed the constructing process of decision-tree data mining models for daily load forecasting and then presented their realization by programming with DSO. It further offered the whole process of load forecasting with the decision tree algorithm and its actual forecasting results. The statistic analysis showed that the data mining models totally reached the applied standard. It is intelligently adaptive, self-learning and automatic, reliable and highly accurate.