摘要
科学合理进行数据处理是提高短期电力负荷预测精度的最基本环节之一。提出从采集样本数据到预测结果最终生成的数据处理流程;应用盆形曲线的体感温度标准化模型和模式识别96点相似采集样本技术进行采样前的标准化处理和样本选取;基于小波局部奇异性理论提出异常数据辨识与处理技术;利用最小平方误差法对预测曲线进行平滑处理。大量实际负荷数据分析和工程应用结果证明了此数据处理技术的有效性与实用性。
To process data scientifically and rationally is the most basic part to improve the load forecasting precision. This paper proposed a set of practical data processing flow from gathering sample data to the final forecasting result. The method applied the feeling temperature normalization models of the basin curve and sample technology of the pattern-recognition based on 96 points, unusual data processing approach based on singularity theory of wavelet. Finally, the curve was dealt with the minimal square error method. The validity and practicability of the approach were verified by the massive actual data analysis and project application.
出处
《电力需求侧管理》
北大核心
2007年第1期11-14,共4页
Power Demand Side Management
关键词
数据处理
相似度
模式识别
小波奇异性检测
曲线平滑
data processing
similar degree
pattern recognition
wavelet singularity detection
curve smoothness
