基于小波分析的母线负荷预测不良数据检测

Bad data identification based on wavelet in bus load forecasting

母线负荷预测是电力系统调度运行的重要基础,而母线负荷预测中可能存在较多的不良数据,极大地影响预测精度。以准确检测母线负荷预测中的不良数据为目标,分析不良数据的来源,指出了数据奇异点与不良数据的关系,提出了基于小波分析理论进行不良数据检测的方法。理论分析表明,该方法具有完整、精确、准确及简单的特点,通过实验证明了该方法能够很好地识别增幅点、降幅点、突变点等典型的不良数据,有效提高预测精度。希望能够对母线负荷预测工作提供有益的参考。

Bus load forecasting is an important basis for power system operation. Bad data are quite common in bus load, which may greatly affect the prediction accuracy. Aimed at accurate detection of bad data in historical bus load, the source of bad data are first analyzed; then the relationship between the singular points and bad data are pointed out. A method based on wavelet analysis for bad data detection is proposed. Theoretical analysis shows that this method has several features such as integrality, precision, accuracy and simplicity. Experiments show that the method can identify typical bad data such as the growth points, drop points and the mutation points, thus effectively improve the prediction accuracy. The work will be helpful to the future development of the bus load forecasting.