小波变换阈值降噪在电力负荷管理终端中的应用

Application of Wavelet Transform Threshold Noise Reduction in Load Management Terminal

负荷管理终端所采集的用户侧数据是必不可少的一手基础资料,而且要及时准确。但是,由于用户侧会发生如采集通道故障、电磁干扰、冲击负荷等原因会使负荷数据产生脉冲状尖刺、阶跃状或锯齿状波动等噪声数据。目前对这些异常数据,终端会不加辨别地按照冻结时刻冻结并送给主站,而有可能将合理的数据丢掉。这些异常数据将以伪信息、伪变化规律的方式提供给负荷预测作为参考,必然误导负荷预测模型的建立,影响预测结果的精确度及可靠性。应用小波理论对电力系统负荷管理终端采集的负荷数据进行降噪处理,采用不同的阈值量化方法对降噪后的信号进行了比较和分析。通过实验证明,Sqtwolog阈值降噪方法未出现明显的失真又较好地保留了信号中的有用部分,负荷管理终端将这些数据进行必要的降噪处理,效果明显,为提高负荷预测的精度奠定了基础。

The basic function of load management terminal is data collection,but the data are always stained when the collection channels have malfunction or are interfered by electromagnetic waves, sometimes impact load will make pulse, step or zigzag wavcform noise. Currently these abnorroal data will be selected by time interval and send to the master station and not to discern true or false, some normal data may be abandoned. All these abnormal data can give false information to period load prediction, which will firstly mislead the setting up of load prediction module by all means, then will affect the precision and reliability of prediction result. In this paper, wavelet theory is used to reduce noise of power load data and several threshold quantization methods are analyzed and compared. In the end,draw a conclusion that Sqtwolog threshold noise reduction method can keep the useful information of old data and have not via experiment.