考虑高频数据V-I特性的电力负荷异常值自动识别系统

Automatic identification system of power load outliers considering V-I characteristics of high-frequency data

为解决现存电力负荷异常值自动识别系统因低频采样数据包含的数据信息较少、容易陷入过度训练的问题,设计了一种考虑高频数据V-I特性的电力负荷异常值自动识别系统。系统分为硬件和软件两部分,硬件设计中对信号数据的处理电路进行优化,结合放大电路、过零检测电路、整形电路,完善了系统负荷数据的初步处理;软件设计中对负荷高频数据V-I进行聚类、清洗、归一化处理,得到负荷特征曲线,求出带通矩阵和上下阈值,构建异常负荷识别流程,完成电力负荷异常值自动识别系统设计。仿真实验结果表明,所设计系统的识别错误率和识别相似率较低,但与原系统相比得到了提升,能够有效缓解过度训练问题。

In order to solve the problem that the existing automatic identification system of power load outliers is easy to fall into over training and cause to the low-frequency sampling data contains less data information, an automatic identification system of power load outliers considering the V-I characteristics of high-frequency data is designed. The system is divided into two parts: hardware and software. In the hardware design, the signal data processing circuit is optimized, and the preliminary processing of system load data is improved by combining amplification circuit, zero crossing detection circuit and shaping circuit. In the software design, the load high-frequency data V-I is clustered, cleaned and normalized, the load characteristic curve is obtained, the band-pass matrix and upper and lower thresholds are obtained, the abnormal load identification process is constructed, and the design of power load abnormal value automatic identification system is completed. The recognition error rate and recognition similarity rate of the designed system are low, but compared with the original system, improved, which can effectively alleviate the problem of over training.