基于梯度算法的低压台区线损异常实时检测方法

Real-time Detection Method of Line Loss Anomaly in Low-voltage Substation Area Based on Gradient Algorithm

针对传统线损异常检测方法缺乏规范的体系、智能化程度较低,为电网台区带来一定风险的问题,提出基于梯度算法的低压台区线损异常实时检测方法。利用数据挖掘算法挖掘低压台区线损异常数据,作为检测方法的数据支持,通过预处理线损异常数据恢复缺失值,引入梯度算法得到用于识别线损状态的梯度计算公式作为检测依据,建立一个低压台区线损异常实时检测模型。实例应用结果表明,该方法可准确识别低压台区线损异常原因,且具有96.6%的异常检测查全率。

The lack of standard system and low intelligence level of traditional line loss anomaly detection methods poses certain risks to power grid substations.In consideration of this problem,a real-time detection method for line loss anomaly in low-voltage substations based on gradient algorithm is proposed in this paper.Using the data mining algorithm,the abnormal line loss data of low-voltage substation area are mined,which then serve as the data support of the detection method.And the missing values are recovered by preprocessing the abnormal line loss data.By introducing the gradient algorithm,the gradient calculation formula is obtained,which can identify the line loss state and be used as the detection basis,and a real-time detection model for line loss anomaly of low-voltage substation area is established.The results of practical application show that the proposed method can accurately identify the cause of line loss anomaly in low-voltage substation area,and has a recall rate of 96.6%for anomaly detection.