基于k-means聚类和模糊神经网络的母线负荷态势感知

Bus load situation awareness based on the k -means clustering and fuzzy neural networks

为顺应电力调度计划朝更精细化方向发展,提出基于k-means聚类和模糊神经网络的母线负荷态势感知方法。首先提出表征母线负荷状态参量和体现其状态参量变化趋势的母线负荷静动态势概念,然后建立母线负荷态势感知方法,包括:在态势觉察阶段,对母线历史负荷态势信息进行采集和处理;在态势理解阶段,采用基于手肘法的k-means聚类算法对考虑母线环境因素和负荷因素的母线历史负荷态势信息进行聚类分析;在态势预测阶段,采用费歇尔判别分析针对待测日动态势信息进行分类预测匹配待测日所属历史数据聚类类别,将所属类别的历史静态势数据代入模糊神经网络预测模型,建立基于k-means聚类的模糊神经网络预测方法,对待感知日母线负荷进行态势预测。最后应用该文方法进行算例仿真,结果表明所提方法的有效性和可行性,同时与传统模糊神经网络预测相比,该文母线负荷态势感知方法具有更高的态势预测精度。

In order to refine the power dispatching plan,a load situational awareness method is proposed for the bus in the basis of the k-means clustering and fuzzy neural networks.Firstly,the concept for the static dynamic potential of bus load is proposed.It characterizes the bus load state parameter and the trend of its state parameter change,and then the bus load situational awareness method is established.This method collects and processes the historical load situation information of the bus in the situational awareness stage.In the situation understanding stage,it adopts the k-means clustering algorithm based on the elbow method which clusters the historical load situation information of the busbar considering the bus environmental factors and load factors.In the situation prediction stage,the Fisher discriminant analysis is utilized to classify the dynamic information of the day to be measured and predict its category of historical data clustering.Then,the historical static potential data of the category is substituted into the fuzzy neural network prediction model to predict the situation of the perceived daily bus load.Finally,a simulation is included to verify the effectiveness and feasibility of the proposed method.It is shown that comparing with the traditional fuzzy neural network prediction,the proposed bus load situational awareness method has the higher situation prediction accuracy.