基于半监督机器学习算法的智能电网调度控制系统健康度评价

Health Evaluation of Smart Grid Dispatching Control System Based on Semi-supervised Machine Learning Algorithm

智能电网调度控制系统实时监控电力系统的发、输、变和配电等全过程,保证了电网安全稳定的运行,其健康度评价反映了整个电网运行情况。采用基于专家经验的评价方法虽然整体计算简单,但所得结果过于依赖主观因素,准确度不高。采用LinkedCop K-means (LCop K-means)算法对智能电网调度控制系统的健康度进行评价。首先介绍了智能电网调度控制系统的组成和主要功能,以及其健康度评价模型。LCopK-means算法属于机器学习算法中的半监督学习算法,因此在计算过程中并不会过分依赖于标签信息,具有更高的适用性及准确度。算例分析基于试验数据和实测数据,对比了LCop K-means算法和传统K-means算法,验证了所引入算法的高效性。

Smart grid dispatching control system monitors the whole process of power system, such as generation, transmission, transformation and distribution, to ensure the safe and stable operation of power grid, and its health evaluation reflects the operation of the whole power grid. Although the evaluation method based on expert experience is simple in overall calculation, the obtained results are too dependent on subjective factors, thus causing lower accuracy. This paper introduced the Linked Cop K-Means(Lcop K-Means) algorithm, and evaluated the health of smart grid dispatching control system. Firstly, the composition and main functions of smart grid dispatching control system and its health evaluation model were introduced. The LCop K-means algorithm belongs to the semi-supervised learning algorithm of machine learning algorithm, and does not rely too much on label information in the calculation process, so it has higher applicability and accuracy. Based on the experimental data and the measured data, the LCopK-means algorithm was compared with the traditional K-means algorithm, and the results verified the efficiency of the introduced algorithm.