基于稀疏自编码降维的用电模式识别方法

Electricity consumption pattern recognition based on dimension reduction using sparse autoencoder

研究电力用户的用能模式,制定相应的激励政策,引导用户参与需求侧响应,对保证电网的安全经济运行和提升用户服务质量具有重要的意义。通过稀疏自编码器将高维的电力负荷数据进行降维,在二维空间中通过DBSCAN聚类发现离群点以剔除随机事件导致的数据干扰,基于动态时间规整距离的改进k-均值聚类分析,从而实现更准确和可靠的用电模式识别。通过3个指标对不同聚类方法的有效性进行了对比分析,结果表明:该方法能够提供更准确和稳定的聚类效果。

Study the energy consumption mode of power users and formulate corresponding incentive policies to guide users to participate in demand side response,so as to ensure the safe and economic operation of power grid and improve the quality of customer service is of great significance to.The dimension of high-dimensional electricity load data is reduced by sparse autoencoder,and outliers are found by DBSCAN clustering in two-dimensional space to eliminate the data interference caused by random events.An improved k-means clustering analysis based on dynamic time warping distance is carried out to realize more accurate and reliable power consumption pattern recognition.The effectiveness of different clustering methods is compared and analyzed through three indexes.The result shows that the proposed method can provide more accurate and stable clustering results.