Search

Summary: With the increasing access of distributed generation (DG), the investment decision of distribution network (DN) has become a large‐scale portfolio optimization problem with various reconstruction strategies, which reduces the applicability of the traditional investment decision optimization model. Therefore, aiming at reliability of DN, a novel deep belief networks (DBN)‐based correlation constraint‐integrated investment decision model is proposed in this paper. With the DBN‐based correlation constraint replacing the nonlinear and nonconvex constraints in the traditional model, a new investment decision model is established aiming at maximizing the reliability index and minimizing the total investment cost. In this way, the effects of different reconstruction strategies can be analysed, from which the optimal investment reconstruction plans are identified. Finally, an example of a regional distribution network in a city is provided to verify the rapidity, feasibility, and effectiveness of the investment decision model. [ABSTRACT FROM AUTHOR]

Cluster analysis is used to study the group of load profiles from smart meters to improve the operability in distribution network. The traditional K-means clustering analysis method employs Euclidean distance as similarity measurement, which is insufficient in reflecting the shape similarities of load profiles. In this letter, we propose a novel shape cluster method based on the segmented slope of load profiles. Compared with traditional K-means and two improved algorithms, the proposed method can improve the clustering accuracy and efficiency by capturing the shape features of smart metering load profiles. [ABSTRACT FROM AUTHOR]