A Data-Mining-Based Methodology to Identify the Behavioural Characteristics of Prosumers within Active Distribution Networks

Abstract

The integration of the distributed small-scale generation, associated with the prosumers, affects the operational characteristics of the active distribution networks (ADNs), representing in the last years a great challenge for the Distribution Network Operators (DNOs). Their penetration degree and characteristics of the injected active power must be known as good as possible to evaluate the effects on the operation of the ADNs. Based on these considerations, a data mining-based methodology is proposed to identify the real behavioural characteristics of prosumers, represented by the injected active power typical profiles (IAPTPs), within the ADNs. Inside the methodology, the K-means clustering method was used to obtain the IAPTPs of the prosumers. The testing was done using a database with 64 prosumers, having PV systems installed, that inject the active power in the LV electric distribution networks of the DNO from the north-eastern region of Romania. The proposed methodology represents an efficient solution to identify easy and quickly the real behavioural characteristics of the prosumers, helping the DNOs to evaluate the impact on the operation of the ADNs.