Cloud-edge collaborated dust deposition degree monitoring for distributed photovoltaic systems

Abstract

After installation, the long-term efficient operation dominates the profitability of photovoltaic (PV) generation systems. Unlike large-scale PV plants, distributed PV systems are relatively dispersed, and belong to different owners, whose main obstacle for maximizing the profit is the lack of professional maintenance. This paper therefore proposes a cloud-edge collaborated dust deposition degree monitoring scheme for distributed PV systems without employing additional equipment, sensors or meteorological data, which just assumes the operational and historical data of all distributed PV systems in the same area to give the guidance whether the distributed PV system needs cleaning. In implementation, temporal and interactive models are established based on the operation characteristics of distributed PV systems, whose overall classification accuracy could reach more than 98%. In addition, in order to improve the adaptivity of monitoring scheme, this paper proposes a data-based random grouping method, which avoids the model mismatch caused by the addition or withdrawal of PV arrays. Finally, experimental verifications are carried out to prove the effectiveness of the presented scheme.© 2017 Elsevier Inc. All rights reserved.