Reinforcement learning based Islanding detection technique in distributed generation

Jaiswal, Aashish; Chandra, Shubhash; Priyadarshi, Anurag; Sachan, Sulabh; Deb, Sanchari

doi:10.1016/j.egyr.2023.05.069

Cited by 4 publications

(2 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These clean and green energy resources include micro-hydro plants, fuel cells, wind farms, and solar power modules. However, these types of DGs are only reliable and productive in grid-connected mode when the national grid power is adequate [2,3]. Islanding detection requires special attention because unintentional islanding is one of the most dangerous situations in such GCDGNs [4][5][6][7].…”

Section: Motivation and Problem Statementmentioning

confidence: 99%

See 1 more Smart Citation

Triple‐indexed passive islanding detection strategy for grid‐connected distributed generation networks using an extended Kalman filter

Larik,

Li,

2024

IET Generation Trans & Dist

View full text Add to dashboard Cite

Islanding detection is a challenging issue in modern grid‐connected distributed generation networks (GCDGN). Generally, islanding detection has two categories local and remote, local schemes can be categorized into active, passive, and hybrid schemes. This article proposes a triple‐indexed passive islanding detection (TIPID) scheme using an extended Kalman filter (EKF). Initially, the EKF algorithm is applied on voltage signal at the point of the common coupling to estimate the desired fundamental and non‐fundamental features. The first index, known as the cumulative voltage logarithmic index, is computed by taking the natural logarithm of the fundamental voltage features to detect any variations in the GCDGN. The second index, known as the voltage differentiation index (VDI), is calculated from the fundamental voltage features, while the third index, known as the odd‐order harmonic distortion index (OOHDI), is obtained from the non‐fundamental odd‐order harmonics of the PCC voltage. Then, the VDI and OOHDI are compared to pre‐defined threshold to detect/distinguish islanding events. The proposed TIPID method is validated through extensive simulations on the IEEE 13‐bus test bed via MATLAB/Simulink 2022b. Results show that under both balanced/unbalanced load & generation, the proposed TIPID approach detects islanding occurrences with reduced non‐detection zone (NDZ) in less than 5 ms.

show abstract

Section: Motivation and Problem Statementmentioning

confidence: 99%

“…Another author in ref. [34] utilized a reinforcement learning algorithm for islanding detection in renewable energy-based distribution networks. Artificial neural networks in corporations with a rule-based mathematical model were deployed in [35] for islanding detection in microgrids.…”

Section: Extensive Literature Reviewmentioning

confidence: 99%