Detection and diagnosis of islanding using artificial intelligence in distributed generation systems. (March 2022)

• Record Type: Journal Article
• Authors: Mohapatra, Sushree Shataroopa; Maharana, Manoj Kumar; Pradhan, Abhilash; Panigrahi, Pratap Kumar; Prusty, Ramesh Chandra
• Is Part Of: Sustainable energy, grids and networks. Volume 29(2022)
• Publication Date: 2022-03
• Subjects: Distributed generation -- Eradicate Liability Passive Islanding Detection (ELPID) -- Micro grid -- Non-detection zone -- ANN -- Threshold -- Decision tree -- MLPNN; Renewable energy sources -- Periodicals; Smart power grids -- Periodicals; Electric power systems -- Periodicals; 333.794
• DOI: 10.1016/j.segan.2021.100576
• Languages: English
• ISSNs: 2352-4677

Abstract

Abstract: Renewable energy development technology has been profoundly involved and established over recent years. The primary method of using renewable energy is efficient by distributed generation (DG). These capabilities can be combined with a Microgrid as a hybrid energy device that optionally includes cooling or heating for electricity. Islanding is the biggest concern with such a DG. Islanding happens when, after the grid is disconnected, DG supplies load electricity. In this analysis, the strategies for the Eradicate Liability Passive Islanding Detection (ELPID) are used to detect islands from distributed generations that are Over–under voltage/Over–under frequency (OUV/OUF), Conversion rate of voltage (CROV), Conversion rate of phase angle difference (CROPAD), Conversion rate of frequency (CROF) and Point of common connection (PCC). They penetrate to produce enormous NDZ (Non-Detection Zone) and struggle to sense islanding with low or zero power imbalances. So, the islanding detection with imbalance conditions, NDZ, and mal-operation of the fault have to diagnosis, therefore using Artificial Neurological Network (ANN) can obtain the better-quality result. Therefore, the ANN techniques of Decision Tree characteristics and Multistage Perception Neurological Networks (MLPNNs) are used here. The DTC is making a decision and MLPNN will diagnose the fault with the training of the Backpropagation method. To obtain the outcome of power mismatch in reactive and active power without a calculation error, the combination of OUF/OUV, CROF, CROV, CROPAD in the ELPID method is perform. But for diagnosing the fault, the ANN techniques were used to get the result with an accuracy of 99.1%. The existing methods of SVM, Bagging, Random Forest (RF), and Decision Tree algorithm (DTA) has an accuracy range of 97.8%, 98.9%, 98.9%, and 83.33%. The suggested approach is to attain a better outcome when compare the previous approaches. The above process is examined using MATLAB/SIMULINK software tool to get the accurate result respectively.