An innovative real-time framework for probabilistic load flow computation in renewable-based microgrids considering correlation: Integrating automatic data clustering with an enhanced arithmetic optimization algorithm

This paper suggests a new approach, called automatic data clustering, using an enhanced arithmetic optimization algorithm (ADC-EAOA), for improving the probabilistic load flow (PLF) in microgrids (MGs) integrated with distributed generation resources (DGR). The proposed ADC-EAOA method aims to address real-time solution of the PLF in MGs based on DGR with output large fluctuations, across various states of MG operation. The ADC-EAOA approach achieves more robust clustering compared to traditional data clustering methods by optimizing two objectives: a) maximizing cohesion within clusters, and b) enhancing separation between clusters. The correlation modeling between input random variables (IRVs) comprehensively has been considered. Five strategies have been introduced to improve the arithmetic optimization algorithm (AOA) to stop the algorithm from being extremely greedy and avoid trapping in local optima. The effectiveness of the EAOA is firstly assessed utilizing the test functions CEC2005 and CEC2019 in comparison to well-known optimization algorithms. Subsequently, the EAOA and the suggested ADC-EAOA hybrid approach are applied on 69-node, 33-node, and 123-node MGs to optimize the cost function used in the PLF which is equal to the mismatch among computation power with scheduled power in each node. The outcomes of solving the PLF problem were contrasted with those of the Monte-Carlo Simulation (MCS), the AOA, the PSO, the K-means, and as well as the Fuzzy C-means in terms of precision and run-time. The simulation outcomes demonstrate that the suggested ADC-EAOA approach dramatically increases the processing speed while simultaneously maintaining a high degree of accuracy so that it can provide various advantages and facilities for real-time power system studies.