An optimal deep belief with buffalo optimization algorithm for fault detection and power loss in grid-connected system

The recent increase in photovoltaic (PV) power generation and its extensive use worldwide has led to the development of complex distributed generation systems, which has caused an increase in PV faults. These defects lead to considerable power losses, significantly impacting the reliability and performance of the PV system. Several approaches have been implemented, but an accurate solution has not been found. Therefore, an optimal Deep Belief with Buffalo Optimization (DB-BO) algorithm is applied in the grid-connected system for detecting faults and regulating its classes. Moreover, principal component analysis is used to analyze the power loss issues, and linear discriminant analysis is utilized to mitigate the voltage deviation issues. MATLAB or Simulink is used as the implementation process, and simulation outcomes are compared with recent conventional models. It has been revealed that the developed DB-BO algorithm has reduced the power loss to 3.4 mW. Also, total harmonic distortion (THD) is improved compared to the existing security methods. Thus, the efficiency of the model that was built has been proven by getting the best results in accuracy, total harmonic distortion (THD), and power loss. The computation time of the proposed model (0.238 s) is compared with metaheuristic algorithms such as CSE (0.315629 s) and GWA (3.636 s).

Automated summary

The increase in PV power generation has led to complex distributed generation systems and an increase in PV faults. An optimal DB-BO algorithm is applied in a grid-connected system to detect and regulate faults. Principal component analysis and linear discriminant analysis are used to analyze and mitigate power loss and voltage deviation issues. The developed algorithm reduces power loss to 3.4 mW and improves total harmonic distortion compared to existing methods. The proposed model has proven to be efficient and has faster computation time compared to other algorithms.