Sizing optimization of large-scale grid connected photovoltaic system using dolphin echolocation algorithm / Muhammad Zakyizzuddin Rosselan

Despite being used as large-scale power plant, a common issue of Grid-Connected Photovoltaic (GCPV) system is the system sizing. As numerous models of system components are commercially available, the selection of optimal components has frequently become tedious and time consuming for system designers. Hence, optimization methods are regularly incorporated in the sizing algorithm for such system. This study presents the development of Dolphin Echolocation Algorithm (DEA)-based sizing algorithm for sizing optimization of large-scale GCPV systems. DEA was used to select the optimal combination of the system components which are PV module and inverter such that either the Performance Ratio (PR) or Net Present Value (NPV) is correspondingly optimized. Before incorporating the optimization methods, a sizing algorithm for large-scale GCPV systems was developed. Later, an Iterative-based Sizing Algorithm (ISA) was developed to determine the optimal sizing solution which was later used as benchmark for sizing algorithms using optimization methods. Besides DEA, Evolutionary Programming (EP), Firefly Algorithm (FA) and Cuckoo Search Algorithm (CS) were also incorporated in the sizing algorithm for performance comparison. For each sizing algorithm using these optimization methods, the optimal population size and the number of iterations for convergence were investigated. The results showed that the DEA-based sizing algorithm had successfully found the optimal PR and NPV for the system. Apart from that, sizing algorithm with DEA was also discovered to outperform sizing algorithms with selected computational intelligence, i.e. EP, FA and CS in producing the lowest computation time in finding the optimal sizing solution. Besides having more than 200 times faster than ISA, DEA was found to be approximately 2, 2, 13 times faster than EP, FA and CS respectively. Moreover, DEA was the only Computational Intelligence that is capable of finding the optimal PR and NPV as suggested by the benchmarked algorithm ISA.