| Article ID | Journal | Published Year | Pages | File Type |
|---|---|---|---|---|
| 1550732 | Solar Energy | 2012 | 16 Pages |
This paper presents an intelligent sizing technique for sizing grid-connected photovoltaic (GCPV) system using evolutionary programming (EP). EP was used to select the optimal set of photovoltaic (PV) module and inverter for the system such that the technical or economic performance of the system could be optimized. The decision variables for the optimization process are the PV module and inverter which had been encoded as specific integers in the respective database. On the other hand, the objective function of the optimization task was set to be either to optimize the technical performance or the economic performance of the system. Before implementing the intelligent-based sizing algorithm, a conventional sizing model had been presented which later led to the development of an iterative-based sizing algorithm, known as ISA. As the ISA tested all available combinations of PV modules and inverters to be considered for the system, the overall sizing process became time consuming and tedious. Therefore, the proposed EP-based sizing algorithm, known as EPSA, was developed to accelerate the sizing process. During the development of EPSA, different EP models had been tested with a non-linear scaling factor being introduced to improve the performance of these models. Results showed that the EPSA had outperformed ISA in terms of producing lower computation time. Besides that, the incorporation of non-linear scaling factor had also improved the performance of all EP models under investigation. In addition, EPSA had also shown the best optimization performance when compared with other intelligent-based sizing algorithms using different types of Computational Intelligence.
► Intelligent-based algorithm is faster than the iterative-based algorithm. ► Without NLSS, Meta-EP produced lowest error when compared to other EP models. ► With NLSS, Meta-EP produced lowest error when compared to other EP models. ► Incorporation of NLSS had improved the performance of all EP models. ► Meta-EP with NLSS produced lowest error when compared to AIS and GA.
