Optimal planning and sizing of an autonomous hybrid energy system using multi stage grey wolf optimization
The continuous increase in energy demand and the perpetual dwindling of fossil fuel coupled with its environmental impact have recently attracted research focus in harnessing renewable energy sources (RES) across the globe. Representing the largest RES, solar and wind energy systems are expanding du...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2020
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/id/eprint/98225/1/SaniSalisuPSKE2020.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The continuous increase in energy demand and the perpetual dwindling of fossil fuel coupled with its environmental impact have recently attracted research focus in harnessing renewable energy sources (RES) across the globe. Representing the largest RES, solar and wind energy systems are expanding due to the growing evidence of global warming phenomena. However, variability and intermittency are some of the main features that characterize these RES as a result of fluctuation in weather conditions. Hybridization of multiple sources improves the system’s efficiency and reliability of supply due to the varying nature of the RES. Also, the unavailability of solar radiation (SR) and wind speed (WS) measuring equipment in the meteorological stations necessitates the development of prediction algorithms based on Artificial Intelligent (AI) techniques. This thesis presents an autonomous hybrid renewable energy system for a remote community. The hybrid energy system comprises of a photovoltaic module and wind turbine as the main source of energy. Batteries are used as the energy storage devices and diesel generator as a backup energy supply. A new hybrid Wavelet Transform and Adaptive Neuro-Fuzzy Inference System (WT-ANFIS) is developed for the SR prediction, while a hybrid Particle Swarm Optimization (PSO) and ANFIS (PSO-ANFIS) algorithm is developed for the WS prediction. The prediction accuracy of the proposed WT-ANFIS model was validated by comparison with the conventional ANFIS model, Genetic Algorithm (GA) and ANFIS (GA-ANFIS), and PSO-ANFIS models. The proposed PSO-ANFIS for the WS prediction is also compared with ANFIS and GA-ANFIS models. Also, Root Mean Square Error (RMSE), Correlation Coefficient (r) and Coefficient of Determination (R²) are used as statistical indicators to evaluate the performance of the developed prediction models. Additionally, a techno-economic feasibility analysis is carried out using the SR and WS data predicted to assess the viability of the hybrid solar-wind-battery-diesel system for electricity generation in the selected study area. Finally, a new cost-effective Multi Stage – Grey Wolf Optimization (MS-GWO) algorithm is applied to optimally size the different system components. This is aimed at minimizing the net present cost (NPC) while considering reliability and satisfying the load demand. MS-GWO is evaluated by comparison with PSO, GWO and PSO-GWO algorithms. From the results obtained, the statistical evaluators used for model performance assessment of the SR prediction shows that the hybrid WT-ANFIS model’s accuracy outperforms the PSO-ANFIS model by 65% RMSE and 9% R². Also, from the simulation results, the optimal configuration has an NPC of $1.01 million and cost of energy (COE) $0.110/kWh, with an operating cost of $4,723. The system is environmentally friendly with a renewable fraction of 98.3% and greenhouse gas emission reduction of 65%. Finally, a comparison is done between the proposed MS-GWO algorithm with the PSO, GWO and PSO-GWO algorithms. Based on this comparison, the proposed hybrid MS-GWO algorithm outperforms the individual PSO, GWO and PSO-GWO by 3.17%, 2.53% and 2.11% in terms of NPC and reduces the computational time by 53%, 46% and 36% respectively. Therefore, it can be concluded that the proposed MS-GWO technique can be applied for optimal sizing application globally. |
|---|