Abstract
The demand for energy from photovoltaic (PV) system has been rising lately due to the economic and health benefits that solar energy offers. A complete standalone PV sys-tem comprises solar panels(s), dc-dc power converter, charge controller and a resistive load. Despite the merits with PV systems, solar systems are greatly disturbed by partial shading, sudden changes in climate conditions and the prediction of accurate cell pa-rameters that can guarantee maximum power extraction from PV systems is another challenge. These downsides lead to losses in the amount of energy that can be harvested from PV panels. On the other hand, Population-based metaheuristic techniques are branches of optimisation algorithms that are inspired by nature or swarm intelligence. Of recent, optimisation algorithms embedded in solar charger microcontrollers for global maximum power point searching tasks have shown better performances com-pared to online maximum power point tracking (MPPT) algorithms like perturb & ob-serve that suffers from power oscillation and local optima power detection during par-tial shading and offline methods like artificial neural network (ANN) that are problem-specific and expensive.
Still, optimisation algorithms suffer from premature convergence, weak exploration and exploitation, getting trapped at local optima, complexity, random solution per trial, slow convergence speed, and proper fine-tuning of hyperparameters required by some algo-rithms. Furthermore, population-based algorithms that comprise evolutionary algo-rithms and swarm intelligence algorithms currently lack a mathematical standard that can validate if the algorithm has truly searched for the global solution to an optimisation problem except through comparison with other algorithms results for ‘better solution”. Likewise, these limitations with optimisation algorithms tend to lower the efficiency of PV systems which in turn leads to significant loss in power and higher solar panel en-ergy payback period. Moth Flame optimisation (MFO), a recent biological-inspired al-gorithm that has been presented for solving different optimisation problems with high prowess. Despite, MFO algorithm also has its own drawbacks as the algorithm suffers from premature convergence and also fails to create a balance between exploration and exploitation searching scheme.
This thesis presents an improved Moth Flame optimisation (IMFO) algorithm combin-ing the fast-tracking capability of MFO and the Lévy flight long-jump movement in cuckoo search (CS) algorithm as a hybrid technique capable of resolving some of the known issues with optimisation algorithms. To validate the effectiveness of the antici-pated IMFO algorithm, different experiments were conducted using the proposed IMFO algorithm experimental results to compare with ten recent population-based algorithms that include the traditional MFO, CS, Salp swarm algorithm (SSA), Jaya, teaching-learning-based optimisation (TLBO), particle swarm optimisation (PSO), butterfly PSO, flower pollination algorithm (FPA), flying squirrel search optimisation (FSSO) and differential evolution (DE) algorithm results as case studies.
These case studies aim to evaluate the IMFO algorithm performance toward bench-marking of test function for robustness check, parameter extraction for solar cell mod-elling, power maximisation from PV systems under uniform irradiance, non-uniform irradiance and partial shading conditions and the solar panel energy payback period. Results obtained show that the proposed IMFO algorithm displayed the overall best performance and can be recommended for solar cell parameter extraction and maximi-sation of power from PV systems under uniform, partial shading and dynamic weather
Page | iii
conditions. Findings also show that the proposed IMFO algorithm can be used to lessen the solar panel energy payback period on solar panel investment.
Keywords: Global Maximum Power Point Tracking (GMPPT) techniques, Improved Moth Flame Optimization, metaheuristic algorithm, MPPT algorithms, optimization al-gorithms, parameter extraction, partial shading, photovoltaic systems, renewable en-ergy, solar cell.