| Energy demands are increasing rapidly.To satisfy energy demands,Renewable Energy Sources(RES)are emerging as dependable and cleaner alternatives to fossil fuels.The complexity and unreliable availability of RES require robust intelligent control and optimization.Targeting at the problem,this thesis investigates AI-based energy harvesting techniques for Photovoltaic(PV)systems,Thermoelectric Generators(TEG),and Wind Energy Conversion Systems(WECS).Among all RES systems,solar energy,TEG,and WECS are growing fast.The primary objectives of control systems are to maximize power conversion efficiency and enhance the cost parity($/Watt).As modern Artificial Intelligence(AI)techniques are more robust than classical control methods,this thesis focuses on swarm intelligence-based stochastic optimization algorithms and techniques for RES of PV and TEG systems.The proposed optimization algorithms have been modified using stochastic operations for improved control stability and higher power tracking under various dynamic operating conditions.The applications of Swarm Intelligence(SI)to PV,TEG,and WECS power forecasting are an effective way to improve efficiency and performance.The potential of swarm intelligence to improve the efficiency,performance,and reliability of RES systems makes them more viable for widespread adoption.The main contributions of this thesis are summarized as follows.The first research proposes an intelligent Tunicate Swarm Algorithm(TSA)for MPPT control of PV systems in multiple PV array configurations.The mathematical modeling of TSA with a search and skip(SAS)scheme is utilized to minimize the tracking time and search area.The performance of the proposed TSA strategy is compared to some state-of-the-art techniques,including Incremental Conductance(InC),Improved Particle Swarm Optimization(IPSO),Grey Wolf Optimization(GWO),and Cuckoo Search Algorithm(CSA),through detailed case studies.TSA is further validated on a low-cost hardware setup,confirming its superior performance.The results provide insightful validation of the practicality of the proposed TSA strategy in real-world applications.We can observe the power tracking efficiency improvement of 5.8%due to an oscillation reduction of 93.45%.The second study proposes a TEG MPPT control technique using a novel Equilibrium Optimization algorithm(EQO)for thermoelectric systems under non-uniform temperature distribution(NUTD)conditions.Thermoelectric generators have several industrial and energy generation applications.The TEG is used for heat recovery,cooling,and renewable power generation.The generation of nonlinear and multi-maxima electrical behavior due to activation of bypass diode in a series-parallel combination of TEG modules.The equilibrium optimization algorithm has been applied to this task as an intelligent controller to distinguish between local and global maxima in the least number of iterations.This algorithm introduces the balancing coefficient to manage the exploitation and exploration phases of the algorithm which is essential to compensate between GMPP tracking time and tracking efficiency.This allows for superior control action by the proposed technique under nonuniform temperature distribution over heat source surfaces.The proposed EQO algorithm is compared with conventional MPPT techniques such as Perturb and Observe(P&O),improved PSO(IPSO),and GWO.The EQO algorithm is shown to outperform conventional methods in terms of stability,accuracy,and efficiency.The results show that the tracked power efficiency can be increased to 99.68%with 1.8-8%more energy harvest.The stable voltage and current transients can be attained with a min tracking time of 180ms,achieving up to 46%faster GM tracking and 238%faster settling time than some conventional MPPT control techniques.The third study proposes a modified White Shark Optimization Algorithm(MWSO)for short-term wind power forecasting to manage electricity demands for regional grids,using General Regression Neural Networks(GRNN)and Radial Basis Function Neural Networks(RBFNN).The MWSO algorithm is used to train GRNN and RBFNN models effectively in each iteration by accelerating the search speed and strengthening exploration and exploitation for feature-enhanced data.The supervised control and data acquisition(SCADA)data provide the input metrics,and a Hybrid Variation Decomposition Model(HVDM)is used for data preprocessing.The metaheuristic algorithm is improved using the Levy flight function(LFF)and sine-cosine function to enhance the training mechanism developed for the STWF problem.Six seasonal cases from Turkey and Malaysia are used to validate the proposed MWSO algorithm.This thesis is expected to make a significant contribution to the energy harvesting of renewable resources,such as photovoltaics,thermoelectric power generators,and wind energy.Through the development of new techniques,methodologies,and tools,this thesis aims to improve the efficiency,performance,and reliability of renewable energy systems.Its research has the potential to drive significant advancement of renewable energy and promote a more sustainable future for all. |
