Abstract
Most parts of the world are experiencing a high increase in electric demand due to the population increase. For that fact, the current sources of Electric power generation are starting to struggle to meet the demand for electric energy, harming the environment due to air pollution. The ocean, which covers three-quarters of the planet earth, has tremendous energy trapped in its waves. Researchers from different regions have started searching for methods that can be used to convert this vast amount of abundant energy into electric energy to meet the forever increase of its demand. Ocean Wave Energy Converter devices are attracting researchers due to their ability to convert Wave energy into electric energy. However, they are not efficient enough to convert energy into the required form at this stage.
The research aimed to seek an efficient Wave Energy Converter (WEC) system that meets the high electric demand at reasonable costs. The deduced study solution to the objective was to design and optimize the heaving buoy WEC system to achieve maximum annual generated electric energy while lowering the Levelized maintenance cost of energy per unit. The research proposed using Machine Learning and Evolutionary Algorithms to design and optimize the WEC system mentioned above, and two methods were presented. A Hybrid Multi-Objective Optimization Algorithm based on Non-Dominated Sorting and Crowding Distance algorithm was proposed to design and optimize the WEC system, considering the system’s buoy shape and size as the control variables. Machine Learning and Particle Swarm, Inspired Success History Based Adaptive Multi-Objective Differential Evolution Algorithm was also proposed to design and optimize the WEC system. During this method, the variable control parameters were the WEC system’s Frictional coefficient, Velocity-dependent parameter of back-torque, Ratio of power output to the square of omega, and Generator start-up torque parameters. The performance of both algorithms was tested using Multi-Objective benchmark problems, and several Multi-Objective Algorithms were also used for the performance comparison.
Performance test matrices such as the Integral Generational Distance and the Hypervolume were used. The Wilcoxon and the Friedman tests were done to test the rankings of the best
v
algorithms. After analysing the results, the study showed that the developed methods achieved better performance on the benchmark functions and WEC system applications. The study revealed that Machine Learning and Evolutionary Algorithms enhanced energy conversion efficiency in WEC devices. The study further revealed that the incorporation of multiple individual algorithms into a hybrid algorithm achieves better results in solving real-world problems. MATLAB was used in the study to model the proposed solutions and experimental simulations.
KEY TERMS: Wave Energy Converter, Machine Learning, Evolutionary Optimization Algorithms, Heave Buoy Point Absorber, Hybrid Multi-Objective Optimization, and Power Take-Off