Modeling and Simulation of Renewable Energy Sources Using Machine Learning Techniques
Open Access
- Author:
- Borochok, Anastasia
- Area of Honors:
- Electrical Engineering
- Degree:
- Bachelor of Science
- Document Type:
- Thesis
- Thesis Supervisors:
- Yan Li, Thesis Supervisor
Julio Urbina, Thesis Honors Advisor - Keywords:
- photovoltaic
machine learning
next generation reservoir computing
renewable energy
artificial intelligence
energy
reservoir computing
algorithms
python
coding - Abstract:
- With the growing increase in demand for renewable energy sources, there has also been an in- crease in demand for research techniques to improve their efficiency. The cost of fossil fuel energy on the planet’s C02 levels requires an alternative and cleaner energy. One of the most common and popular alternative energy are photovoltaics. This thesis will specifically focus on photovoltaic (PV) systems. Common photovoltaic systems generally lose a lot of energy in the conversion phases from solar to electrical energy. This loss of energy leads to the overall inefficiency of solar panels. Because renewable energy systems are not very efficient there are many existing methods that attempt to compensate for this deficiency. One technique to improve this inefficiency is called next generation reservoir computing (NGRC), a branch of reservoir computing (RC). NGRC is a subsection of machine learning (ML), using artificial intelligence to model human behavior and better predict data of the systems. Machine learning is a subset of artificial intelligence. Within machine learning, models are built based on existing data. NGRC differs from the older technique, RC, because it requires less computational efforts and shows promising results. The algorithm an- alyzes previous data and makes predictions for future outcomes. Given a specific training data set, the algorithm of the ML can analyze this data and make suggestions for future recommendations of PV operation. Based upon these predictions, PVs see an increase in power output, addressing the need and dire demand for alternative energy replacements. NGRC analyzes the system using training data sets and linear optimization; it is very efficient because it does not require large and complicated calculations. NGRC is a promising system that with greater implementation may help to better the efficiency of renewable energy like PVs.