Machine learning for PV performance optimization
Thomas Schutzius, Professor
Mechanical Engineering
Applications for Fall 2025 are closed for this project.
This project aims to apply supervised machine learning techniques to optimize the performance of photovoltaic (PV) systems. By analyzing real-world operational data such as weather conditions, soiling levels, shading patterns, and energy output, we will develop predictive models that can accurately forecast PV performance and identify factors that reduce efficiency. The ultimate goal is to enhance the reliability and energy yield of solar power systems through data-driven insights and smart decision-making tools.
Role: The undergraduate researcher will assist in collecting and preprocessing PV system data, and in training and testing supervised machine learning models such as regression and classification algorithms. Tasks include coding in Python, cleaning data, selecting features, evaluating model accuracy, and interpreting results. The student will gain practical experience in applying supervised learning methods to solve real-world renewable energy challenges and will develop skills in data analysis, model development, and performance optimization.
Qualifications: Required:
Basic programming skills in Python
Interest in renewable energy systems and data-driven research
Good problem-solving and communication skills
Desirable but not essential:
Familiarity with supervised machine learning concepts
Experience using Python libraries such as Pandas, Scikit-learn, or TensorFlow
Coursework in data science, machine learning, or energy systems
Class level / major:
Open to all undergraduate levels; students majoring in Engineering, Computer Science, Data Science, or related fields are preferred but not required.
Day-to-day supervisor for this project: Shuai Li, Post-Doc
Hours: 12 or more hours
Off-Campus Research Site: 2505 Hearst Ave Room3116 Etchevery hall
Related website: https://mtsn.berkeley.edu/