Thesis Topics – Papageorgiou Elpiniki
1 Wind Turbine Energy Production Prediction
1.1 Objective
In this thesis, one year of data generated by the SCADA system of a wind turbine will be analyzed. Through this work, the student will learn the basic principles of data processing using machine learning tools (and beyond) in order to extract insights for a real-world problem. An open dataset will be used. More specifically, the dataset includes 10-minute measurements of:
- Actual Power Generation (kW): The actual energy produced at the time of measurement
- Wind Speed (m/s): The wind speed recorded at that moment
- Wind Direction (°): The wind direction in degrees at the time of measurement
- Theoretical Power Generation: The energy stated by the manufacturer that can be produced by the generator based on the measured conditions
For the prediction, the actual generation column will be the system output, while the remaining columns will be inputs.
1.2 Implementation Methodology
The interested student can visit the Kaggle website1 where the dataset is provided. There, they will find more information about the dataset as well as code offered for analysis and production prediction in English. In the thesis, the student must theoretically describe the operation of wind turbines, as well as the prediction model chosen (e.g., Neural Networks) through literature found. The structure of the thesis will be:
- Introduction – Wind Turbine Operation (Theory)
- Methodology
- Data description
- Description of the prediction algorithm, e.g., Neural Network architecture (layers, neurons, epochs, etc.)
- Approximation results (Absolute error, Mean Absolute Error, Mean Squared Error)
- Conclusions
The student can use any software for the implementation of the prediction algorithms and data processing (matlab, python, excel, etc.). The use of Python is recommended, as they will also find ready-made code snippets on the dataset’s website. If the student is comfortable using English, they can attend the free Python learning course on netacad2 or in Greek via mathesis3.
1 https://www.kaggle.com/datasets/berkerisen/wind-turbine-scada-dataset
2 https://www.netacad.com/courses/programming/pcap-programming-essentials-python
3 https://mathesis.cup.gr/courses/course-v1:ComputerScience+CS1.1+21D/about
2 Water Quality Analysis and Classification Using Python
2.1 Objective
In this thesis, data from quality control in water tanks will be analyzed. Through this work, the student will learn the basic principles of data processing using machine learning tools in order to extract insights for a real-world problem. More specifically, they will implement a classification algorithm where, depending on the measured characteristics of the water, the algorithm will examine whether it is potable or not (binary classification). The dataset includes 9 measurements regarding water quality (columns) from 3,276 tanks (rows). The measurements include:
- pH
- Hardness
- Total Dissolved Solids (TDS) level
- Chemicals (chlorine)
- Sulfate concentration
- Conductivity
- Total Organic Carbon (TOC) level
- Trihalomethanes
- Water purity-clarity (turbidity)
- Potability
2.2 Implementation Methodology
The interested student can visit the Kaggle website4 where the dataset is provided. There, they will find more information about the dataset as well as code offered for analysis and classification in English. In the thesis, the student must describe through literature found the problem of water potability5, the data, and the prediction model chosen (e.g., Neural Networks). The structure of the thesis will be:
- Introduction: Water suitability-potability
- Methodology
- Data description
- Algorithm description
- Classification results with metrics such as accuracy, precision, recall
- Conclusions
The algorithms will be implemented in Python, which is widely used for such types of problems, using libraries such as pandas, tensorflow, scikit-learn, etc., for algorithm development and data processing. If the student is comfortable using English, they can attend the free Python learning course on netacad6 or in Greek via mathesis7.
4 https://www.kaggle.com/datasets/adityakadiwal/water-potability
5 https://el.wikipedia.org/wiki/%CE%A0%CE%BF%CE%B9%CF%8C%CF%84%CE%B7%CF%84%CE%B1_%CF%84%CE%BF%CF%85_%CE%BD%CE%B5%CF%81%CE%BF%CF%8D
6 https://www.netacad.com/courses/programming/pcap-programming-essentials-python
7 https://mathesis.cup.gr/courses/course-v1:ComputerScience+CS1.1+21D/about