SYMBOLIC REGRESSION IN ENERGY ENGINEERING

Abstract

The global call for more sustainable energy development and natural resource management hinges on both the technical ability and social capacity to harness the potentials from these resources.
Symbolic Regression in Energy Engineering explores leveraging machine learning to solve renewable energy challenges arising from the notorious volatility of resources. Symbolic regression, a machine learning technique, uncovers mathematical models from data without predefined structures, thus providing interpretable and accurate models. This thesis investigates symbolic regression's applications in energy engineering, particularly in predicting renewable energy outputs such as wind speed against power output, which are highly variable and unpredictable. The study utilizes genetic programming to evolve symbolic expressions that model complex relationships within wind energy systems. The methodology includes collecting and preprocessing data, training symbolic regression algorithms, and evaluating models using various metrics. The results demonstrate symbolic regression's effectiveness in creating predictive models that outperform traditional regression methods in both accuracy and interpretability. By capturing intrinsic data patterns, symbolic regression offers a promising approach to enhancing the reliability and efficiency of renewable energy systems. The discussion highlights the advantages of symbolic regression over traditional methods, including better model interpretability and reduced human bias, and suggests future research directions to further improve this technique's applicability in energy engineering.
This abstract captures the essence of the thesis, highlighting the importance of symbolic regression in addressing renewable energy challenges, the methodology employed, and the significance of the results obtained.

Popular Abstract

Popular Science Summary: Symbolic Regression for Energy Engineering
Renewable energy sources like wind and solar are essential for sustainable development but are highly unpredictable. This volatility complicates efficient energy harnessing. Traditional methods for predicting energy output often fall short due to their complexity and limited accuracy. Symbolic regression—a machine learning technique—offers a promising solution. This research dives very deep into various systematics approach in addressing energy resource volatility issues using symbolic regression
What is Symbolic Regression?
Symbolic regression discovers mathematical models from data without predefined structures. Unlike traditional methods that fit data to a specified model, it searches through numerous possible mathematical expressions to find the best fit. This results in accurate, interpretable models. Symbolic regression uses genetic programming, inspired by natural selection. It starts with random mathematical expressions and evolves them through iterations. The best-performing models are selected and modified, continuing until an optimal model is found.
Importance in Renewable Energy
Accurate energy output prediction is crucial for integrating renewable sources into the energy grid. For example, knowing the power output from wind turbines based on wind speed aids better planning. Symbolic regression can uncover complex data relationships that traditional methods might miss.
Major Case Study: Wind Energy Prediction
In "Symbolic Regression in Energy Engineering," symbolic regression predicts wind energy output. Data on wind speed and power output are collected, and models are trained and compared to traditional methods. Results show symbolic regression models provide more accurate predictions and are more interpretable. This clarity helps engineers understand and utilize the models better.
Advantages Over Traditional Methods
1. Accuracy: Symbolic regression often outperforms traditional methods in predicting energy output.
2. Interpretability: Models are simpler and more transparent, clarifying the underlying processes.
3. Flexibility: It adapts to various data types and uncovers unexpected patterns without relying on predefined models.
Future Directions
Challenges include the computational resources needed for training and the need for more robust algorithms. Future research could improve algorithm efficiency and explore applications in other renewable energy sources like solar power.
Conclusion
Symbolic regression is a significant advancement in energy engineering, providing accurate, interpretable models to overcome the variability of renewable energy sources. This enhances energy system efficiency and contributes to sustainable development.

• author: Ukachukwu, Christian Chinweike
• supervisor: Rixin Yu
• organization: Department of Energy Sciences
• course: MVKM05 20241
• year: 2024
• type: H2 - Master's Degree (Two Years)
• subject: Technology and Engineering
• keywords: Symbolic Regression, Machine Learning, Genetic Programming, Energy Engineering, Wind Power, Wind Speed
• report number: LUTMDN/TMHP-24/5577-SE
• ISSN: 0282-1990
• language: English
• id: 9163117
• date added to LUP: 2024-06-13 14:38:17
• date last changed: 2024-06-13 14:38:17