Data-Driven Modelling of Non-Domestic Buildings Energy Performance: Supporting Building Retrofit Planning

Preface
Acknowledgements
Contents
Acronyms
1 Introduction
	1.1 Motivation and Background
	1.2 Why This Book Was Needed
	1.3 The Novelty Element and Rationale
	1.4 The Key Questions
	1.5 How Would This Book Answer the Key Questions
	1.6 Summary of Methods Applied in This Work
	1.7 The Book in a Nutshell
	References
2 Building Energy Performance Assessment Methods
	2.1 Introduction
	2.2 Different Approaches
		2.2.1 Engineering Calculation
		2.2.2 Simulation Method
		2.2.3 Statistical Models
	2.3 Classification of Energy Assessment
		2.3.1 Process of Benchmarking
		2.3.2 Rating
		2.3.3 Certification
		2.3.4 Labelling
	2.4 The UK Energy Performance Certificate
	References
3 Multi-objective Optimisation and Building Retrofit Planning
	3.1 Building Retrofit Planning
		3.1.1 Critical Factors of Efficient Retrofit Plan
		3.1.2 Multi-objective Optimisation
		3.1.3 Decision-Making for Retrofit
	3.2 Conclusion
	References
4 Machine Learning for Building Energy Forecasting
	4.1 Introduction
	4.2 Machine Learning Techniques
		4.2.1 Artificial Neural Networks
		4.2.2 Support Vector Machine
		4.2.3 Gaussian Process
		4.2.4 Random Forest
		4.2.5 Gradient Boosted Regression Trees
		4.2.6 Clustering Algorithms
	4.3 Evaluation Methods
		4.3.1 Performance Evaluation Measurements
		4.3.2 Model Validation
	4.4 Feature Engineering
		4.4.1 Feature Importance
	4.5 Machine Learning for Modelling Energy Performance
		4.5.1 Artificial Neural Networks
		4.5.2 Support Vector Machine
		4.5.3 Gaussian Process and Mixture Models
		4.5.4 Ensemble Models
		4.5.5 Clustering Algorithms
	4.6 Summary of ML Models
	4.7 Conclusion
	References
5 Machine Learning Models for Prediction of Building Energy Performance
	5.1 Introduction
	5.2 Supervised Machine Learning Models
		5.2.1 Models Hyper-parameters
	5.3 Selected Datasets for Case Study
		5.3.1 Ecotect Dataset
		5.3.2 EnergyPlus Dataset
	5.4 Comparison of Models Accuracy
	5.5 Performance of the Best Model
		5.5.1 Effect of Increasing Size of Training Data
	5.6 Feature Selection and Importance
	5.7 Summary
	References
6 Building Energy Data-Driven Model Improved by Multi-objective Optimisation
	6.1 Introduction
	6.2 Optimisation Based on Evolutionary Algorithms
		6.2.1 Evaluation Criteria and Optimisation Variables
	6.3 Performance of Intelligent Tuning Method
	6.4 Evaluation of the Trained Model
	6.5 Feature Space Reduction
	6.6 Summary
	References
7 Modelling Energy Performance  of Non-domestic Buildings
	7.1 Introduction
	7.2 Building Energy Performance Modelling
	7.3 Model Parameter Optimisation
	7.4 Data Engineering
		7.4.1 Feature Extraction
		7.4.2 Input Feature Selection
	7.5 Evaluation of Energy Performance Model
		7.5.1 Extracted Features
		7.5.2 Model Optimisation
	7.6 Sensitivity Analysis
	7.7 A Case Study
	7.8 Summary
	References
8 Conclusion: Contributions, Impacts and Recommendations for Future
	8.1 Introduction
	8.2 How Main Research Tasks Are Delivered
		8.2.1 Task 1
		8.2.2 Task 2
		8.2.3 Task 3
		8.2.4 Task 4
		8.2.5 Task 5
	8.3 Implication for Practice
	8.4 Research Limitations
	8.5 Further Recommendations
	References
Appendix A Detailed Results for Tuning Machine Learning Model
Appendix B Non-domestic Building Records Distribution
Appendix C Calculating Features for Retrofit-Specific Energy Modelling
Appendix  Glossary