
Machine Learning and Data Science in the Power Generation Industry
Best Practices, Tools, and Case Studies
- 1st Edition - January 14, 2021
- Imprint: Elsevier
- Editor: Patrick Bangert
- Language: English
- Paperback ISBN:9 7 8 - 0 - 1 2 - 8 1 9 7 4 2 - 4
- eBook ISBN:9 7 8 - 0 - 1 2 - 8 2 2 6 0 0 - 1
Machine Learning and Data Science in the Power Generation Industry explores current best practices and quantifies the value-add in developing data-oriented computational progra… Read more

Purchase options

Institutional subscription on ScienceDirect
Request a sales quoteMachine Learning and Data Science in the Power Generation Industry explores current best practices and quantifies the value-add in developing data-oriented computational programs in the power industry, with a particular focus on thoughtfully chosen real-world case studies. It provides a set of realistic pathways for organizations seeking to develop machine learning methods, with a discussion on data selection and curation as well as organizational implementation in terms of staffing and continuing operationalization. It articulates a body of case study–driven best practices, including renewable energy sources, the smart grid, and the finances around spot markets, and forecasting.
- Provides best practices on how to design and set up ML projects in power systems, including all nontechnological aspects necessary to be successful
- Explores implementation pathways, explaining key ML algorithms and approaches as well as the choices that must be made, how to make them, what outcomes may be expected, and how the data must be prepared for them
- Determines the specific data needs for the collection, processing, and operationalization of data within machine learning algorithms for power systems
- Accompanied by numerous supporting real-world case studies, providing practical evidence of both best practices and potential pitfalls
Power industry experts and practitioners working either in engineering, production, maintenance or management. Individual contributors in charge of actually carrying out a project or managers at all levels who want to create a project, product, or service based on ML. Graduate students and early career researchers working in power systems and power generation, or in computational aspects of power
- Cover image
- Title page
- Table of Contents
- Copyright
- Contributors
- Foreword
- 1: Introduction
- Abstract
- 1.1: Who this book is for
- 1.2: Preview of the content
- 1.3: Power generation industry overview
- 1.4: Fuels as limited resources
- 1.5: Challenges of power generation
- 2: Data science, statistics, and time series
- Abstract
- 2.1: Measurement, uncertainty, and record keeping
- 2.2: Correlation and timescales
- 2.3: The idea of a model
- 2.4: First principles models
- 2.5: The straight line
- 2.6: Representation and significance
- 2.7: Outlier detection
- 2.8: Residuals and statistical distributions
- 2.9: Feature engineering
- 2.10: Principal component analysis
- 2.11: Practical advices
- 3: Machine learning
- Abstract
- 3.1: Basic ideas of machine learning
- 3.2: Bias-variance-complexity trade-off
- 3.3: Model types
- 3.4: Training and assessing a model
- 3.5: How good is my model?
- 3.6: Role of domain knowledge
- 3.7: Optimization using a model
- 3.8: Practical advice
- 4: Introduction to machine learning in the power generation industry
- Abstract
- 4.1: Forecasting
- 4.2: Predictive maintenance
- 4.3: Integration into the grid
- 4.4: Modeling physical relationships
- 4.5: Optimization and advanced process control
- 4.6: Consumer aspects
- 4.7: Other applications
- 5: Data management from the DCS to the historian and HMI
- Abstract
- 5.1: Introduction
- 5.2: Sensor data
- 5.3: How control systems manage data
- 5.4: Data visualization of time series data—HMI
- 5.5: Data management for equipment and facilities
- 5.6: How to get data out of the field/plant and to your analytics platform
- 5.7: Conclusion: Do you know if your data is correct and what do you plan to do with it?
- 6: Getting the most across the value chain
- Abstract
- 6.1: Thinking outside the box
- 6.2: Costing a project
- 6.3: Valuing a project
- 6.4: The business case
- 6.5: Digital platform: Partner, acquire, or build?
- 6.6: What success looks like
- Disclaimer
- 7: Project management for a machine learning project
- Abstract
- 7.1: Classical project management in power—A (short) primer
- 7.2: Agile—The mindset
- 7.3: Scrum—The framework
- 7.4: Project execution—From pilot to product
- 7.5: Management of change and culture
- 7.6: Scaling—From pilot to product
- 7.7: Further reading
- 8: Machine learning-based PV power forecasting methods for electrical grid management and energy trading
- Abstract
- 8.1: Introduction
- 8.2: Imbalance regulatory framework and balancing energy market in Italy
- 8.3: Data
- 8.4: ML techniques for PV power forecast
- 8.5: Economic value of the forecast of “relevant” PV plants generation
- 8.6: Economic value of PV forecast at national level
- 8.7: Conclusions
- 9: Electrical consumption forecasting in hospital facilities
- Abstract
- 9.1: Introduction
- 9.2: Case study description
- 9.3: Dataset
- 9.4: ANN architecture
- 9.5: Results of simulation
- 9.6: Conclusions
- 9.7: Practical utilization
- 10: Soft sensors for NOx emissions
- Abstract
- 10.1: Introduction to soft sensing
- 10.2: NOx and SOx emissions
- 10.3: Combined heat and power
- 10.4: Soft sensing and machine learning
- 10.5: Setting up a soft sensor
- 10.6: Assessing the model
- 10.7: Conclusion
- 11: Variable identification for power plant efficiency
- Abstract
- 11.1: Power plant efficiency
- 11.2: The value of efficiency
- 11.3: Variable sensitivity
- 11.4: Measurability, predictability, and controllability
- 11.5: Process modeling and optimization
- 12: Forecasting wind power plant failures
- Abstract
- 12.1: Introduction
- 12.2: Impact of damages on the wind power market
- 12.3: Vibration spectra
- 12.4: Denoising a spectrum
- 12.5: Properties of a spectrum
- 12.6: Spectral evolution
- 12.7: Prediction
- 12.8: Results on turbine blades
- 12.9: Results on the rotor and generator
- Index
- Edition: 1
- Published: January 14, 2021
- Imprint: Elsevier
- No. of pages: 274
- Language: English
- Paperback ISBN: 9780128197424
- eBook ISBN: 9780128226001
PB
Patrick Bangert
Dr. Patrick Bangert is the Vice President of Artificial Intelligence at Samsung SDS where he leads both the AI software development and AI consulting groups that each provide various offerings to the industry. He is the founder and Board Chair of Algorithmica Technologies, providing real-time process modeling, optimization, and predictive maintenance solutions to the process industry with a focus on chemistry and power generation. His doctorate from UCL specialized in applied mathematics, and his academic positions at NASA’s Jet Propulsion Laboratory and Los Alamos National Laboratory made use of optimization and machine learning for magnetohydrodynamics and particle accelerator experiments. He has published extensively across optimization and machine learning and their relevant applications in the real world.
Affiliations and expertise
Vice President of Artificial Intelligence at Samsung SDSA, San Jose, CA, United States, and Founder and Board Chair of Algorithmica Technologies GmbH, Bad Nauheim, GermanyRead Machine Learning and Data Science in the Power Generation Industry on ScienceDirect