2021 | Big Data in Energy: a practical introduction

Course description and aims

In this course, “Data science and AI analysis” relating to energy is defined as “Energy big data science” and the course focuses on applying “Energy big data science” to energy conversion and storage technologies. It will explain why big data science is useful in energy science, and various types of data processing/analyzing methods. Furthermore, the course will introduce a example applying data-scientific analysis to the development of energy materials as “Energy-material Informatics” and the development of future energy system toward carbon-free society. The course includes following topics using real energy data (“Ene-Swallow” data) in Tokyo Tech.:

1) Perspective of energy technologies toward carbon-free society (present and future): Brief overview of traditional energy and storage technologies and novel, more sustainable technologies. Highlights an increased role of data with the new technologies.
2) Key data analysis techniques including selected machine learning techniques in “Energy big data science” (“Data science and AI analysis” relating to energy).
3) Functional materials as key to the development of advanced energy conversion and storage technologies. Informatics aspects of materials development.
4) Utilization of energy big data in Grid cooperated/distributed energy system toward carbon-free society (the energy data of Tokyo Tech EEI (Environmental Energy Innovation) building, the data of electric vehicles etc.)

The student will work with real data from “Ene-Swallow” in Tokyo Tech. The students will perform simulations on their own computers connected to the campus network. The course will be arrange even for the student non-familiar to a programing
The course enrolment is limited to 40 students, with priority given to the students belonging to Tokyo Tech Academy of Energy and Informatics.

Student learning outcomes

Students will be able to understand and apply data analysis processes to solution of problems of energy informatics, understand and use machine learning tools (methods and software) and procedures useful in the analysis of data generated by energy technologies. Students will also learn the basics of key energy generation technologies and their interactions and of related functional materials.

Keywords

Energy system, Renewable energy, energy storage, data analysis, big data, machine learning

Competencies that will be developed

✔ Specialist skills	Intercultural skills	Communication skills	✔ Critical thinking skills	✔ Practical and/or problem-solving skills
✔ Specialist skills, Practical and/or problem-solving skills

Class flow

Each class lasting about 100 min will feature a lecture period (about ½ to 2/3 of the class time), an exercise period (about ¼ to 1/3 of the class time), and an assessment period (about 15 min). Some of the lectures will be given as master-classes (i.e. with students repeating on their computers the procedures performed by the professor on their computer) and some will be delivered traditionally. Exercises will involve data analysis using methods and tools introduced in the lectures. Each class will have a brief quiz to provide constant feedback to the professor and to the student about the degree of absorption of the material and to reveal any need for review or adjustments.

Course schedule/Required learning

	Course schedule	Required learning
Class 1	The energy landscape and overview of the course: present and future energy mix (Prof. M. Ihara): Brief overview of traditional energy conversion and storage technologies and novel, more sustainable technologies. Highlights an increased role of data with the new technologies. Highlights on the concept of grid cooperative/distributed real time energy system and increases role of functional materials in renewable technologies.	Students will learn about the contributions of major energy conversion technologies (oil, natural gas, nuclear, wind, solar, fuel cells) and storage technologies (pumped hydro, batteries to the energy mix and how they are expected to evolve. They will learn key in-principle differences between modern and traditional technologies and associated with it increased need for data analysis and management as well as for new functional materials.
Class 2	Key data analysis techniques and introduction to selected machine learning techniques (Assoc. Prof. S Manzhos, Prof. Ihara): Recapitulation of linear regression methods, correlation analysis, and PCA. Neural networks and deep learning. Kernel methods on the example of Gaussian process regression (GPR). Linear and nonlinear dimensionality reduction. Clustering. Pros and cons of all these techniques. Suitability for big data. Exercise 1: Students learn invocation in the Python environment of techniques covered in the lecture.	Students are expected to have been exposed to linear methods - this material will be briefly refreshed. The students will learn practical basics of neural networks including deep networks as well as the basics of kernel methods such as GPR and their pros and cons. The advantages of non-linear analysis in multidimensional spaces of features will be highlighted, including those of nonlinear dimensionality reduction and clustering.
Class 3	Functional materials as a key to the development of advanced generation and storage technologies. Part 1(Assoc. Prof. S Manzhos, Prof. Ihara): introduction to the materials aspect of renewables. Types of functional materials and interfaces used in renewable energy technologies: Electronic materials, intercalation materials, catalysts, nanomaterials and low-dimensional materials and interfaces. Key phenomena responsible for core functionality, including light absorption, electron and ion transport and key reactions. Exercise 2: Students learn how to source data from the web, analyze them statistically and preprocess for further analysis.	Students will familiarize themselves with kinds of data encountered in materials informatics and how they can be used for rational design of functional materials. In a hands-on exercise, students will understand the limits of simple linear regression based analysis, and the need for more involved and in particular machine learning based techniques.
Class 4	Functional materials as a key to the development of advanced energy conversion and storage technologies. Part 2 (Assoc. Prof. S Manzhos, Prof. Ihara): Informatics aspects of materials development. Mapping between descriptors and desired properties. Choices of descriptors. Examples from the design of fuel cell catalysts and solar cell materials. Exercise 3: Students will work with a provided dataset containing data on composition and performance of catalytic materials and will try to machine-learn performance characteristics from the composition.	Students will familiarize themselves with kinds of data encountered in materials informatics and how they can be used for rational design of functional materials. In a hands-on exercise, students will understand the limits of simple linear regression based analysis, and the need for more involved and in particular machine learning based techniques.
Class 5	Energy meets big data in advanced energy system 1: Introduction of present ICT structure of energy system “Ene-Swallow”(Prof. Ihara, Assoc. Prof. Manzhos): Exercise 4: Students will acquire Tokyo Tech EEI building data and electric vehicle (EV) data from the cloud data server of “Ene-Swallow” and the connected EVs, including multiple types of data on a multitude of solar cells, fuel cell, gas engines, butteries as well as of energy consuming devices and location information of EVs. Furthermore, the students will analyze the location information of EVs based on Google Maps data.	Students will experience data generated in a real distributed energy system consistent of different devices from the realm of renewable energies.
Class 6	Energy meets big data in advanced energy system 2: Summary and final review (Prof. Ihara, Assoc. Prof. Manzhos): Exercise 5: Students will analyze the acquired data on EEI building. Students will work with very high-dimensional, unlabeled data and perform inference of data meaning. Clustering with methods such as K means will be used to identify collective variables.	Students will experience work with very high dimensional data. They will learn to identify the meaning of the data from data statistics. They will apply different statistical and clustering techniques to identify a tractable number of underlying variables. With a summary review of all classes the students will acquire a cohesive picture of the interplay between physical principles and data in renewable energy conversion and storage technologies.

Out-of-Class Study Time (Preparation and Review)

To enhance effective learning, students are encouraged to spend approximately 100 minutes preparing for class and another 100 minutes reviewing class content afterwards (including assignments) for each class.
They should do so by referring to textbooks and other course material.

Textbook(s)

None

Reference books, course materials, etc.

Chapters from:
Fundamentals of Materials for Energy and Environmental Sustainability, Eds. D. S. Ginley, D. Cahen, Cambridge University Press, 2011, ISBN:9781107000230
A. Vieira da Rosa, Fundamentals of Renewable Energy Processes (Third Edition), Academic Press / Elsevier, 2013, ISBN: 978-0-12-397219-4
Articles from research literature.
Reference books, agency reports, and original course materials.

Assessment criteria and methods

Students' understanding will be assessed by exercises and reports.

Related courses

ENR.A401 ： Interdisciplinary scientific principles of energy 1
ENR.A403 ： Interdisciplinary principles of energy devices 1
ENR.A404 ： Interdisciplinary principles of energy devices 2
ENR.A405 ： Interdisciplinary Energy Materials Science 1
ENR.A406 ： Interdisciplinary Energy Materials Science 2
ENR.A407 ： Energy system theory
ENR.B431 ： Recent technologies of fuel cells, solar cells butteries and energy system

Prerequisites (i.e., required knowledge, skills, courses, etc.)

None

Contact information (e-mail and phone) Notice : Please replace from "[at]" to "@"(half-width character).

Manabu Ihara, mihara[at]chemeng.titech.ac.jp
Sergei Manzhos, Manzhos.s.aa[at]m.titech.ac.jp

TOKYO INSTITUTE OF TECHNOLOGY