IERE | 5th IERE Webinar on AI Collaboration Project

5th IERE Webinar on AI Collaboration Project

March 24, 2022

Organized by IERE

In recent years, Artificial Intelligence (AI) technology has been developing rapidly. AI technology has widely supported our daily lives. The application of AI technology is also advancing in the field of the electric power industry. The purpose of this webinar is to share actual examples of AI applied technology development for solving problems in the field through joint research by IERE members.

Opening, IERE's Activity related to AI
Project Outline
Summary of Results
- Defect detection of high voltage electric transmission insulators (CLP & EPRI)
- Detection of abnormal signs in hydropower plants (CRIEPI & Shikoku EPCO)
- Short term load forecasting (NARI & PLN)
- Optimum Pole Transformer Capacity (Chubu EPCO & Chugoku EPCO)
Next Project

Speakers	Moderator, Dr. Cheuk Wing Lee
Lecture by Mr. Edward Kam Wong Chan	Lecture by Dr. Teruhisa Miura
Lecture by Mr. Qipei Zhang and Mr. Jixiang Lu	Lecture by Mr. Masaya Oirase

Q&A — Typical QuestionsThe day was filled with many questions and a lively Q&A session.

(W-1) Defect detection of high voltage electric transmission insulators

In this presentation, you showed us in the case of electric transmission insulators. Is StyleGan2 used elsewhere in electric utilities?
Did you examine another approach apart from StyleGen2 (deep learning)?

(W-2) Detection of abnormal signs in hydropower plants

Could you comment further on (i) the kinds of anomalies that you studied and (ii) the relative importance of the anomalies that you were able to detect. For example, were you able to detect the most serious troubles?
How many troubles have been observed in 10 years?

(W-3) Short term load forecasting

During data processing, how do you treat outliers in the loading dataset?
Do covid pandemic affects the accuracy of your AI? Assuming you are using data pre-pandemic.
You mention the challenge of including "extreme events." Is it possible to include these probabilistically.

(W-4) Optimum Pole Transformer Capacity

What AI method did you use? How much economic benefit can you expect from these results? You mentioned that you are considering an alternative approach in future work, what is it like?
Do you consider the day type, such as working day or holiday ? it will be impact human power consumption behavior and load peak as well.