This laboratory is in charge of part of the education in the field of electronic engineering in the Department of Information and Electronic Engineering. Not only electrical and electronic circuits but also mathematical ability is indispensable for understanding and designing electronics. In the second year, we teach electronic circuits, basic electronics experiments 1, and mathematical statistics, in the third year we teach operations research, and in the first year of the Graduate School, we teach multivariate analysis. He also teaches mathematical statistics and electronic circuits 1 and 2, operations research, and multivariate analysis in the Division of Informatics Science Department of Information Science, a Correspondence Course. In my graduation research, I deal with electronic engineering and information science comprehensively, and I am working on the theme of application to abnormality detection and diagnosis of photovoltaic power generation systems using statistical machine learning.
Basic Information
Faculty name/Affiliation
Yasuyuki Kobayashi / Department of Information and Electronic Engineering, Faculty of Science and Engineering
Specialized Fields
Soundness diagnosis, applied statistical science
Research theme
Researching the basics and applications of statistical machine learning (improvement of classifiers, abnormality detection and diagnosis of photovoltaic power generation systems)
Improved research on statistical machine learning algorithms
Recently, machine learning, in which a computer learns and judges multivariable data, is attracting attention. Mahalanobis distance is one of the statistics by the statistical machine learning algorithm that is attracting attention in Japanese industry, but there is a problem that the calculation result becomes inaccurate and unstable. Therefore, we are improving the algorithm by theory and computer simulation so that the Mahalanobis distance can be calculated accurately and stably. As a result of the examination so far, the Mahalanobis distance correction method when the number of training samples is insufficient, the model of the magnitude of the influence of the numerical error of the computer, the condition of the training data not affected by the numerical error of the computer, etc. The results have been obtained.
Development and research of anomaly detection technology for photovoltaic power generation systems
Construction of solar power generation systems is progressing all over Japan and around the world, but the problem is that the amount of power generation cannot be ignored due to the occurrence of power generation defects. Therefore, we are conducting research on technological development to discover solar panels with power generation defects. Examining the relationship between the abnormal voltage of the solar cell in which the power generation failure occurred in the solar cell and the abnormal heat generation of the corresponding cell, and irradiating the defective solar cell with modulated laser light without making electrical contact inside the sealed module. We are proposing a technology to estimate the voltage of defective cells just by doing so. In particular, in order to make the latter technology applicable to actual photovoltaic power generation systems, we are aiming to conduct tests on photovoltaic power generation systems in operation on campus.
Papers and Conferences Presentation
Paper presentation
Title
Laboratory
Contents
New precise model of studentized principal components
Introducing the FacultyThis section introduces details about the research and the passionate thoughts around the research.
Discover the fascinating and untold stories behind the cutting-edge research being conducted by our esteemed professors. Get an in-depth look at the motivations and passions driving their work, and learn how their findings are making a real impact on society. Don't miss the opportunity to gain a deeper understanding of the amazing work being done right here at our laboratory!
