An introduction to multiobjective optimization problems in nonlinear circuits and systems

IEICE English Webinar Distinguished Lecturer Program Series August 2024

"An introduction to multiobjective optimization problems in nonlinear circuits and systems"
Lecturer: Prof. Toshimichi Saito
Electrical and Electronics Engineering Department, Hosei University

Biography:
Toshimichi Saito received the B. E., M. E., and Ph. D. degrees in electrical engineering from Keio university, Yokohama, Japan, in 1980, 1982 and 1985, respectively. He is currently a full professor at Hosei university, Tokyo. His research interests include chaos and bifurcation, analysis/synthesis of artificial neural networks, and power electronics. He served in several editorial boards including the IEEE Trans. Circuits Syst. I (2000-2001), the IEEE Trans. Circuits Syst. II (2003-2005), and the Elsevier Intl. J. Electronics and Communications (2010-2014). He is a life senior member of the IEEE and a fellow of the IEICE.

Summary：
Multiobjective optimization problems require simultaneous optimization of multiple objectives (e.g., cost and performance). In the problem, we often encounter trade-offs: an improvement in one objective causes deterioration in other objectives. The trade-off is characterized by the Pareto front. Efficient evolutionary algorithms have been presented to calculate the Pareto front. The problems have been studied intensively in benchmarks (e.g., the multiobjective 0–1 knapsack problem). However, the studies in nonlinear circuits and systems are not sufficient. The reasons include huge parameter space and complicated calculation of the objectives. For simplicity, we introduce two example problems.

(1) Biobjective optimization problems in switching power converters with photovoltaic inputs. The circuits play important roles in renewable energy supply in carbon neutral technology. The first objective evaluates power efficiency and the second objective evaluates circuit stability. Applying piecewise linear modeling, the two objectives are described theoretically and the Pareto front is obtained exactly. The Pareto front guarantees existence of a trade-off between the two objectives
(2) Biobjectve optimization problems in discrete-time recurrent neural networks as associative memories. The first objective evaluates memory stability and the second objective evaluates connection sparsity. In order to realize precise calculation, we consider the case where the networks are characterized by the signum activation function and ternary connection parameters. Performing precise numerical analysis in typical examples, the Pareto-fronts are obtained.

Based on the two examples, several research directions/themes are suggested. This lecturer is grateful if the suggestions give some hint to studies in young researchers/students.

Send Feedback:

Follow us on Twitter: @Ieice_trans