2023 |
Aloui, Messaoud; Hamidi, Faical; Jerbi, Houssem; Aoun, Mohamed Estimating and enlarging the domain of attraction for polynomial systems using a deep learning tool Conférence 2023. Résumé | Liens | BibTeX | Étiquettes: Deep learning, Domain of attraction, Learning systems, Learning tool, Linear matrix inequalities, Linear polynomials, Lyapunov functions, Lyapunov’s functions, Neural networks, Non linear, Particle swarm, Particle swarm optimization, Particle swarm optimization (PSO), Polynomial systems, Polynomials, Swarm intelligence, Swarm optimization @conference{Aloui2023b,This Paper deals with the topic of non linear polynomial systems. It explains a way to estimate and enlarge the region of attraction of nonlinear polynomial systems. It provides a deep learning method for estimating the domain of attraction and uses the Particle Swarm Optimization Algorithm to enlarge this domain. Based on an analytic method found in literature, a dataset is generated, used then to train an artificial neural network, which will be an objective function of an optimization algorithm. This method dives an imitation to a previous complicated method, with less complexity and les elapsed time. The benchmark examples show the efficiency of the method and compare results with those obtained with the one using linear matrix inequalities. © 2023 IEEE. |
Saafi, Omar; Dabbaghi, Boudour; Hamidi, Faical; Aoun, Mohamed 2023. Résumé | Liens | BibTeX | Étiquettes: Autonomous switching sequence, Hybrid switched systems, Hybrid switching systems, Hybrid systems, Identification problem, Numerical methods, Optimization techniques, Particle swarm, Particle swarm optimization, Particle swarm optimization (PSO), Swarm optimization, Switching instants, Switching sequence @conference{Saafi2023b,In this paper we propose a new optimization technique based on an evolutionary method known as particle swarm optimization (PSO) for solving and specifying the switching instants of hybrid systems. The main objective is to minimize a performance measure that depends on these switching instants within a finite time interval. Our approach assumes that there is a predefined sequence of system modes and, at each switching instant, it is possible for a state-space variable to jump from one mode to another, resulting in an additional associated cost. Our approach is justified by numerical examples and compared with the results obtained by gradient-based methods. The results obtained by PSO are very promising, without requiring any a priori assumptions about the regularity of the objective function to be minimized. © 2023 IEEE. |
Aloui, Messaoud; Hamidi, Faical; Jerbi, Houssem; Aoun, Mohamed Estimating and enlarging the domain of attraction for polynomial systems using a deep learning tool Conférence 2023. Résumé | Liens | BibTeX | Étiquettes: Deep learning, Domain of attraction, Learning systems, Learning tool, Linear matrix inequalities, Linear polynomials, Lyapunov functions, Lyapunov’s functions, Neural networks, Non linear, Particle swarm, Particle swarm optimization, Particle swarm optimization (PSO), Polynomial systems, Polynomials, Swarm intelligence, Swarm optimization @conference{Aloui2023,This Paper deals with the topic of non linear polynomial systems. It explains a way to estimate and enlarge the region of attraction of nonlinear polynomial systems. It provides a deep learning method for estimating the domain of attraction and uses the Particle Swarm Optimization Algorithm to enlarge this domain. Based on an analytic method found in literature, a dataset is generated, used then to train an artificial neural network, which will be an objective function of an optimization algorithm. This method dives an imitation to a previous complicated method, with less complexity and les elapsed time. The benchmark examples show the efficiency of the method and compare results with those obtained with the one using linear matrix inequalities. © 2023 IEEE. |