Optimal design of smart structures with intelligent control

At Cnam, Paris, November 30th 2017, 2 p.m.

Georgios Tairidis
Technical University of Crete, TEI of Epirus, Chania, Greece

The presentation deals with optimization in the design of smart structures along with implementation and incorporation of intelligent fuzzy and neuro-fuzzy control. Different models of smart multilayer composite structures such as beams and plates with sensors and actuators of piezoelectric materials are studied under low frequency dynamic external loadings. The discretization of the host structures is made by using the finite element method.

The first purpose of the study is the development of reliable control systems and their connection with numerical integration algorithms for the study of dynamic systems. For the development of the controllers using fuzzy inference tools and artificial neural networks are used. In the smart structures which are considered here, a significant degree of uncertainty, due to imperfections and errors of both the control mechanisms and the model itself, is always involved. Especially in multilayer structures, several failures, such as delamination between the different layers, fatigue or other damages, may appear.

Thus, the second purpose of the study is the implementation of suitably defined robust controllers which should be able to function well, despite the existence of failures and errors, not only in the controllers components (sensors, inadequate measurements), but in the mechanical model itself (simplifications of the theory, imperfections, material fatigue).

For the investigation of these phenomena, as well as for the minimization of the errors mentioned above, a generalization of the classical theories in order to include the energy losses in the various layers with adhesives is attempted. Regarding control, classic monitoring tools often encounter several limitations to the study of such problems. For this reason, the use of intelligent fuzzy and neuro-fuzzy control techniques is suggested. Indeed, to maximize the efficiency of the proposed controllers, their characteristics are fine-tuned using global optimization. Namely, the controllers are first designed and subsequently they are subjected to an optimization process, by using either adaptive neural fuzzy techniques or global optimization methods such as genetic algorithms, particle swarm optimization, etc...

As a by-product of this work, an exemplary application of the field of micromechanical coordinators for energy harvesting is presented.