Optimal Structural Control Using Neural Networks

• Published in: Journal of engineering mechanics Vol. 126; no. 2; pp. 201 - 205
• Main Authors: Kim, Ju-Tae, Jung, Hyung-Jo, Lee, In-Won
• Format: Journal Article
• Language: English
• Published: Reston, VA American Society of Civil Engineers 01.02.2000
• Subjects: Applied sciences; Buildings. Public works; Exact sciences and technology; Fundamental areas of phenomenology (including applications); Geotechnics; Physics; Solid mechanics; Structural and continuum mechanics; Structure-soil interaction; TECHNICAL PAPERS; Vibration, mechanical wave, dynamic stability (aeroelasticity, vibration control...); Vibrations and mechanical waves; Equation of motion; Active system; Optimal control; Non linear control; Vibration control; Neural network; Non linear effect; Modeling; Mechanical hysteresis
• ISSN: 0733-9399; 1943-7889
• DOI: 10.1061/(ASCE)0733-9399(2000)126:2(201)

An optimal control algorithm using neural networks is proposed. The controller neural network is trained by a training rule developed to minimize cost function. Both the linear structure and the nonlinear structure can be controlled by the proposed neurocontroller. A bilinear hysteretic model is used to simulate nonlinear structural behavior. Three main advantages of the neurocontroller can be summarized as follows. First, it can control a structure with unknown dynamics. Second, it can easily be applied to nonlinear structural control. Third, external disturbances can be considered in the optimal control. Examples show that structural vibration can be controlled successfully.