Loading...

Design of Neural Network Controller for MR Dampers Considering Minimization of their Energy Consumption

Khoshnoud, Armin | 2022

72 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 55619 (09)
  4. University: Sharif University of Technology
  5. Department: Civil Engineering
  6. Advisor(s): Rahimzadeh Rofooei, Fayaz
  7. Abstract:
  8. Classic controllers are designed to control the structure response at the desired thresholds concerning the minimum applied force by control elements. In some cases, the minimum applied force does not lead to the minimum energy consumption of the controlling system. Especially, the consumption energy of the semi-active control system is not linearly related to their applied force and their energy consumption is a nonlinear function of system dynamic variables. On the other hand, classic control algorithms are not optimized in the time domain of earthquakes and those are sub-optimal algorithms based on the structure's inherent characteristics. Time domain optimized algorithm control is an unsolved problem that can be solved using new numerical optimization methods. The obtained optimized states from the modeling of structure and optimization of structure response, with the genetic algorithm, used for training neural network controller. In this study, a single and three degrees of freedom structure equipped with a magnetorheological damper are modeled. its response is measured with different arrays of voltages and its array of voltages is optimized with respect to the minimization of structure response and energy consumption of the magnetorheological damper. The minimized array of voltages is used to train a neural network controller for the considered structure
  9. Keywords:
  10. Magnetorheological Damper ; Semiactive Control ; Controller ; Optimization ; Energy Consumption ; Machine Learning ; Neural Network ; Energy Minimization

 Digital Object List

 Bookmark

No TOC