Johan Der Hagopian

Modal Active Vibration Control of a Rotor Using Piezoelectric Stack Actuators

This work deals with active vibration control of a rotating machine in both steady state and transient motion. Two stacks of piezoelectric ceramic orthogonally arranged in a plane localized at one of the bearings were used as the control actuators, using a modal control strategy. The optimal controller LQR was used to calculate the control gain, working together with an LQE observer that estimates the modal state. Simulation carried out on FEM model suggested the feasibility of the control strategy, and experimental tests using a physical test rig show good agreement with the numerical results, and confirm the efficiency of the strategy.

Fuzzy Active Control of Flexible Structures by Using Electromagnetic Actuators

Numerical and experimental investigations are carried out to assess the possibility of controlling a flexible beam by using an electromagnetic actuator (EMA). The advantage of EMAs is that they do not require contact with the structure so they can be applied to light and small mechanisms. Nevertheless, their open-loop instability and nonlinear dynamic behavior relating to excitation frequency can limit their fields of application. The EMA is designed and dimensioned as a function of the structure to be controlled. The effect of the EMA is considered a restoring force; consequently, the structure is still linear, which enables the calculation of the modal matrices of the structure. Moreover, an inverse model of the EMA is proposed to implement a linear action block for the frequency range used. The gap distance is estimated by using a modal approximation of the displacements resulting from the measurements. The control strategy is a fuzzy controller with displacements and velocities as inputs. Fuzzy contro...

Balancing of a Highly Flexible Rotor by Using Artificial Neural Networks

The present work is an alternative methodology in order to balance an unsymmetrical damped highly flexible rotor by using neural networks. This procedure was developed aiming at improving the performance of classical balancing methods, which are not well adapted to these situations. The approach developed is based on the influence coefficient method and is adequate to integrate an active balancing system. The methodology is tested successfully experimentally.Copyright

Neuro-Fuzzy Active Control of Mechanical Structures

The abilities of neural networks combined with fuzzy logic offer interesting prospects for the active control of structures. By identification, they permit discarding the often delicate modeling step and they also permit the automatic regulation of the controllers that have non-linear characteristics. This study describes the application of neuro-fuzzy control to the dynamic behavior of structures. The study first explains the process chosen, which consists of two parts: • the first part is essential for the adjustment of the associated controller and concerns the neural identification of the structure studied; • the second part describes the controller development and the training stage; the controller is based on the simplest neural network model possible. This network is also able to translate Sugeno’s fuzzy function and optimize its performances according to a reference response. The study then presents two applications: the first deals with the identification and control of a linear mechanical system with two degrees of freedom. The second deals with the identification and control of the non-linear dynamic behavior of active electromagnetic actuators along one acting axis. In both cases, the results show the abilities and the efficiency of this process and underline the main advantage of this type of controller operating even on in a strongly non-linear system.Copyright