220 likes | 418 Views
Self-Sensing Active Magnetic Dampers for Vibration Control. Presenting by, JITHIN.K M-Tech, Machine Design Roll No: 9. Guided by, Dr. K.G.Jolly H.O.D Mechanical Dept. INTRODUCTION. Viscoelastic and fluid film dampers. Passive, semi-active and active dampers.
E N D
Self-Sensing Active MagneticDampers for Vibration Control Presenting by, JITHIN.K M-Tech, Machine Design Roll No: 9 Guided by, Dr. K.G.Jolly H.O.D Mechanical Dept.
INTRODUCTION • Viscoelastic and fluid film dampers. • Passive, semi-active and active dampers. • Electromechanical dampers • Absence of all fatigue and tribology issues. • Smaller sensitivity to the operating conditions. • Wide possibility of tuning even during operation. • Predictability of the behavior. • Active magnetic bearings • Shaft is completely supported by electromagnets
Active magnetic dampers • Rotor is supported by mechanical means and the electromagnetic actuators are used only to control the shaft vibrations. • The combination of mechanical suspension with an electromagnetic actuator is advantageous. • The system can be designed to be stable even in open loop. • Actuators are smaller compared to AMB configuration. • Our aim is to investigate self sensing approach in the case of AMD configuration. • The self sensing system is based on the Luenberger observer. • Parameters can be obtained in two different ways • Nominal ones and identified ones.
Modeling and Experimental Setup Nominal model • A single degree of freedom mass spring oscillator actuated by two opposite electromagnets. • Adoption of mechanical stiffness in parallel to electromagnets allows to compensate the -ve stiffness induced by electromagnets. • The back-electromotive force produced can be exploited to estimate mechanical variables from the measurement of electrical ones. Fig. 1 Model
This leads to the so-called selfsensing configuration that consists in using the electromagnet either as an actuator and a sensor. • Voltage and current are used to estimate the airgap. • Each electromagnet can be considered as a two-port element (electrical and mechanical). • The energy stored in the electromagnet j is expressed as: (1) where the force can be obtained as (2)
The total flux and the coil current are • related by a nonlinear function (3) where is the radial airgap of electromagnet j (4) where is the nominal airgap • Owing to Newton’s law in mechanical domain, the • Faraday and Kirchoff law in the electrical domain, the • dynamic equations of the system are (5)
where, R = coil resistance = voltage applied to electromagnet j = disturbance force applied to the mass • The system dynamics is linearized around a working • point corresponding to a bias voltage imposed to both • electromagnets (6) where is the initial force generated by the electromagnet due to the current .
The resulting linearized state space model is (7) where A,B and C are dynamic, action and output matrices respectively, defined as (8) with the associated input and output state vectors and .
The terms in the matrices derive from the linearization of • the nonlinear functions defined in eqs. (2) and (3) (9) • where are the inductance, the current-force factor, the back-electromotive force factor, and the negative stiffness of one electromagnet respectively. • Assuming that ferromagnetic material of the actuator does not saturate, has infinite magnetization and there is no magnetic leakage in the air gap, (10)
Where , characteristic factor of electromagnets. S = cross-sectional area of the magnetic circuit. • The presence of a mechanical stiffness large enough to overcome the negative stiffness of the electromagnets makes the linearization point stable and compels the system to oscillate about it. • As far as the linearization is concerned, the larger is stiffness k relative to | |, the more negligible the nonlinear effects become.
Identified Model • The system used is a test rig used for static characterization of radial magnetic bearings. • This rig consists in a horizontal arm hinged at one extremity with a pivot and actuated with a single axis magnetic bearing. • Six springs in parallel are placed to provide a stabilizing stiffness to the system. Fig. 2 Photo of the test rig Fig. 3 Test rig scheme
It consist of two electromagnets, power amplifier, BentlyProximitor eddy current sensor and current sensor. • Damping may be introduced into the structure by simply feeding back the position sensor signal by means of a proportional-derivative controller. • Two sets of parameters have been used to build the models. • Based on expression • Have been identified experimentally under two assumptions. • k, c, and m are determined from physical dimensions, direct measurements, and impact response in open-circuit electromagnets conditions. • The electromechanical parameters and are equal.
The proposed identification procedure is • Obtain the transfer function admittance in Fig. 4. • Measure the resistance value R at low frequency 1 Hz in our case. • Identify based on the high frequency slope of • Identify such that the zero-pole pair due to the mechanical resonance corresponds to the experimental ones.
The good correlation between the experimental and • identified plots validates the proposed procedure.
Controller unit • To introduce active magnetic damping into the system. • The control is based on the Luenberger observer approach. • It consists in estimating in real-time the unmeasured states • - displacement and velocity from the processing of the • measurable states i.e. the current.
Experimental results • The open-loop voltage to displacement transfer function obtained from the model and experimental tests are compared. • The same transfer functions in closed-loop operation with the controller designed are compared in the case of identified parameters. • In this case, the correspondence is quite good, which corroborates the control approach, and validates the whole procedure.
The damping performances are evaluated by analyzing the time response of the closed-loop system when an impulse excitation is applied to the system. • The controller based on the identified electromechanical parameters give better results than the nominal model. • Good damping can be conveniently achieved for active magnetic dampers obtained with the simplified model. • This controller does not destabilize the system, as it is the case for full suspension self-sensing configurations.
CONCLUSION • The study of an observer-based self-sensing active magnetic damper has been presented both in simulation and experimentally. • The closed-loop system has good damping performances than open-loop system. • The modeling approach and the identification procedure have been validated experimentally comparing the open-loop and the closed-loop frequency response to the model. • The self-sensing configuration provides good robustness performances even for relatively large parameter deviations.
References • A.Tonoli, N. Amati, M. Silvagni, 2008, “Transformer Eddy Current Dampers for the Vibration Control,” ASME J. Dyn. Syst., Meas., Control, 130, p.031010. • E. H. Maslen, D. T. Montie and T. Iwasaki, 2006, “Robustness Limitations in Self-Sensing Magnetic Bearings,” ASME J. Dyn. Syst., Meas., Control, 128, pp. 197–203. • V.P.Singh, “Mechanical Vibrations”.