Vibration Control of Webbed and Rapidly Moving Continuous Media

My particular research involves the vibration control of a webbed and rapidly moving continuous media.  So what does that mean?  The word "webbed" just means something with width greater than its thickness, in my case I am using an endless woven belt made by Belt Power, Inc.  The "rapidly moving" means the belt will be moving around 3 pulleys.  The term "continuous media" refers to the fact the belt is one continuous object, here it is a continuous loop. 

Here is the mechatronics workstation I have put together, consisting of a Pentium 133 computer, the vibration control setup, and a signal amp/buffer stand. The computer will be running the QNX real time operating system.  A specialized environment for this operating system was created within our mechatronics group to manage the various signals needed to control our experiments.   If you want more information in this area, see the page devoted to our Real Time Control Environments.  The computer - experiment interface is achieved through the MultiQ I/O board. 

The signal amp/buffer stand consists of two Techron amplifiers and a buffering/scaling box.  The box protects the equipment from voltage and current spikes, and also amplifies the power of the signal sent from the computer to a level necessary to run to experiment's motors. 
In order to induce vibration, the belt will be moving at or near its critical velocity.  This velocity is a function of the belt tension and density, and is simply the point where vibration will be at its worst.  In order to control the vibration, the motion and changes in the belt must be able to be monitored.  An NAiS optical displacement sensor will be used to sense the vibration of the belt.

The displacement sensor sends out a laser to hit the belt at a specified distance away.  The light reflected off the belt is then analyzed by the sensor to determine the change in the belt's position.  The displacement sensor then returns a voltage based on the belt's change in position.

This is what the existing setup now looks like.  The motor on the right drives the right pulley and belt.  The two dancer arms are on either side of the center post and are turned by the motors on which they are mounted.  The position of either dancer arm is measured by a 2500 count encoder at the base their motor.  (I'll try to put some arrows on here later to help with the explanation).  
I will be using a single dancer arm for the implementation of an isolation control we are developing.  An isolation control seeks to cancel the vibration coming in to a point.  In my experiment the dancer arm used to cancel the vibration will be near the middle of the belt.  The control will cancel the vibration on one side of the belt while the other side vibrates. 
 
The two controllers I will use for this will be an exact model knowledge controller and an adaptive controller.  In an exact model knowledge controller, the various characteristics of the system are calculated or approximated and then tweaked to give a good result.  This can be very time consuming.  In an adaptive controller, the characteristics do not need to be known, only guessed, and the controller will then compare the desired outcome (no vibration) with the actual system and continue adapt itself until the system is satisfactory. 
 
The isolation control theory will be done by myself and a doctoral student, while the theory for the exact model knowledge and adaptive controller has already been done by some of our other students.  My experiment will provide experimental verification of their theories. 
 
I will post more information, as well as photos of the existing setup, as the work progresses and when I get some extra time.