Still have to receive the final get go but it deals with, ... wait for it :

Active damping of vibrations in very large wind turbines using the smart rotor concept.

See picture :



We are talking about wind turbines that have a rotor diameter of over 100 meters, here the vibrations (even small ones) are becoming quite problematic. They distroy highly loaded parts like the bearings rather quickly and now these turbines have to be shut down when the wind is not quite right, to safe some lifespan of the bearings for perfect wind days. By reducing these vibrations it is intended to increase the lifespan of the highly loaded parts to such an extend that these turbines can just be left on for very long periods of time between maintaince or repair jobs. This is very important economically. The more hours such an expensive turbine can produce energy the quicker the investment is won back and the more interesting it becomes to invest in one and the more the price per KWh reduces.

I'm going to try to get a new type of actuator (to move the control flaps) to work well. This actuator uses memory metal to move the flap and that is a highly non linear and non repetitive actuator. My task will be to design and implement a compensator that will remove all these non-linearities from its behaviour when the combination of compensator and actuator are viewed as a single blackbox. Some of this non-linearities will be linked to the aerodynamic behaviour of the flap itself.

My initial expectation is to do this using an Artificial Neural Network (multi-input ; single-output) fed with all the influence factors and having trained itself to compensate for them.

Everything is intended to go into a windtunnel setup to be tested. Sadly the (old) test setup came loose and was distroyed in the windtunnel some months ago. So somebody has to build a complete new setup before we can proceed on that, but arrangements are being made to that effect.

I would really love to be there when part of my work is tested with that of others in the windtunnel.

Wouter