DC brush sensorless control

[ligafx]

Has anyone any experience or technical reference in sensorless control of (100W) DC brush motors?
I've to control the position of a DC brush motor without the use of encoders, extracting the information from the motor current (the DC power is not stabilized so there is a voltage ripple at 100Hz to consider and digitally remove).
I'm trying to use an LMS algorithm to remove the voltage ripple, I don't know if it is the best choice.
I don't know how to recognize the sinusoidal signal which indicates the rotor movement. (Is it better to operate in time or frequency domain? LMS style algorithms or neural networks?)
suggestions will be appreciated!
thanks

 

[Highspeed]

Usually , it's easier to have a position feed back of a DC brush motor by monitoring Back emf rather than current
What do you want to control, speed, position?

 

[ligafx]

the main goal is to realize a cheap system to estimate(with a reasonable precision, not the absolute precision achievable with an encoder)the position of one or more linear actuators. It should be very useful to operate in PWM, controlling the speed (useful to sincronize the positions of two actuators) but it is important that the system can work with a relay controlled motor. we already monitor the motor current and I've seen a quasi sinusoidal wave superposed to the main wave, every period representing the passage of brushes to a new coil.
I've seen application on brushless motors but there is always an unexcited coil in that cases to measure back emf. In brush motors I think I shuold periodically turn off the motor for a few time to measure back EMF, but it will give me information only on speed (is it correct?) and from measurements done in the past that voltage is not so clearly identifiable (100W motors with LC noise suppression filters)so I think it should lead to a considerable error in position estimation (am I wrong?).
So I've thought that the cheapest and most accurate way to estimate position is to apply the theory of signal identification on motor current, but I've not found any reference in literature so I'm a little bit discouraged!
I hope I've been clear enough.