This model can be found in the 20-sim demo directory.
You can open this model and simulate it yourself using the free 20-sim demo/viewer!

Linear Motor

Description

The LIMMS positioning system

The Philips Linear Motor Motion System (LIMMS) is a high performance positioning system. The direct drive linear motor delivers good performance in terms of accuracy, velocity and acceleration compared to conventional rotating systems with toothed belts or screw spindles. The carriage can reach a maximum velocity of 2 m/s, an acceleration up to 60 m/s² and can support a load to a maximum of 100 kg. 

The LIMMS linear motor is a brushless, three phase, moving coil, linear servo-motor. The motor consists of a coil block and a magnet unit. The magnet unit is used as a stator. The coil block is the translator of the motor and is integrated, together with all the control electronics, in the carriage of the LIMMS.

The advantages of this linear motor in the application are:

• No transmission elements required (direct drive, no rotation to translation conversion).
  

• Motor is robust and free of maintenance (because of the absence of brushes).

The disadvantages of the motor are:

• The system is subjected to Cogging disturbance-forces.
  

• A minor source of disturbance is due to the ball bearing.  

The standard controller configuration of this LIMMS axis consists of a standard feedback loop in combination with acceleration-, velocity and friction feedforward.

At the Control Laboratory of the University of Twente, a learning feedforward component has been implemented and used next to the feedback controller (see 20-sim model below). In this controller structure, a simple model of the linear motor is used for the design of the feedback controller, realizing a stable system. A learning feedforward component is added to this feedback controller to compensate for model deficiencies. In this way. an inverse process model is created in the learning feedforward. When the inverse model is perfect, the feedback control signals are zero (unless disturbances arise). As long as the feedback signal is not zero, the inverse model apparently is not perfect. The feedback control signal can be interpreted as an error signal for the current inverse model output. So, this signal can be used to improve the inverse model, briefly to learn. The part of the feedback controller in the control signal will decrease, while the feedforward part will increase. 

In this direct adaptive controller structure, a neural network is used for the feedforward controller component. From a number of possibilities a single-layer network with second order B-spline functions has been selected and implemented, mainly because of its computational efficiency.

1. You can load and view a model of the complete system by opening the file LinearMotor.em. (choose File and Open from the menu).

2. Perform a simulation run by selecting Model and Start Simulator from the menu.

    

20-sim will check the model, process it, and open the Simulator with the model and predefined experiment LinearMotor.exp loaded.