Based on the 1997 diploma and project theses we had to improve the existing time discrete state controller such that not only the x-position (left, right) but also the y-position (up, down) of the model crane could be controlled. Thus the load y-position becomes the variable to be controlled, in contrast to just controlling the rope length. This means we had to build a new mathematical model of the crane in which the rope length is no longer just a parameter but a new state variable. Thus we have to deal with a multivariable system which has two inputs and two outputs.

Alphasem is a manufacturer of special machines, which are used to attach the semiconductor dies to their protective packages. This process depends on a very high throughput and on a high accuracy. This requires very efficient position drives.

In our thesis we had to implement a high-dynamic position control loop for an air-beared linear motor. We used a linear motor in combination with a servo controller, made by ETEL SA. We had to carry out position regulations to fulfil the demands set by Alphasem. They prescribed that a positioning process of 400mm with a S-type movement (without jerk) has to process within 140ms precise to 5mm.

Our working environment consisted of an air-beared linear motor and a servo controller. The servo controller contains the entire power electronics, the automatic control loops and various protective functions. A computer with installed software controlled and monitored the entire linear drive.

Because of the air-bearing the damping of the system is very small. A unfavourable selection of the control loop parameters can lead to instability. First, we optimized the current loop, so that this became as fast as possible. Afterwards we integrated the current loop into the position loop and optimized this with various measurements and simulations. We developed a mathematical model, which easily allows to calculate optimal automatic controller parameters for different masses. We acknowledged this by various measurements. We succeeded in fulfilling the demands set by Alphasem concerning the positioning time and the accuracy.

For experimental purposes the University of Applied Sciences in Winterthur has a brine/water heat pump plant whose rotation speed is freely adjustable. Energy is extracted from an electrically heated brine circulator and is transfered to the consumer supply which in turn gives off the heat to cool water, thus simulating the house to be heated.

The energy supply is determined by a mathematical model of a house. Changes in external temperature and solar radiation can be simulated by programming this model.

A control system has been developed for the whole heat pump plant which maintains the internal temperature of the house at a constant given level, and in doing this optimizes energy use and costs. Information from weatherforcasting systems is also integrated in this control system. All control units of the system operate by Fuzzy Control. By means of a simulation of the whole system the main control units can be optimized and have shown the expected results in the first test runs.

The model "pendulum on an inclined plane" is an idea conceived and implemented by a team of students. From the lower surface of a moving positionable car hangs a swing pendulum. The car is to be moved in such a manner along the inclined plane from A to B, so that the pendulum sets gently on the target object at B. The pendulum may not overshoot at the target at B. At the beginning of the thesis, the mechanical model was assembled. Afterwards the model was wired, the sensors were installed and the necessary electronic circuits were sketched. For the regulation of the nonlinear unstable pendulum model we used a automatic status controller. Since only three can be measured by four statuses, we sketched a reduced observer for the estimation of the missing status. For us as beginning electrical engineers this thesis was a great challenge, since, for its solution, a drawer-spreading knowledge was needed.

Every body has tried to stand a rod vertically in the palm of their hand. The same principle is founded in our diploma model "inverted pendulum", which was developed by Quanser Consulting for education and research in control system design. An axis is mounted on a cart and a rod can turn around this axis. There is no drive between the cart and the rod. An potentiometer mounted on the axis of rotation allows you to measure the angle of the rod with the vertical axis. The objective of this experiment is to design a controller that would stabilize the rod and keep the cart in a desired position.

Because the system is unstable it was necessary to design a mathematical model of this physical system. Based on this model we have been able to develop several controllers.At first we developed a conventional controller which stabilize the rod in the vertical position. The position of the cart is controlled by a second conventional controller. With this structure of control we obtained satisfactory results.

Modern control theory is based on the state feedback controller. Because of this, we designed such a controller. If you have the state variable, it is possible to design and build a state feedback controller very easily and quickly (if you can measure all states). With this controller we obtained almost the same results as with the conventional controllers.

A second objective of this experiment is to design a self-erecting mode. The controller is consists of two main parts. One is the "swing up" controller while the second is the "balance" controller. The swing up controller oscillates the cart until the rod is almost upright, at which point the "balance" controller is activated and keeps the rod in the vertical position. If you tap the rod, in its vertical position the balance controller will try to bring the rod back to its old vertical position. If the tap is, too hard, the rod will fall down and the "swing up" controller will start to operate again.

This thesis describes the development of different conventional controls for the helicopter model. The work is based on project Wl_PA1_98_3 and comprises the examination of the cross coupling.

We discovered that a non-linear function had been wrongly identified in our project. A further fault was the linearisation of the non-linear model at an incorrect position. These inadequacies needed to be rectified.

After these corrections had been made a non-linear simulation could be designed. This simulation was a close approximation to reality. It enabled the presentation of all desired physical magnitudes, which helped in the understanding of the model.With the aid of the linearised model various conventional controls could be calculated. These control systems could be simulated on the non-linear model. This allowed the evaluation of whether the control system worked within the linear calculable range or in the non-linear range. A control system working in the non-linear range can improve the reference reaction of the control loop considerably.

In the biaxial helicopter model one axle influences the other axle. The phenomenon of both axles influencing each other is called cross coupling. A feedforward control efficiently reduces this effect.As the disturbance signal could not be measured directly, it needed to be calculated with our linearised model. The feedforward controls were optimised for the control systems utilised.

With these conventional control structures an illustration of the limitations of the stabilisation of the unstable helicopter model was possible.

The various control system including feedforward control were calculated using Matlab. The non-linear helicopter model was implemented and simulated in Simulink. The control processor could be programmed easily with RealLink32.