Supervisors:

Prof. Dr. Sven Pawletta, Wismar University, Germany
Prof. Dr. Thorsten Pawletta, Wismar University, Germany
Dr. Krzysztof Oprzedkiewicz, Tarnow Polytechnic University, Poland

Thanks to Dr. Lutz Mohr from the University of Rostock, who constructed the railroad system hardware.

General problem

In many industrial areas the production strategy is changing to a customer oriented production. That means, manufacturing systems have to support production of very different products. Such manufacturing systems are called high flexible manufacturing systems. High flexible manufacturing systems have high demands on the performance and variability of their integrated material handling systems. This necessary degree of variability can often not be realized with classical control techniques like PLC programming. Goal of the entire research project is the development of a computer based approach, that realizes the necessary flexible and intelligent process control. Main idea of this approach is the integration of dispositive and operative control tasks. Process coupled simulation models consist of different layers. The lowest layer models the operative control layer with hardware connections to plant sensors and actors. This layer should replace the classical PLC programming. The upper layers are built by dispositive simulation models for long term production planning. Both layers are connected by a feedback loop under real time conditions. In this structure the decision oriented upper model layers are realizing a controller for the operative layer. Typical control tasks are workpiece scheduling and production resource management.

Figure 1: Classical realization of a PLC based process control (left) vs. process coupled simulation approach (right)

Testing of developed control strategies on real systems is expensive and can be dangerous, so we need to create a model of the plant and its control. That is the motivation of the thesis, which contains a short introduction in process coupled simulation concept, a description of used hardware and experiments with developed software layers. An experimental transportation system is used as plant model in this project, that corresponds to a part of a real bike production line.

Main goals

Analysis of the experimental transportation system (the plant model) and its controlling interfaces

Development of an interface between the experimental system and MATLAB®

Development of an interface between the experimental system and Simulink®

Development of a Simulink® model to control the system with a single autonomous transportation unit

Development of a Simulink® model to control the system with multiple autonomous transportation units

Structure of the system and communication layer

A railroad system model is used, because it is probably least expensive and most efficient due to its similarity to the real system, where autonomous transportation units are used. The model is built in the N scale with use of DCC NMRA standard for digital control.

Construction layer

N-scale track

Turnouts (two-coil devices)

Double turnouts (double two-coil devices)

Turntable (slow motion motor device)

Locomotives

Tillig AC power supply (sec. I 16,5V, max. 2,73A used)

Figure 2: Structure of the railroad system

Control and feedback layer

Sensors (buffers and solenoid switch)

S88 feedback bus modules

Lenz LS100 DCC accessory decoders (turnouts and turntable control)

Viessmann 5551 universal relays for additional use with turntable control

Lenz LE080XS DCC locomotive decoders

Intellibox digital control unit by Uhlenbrock

Communication between the PC and Intellibox is realized with Maerklin 6050/6051 protocol (Motorolla P50)

Figure 3: Control and feedback layer

MATLAB® level interface

After analysis of the existing low-level MATLAB® interface, that maps Motorolla P50 protocol commands, we added a more abstract interface layer as shown in figure 4.

Figure 4: MATLAB® interface structure

Simulink® level interface

To enable further research in creating control simulation we moved the interface to the real simulation environment Simulink®. Experiments with this layer allowed us to achieve the most comfortable and probably most suitable interface based on information exchange between control and interface layers over global variables.

Figure 5: Simulink® layer interface blockset

On this layer we based our PLC style control strategy to test the interface functionality with multiple transportation units.

Figure 6: Simulink® control system structure

Summary and conclusions

Global variable based Simulink® interface layer seems to be the most suitable way for the purposes of the project task. Although, it can be simplified by using another type of variables (not binary) or using connections instead of some global variables. Its design also depends on control strategy, if the interface should be more sophisticated or simplified. PLC control strategy developed seems to be not flexible enough for such a big system and is created only for interface testing purposes. Now, when model and interface are developed and investigated, creating of top level control layer, which is the goal of the whole project, and testing it, are possible.

---

Pictures of the model plant and associated links can be found here

---

Maciej Salamon, Wismar, April 2002