0,1

0,05

0

-0,05 0 50 100 150 200 250 300 350 400 450 500

Fig.7. Relative errors for the determination of parameter t1 with net Fig.6a

Relative errors of time t3

0,025

0,02

0,015

0,01

0,005

0

-0,005 0 50 100 150 200 250 300 350 400 450 500

Fig.8. Relative errors for the determination of parameter t3 with net Fig.6b

3. LABORATORY EXPERIMENTS

The final evaluation of each experimental approach can be done only in a laboratory. The proper laboratory development system has been built and equipped in Institute of Automation and Robotics at Warsaw University of Technology, Fig. 9.

Fig. 9. Testing stand for the developed control approach.

The investigated system was composed of a Festo pneumatic cylinder DSNU-25-400 PPV-A, four fast switching valves Festo MHE4_MS1h-3/2G and 10 dcm buffer volume. The valves were controlled by PC equipped with dSPACE 1102 board.

Many experiments has confirmed quite good properties of the proposed approach. On Fig 10a and 10b are presented some transients received in experiments.

a. displacement velocity acceleration

4. CONCLUSIONS

The paper presents an approach for pneumatic positioning system and only some main problems investigated at research, but it shows an important conclusion: the fast pneumatic posi-tioning system with a low air consumption and low price can be technically realized. It has cheap components, is able to perform the positioning with the acceptable accuracy. The problems with displacements of short range and close to the cylinder covers will be next tested with some adaptive ap-proach for removing of the resulting errors.

An application of such systems can be very attractive for all of the manipulation or packaging tasks in which electrical systems are not allowed or are to expensive. Recently some reduction of application number of pneumatic positioning drives due to high cost of the compressed air is observed. The proposed approach can reduce at least this disadvantage and allow the pneumatic systems to keep its competitive position on the market.

The limited accuracy of the described approach results from the open-loop control, that does not have a feed-back action within the same piston displacement, but for applications working in repetitive cycles, the feed-back can be replaced in part by the adaptive action of algorithm.

5 . REFERENCES

b.

Control of valve I. & II.

Control of valves III. &

Fig.10 Example of transients recorded at experiment: a) piston displacement, velocity and acceleration, b) controls settled to the valves (shown in pairs) for

each cylinder volume.