Adaptive and Intelligent Systems
Applications 1994-1999 - Final Report

10. Tool environments and development methods 2 (TOOLMET-2)

10.1 Applying optimization methods for production planning in steel and electronic manufacturing (PROPOS)

Abstract

Optimization and planning systems were created in an electronic manufacturing case and for a steel manufacturing case. The purpose of the systems was to help the production planners quickly prepare good production plans, which are within the constraints and could be easily tested under varying situations. The developed systems were case dependant, but used architecture is general and could be used also in other cases, some of which are already in progress. The systems use stochastic local search with linear programming, heuristics and tabu list as methods, but don`t exclude any method that is seen as appropriate. The results that have been obtained this far have been good and some further development will be carried out to finalize the system.

Results

Business achievements

The developed optimization tools are supporting their users with following benefits:

Case 1: Better service to the customer (more products delivered on time), better knowledge of the whole process, better plans (faster throughput time).
Case 2: Good plans can be made very quickly by computer, which gives good possibility to react to the customer needs and to the sudden changes. The optimization tool gives lots of time from routine work to developing activities.

Engineering achievements

Every participating company is using the developed technology and the learnt skills to make their business more effective: two new cases are already in progress. VTT and Kaski Tech Oy will both conduct further development to make the planned software architecture of use in new applications.

Scientific achievements

The results show that simple methods together with intelligent application and today’s effective PC-computers can provide very good and useful benefits in production planning in industry.

Many algorithms have been tested and their suitability in these cases has been evaluated. It has been learned that stochastic methods are well-suited to this kind of problems, because a large number of constraints can be added. The user can easily understand the working method of the tool and therefore he can later finr-tune and even change the application without changing the source files.

It has been shown that different methods should be connected to hybrid solution to quickly identify the best solution. For example, in one case four different stages were used to reach the best solution: linear programming in preliminary allocation, first a genetic algorithm, then a local search in main optimization and finally end allocation to limit the data of the earlier stages.

Project information

Participants

Companies:

Outokumpun Polarit Oy (Steel manufacture)
Salcomp Oy (Electronic manufacturing)
Kaski Tech Oy (Information tecnology)
Rautaruukki Steel (Steel manufacture)

Research institute:

VTT Electronics (Technical Research Centre of Finland, VTT)

Project dates

Starting date: 1.3.1998, ending date: 31.12.1999

Project volume

Total budget: FIM 2.200.000, (55 man months)

Project manager

Jorma Hintikka
VTT Elektroniikka
Kaitoväylä 1, PL 1100, 90571 Oulu
Tel: (08) 551 2285
Fax: (08) 551 2320
E-mail: jorma.hintikka@vtt.fi

Publications

Riikonen, J., Haataja, K. Interactive Planning Environment for Cold-Rolling Mill., UK PLANNING AND SCHEDULING SPECIAL INTEREST GROUP, 17^TH WORKSHOP. 9-10.9.1998. Huddersfield, UK.

Hintikka, J., Haataja, K., Kyllönen, V., Applying Optimization Methods for Production Planning in Electronic Manufacturing. In: Proceedings of TOOLMET'99 Symposium - Tool Environments and Development Methods for Intelligent Systems. Oulu, Finland, 15-16 April, 1999. University of Oulu. pp. 135-142. ISBN 951-42-5176-8.

Haataja, K., Hintikka, J., Kyllönen V., Simple approach to production planning in electronic manufacturing. Newsletter of the Finnish Artificial Intelligence Society (FAIS), Special issue 1/99 on the IJCAI'99 affiliate event Networks'99. 8-10.8.1999. Helsinki, Finland.

Report of the PROPOS project.

Hautamäki T., Pro Gradu thesis (in Finnish): Aikataulutusongelman ratkaisumenetelmistä: sovelluksena simulated annealing -menetelmän käyttö töiden aikatauluttamisessa. s. 48 + encl., University of Oulu.Finland. 1998. (Approach to solution of timing problems, case: simulated annealing in timing.)

Hautamäki T. Special study: Verifying optimization methods to timing problem. University of Oulu. Finland. 1999.

To Appear: Diploma thesis: Saari M., Applying Genetic Algorithm to timing problem in production planning. University of Oulu. Finland.

More information about VTT Electronics can be found at the internet address www.ele.vtt.fi .

10.2 Model-based process analysis (MODIPRO)

Abstract

The MODIPRO subproject continued the work started in the earlier USTI subproject. Its goal was to develop methods that combine different sources of knowledge and to apply them to fuzzy systems and multilevel expert system type of reasoning. The subproject aimed to develop generic tools and tool environments for different types of adaptive and intelligent systems applications and to integrate these into design environments and with actual control systems. Four industrial case studies were carried out: Intelligent cooking liquor analyser (ABB Industry), Web break sensitivity indicator (Valmet Oyj), Expert system for X-ray inspection (Nokia) and Quality forecasting tool (Nokia).

Results

Business achievements

All case studies have lead at least to industrial prototypes. Utilisation remains the responsibility of the industrial partners. Benefits of the developed systems include better quality, less human work, and higher production. X-ray inspection system is already in industrial use in production of printed circuit boards for telecommunication devices. Prototypes from other cases are undergoing industrial testing.

Engineering achievements

Intelligent X-ray inspection system reduces false alarms very efficiently. Models for cooking liquor analyser have been tested in one digester house and testing with another has started. The LE based soft sensor provides accurate Kappa-number estimates under variable operating conditions. Web break sensitivity indicator was tested during summer 1999 on one paper machine and development continues. The resulting break sensitivity value correlates well with actual break levels. Quality forecasting tool utilises huge testing and production databases. According to the tests, critical packages can detected, and the results are being used to include intelligence in the system.

Scientific achievements

The development of the Linguistic Equation approach was continued and the method was used in cooking and paper machine applications. Original method was extended to the direction of Case-Based Reasoning systems and multiple area modelling. Tuning techniques have been improved. Correlation analysis, principal component analysis and expertise are combined with linguistic equations in variable selection. Delays and time constants are handled with fuzzy numbers. Strongest interactions with a considerable difference between break categories are selected to the models. Case detection is based on these models and fuzzy logic.

Intelligent cooking liquor analyser was developed by combining automatic data analysis with expert knowledge. Wavelets provided good results in data pre-processing. Both single and multiple LE models were studied, and the LE approach has been compared with statistical methods, neural networks and fuzzy set systems. Similar accuracy can be achieved with various techniques but the LE approach provides strong benefits in its adaptability to different operating conditions.

In the X-ray application, a new method for the online generation of membership functions was developed. This method is based on analysing differences between acceptable and defective joints. For the quality forecasting tool, statistical analysis has been used in detecting critical packages. Variations in fault levels have been analysed in the development of an intelligent tool. A study on hardware realisation of the linguistic equation was started in the MODIPRO subproject and continued as a separate project.

Project information

Participants

Companies:

ABB Industry, Pulp and Paper
Valmet Oyj, Paper Machines
Nokia Access Systems Oy

Research institutes:

University of Oulu, Control Engineering Laboratory

Project dates

Starting date: 1.3.1998, ending date: 31.12.1999

Project volume

Total budget in FIM 2.100.000, 103 man months

Project manager

DI Esko Juuso
University of Oulu
Control Engineering Laboratory
P.O. Box 4300, FIN-90401 OULU, FINLAND
Tel: +358 8 5532463
Fax: +358 8 5532466
E-mail: Esko.Juuso@oulu.fi

Publications

Tool and methodology development

Juuso, E. K.:.Fuzzy Control in Process Industry: The Linguistic Equation Approach. In: Verbruggen, H. B., H.-J. Zimmermann and R. Babuska, editors, Fuzzy Algorithms for Control, International Series in Intelligent Technologies, pp. 243-300. Kluwer, Boston.

Juuso E. K.: Modelling and Control with Linguistic Equations. European Symposium on Intelligent Techniques (ESIT ’99). Orthodox Academy of Crete, Greece, June 3-4, 1999. (CD-ROM).

Juuso E. K.: Fuzzy Expert Systems. European Journal of Pharmaceutical Sciences, 1999, Volume 8, pp. S15-16.

Juuso E. K.: Linguistic Equation Based Data Analysis for Fault Diagnosis and Forecasting. Proceedings EUFIT’99, CC-6-4, 8 pp. (CD-ROM).

Juuso E. K.: Multiple Model Intelligent Simulation with Linguistic Equations and Fuzzy Logic. In: SIMS’99 – the 1999 Conference of the Scandinavian Simulation Society, Linköping, Sweden, October 18-19, 1999, Peter Fritzson (ed.), Linköpings Universitet, Linköping, Sweden, 1999, pp. 192-199.

Juuso E. K.: Intelligent Systems Design with Linguistic equations. 9^th Workshop Fuzzy Control des GMA-FA 5.22 am 4/5.11.1999, Dortmund, Deutschland. Forschungsberict Nr. 0449, Universität Dortmund, Facultät fur Electrotechnik, Dortmund, 1999, pp. 177-196.

Juuso E., Kangas P.: Compacting Large Fuzzy Set Systems into a Set of Linguistic Equations. Report A No X, March 2000, University of Oulu, Control Engineering Laboratory, 23 pp.

Applications

Juuso E., Leiviskä K.: Intelligent Systems in Electronics Manufacturing. Proceedings of TOOLMET'99 Symposium - Tool Environments and Development Methods for Intelligent Systems, Oulu, April 15-16, 1999 (L. Yliniemi & E. Juuso, eds), Oulun yliopistopaino, Oulu, 1999, pp. 143-154.

Juuso E., Leiviskä K.: Oppivat ja älykkäät menetelmät elektroniikkatuotannossa. Elektroniikkatuotannon oululaista osaamista (M. Berg, toim.), ProElectronica, Oulu, 1999, pp.36-55.

Leiviskä K.: Industrial Applications of Intelligent Methods. Proceeding of EUFIT’98, Aachen, Germany, September 7-10, 1998. Volume 3, pp. 1449-1457.

Leiviskä K.: Applications of Intelligent Methods in Primary and Process Industries. European Symposium on Intelligent Techniques (ESIT ’99). Orthodox Academy of Crete, Greece, June 3-4, 1999. (CD-ROM).

Modelling of pulp characteristics

Ikäheimonen J., Juuso E., Leiviskä K., Murtovaara S.: Sulfaatisellun vuokeitto - menetelmät, keiton ohjaus ja massan pesu. Raportti B, Maaliskuu 2000, Säätötekniikan laboratorio, Oulun yliopisto, Oulu. 48 s.

Ikäheimonen J., Juuso E., Leiviskä K., Murtovaara S., Sutinen R.: Keittolipeä- ja massa-analyysi sellun keitossa ja pesussa. Raportti B, Maaliskuu 2000, Säätötekniikan laboratorio, Oulun yliopisto, Oulu. 35 s.

Isokangas A., Juuso E.: Fuzzy Modelling with Linguistic equations. Report A No 11, February 2000, Control Engineering Laboratory, University of Oulu, Oulu, 25 p.

Lemmetti A., Murtovaara S., Sutinen R. and Leiviskä K.: Cooking Variables Affecting the Kraft Pulp Properties. Report A No 7, June 1999, Control Engineering Laboratory, University of Oulu, Oulu. 27 pp.

Murtovaara S., Leiviskä K., Juuso E. K., Sutinen R.: Modelling of Pulp Charasteristic in Kraft Cooking. Report A No 9, December 1999, Control Engineering Laboratory, University of Oulu, Oulu. 20 pp.

Murtovaara S., Juuso E.K., Leiviskä K.: Keittopitoisuusmittausten käyttäminen kappaluvun määrittämisessä. Automaatio '99, Helsingin messukeskus, 14. - 16.9.1999, Helsinki 1999, ss. 265 - 269.

Murtovaara S., Juuso E. K., Leiviskä K., Ikäheimonen J.: Kappa Number Prediction Based on Cooking Liquor Measurements. In Proceedings of TOOLMET'99 - Tool Environments and Development Methods for Intelligent Systems, Oulu, April 15 - 16, 1999 (L. Yliniemi and E. K. Juuso, ed.), University of Oulu, Control Engineering Laboratory, Oulu 1999, pp. 210 - 216.

Murtovaara S., Juuso E., Sutinen R., Leiviskä K.: Modelling of Pulp Cooking Characteristics by Using Fuzzy Logic and Linguistic Equations, Preprints of TOOLMET'98 Symposium - Tool Environments and Development Methods for Intelligent Systems, Oulu, April 16-17, 1998 (L. Yliniemi & E. Juuso, eds), Oulun yliopistopaino, Oulu, 1998, pp. 90-96.

Murtovaara S. Juuso E. K. Sutinen R. and Leiviskä K.: Neural Networks Modelling of Pulp Digester. In Proceedings of CONTROLO'98, 3^rd Portugese Conference on Automatic Control, Coimbra, Portugal, September 9 – 11, 1998 (A. Dourado et al., eds), APCA, Portugal, 1998, pp. 627 - 630.

Murtovaara S., Lemmetti A., Juuso E. K., Sutinen R., Leiviskä K.: Kappa Number Prediction Based on Cooking Liquor Measurements. In Proceedings of 6th International Conference on New Available Technologies - The World Pulp and Paper Week, June 1 - 4, 1999, Stockholm, Sweden, SPCI - The Swedish Associations of Pulp and Paper Engineers, Stockholm, Sweden 1999, p. 633. (Poster)

Web break indicator

Ahola T., Ruuska J., Juuso E., Leiviskä K.: Web Break Sensitivity Indicator. Proceedings of TOOLMET'99 Symposium - Tool Environments and Development Methods for Intelligent Systems, Oulu, April 15-16, 1999 (L. Yliniemi & E. Juuso, eds), Oulun yliopistopaino, Oulu, 1999, pp. 202-209.

Ahola T., Ruuska J., Juuso E., Leiviskä K.: Paperikoneen katkoherkkyysindikaattori. Raportti B No 14, Helmikuu 2000, Säätötekniikan laboratorio, Oulun yliopisto, Oulu, 35 s.

Juuso E., Ahola T., Oinonen K.: Case-Based Reasoning with Linguistic equations for Web Break Sensitivity Indicator, Preprints of TOOLMET'98 Symposium - Tool Environments and Development Methods for Intelligent Systems, Oulu, April 16-17, 1998 (L. Yliniemi & E. Juuso, eds), Oulun yliopistopaino, Oulu, 1998, pp. 97-106.

Juuso E., Ahola, T., Oinonen, K. and Leiviskä, K.: Web Break Sensitivity Indicator for a Paper Machine. Proceedings of 6th European Congress on Intelligent Techniques and Soft Computing, Aachen, Germany, September 7-10, 1998. Volume 3, pp. 1549-1553.

Ruuska J.: Paperikoneen katkoherkkyysindikaattori. Diplomityö.Oulun yliopisto, Oulu, 1999. 57 s.

X-ray solder joint inspection

Xu Wei, Frantti T., Verkasalo M., Lappalainen K., Juuso E., Leiviskä K.: A Real Time Fuzzy Expert System for X-ray Solder Joint Inspection, Preprints of TOOLMET'98 Symposium - Tool Environments and Development Methods for Intelligent Systems, Oulu, April 16-17, 1998 (L. Yliniemi & E. Juuso, eds), Oulun yliopistopaino, Oulu, 1998, pp. 114-120.

Xu Wei, Verkasalo M., Lappalainen K., Juuso E., Leiviskä K.: Building A Fuzzy Expert System For Solder Joint X-Ray Inspection From On-Line Data. Proceedings EUFIT’99. (CD-ROM).

Quality forecasting tool

Juuso E., Jokinen T., Ylikunnari J., Leiviskä K.: Quality Forecasting Tool for Electronics Manufacturing. Report A No X, March 2000, University of Oulu, Control Engineering Laboratory, 38 p.

Jokinen T.: Elektroniikkatuotteen laadunennustustyökalun käyttöliittymä. Oulun yliopisto, Oulu, 1999. 70 s.

Rönkä K., Juuso E., Jokinen T., Lotvonen S.: Elektroniikkatuotteen laadunennustustyökalu. Automaatio '99, Helsingin messukeskus, 14. - 16.9.1999, Helsinki 1999, ss. 320-325.

Statistical methods and SPC

Gebus S.: Process Control Tool for a Production Line at Nokia. Report A, March 2000, University of Oulu, Control Engineering Laboratory, 23 p.

Sievola H., Juuso E.: Tilastollisten menetelmien käyttö piirilevyn asettelu- ja juotostarkastuksen simuloimiseen elektroniikkateollisuudessa. Raportti B, Maaliskuu 2000, Säätötekniikan laboratorio, Oulun yliopisto, Oulu. 15 s.

Hardware realisation of linguistic equations

Donnini P. A.: Linguistic equations and their hardware realisation in image analysis. Oulu, June 1999, University of Oulu, Control Engineering Laboratory, Report A No 8. 23 p.

Forecasting

Alajärvi K., Juuso E.: Aikasarjoihin pohjautuva älykäs ennustaminen DataEngine ohjelmistoa käyttäen. Raportti B, Maaliskuu 2000, Säätötekniikan laboratorio, Oulun yliopisto, Oulu, 23 s.

Juuso E., Alajärvi K.: Time Series Forecasting with Intelligent Methods. Report A, March 2000, University of Oulu, Control Engineering Laboratory, 25 p.

TOOLMET Symposium

Yliniemi L., Juuso E.: Preprints of TOOLMET'98 Symposium - Tool Environments and Development Methods for Intelligent Systems, Oulu, April 16-17, 1998, Oulun yliopistopaino, Oulu, 1998, 156 pp.

Yliniemi L., Juuso E.: Proceedings of TOOLMET'99 Symposium - Tool Environments and Development Methods for Intelligent Systems, Oulu, April 15-16, 1999, Oulun yliopistopaino, Oulu, 1999, 230 p.

jukka.iivarinen@hut.fi
http://www.cis.hut.fi/neuronet/Tekes/10.shtml
Monday, 27-Nov-2000 16:18:00 EET