Stochastic Methods

Research & Projects

2004-...: Bioinformatics

2002-2003: Activity report (in french)

1986-...: List of publications (partial)

This page is divided into four parts :

Research topics

• Automatic Learning
• Data Mining
• Power Systems

Collaborations and contracts

Events, conferences, work groups

Publications

---

Research topics

Automatic Learning

The research followed in the service is mainly in the field of automatic learning and more particularly in machine learning. The described methods in what follows were applied to various real problems of automatic learning (see industrial collaborations). The contributions to every method are indicated by means of pointers towards the corresponding publications. 

Symbolic Learning

Symbolic Learning aims at extracting from a database information in the form of rules of type: 'If <condition> then <conclusion>'. These methods have the advantage of providing interpretable results (that can be then validated by experts). In addition to their interpretability, the advantages of the methods based on construction of trees are their effectiveness (they are applicable on very large databases) and the possibility to provide additional information (selection of attributes, correlation between attributes...). 
The first work of our group in this field goes back to the years 1985-86. 

1. Classical decision trees (criteria of partitioning, stop splitting, pruning based on a measure of information quality, variance study and reduction)
2. Regression trees 
3. Fuzzy trees
4. Temporal decision trees

Generalized additive models

In this family one finds methods such as certain models of artificial neural networks, as well as a great number of statistical methods. The objective of these methods is to extend to the field of non-linear, the well known methods of linear regression and discrimination. These methods are well adapted to the modeling of the non-linear functions of R^n in R^p. 

1. Multilayer perceptrons
2. Radial basis functions
3. Projection pursuit : orthogonal and oblique models
4. Unidimensional approximation (scatter-plot fitting): combination of linear or cubic models per pieces with techniques of filtering (moving average, supersmoother)

Models based on prototypes

These techniques are based on the exploitation of a distance measure or a similarity measure, and consist in extrapolating unknown information for an object starting from known information relating to the objects which resemble the most to it. 

1. KNN approach with adaptive metric 
2. Clustering techniques : Kmeans method and ascending hierarchical regrouping

Hybrid approaches

The various methods described above being of limited range, it is very often interesting to combine them. One can thus formulate various ' hybrid' approaches

1. Side by side   (toolbox approach) 
2. One after the other (one uses the information provided by a method before applying the following one) 
3. Integrated 

 

Data mining

Data mining (data exploration) indicates on the one hand, an activity which consists in using the methods of automatic learning on well defined problems to build interesting models, and on the other hand, a field of research, the purpose of which is to develop methodologies facilitating this activity. 

GTDIDT is a data mining software written in GCL (Gnu Common Lisp). It was developed for research purposes, but is also used by the students within the framework of a course on automatic learning and is also used as tool in all the applications of data mining presented on this page. It contains various automatic learning algorithms (induction of decision and regression trees, linear and generalized regression, correlation analyses and clustering, search based on the nearest neighbor, tools for visualization). 

Some practical applications of data mining on which we worked: extraction of rules of conduct for the electric power systems, identification of the risks of instability in a power system, analysis of costs, fault detection, load forecasting. 

Power Systems

Automatic learning applications and data mining

This research, initialized by the University of Liege in 1985, is currently carried out in collaboration with various industrial partners (see collaborations). The general objective of this research is to develop decision-making tools for engineers responsible of the design and operators responsible of the operation of power systems. The developed methodology combines techniques of Monte-Carlo simulation (generation of databases) with automatic learning (exploitation of the databases). 

The objective of this research is to develop a method for the evaluation of risks of system insecurity and associated consequences (severity). 

The probabilistic approaches and/or based on automatic learning, described above, make use of the Monte-Carlo simulation. Several software tools for Monte-Carlo simulations were developed within the framework of industrial collaborations. 

Development of tools for control assistance (CCLAB)

Coupling automatic learning with the platforms of control in real time (project ALDYSA). See also Web page of the CCLab.

< <

---

Collaborations and contracts with the industry

• Electricité de France (1990-) 
• Hydro-Québec (1992-1994; 1998-) 
• Electrabel (1997-) 
• Alstom/ESCA (1996-) 
• Automation & Robotics (1999-) 
• Cockerill-Sambre (1999-) 
• CRM (1999-) 

< <

---

Events, conferences, work groups

Seminars 

The service of stochastic methods in common with the service of electric circuits organizes in Montefiore Institute a series of seminars aiming at presenting the progress of the work of the researchers of each one of these services. These seminars are opened to all and are announced by electronic mail (twice a month). . 

Symposiums and conferences

• ELECTRIC UTILITIES RESTRUCTURING

Wednesday December 8, 1999.
• Chair IBM

on Data Mining, organized in common with University of Anvers.
• 'Power system operation and control : A new beginning', Montefiore/CClab/Seminar, from 22 to 25 March 1999, Liège, Belgium.
• Workshop Decision Trees (in 2000)

Work groups

• Task-Force Cigré 38.02.21 "Security assessment"

< <

---

Publications

Books 

Automatic learning techniques in power systems;
L. Wehenkel.
Kluwer Academic Publishers , Boston, 1997, 310 pages.

---

Theses 

Une approche de l'intelligence artificielle appliquée à l'évaluation de la stabilité transitoire des réseaux électriques;
L. Wehenkel.
Ph.D. thesis. Faculty of Applied Sciences. University of Liege, May 1990. 

Machine learning approaches to power system security assessment;
L. Wehenkel.
Aggregation thesis. Faculty of Applied Sciences. University of Liege, May 1994.

---

Articles 

The publications are grouped by theme of research, certain articles being able to appear in several sections. Within a given section, the articles are classified by reversed chronological order. To obtain a postscript version for certain articles, go to the home pages of their authors.

Automatic Learning Methods

Application of Automatic Learning Methods and Data Mining

Power System Stability

---

Automatic Learning Methods

Investigation and reduction of discretization variance in decision tree induction; 
P.Geurts and L. Wehenkel;
To appear in Proc. of ECML'2000, European Conference on Machine Learning, Barcelone, juin 2000. 

Automatic induction of fuzzy decision trees and its application to power system security assessment;
X. Boyen, L. Wehenkel.
Int. Journal on Fuzzy Sets and Systems, Vol. 102/1, pp. 3-19, 1999. 

Distance Rejection in the Context of Electric Power System Security Assessment Based on Automatic Learning;
I. Houben, L. Wehenkel.
Proc. Int. Conf. Statistical Pattern Recognition, Sidney, Aug. 1998. 

Early prediction of electric power system blackouts by temporal machine learning; 
P. Geurts, L. Wehenkel.
Proc. of ICML98-AAAI98 Workshop on "Predicting the Future: AI Approaches to Time-Series Problems", 1998. 

On neurofuzzy and fuzzy decision trees approaches;
C. Olaru, L. Wehenkel (invited paper).
Proc. of IPMU'98, 7th Int. Congr. on Information Processing and Management of Uncertainty in Knowledge based Systems, Paris, 1998

The hinges model : a one-dimensional continuous piecewise linear model;
E. F. Sánchez-Úbeda, L. Wehenkel.
Proc. of IPMU'98, 7th Int. Congr. on Information Processing and Management of Uncertainty in Knowledge based Systems, Paris, 1998

Genetic algorithm-based k-NNs;
I. Houben, L. Wehenkel, M. Pavella.
Proc. of IFAC-IFIP-IMACS Conf. on Control of Industrial Systems, Belfort, May 1997.

Discretization of continuous attributes for supervised learning - Variance evaluation and variance reduction; 
L. Wehenkel;
Proc. of IFSA'97, International Fuzzy Systems Association World Congress, (invited paper), Prague, June 1997. 

Automatic learning. Overview of methods and potentials;
L. Wehenkel.
Special number of the Journal E - SRBE, ``Apprentissage automatique - Applications aux réseaux d'énergie électrique'', December 1996. 

On uncertainty measures used for decision tree induction;
L. Wehenkel.
Proc. of IPMU'96, 6th Int. Congr. on Information Processing and Management of Uncertainty in Knowledge based Systems, Granada, July 1996, pp. 413-418. 

Coupling of K-NN with decision trees for power system transient stability assessment;
I. Houben, L. Wehenkel, M. Pavella.
Proc. of IEEE CCA, 4th Int. Conf. on Control Applications, Albany (NJ), Sept. 1995, pp. 825-832. 

Decision tree pruning using an additive information quality measure; 
L. Wehenkel.
Part of the chapter 5 : Measures of information, Intelligent Systems with Uncertainty, Eds. B. Bouchon-Meunier, L. Valverde, R.R. Yager, Elsevier, 1993 , pp. 397-411. 

A hybrid decision tree - neural network approach for power system dynamic security assessment;
L. Wehenkel, V.B. Akella.
Proc. of the 4th Int. Symp. on Expert System Application to Power Systems, Melbourne, Australia, Jan. 4-8, 1993, pp. 285-291. 

An information quality based decision tree pruning method; 
L. Wehenkel. 
Proc. of IPMU'92 - 4th Int. Congr. on Information Processing and Management of Uncertainty in Knowledge based Systems, Palma de Mallorca, July 6-10, 1992, pp. 581-584. 

---

Application of Automatic Learning Methods and Data Mining

Probabilistic design of power system special stability controls;
L. Wehenkel, C. Lebrevelec, M. Trotignon, J. Batut.
Control Engineering Practice, Vol. 7, No. 2, pp. 183-194, 1999. 

Distance Rejection in the Context of Electric Power System Security Assessment Based on Automatic Learning;
I. Houben, L. Wehenkel.
Proc. Int. Conf. Statistical Pattern Recognition, Sidney, Aug. 1998. 

Automatic induction of fuzzy decision trees and its application to power system security assessment;
X. Boyen, L. Wehenkel.
Int. Journal on Fuzzy Sets and Systems, Vol. 102/1, pp. 3-19, 1999. 

Early prediction of electric power system blackouts by temporal machine learning; 
P. Geurts, L. Wehenkel.
Proc. of ICML98-AAAI98 Workshop on "Predicting the Future: AI Approaches to Time-Series Problems", 1998. 

Visualizing dynamic power system scenarios for data mining;
P.Geurts, L.Wehenkel.
Proc. of LESCOPE'98, Large Engineering Syst. Conf. on Power Engineering, Halifax, 1998. 

Hybrid adaptive nearest neighbor approaches to dynamic security assessment;
I. Houben, L. Wehenkel, M. Pavella.
Proc. of CPSPP'97 IFAC/CIGRE Symposium on Control of Power Systems and Power Plants, Beijing, August 1997, pp. 685--690. 

A probabilistic approach to the design of power system protection schemes against blackouts;
L. Wehenkel, C. Lebrevelec, M. Trotignon, J. Batut.
Proc. of CPSPP'97, IFAC/CIGRE Symposium on Control of Power Systems and Power Plants, Beijing, August 1997. 

Machine-learning approaches to power-system security assessment;
L. Wehenkel. 
IEEE Expert, Intelligent Systems & their Applications, Vol. 12, No. 5, 1997. 

A new probabilistic approach to improve security assessment;
C. Lebrevelec, L. Wehenkel, P. Cholley, J. P. Clerfeuille.
Proc. of IERE Workshop on Future Directions in Power System Reliability,, Palo Alto, CA, May 1997. 

Practical contribution of a statistical methodology to voltage security criteria determination;
Y. Jacquemart, L. Wehenkel, P. Pruvot.
Proc. of PSCC'96, 12th Power System Computation Conference, Dresden, Aug. 1996, pp. 903-910. 

Why and which automatic learning approaches to power systems security assessment;
L. Wehenkel, M. Pavella.
Proc. of the IEEE-SMC IMACS Multiconference CESA'96, Symposium ``Modelling, Analysis and Simulation'', Lille, July 1996, pp. 1072-1077. 

Contingency severity assessment for voltage security using non-parametric regression techniques; 
L. Wehenkel. 
IEEE Trans. on PS, Vol. 11, No. 1, 1996, pp. 101-111. 

Automatic learning. Overview of methods and potentials;
L. Wehenkel.
Special number of the journal E - SRBE, ``Apprentissage automatique - Applications aux réseaux d'énergie électrique'', Dec. 1996. 

Fuzzy decision tree induction for power system security assessment;
X. Boyen, L. Wehenkel.
Proc. of SIPOWER'95, IFAC Symp. on Control of Power Plants and Power Systems, Mexico, Dec. 1995, pp. 151-156. 

Statistical approaches to dynamic security assessment: the data base generation problem;
Y. Jacquemart, L. Wehenkel, T. Van Cutsem, P. Pruvot.
Proc. of SIPOWER'95, IFAC Symp. on Control of Power Plants and Power Systems, Mexico, Dec. 1995, pp. 243-246. 

Automatic learning approaches for on-line transient stability preventive control of the Hydro-Québec system - Part I. Decision tree approaches;
L. Wehenkel, I. Houben, M. Pavella, L. Riverin, G. Versailles.
Proc. of SIPOWER'95, IFAC Symp. on Control of Power Plants and Power Systems, Mexico, Dec. 1995, pp. 231-236. 

Automatic learning approaches for on-line transient stability preventive control of the Hydro-Québec system - Part II. A tool box combining decision trees with neural nets and nearest neighbor classifiers optimized by genetic algorithms;
L. Wehenkel, I. Houben, M. Pavella.
Proc. of SIPOWER'95, IFAC Symp. on Control of Power Plants and Power Systems, Mexico, Dec. 1995, pp. 237-242. 

Use of Kohonen feature maps for the analysis of voltage security related electrical distances;
L. Wehenkel, Y. Jacquemart.
Proc. of ICANN'95, Int. Conf. on Artificial Neural Networks, Paris, Oct. 1995, Industrial conference, pp. 8.3.1-8.3.7. 

Coupling of K-NN with decision trees for power system transient stability assessment;
I. Houben, L. Wehenkel, M. Pavella.
Proc. of IEEE CCA, 4th Int. Conf. on Control Applications, Albany (NJ), Sept. 1995, pp. 825-832. 

Machine learning approaches to power system security assessment;
L. Wehenkel. 
Proc. Workshop on Machine Learning in Engineering - International Joint Conference on Artificial Intelligence, IJCAI'95, Montréal, Aug. 1995, pp. 1-20. 

A statistical approach to the identification of electrical regions in power systems;
L. Wehenkel.
Proc. of Stockholm Power Tech, Joint IEEE-KTH International Power Conference, (IEEE paper # SPT PS 17-03-0237), Stockholm, June 1995, pp. 530-535. 

Decision tree based transient stability method - A case study; 
L. Wehenkel, M. Pavella, E. Euxibie, B. Heilbronn. 
IEEE Trans. on PS, Vol. 9, No. 1, 1994, pp. 459-469. 

Machine learning, neural networks and statistical pattern recognition for voltage security : a comparative study; (Extended version)
L. Wehenkel, T. Van Cutsem, M. Pavella, Y. Jacquemart, B. Heilbronn, and P. Pruvot.
Engineering Intelligent Systems for Electrical Engineering and Communications, Vol. 2, No 4, 1994, pp. 233-245. 

Multicontingency decision trees for transient stability assessment;
V.B. Akella, L. Wehenkel, M. Pavella, M. Trotignon, A. Duchamp, B. Heilbronn.
Proc. of PSCC'93, 11th Power System Computation Conference, Avignon, Aug., 1993, pp. 113-119. 

Decision tree approaches to voltage security assessment;
T. Van Cutsem, L. Wehenkel, M. Pavella, B. Heilbronn, M. Goubin.
Proc. of the IEEE, Part C, Vol. 140, No. 3, 1993, pp. 189-198. 

Decision tree approach to power systems security assessment; 
L. Wehenkel, M. Pavella. 
EPES, Vol. 15, No. 1, 1993, pp. 13-36. 

Evaluation de la sécurité en temps réel : approche par arbres de décision;
L. Wehenkel.
Actes de la journée d'études SEE, Intégration des techniques de l'intelligence artificielle dans la conduite et la gestion des réseaux électriques, Paris, Mar., 1993, pp. 11-20. 

A hybrid decision tree - neural network approach for power system dynamic security assessment;
L. Wehenkel, V.B. Akella.
Proc. of the 4th Int. Symp. on Expert System Application to Power Systems, Melbourne, Australia, Jan. 4-8, 1993, pp. 285-291. 

Decision trees for detecting emergency voltage conditions; 
T. Van Cutsem, L. Wehenkel, M. Pavella, B. Heilbronn, M. Goubin. 
Proc. of the Int. workshop on ``Bulk Power system Voltage Phenomena, Voltage Stability and Security'', Deep Creek Lake, Ma., USA, Aug. 1991, pp. 229-240. 

Decision trees and transient stability of electric power systems; 
L. Wehenkel, M. Pavella. 
Automatica, Vol.27, No.1, pp. 115-134, 1991. 

Decision trees for preventive voltage stability assessment; 
L. Wehenkel, T. Van Cutsem, M. Gilliard, M. Pavella, B. Heilbronn, M. Goubin. 
Proc. of the Int. workshop on ``Bulk Power system Voltage Phenomena, Voltage Stability and Security'', Deep Creek Lake, Ma., USA, Aug. 1991, pp. 217-228. 

An artificial intelligence framework for on-line transient stability assessment of power systems; 
L. Wehenkel, Th. Van Cutsem, M. Ribbens-Pavella.
IEEE Trans. on PS, Vol. PWRS-4, No.2, 1989, pp. 789-800. 

Inductive inference applied to on-line transient stability assessment of electric power systems; 
L. Wehenkel, Th. Van Cutsem, M. Ribbens-Pavella. 
Automatica, Vol.25, No.3, 1989, pp. 445-451. 

Artificial intelligence methods for on-line transient stability assessment of electric power systems; 
L. Wehenkel.
Proc. ESAP'89, 1st Int. Symp. on Expert Systems Application to Power Systems, Stockholm, Sweden, Aug. 22-26, 1988, pp. 5.1-5.8. 

Decision trees applied to on-line transient stability assessment of power systems; 
L. Wehenkel, Th. Van Cutsem, M. Ribbens-Pavella.
Proc. IEEE Int. Symp. on Circuits and Systems, Helsinki, Vol.2, June 1988, pp. 1887-1890. 

Artificial intelligence applied to on-line transient stability assessment of electric power systems; 
L. Wehenkel, Th. Van Cutsem, M. Ribbens-Pavella.
Proc. of the 10th IFAC World Congress, Münich, July 27-31, 1987, pp. 308-313. 

Machine learning applied to power systems transient security functions; 
L. Wehenkel, Y. Xue, Th. Van Cutsem, M. Ribbens-Pavella.
Proc. of the IMACS Int. Symp. on ``AI, Experts Systems and Languages in Modelling and Simulation'', Barcelona, Spain, 2-4 June, 1987, pp. 243-248. 

Artificial intelligence applied to on-line transient stability assessment of electric power systems; (Short paper) 
L. Wehenkel, Th. Van Cutsem, M. Ribbens-Pavella.
Proc. of the 25th IEEE Conf. on Decision and Control (CDC), Athens, Greece, December 10-12, 1986, pp. 649-650. 

---

Power System Stability

Transient and voltage stability criteria for real-time emergency control; 
L. Wehenkel, Y. Zhang, M. Pavella.
CIGRE, Conf. Int. des Grands Réseaux Electriques, Paris, Aug.-Sept. 1998. 

Generation allocation for transient stability-constrained maximum power transfer;
A. L. Bettiol, L. Wehenkel, M. Pavella.
Proc. of SEPOPE'98, Rio de Janeiro, 1998

Real-time transient stability emergency control of the south-southeast brazilian system;
D. Ernst, A. L. Bettiol, Y. Zhang, L. Wehenkel, M. Pavella.
Proc. of SEPOPE'98, Rio de Janeiro, 1998

An Approach to Real-Time Transient Stability Assessment and Control;
M. Pavella, L. Wehenkel, A. Bettiol, D. Ernst.
IEEE publication on Power System Stability, 1998

Electric power system dynamic security assessment; 
M. Pavella, L. Wehenkel.
to be published in ``Revue Internationale Générale d'Electricité'', 1998.

SIME : a comprehensive approach to fast transient stability assessment;
Y. Zhang, L. Wehenkel, M. Pavella.
Transactions of the IEE of Japan (Section B), Vol. 118-B, Jan. 1998. 

SIME : A hybrid approach to fast transient stability assessment and contingency selection;
Y. Zhang, L. Wehenkel, P. Rousseaux, M. Pavella.
EPES, Vol. 19, No. 3, 1997, pp. 195-208. 

Power system security assessment : a position paper;
R. J. Marceau, J. Endrenyi (convenors) and Cigré TF 38.03.12.
Electra, No. 175, 1997, pp. 49-78. 

Transient stability investigations on a Brazilian network by SIME;
A. L. Bettiol, Y. Zhang, L. Wehenkel, M. Pavella.
Proc. of APSCOM'97, Advances in Power Systems Control, Operation and Management, Hong Kong, Novembre 1997

A method for real-time transient stability emergency control; 
Y. Zhang, L. Wehenkel, M. Pavella.
Proc. of CPSPP'97, IFAC/CIGRE Symposium on Control of Power Systems and Power Plants, Beijing, Aug. 1997. 

Transient stability assessment of the South Brazilian network by SIME. A preliminary study. ;
A.L. Bettiol, Y. Zhang, L. Wehenkel, M. Pavella.
Proc. of NAPS'28, 28th North American Power Symposium, MIT, Boston, Nov. 1996. 

SIME : a comprehensive approach to fast transient stability assessment;
Y. Zhang, P. Rousseaux, L. Wehenkel, M. Pavella.
Proc. of IEE Japan, Aug. 1996, pp. 177-182. 

First- and multi-swing transient stability limits of the Hydro-Québec system using the SIME method;
Y. Zhang, L. Wehenkel, P. Rousseaux, M. Pavella.
Proc. of Melecon'96, 8th Mediterranean Electrotechnical Conference, Bari, May 1996, pp. 809-815. 

Hybrid extended equal-area criterion for fast transient stability assessment with detailed power system models;
Y. Zhang, P. Rousseaux, L. Wehenkel, M. Pavella, Y. Xue, B. Meyer, M. Trotignon.
Proc. of SIPOWER'95, IFAC Symp. on Control of Power Plants and Power Systems, Mexico, Dec. 1995, 6 pages. 

Developments in fast transient stability methods;
B. Garnier, J.N. Marquet, M. Trotignon, P. Rousseaux, Y. Xue, Z. Gao, L. Wehenkel, M. Pavella.
Paper presented at the CIGRE Study Committee 38, Colloquium on Power System Dynamic Performance, Brazil, Sept., 1993. 

Dynamic extended equal area criterion - Part 1. Basic Formulation;
Y. Xue, P. Rousseaux, Z. Gao, L. Wehenkel, M. Pavella, R. Belhomme, E. Euxibie, B. Heilbronn.
Proc. of Athens Power Tech, Joint IEEE-NTUA International Power Conference, Athens, Sept., 1993, pp. 889-895. 

Dynamic extended equal area criterion - Part 2. Embedding fast valving and automatic voltage regulation;
Y. Xue, Y. Zhang, Z. Gao, P. Rousseaux, L. Wehenkel, M. Pavella, M. Trotignon, A. Duchamp, B. Heilbronn.
Proc. of Athens Power Tech, Joint IEEE-NTUA International Power Conference, Athens, Sept., 1993, pp. 896-900. 

Extended equal area criterion revisited; 
Y. Xue, L. Wehenkel, R. Belhomme, P. Rousseaux, M. Pavella, E. Euxibie, B. Heilbronn, J.F. Lesigne. 
IEEE Trans. on PS, Vol. PWRS-7, No.3, 1992, pp. 1012-1022. 

Prospects of application to the French system of fast methods for transient stability and voltage security assessment; 
E. Euxibie, M. Goubin, B. Heilbronn, L. Wehenkel, Y. Xue, T. Van Cutsem, M. Pavella. 
CIGRE Paper 38-208, Paris, Aug.-Sept. 1992. 

Recent progress in real time electric power systems transient stability assessment; (Invited) 
M. Pavella, L. Wehenkel, R. Belhomme. 
Proc. of the 8th Brazilian Congr. of Autom. Control, Belem-PA, Sept. 10-14, 1990. 

Trends on real-time transient stability assessment of electric power systems; 
M. Ribbens-Pavella, R. Belhomme, L. Wehenkel.
Electrical and Power Systems Modelling and Simulation, J.C. Baltzer AG, Scientific Publishing Co, IMACS 1989, pp. 3-10.