OPTEC - K.U.Leuven Center of Excellence: Optimization in Engineering

For the first four years the reserach activity inside OPTEC has been organized within five workpackages:

• Workpackage 1: Conceptual and theoretical challenges (responsible Moritz Diehl). This work package addresses fundamental questions of optimization theory, in support of the other work packages which are application or algorithm oriented. It focus on fundamental methodological questions in convex and other structure exploiting algorithms. It will provide new methodologies for the application areas in work packages 2, 3 and 4 and will interact intensely with the numerical algorithms and software development in work package 5.
  • A convex optimization approach for the initialization of parameter estimation and optimal control problems (Julian Bonilla) PDF
    
  • Positive Polynomial Constraints for POD-based Model Predictive Controllers (Oscar Mauricio Agudelo) PDF
    
  • A new decomposition method for multi-agent convex optimization (Ion Necoara) PDF
    
  • Coupled Simulated Annealing (Samuel Xavier-de-Souza) PDF
    
  • The Lifted Newton Method for Optimization (Jan Albersmeyer) PDF
    
  • Accelerating accelerated schemes (Michel Baes) PDF

• Workpackage 2: System identification and approxiamtion (responsible Johan Suyksen). In this workpackage generic methods for obtaining linear or nonlinear mathematical models from given data are investigated for various purposes in engineering designs. Applications are studied in the areas of electricity load forecasting, sensor-based robot systems, bioprocess modelling, optimal fed batch experiment design, magnetic resonance spectroscopy metabolite quantification and modelling for control of glycemia in critically ill patients.

  • Electric Load Forecasting using kernel based modeling for nonlinear system identification (Marcelo Espinoza) PDF
    
  • Quadratically constrained quadratic programming for subspace selection in kernel regression (Marco Signoretto) PDF
    
  • SOCP and least squares approaches in kernel-based modelling and perturbation analysis (Tillmann Falck) PDF
    
  • Survival analysis: a convex optimization approach (Vanya Van Belle) PDF
    
  • Low rank updates for feature selection (Fabian Ojeda) PDF
    
  • Large scale multi-class kernel logistic regression: stable numerical implementation and application to phoneme classification (Peter Karsmakers) PDF
    
  • Kalman smoothing improves human motion estimation (Friedl De Groote) PDF

• Workpackage 3: Control system design (responsible Ilse Smets). This workpackage deals with optimization in control system design and focuses on the following (engineering optimization relevant) application domains: optimal distributed control of noise and vibrations and of tubular reactors, robust controllers for linear parameter varying (LPV) mechatronic systems,optimal adaptive control and model based predictive control (MPC) of (bio)chemical (stirred and tubular) reactors.

  • Time-optimal path tracking for robots: a convex optimization approach (Diederik Verscheure) PDF
    
  • Model Predictive Control of Automotive Powertrains (Bart Saerens) PDF
    
  • Periodic Stability Optimization for Power Generating Kites (Boris Houska) PDF
    
  • Model Predictive Prefilter (Lieboud Van den Broeck) PDF
    
  • Optimal performance trade-offs in repetitive control (Goele Pipeleers) PDF

• Workpackage 4: Design optimization (responsible Bram Demeulenaere). This work package focuses on optimal design of products and systems. In particular, it focuses on two example applications: optimal design of tubular chemical reactors and dynamic balancing of machines. Other applications in various engineering disciplines (e.g. electrical engineering, civil engineering, agricultural machines and systems) will be worked out in collaboration with the peripheral partners.

  • Counterweight Balancing of Mechanisms (Bram Demeulenaere) PDF
  • Shaking Force and Shaking Moment Balancing of Cam Mechanisms (Myriam Verschuure) PDF
  • Multiobjective Optimization Across Thousands of Circuit Structures Simultaneously (Trent McConaghy) PDF
  • Model based optimisation of jacketed tubular reactors (Filip Logist) PDF

• Workpackage 5: Numerical algorithms and software (responsible Wim Michiels). The aim of this workpackage is to develop and analyze efficient and robust numerical algorithms for complex and large-scale optimization problems appearing in engineering applications. These algorithms will be implemented and made publicly available to facilitate dissemination of the new computational techniques and their integration in new applications.

  • Geometric Optimization on the Manifold of Rank Constrained Symmetric Semidefinite Matrices (Bart Vandereycken) PDF
  • Cutting plane methods for solving convex optimization problems over nonnegative polynomials (Iurie Caraus) PDF
  • Robust Stability Optimization with the Smoothed Spectral Abscissa (Joris Vanbiervliet) PDF
  • Sparse Kernel Models for Spectral Clustering Using the Incomplete Cholesky Decomposition (Carlos Alzate) PDF
  • Distributed Spectrum Management for Digital Subscriber Lines (Paschalis Tsiaflakis) PDF
  • Constrained Pole-Zero Linear Prediction: Optimization of Cascaded Biquadratic Notch Filters for Multi-Tone Frequency Estimation (Toon van Waterschoot) PDF

Across the workpackages, some OPTEC research is organized in groups around topics of critical importance:

• Fast MPC Group (see the new Dynamic and Embedded Optimization working group);

• Kite Power Project (the activities of this project converged in the new Kite Power Project).