Model Predictive Parameter and State Estimation and Optimal Sensor Placement


Project Description

In this subproject we consider the online and self-calibrating identification of relevant parameters in transient thermo-elastic FE models. These parameters can include, for instance, heat transfer coefficients as well as heat inputs due to the machining process and electrical drives. Their knowledge is of paramount importance in particular for the subprojects developing online correction strategies. We will develop model-predictive parameter and state estimation schemes, which are using a moving time horizon to adapt the estimated parameters and state to measured temperature (or, when available, displacement) data.

These techniques will be developed under the specific operating conditions of machine tools, i.e., having available only a limited number of measurements, high dimensional thermo-elastic FE models, and they will take into account the pose dependence. Parameter models developed in other subprojects will serve as so-called background information and stabilize the estimation process. Initial experiments are showing this procedure to be very robust and capable of reconstructing time-dependent changes in heat transfer coefficients by means of only a few temperature measurements. In actual machine geometries, it will be important to place the temperature sensors intelligently, and we will further develop our optimal sensor placement techniques in order to make best use of the available sensors also in these nonlinear identification settings.

Associated Publications

  • Comparison of model order reduction methods for optimal sensor placement for thermo-elastic models Article
    Peter Benner, Roland Herzog, Norman Lang, Ilka Riedel and Jens Saak
    Engineering Optimization 51(3), p.465-483 · 2019
  • Optimal sensor placement for joint parameter and state estimation problems in large-scale dynamical systems with applications to thermo-mechanics Article
    Roland Herzog, Ilka Riedel and Dariusz Uciński
    Optimization and Engineering 19(3), p.591-627 · 2018
  • Optimal sensor placement based on model order reduction Proceedings
    Peter Benner, Roland Herzog, Norman Lang, Ilka Riedel and Jens Saak
    Conference on Thermal Issues in Machine Tools, p.355-365 · 2018
  • Optimale Sensorplatzierung und Online-Zustands- und Parameter-Identifikation Proceedings
    Proceedings of the 5th Colloquium of SFB/Transregio 96 · 2017
  • Simulation-based correction approach for thermo-elastic workpiece deformations during milling processes Proceedings
    Janine Glänzel, Roland Herzog, Steffen Ihlenfeldt, Arnd Meyer and Roman Unger
    Proceedings of the 7th HPC 2016 - CIRP Conference on High Performance Cutting 46, p.103-106 · 2016
  • Parameteridentifikation in thermo-elastischen Systemen Proceedings
    Roland Herzog, Ilka Riedel, Bernd Kauschinger and Steffen Schroeder
    Proceedings of the 4th Colloquium of SFB/Transregio 96, p.55-82 · 2016
  • Identifikation der Topologie eines Wärmenetzwerkes (ein Vergleich zwischen kontinuierlichen und diskreten Optimierungstechniken) Thesis
    Saber Jalilzadeh-Galaeh
    mathesis, Technische Universität Chemnitz, Germany · 2016
  • Verfahren zur optimalen Platzierung von Messpunkten zur Parameteridentifikation in zeitabhängigen Systemen Thesis
    Luca Landwehrjohann
    bathesis, Technische Universität Chemnitz, Germany · 2016
  • Identifikationsaufgaben auf Wärmenetzwerken Thesis
    Hussain Obaid
    mathesis, Technische Universität Chemnitz, Germany · 2016

Agency logo