Browsing by Subject "System identification"

Now showing 1 - 6 of 6
  • Thumbnail Image
    Item
    Accurate Mathematical neuron models.
    (2012-08) Miranda Dominguez, Oscar
  • Thumbnail Image
    Item
    Real time identification of local surface properties of material using atomic force microscope - An FPGA based implementation
    (2017-02) Pradhan, Sourav
    System identification is widely employed for building mathematical models of manifold systems using statistical techniques. In this thesis, the application of system identification to atomic force microscopy using a real-time embedded solution has been reported. Atomic force microscopes are prevalent instruments utilized to explore material properties at the micro/nanometer level. A Field Programmable Gate Array has been chosen to harbor the design of the system identification module. The reported module has been successfully cascaded with an atomic force microscope to estimate local surface mechanical properties of materials. The design layout described in this thesis is not just applicable to commercially available atomic force microscopes, but to a large group of real-time signal processing units. Numerous simulations over multiple platforms and experimental results are presented to validate the accuracy and performance of the designed system identification module.
  • Thumbnail Image
    Item
    System identification for the Clipper Liberty C96 wind turbine
    (2014-06) Showers, Daniel
    System identification techniques are powerful tools that help improve modeling capabilities of real world dynamic systems. These techniques are well established and have been successfully used on countless systems in many areas. However, wind turbines provide a unique challenge for system identification because of the difficulty in measuring its primary input: wind. This thesis first motivates the problem by demonstrating the challenges with wind turbine system identification using both simulations and real data. It then suggests techniques toward successfully identifying a dynamic wind turbine model including the notion of an effective wind speed and how it might be measured. Various levels of simulation complexity are explored for insights into calculating an effective wind speed. In addition, measurements taken from the University of Minnesota's Clipper Liberty C96 research wind turbine are used for a preliminary investigation into the effective wind speed calculation and system identification of a real world wind turbine.
  • Thumbnail Image
    Item
    System identification of the Brompton bicycle
    (2015-01) Hladun, Monique Victoria Teresa
    The Brompton (a European folding design) bicycle was instrumented with a variety of sensors including acceleration, angular rate, speed, and steering sensors. A bicycle state estimator was designed to obtain additional information from this data including heading, turn rate, lean angle, steer rate, and positions of the wheels during a trajectory. The first part of the thesis describes the model setup for system identification including the Steer-to-Lean dynamics and Lean-to-Steer dynamics reduced models. CIFER software was used in the system identification process of these models. The second part describes the validation of the Empirical model by using the Rider Control model ([1]) and the Complete Rider/Vehicle model ([1]) to determine the feedback gains. The Theoretical model feedback gains were also determined by using the Rider Control model ([1]) and the Complete Rider/Vehicle model ([1]).
  • Thumbnail Image
    Item
    System identification via Nuclear norm regularization
    (2014-07) Deshmane, Harshad
    We study the subspace method for system identification, and look at algorithms that rely on nuclear norm regularization for solving this problem. We introduce our own algorithm for solving the problem, based on the alternating direction method of multipliers (ADMM). Our algorithm involves an iterative minimization step, which is solved using line search methods. We demonstrate the effectiveness of our algorithm on a particular real world example, as well as two benchmark examples. In addition, we compare the computational efficiency of our algorithm to that of other existing algorithms for solving the nuclear norm system identification problem, observing that for single-input single-output systems, our algorithm is faster than an existing interior-point method. We also note that our algorithm converges the fastest when we use a gradient descent direction in the iterative minimization step of the ADMM.
  • Thumbnail Image
    Item
    Test platforms for model-based flight research
    (2013-09) Dorobantu, Andrei
    Demonstrating the reliability of flight control algorithms is critical to integrating unmanned aircraft systems into the civilian airspace. For many potential applications, design and certification of these algorithms will rely heavily on mathematical models of the aircraft dynamics. Therefore, the aerospace community must develop flight test platforms to support the advancement of model-based techniques. The University of Minnesota has developed a test platform dedicated to model-based flight research for unmanned aircraft systems. This thesis provides an overview of the test platform and its research activities in the areas of system identification, model validation, and closed-loop control for small unmanned aircraft.