|
STP Method for Solving the Minimal Norm Centrosymmetric Solutions of MX-XN=GY+R |
|
Weihua Chen, Caiqin Song(University of Jinan, China) |
| In this paper, STP method for solving the minimal norm (anti)-centrosymmetric solution of nonhomogeneous Yakubovich quaternion matrix equation are presented. The structural matrix of the quaternion matrix is constructed by using STP, and the expressions of the (anti)-centrosymmetric solution of this quaternion equations. Furthermore, the sufficient and necessary conditions having minimal norm (anti)-centrosymmetric solution for this equations are provided. Finally, a numerical example is given for demonstrating the efficiency and superity of the STP approach. |
|
|
Data-Driven Design for Model-Referenced Model-Free Controller |
|
Shuichi Yahagi(ISUZU Advanced Engineering Center, Japan), Itsuro Kajiwara(Hokkaido University, Japan) |
| This paper presents the data-driven design of model-free control (MFC) based on an ultra-local model to realize model-matching. The proposed method provides the MFC design parameters from single-experiment time-series data without repeated experiments and a plant model. The effectiveness of the proposed method was verified through simulations with a linear parameter varying system. |
|
|
Comparison of Different Gaussian Process Models and Applications in Model Predictive Control |
|
Florian Diepers, Dominik Polke, Elmar Ahle(University of Applied Sciences Niederrhein, Germany), Dirk Söffker(University of Duisburg-Essen, Germany) |
| In this work, different dynamic models using Gaussian processes (GPs) and the application of their uncertainty prediction in stochastic MPC (SMPC) are evaluated and compared. The main advantage of GPs is the construction of a dynamic system model with the possibility of determining the uncertainty of its own prediction. Three different utilizations of prediction uncertainty in SMPC are compared. Additionally, two different implementations of dynamic GPs, nonlinear autoregressive exogenous GPs (GP-NARX) and state-space models based on GPs (GP-SSM), are considered. All six test cases are compared based on an evaluation criterion. A simulated inverted pendulum is used as the test system. |
|
|
Equivalent Transformations in the Data-Informativity Framework and Its Applications: Homological Algebraic Approach |
|
Yuki Tanaka, Osamu Kaneko(The University of Electro-Communications, Japan) |
| The data informativity approach has been proposed as a framework for considering the model-based control and the data-driven control from a unified perspective.
This framework emphasizes the advantages of data-driven control and the mathematical duality between time series data and functions.
In this study, we focus on this mathematical aspect and study equivalent transformations in the data informativity approach.
Equivalent transformations in the data informativity approach require consideration of the correspondence between sets of models. To address this point, we introduce the homology theory and discuss the data-informativity for observability through the equivalence transformation. |
|
|
On set-invariance for nonlinear output-feedback control systems: a modified control barrier function |
|
Hyung Tae Choi, Jung Hoon Kim(POSTECH, Korea) |
| This paper is concerned with establishing the set-invariance property for nonlinear output-feedback control
systems by developing a new type of control barrier function. More precisely, the overall architecture consists of
a Luenberger-like state observer and a controller equipped with such a new control barrier function, by which the relevant
invariance property can be ensured even for the existence of an estimation error from the observer; the proposed
control barrier function is a modified version of the conventional one whose employment was confined to the case of
full-state feedback. We further show that the controller can be obtained by solving a quadratic programming problem. |
|
|
Invariance Guarantees using Continuously Parametrized Control Barrier Functions |
|
Inkyu Jang, H. Jin Kim(Seoul National University, Korea) |
| This paper introduces the notion of parametrized control barrier function (PCBF), a continuous spectrum of CBFs whose parameter space is a differentiable manifold. An optimization-based control strategy that uses PCBF and quadratic programming (QP), namely PCBF-QP, is proposed. PCBF-QP is capable of ensuring invariance of its level set, and it makes the best use of the differentiable structure of the parameter space to expand the feasible region of the optimization, allowing the system to select its control input from a broader range. The proposed control strategy is validated in simulation experiments. |
|