http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Optimal Fault Tolerant Error Governor for PID Controllers
Luca Cavanini,Francesco Ferracuti,Sauro Longhi,Andrea Monteriù 제어·로봇·시스템학회 2022 International Journal of Control, Automation, and Vol.20 No.6
The Error Governor (EG) paradigm considers the issue of dynamically changing the error which drives a feedback controller featured by bounded control action magnitude to prevent the actuators’ saturation and to avoid the slow wind-up effects due to integrator or slow dynamics. Fault Tolerant (FT) policies are control methods permitting to mitigate the effect of faults occurring on driven actuators by modifying the structure of the controller which provides the reference signal for such actuators. In this paper, a FT policy based on an optimal EG approach is proposed. The policy, termed Fault-Tolerant Error Governor (FT-EG), permits to introduce a FT action in a closedloop system driven by PID controllers neglecting changes in the controller structure and, further, the wind-up issue given by nominal actuator saturation. The FT-EG is based on the solution of a constrained optimization problem and a computationally efficient version of the algorithm is presented. An analysis of control performance and the computational burden is provided, comparing in simulation studies the optimal FT-EG scheme performance with respect to control results provided by the baseline EG policy and saturated PID controller in the fault-free and the faulty scenario.
Fixed-size LS-SVM LPV System Identification for Large Datasets
Luca Cavanini,Riccardo Felicetti,Francesco Ferracuti,Andrea Monteriù 제어·로봇·시스템학회 2023 International Journal of Control, Automation, and Vol.21 No.12
In this paper, we propose an efficient method for handling large datasets in linear parameter-varying (LPV) model identification. The method is based on least-squares support vector machine (LS-SVM) identification in the primal space. To make the identification computationally feasible, even for very large datasets, we propose estimating a finite-dimensional feature map. To achieve this, we propose a two-step method to reduce the computational effort. First, we define the training set as a fixed-size subsample of the entire dataset, considering collision entropy for subset selection. The second step involves approximating the feature map through the eigenvalue decomposition of the kernel matrices. This paper considers both autoregressive with exogenous input (ARX) and state-space (SS) model forms. By comparing the problem formulation in the primal and dual spaces in terms of accuracy and computational complexity, the main advantage of the proposed technique is the reduction in space and time complexity during the training stage, making it preferable for handling very large datasets. To validate our proposed primal approach, we apply it to estimate LPV models using provided inputs, outputs, and scheduling signals for two nonlinear benchmarks: the parallel Wiener-Hammerstein system and the Silverbox system. The performances of our proposed approach are compared with the dual LS-SVM approach and the kernel principal component regression.