Learning from adaptive neural network output feedback control of a unicycle-type mobile robot

Characteristic analysis and mitigation strategy for SSCI in series-compensated DFIG-based wind farm controlled by a virtual synchronous generator

The proliferation of virtual synchronous generator (VSG) technology within series-compensated double-fed induction generator (DFIG)-based wind farms is substantially hampered by the attendant risk of subsynchronous control interaction (SSCI), resulting in a significant research deficiency on systematic control interaction analysis and the development of mitigation strategies. The paper proposes an advanced active disturbance rejection control (ADRC) framework, incorporating real-time compensation mechanisms to mitigate the inadequate suppression efficacy attributable to the VSG's diminished output impedance. Initially, the mathematical expression for the VSG output impedance is rigorously deduced, and the positive damping attributes of the VSG in relation to SSCI are elucidated from the perspective of underlying mechanistic principles. Subsequently, the suppressive mechanism of SSCI by the ADRC is revealed in the context of VSG involvement, and the consequent augmentation of SSCI attributed to PI control is systematically derived. In immediate succession, the quanta of oscillation and inductive cross-coupling are encapsulated as the system's aggregate disturbance, thereby streamlining the ADRC to its primary order configuration, permitting the utilization of an extended state observer (ESO) for the dynamic estimation of said disturbance. Furthermore, a fractional-order filter function is instituted to engineer an augmented ESO, which refines the output voltage of the grid-side converter. Concurrently, a meticulous discourse on the rectification strategy for the proposed ESO parameters and its stability ensues. Ultimately, the efficacy of the mechanism analysis, alongside the robustness of the proffered control strategy for SSCI mitigation under diverse perturbation conditions, is corroborated via impedance evaluation and time-domain simulation.

Robust Output Feedback Stabilization and Tracking for an Uncertain Nonholonomic Systems with Application to a Mobile Robot

This paper presents a novel robust output feedback control that simultaneously performs both stabilization and trajectory tracking for a class of underactuated nonholonomic systems despite model uncertainties, external disturbance, and the absence of velocity measurement. To solve this challenging problem, a generalized normal form has been successfully created by employing an input-output feedback linearization approach and a change in coordinates (diffeomorphism). This research mainly focuses on the stabilization problem of nonholonomic systems that can be transformed to a normal form and pose several challenges, including (i) a nontriangular normal form, (ii) the internal dynamics of the system are non-affine in control, and (iii) the zero dynamics of the system are not in minimum phase. The proposed scheme utilizes combined backstepping and sliding mode control (SMC) techniques. Furthermore, the full-order high gain observer (HGO) has been developed to estimate the derivative of output functions and internal dynamics. Then, full-order HGO and the backstepping SMC have been integrated to synthesize a robust output feedback controller. A differential-drive type (2,0) the wheeled mobile robot has been considered as an example to support the theoretical results. The simulation results demonstrate that the backstepping SMC exhibits robustness against bounded uncertainties.

Observer-based dynamical pattern recognition via deterministic learning

In this paper, based on the sampled-data observer and the deterministic learning theory, a rapid dynamical pattern recognition approach is proposed for univariate time series composed of the output signals of the dynamical systems. Specifically, locally-accurate identification of inherent dynamics of univariate time series is first achieved by using the sampled-data observer and the radial basis function (RBF) networks. The dynamical estimators embedded with the learned knowledge are then designed by resorting to the sampled-data observer. It is proved that generated estimator residuals can reflect the difference between the system dynamics of the training and test univariate time series. Finally, a recognition decision-making scheme is proposed based on the residual norms of the dynamical estimators. Through rigorous analysis, recognition conditions are given to guarantee the accurate recognition of the dynamical pattern of the test univariate time series. The significance of this paper lies in that the difficult problems of dynamical modeling and rapid recognition for univariate time series are solved by incorporating the sampled-data observer design and the deterministic learning theory. The effectiveness of the proposed approach is confirmed by a numerical example and compressor stall warning experiments.

Full-state constrained neural control and learning for the nonholonomic wheeled mobile robot with unknown dynamics

The adaptive learning and control are proposed for the full-state(FS) constrained NWMR system with external destabilization. First, the constrained state is reformulated as the unconstrained state. Then, approximating the unknown dynamics in the closed-loop (CL) system is conducted via radial basis function (RBF) NN. Also, a sliding term is designed to deal with the external destabilization and the neural network training error. The derived adaptive neural controller can realize the asymptotic stability of a robot system without violating FS constraints. Moreover, the neural weights are converged so that the unknown dynamics are expressed by the constant weights in the CL system. It is also applicable to other similar control tasks. Lastly, the proposed algorithm is simulated and validated.

Reference modification for trajectory tracking using hybrid offline and online neural networks learning

In this paper, we propose a hybrid offline/online neural networks learning method, which combines complementary advantages of two types of neural networks (NNs): deep NN (DNN) and single-layer radial basis function NN (RBFNN). Firstly, after analyzing the mechatronic system’s model, we select reasonable features as the input of the DNN to learn the inverse dynamic characteristics of the closed-loop system offline, so as to establish the mapping between the desired trajectory and the reference trajectory of the system. The trained DNN is used to generate a new reference trajectory and compensate for the tracking error in advance, which can speed up the convergence of online learning control based on RBFNN. This reference trajectory is further modified iteratively when the tracking task is repeated. For this purpose, a single-layer RBFNN model is established, and an online learning algorithm is developed to update the RBFNN parameters. The proposed hybrid offline/online NN method can improve the tracking performance of mechatronic systems by modifying the reference trajectory on top of the baseline controller without affecting the system stability. To verify the effectiveness of this method, we conduct experiments on a piezoelectric drive platform.

Robust composite nonlinear feedback control for uncertain robot manipulators

On the basis of the classical computed torque control method, a new composite nonlinear feedback design method for robot manipulators with uncertainty is presented. The resulting controller consists of the composite nonlinear feedback control and robust control. The core is to use the robust control for online approximation of the system’s uncertainty as a compensation term for the composite nonlinear feedback controller. The design method of the new controller is given, and the convergence of the closed-loop system is proved. The simulation results show that the proposed scheme can make the uncertain robot system have strong robustness and anti-interference ability.

A multiple kernel classification approach based on a Quadratic Successive Geometric Segmentation methodology with a fault diagnosis case

This work presents a new approach for solving classification and learning problems. The Successive Geometric Segmentation technique is applied to encapsulate large datasets by using a series of Oriented Bounding Hyper Box (OBHBs). Each OBHB is obtained through linear separation analysis and each one represents a specific region in a pattern's solution space. Also, each OBHB can be seen as a data abstraction layer and be considered as an individual Kernel. Thus, it is possible by applying a quadratic discriminant function, to assemble a set of nonlinear surfaces separating each desirable pattern. This approach allows working with large datasets using high speed linear analysis tools and yet providing a very accurate non-linear classifier as final result. The methodology was tested using the UCI Machine Learning repository and a Power Transformer Fault Diagnosis real scenario problem. The results were compared with different approaches provided by literature and, finally, the potential and further applications of the methodology were also discussed.