Search

This article investigates state feedback design for achieving almost-sure robust stabilization of randomly switched linear systems whose subsystems are uncontrollable and whose models are subject to nonlinear modeling errors due to, for example, linearization. Complex systems are often uncontrollable under a fixed configuration from a single control input. However, when control actions can be used sequentially and collaboratively through different system configurations, stabilization can be potentially achieved. This design problem encounters some fundamental issues that must be resolved, involving mostly suitable coordinated implementations of state decomposition, feedback pole-placement design, usage of stochastic information on the switching process, coupling of controllable and uncontrollable substates, subsystem interactions, and modeling errors. The common state feedback on controllable substates can lead to unstable closed-loop systems, due to substate coupling. A modified control algorithm is introduced that decouples substates and designs feedback gains simultaneously. Further complications arise when subsystem interaction destabilizes the system. Some structural conditions are shown to be essential for achieving almost-sure stabilization. A design procedure that integrates feedback gain selection and switching information is introduced to achieve almost-sure stability for the closed-loop system. Robustness of the design procedure is established. Examples and simulation case studies are presented to illustrate the main algorithms and stabilization properties.

This article investigates the problem of joint continuous and discrete state estimation of randomly switched linear systems in which subsystems may not be observable. Estimation of both continuous state and discrete sequence simultaneously based on the same output observations is a challenging task that is inherently nonlinear and often infinite dimensional. This article presents necessary and sufficient conditions when joint estimation is possible without using a probing input. When such conditions are not satisfied, a suitably designed input must be used to achieve the goal of jointly detecting the subsystem and estimating the internal state. This article employs certain structures of randomly switched linear systems to develop algorithms that use finite-dimensional estimators for continuous states and sampled data for detecting the discrete states. The convergence analysis shows that this framework can achieve convergence. Examples and simulation case studies are presented to illustrate the main results of this article. The findings of this article can be used to form a supporting foundation for robust control.

In this article, we model a hybrid system by using a hybrid machine consisting of an automaton and a set of continuous-variable subsystems. We investigate the problem of determining the (current) discrete state of the hybrid system. Extending the terminologies in discrete-event systems, a hybrid system is detectable if its discrete state can be uniquely determined. We derive conditions for a hybrid system to be detectable. If a hybrid system is detectable, we determine its current discrete state using a two-level approach. At the discrete-event level, we construct a discrete-event observer to obtain a discrete state estimate that consists of all possible discrete states that the hybrid system may be in. At the continuous-variable level, a set of continuous-variable observers are constructed, if necessary, to uniquely determine the discrete state. An example of a dc microgrid is used to illustrate the theoretical results.

This paper proposes a deep learning approach to overcome the impacts of communication failures on the performance and convergence rate of the distributed optimal frequency control (DOFC) for power systems. Novel features of the proposed framework are fourfold. First, the nonlinear model of the DOFC is developed to consider for nonlinearities of power flows. Second, the long short-term memory (LSTM) algorithm is used for dynamic model estimation during communication failures. Next, the LSTM-based DOFC method is introduced to cope with the impact of communication failures on the performance of the distributed control strategy. Finally, we prove the convergence of LSTM-DOFC and show that the algorithm has superior performance compared to the linearized prediction methods, such as autoregressive-moving-average models. Simulations on two real-world power systems are carried out to demonstrate the effectiveness of the proposed framework.

This paper analyzes the system-level state of health (SOH) and its dependence on the SOHs of its component battery modules. Due to stochastic natures of battery aging processes and their dependence on charge/discharge rate and depth, operating temperature, and environment conditions, capacities of battery modules decay unevenly and randomly. Based on estimated SOHs of battery modules during battery operation, we analyze how the SOH of the entire system deteriorates when battery modules age and become increasingly diverse in their capacities. A rigorous mathematical analysis of system-level capacity utilization is conducted. It is shown that for large battery strings with uniformly distributed capacities, the average string capacity approaches the minimum, implying an asymptotically near worst-case capacity utility without reorganization. It is demonstrated that the overall battery usable capacities can be more efficiently utilized to achieve extended operational ranges by using battery reconfiguration. An optimal regrouping algorithm is introduced. Analysis methods, simulation examples, and a case study using real-world battery data are presented.

Optimization methods in cyber-physical systems do not involve parameter uncertainties in most existing literature. This paper considers adaptive optimization problems in which searching for optimal solutions and identifying unknown parameters must be performed simultaneously. Due to the dual roles of the input signals on achieving optimization and providing persistent excitation for identification, a fundamental conflict arises. In this paper, a method of adding a small deterministic periodic dither signal to the input is deployed to resolve this conflict and provide sufficient excitation for estimating the unknown parameters. The designing principle of the dither is discussed. Under dithered inputs, the authors show that simultaneous convergence of parameter estimation and optimization can be achieved. Convergence properties and convergence rates of parameter estimation and optimization variable updates are presented under the scenarios of uncertainty-free observations and systems with noisy observation and unmodeled components. The fundamental relationships and trade-off among updating step sizes, dither magnitudes, parameter estimation errors, optimization accuracy, and convergence rates are further investigated.

This paper is devoted to numerically solving a class of optimal stopping problems for stochastic hybrid systems involving both continuous states and discrete events. The motivation for solving this class of problems stems from quickest event detection problems of stochastic hybrid systems in broad application domains. We solve the optimal stopping problems numerically by constructing feasible algorithms using Markov chain approximation techniques. The key tasks we undertake include designing and constructing discrete-time Markov chains that are locally consistent with switching diffusions, proving the convergence of suitably scaled sequences, and obtaining convergence for the cost and value functions. Finally, numerical results are provided to demonstrate the performance of the algorithms.

The purpose of this study was to investigate the effects of secreted protein acidic and rich in cysteine (SPARC) on the maintenance of limbal epithelial stem cell (LESC) stemness and restoration of ocular surface. To determine the suitable concentration of SPARC for LESC culture, the marker expression, mitogenic effect, and holoclone‐forming capacity of LESCs treated with different concentrations of SPARC were analyzed. To investigate the mechanism of SPARC's action on the preservation of LESCs stemness, the phosphorylation of related signaling pathways was evaluated by Western blotting. A corneal wound model was established to verify the function of SPARC in ocular surface repair. Consecutive subculturing, colony‐forming efficiency, immunofluorescence, and 5‐ethynyl‐2‐deoxyuridine incorporation assays indicated that 1 μg/mL SPARC was a suitable concentration to stimulate LESC proliferation and preserve their proliferative potential. Compared with a control group, 1 μg/mL SPARC effectively increased the expression of ABCG‐2, Bmi‐1, and Ki67, while decreasing that of CK3/12. The mitogenic effect of SPARC on LESCs was found to be mediated by the phosphorylation of c‐Jun N‐terminal kinase (JNK) and p38‐MAPK signaling pathways, whereas the inhibitors of JNK and p38 MAPK reduced the marker expression and mitogenic capacity of LESCs. In a corneal injury model, SPARC facilitated corneal epithelial wound healing and promoted the proliferation of p63α‐positive cells both in the limbus and in the epithelial healing front. SPARC promotes proliferation while suppressing spontaneous differentiation of LESCs through JNK and p38‐MAPK signaling pathways, suggesting that SPARC is a promising factor for the improvement of LESCs culture in vitro and in vivo. When cultured in medium containing1μg/mL secreted protein acidic and rich in cysteine (SPARC) in vitro, limbal epithelial stem cells (LESCs) showed a greater proliferative and clonogenic capacity and a higher expression of LESCs marker through c‐Jun N‐terminal kinase/p38 MAPK signal pathway. In corneal wounding model, exogenous SPARC accelerated corneal epithelial wound healing and promoted the proliferation of p63α‐positive cells both in limbus and in epithelial healing front.

The current research of state of charge (SoC) online estimation of lithium-ion battery (LiB) in electric vehicles (EVs) mainly focuses on adopting or improving of battery models and estimation filters. However, little attention has been paid to the accuracy of various open circuit voltage (OCV) models for correcting the SoC with aid of the ampere-hour counting method. This paper presents a comprehensive comparison study on eighteen OCV models which cover the majority of models used in literature. The low-current OCV tests are conducted on the typical commercial LiFePO4/graphite (LFP) and LiNiMnCoO2/graphite (NMC) cells to obtain the experimental OCV-SoC curves at different ambient temperature and aging stages. With selected OCV and SoC points from experimental OCV-SoC curves, the parameters of each OCV model are determined by curve fitting toolbox of MATLAB 2013. Then the fitting OCV-SoC curves based on diversified OCV models are also obtained. The indicator of root-mean-square error (RMSE) between the experimental data and fitted data is selected to evaluate the adaptabilities of these OCV models for their main features, advantages, and limitations. The sensitivities of OCV models to ambient temperatures, aging stages, numbers of data points, and SoC regions are studied for both NMC and LFP cells. Furthermore, the influences of these models on SoC estimation are discussed. Through a comprehensive comparison and analysis on OCV models, some recommendations in selecting OCV models for both NMC and LFP cells are given.

This paper treats the problem of existence of a controller for networked-timed discrete-event systems. The results are new and useful for any networked discrete-event system for which it is important that time be explicitly considered in the model. We assume that a controller (supervisor) communicates with the system to be controlled via a shared communication network. Delays and losses in communication systems and their impacts on control are investigated. Under the assumption that the delays and losses are bounded, we derive a necessary and sufficient condition for the existence of a controller. The condition is expressed as network T-controllability and network T-observability, which are defined in this paper. We apply the results to power distribution network management, and illustrate the application using a 33-node (bus) test system, where the objective is to ensure that the total substation transformer power stays within prespecified safety limits.

The paper considers the recursive identification of Hammerstein systems with noisy output, while its input may or may not be corrupted by noise. The system in the latter case is usually called as EIV system. The conditions required in this paper are considerably weaker than those used in previous works, e.g., the orders of the linear subsystem are allowed to be unknown and no additional conditions are imposed on its moving average part. The nonlinearity is general in the sense that a certain class of functions is out of consideration in some previous papers. In the paper, the almost sure convergence together with convergence rate are established for the estimates for coefficients of the linear part, and then the almost sure convergence for the estimates for the nonlinearity at any given points are derived by using kernel functions. The convergence rate for the nonlinearity is also obtained for the case where the system input is available without noise. A numerical example is provided, and the simulation results are consistent with the theoretical analysis.

This paper introduces identification algorithms for finite impulse response systems under quantized output observations and general quantized inputs. While asymptotically efficient algorithms for quantized identification under periodic inputs are available, their counterpart under general inputs has encountered technical difficulties and evaded satisfactory resolutions. Under quantized inputs, this paper resolves this issue with constructive solutions. A two-step algorithm is developed, which demonstrates desired convergence properties including strong convergence, mean-square convergence, convergence rates, asymptotic normality, and asymptotical efficiency in terms of the Cramér-Rao lower bound. Some essential conditions on input excitation are derived that ensure identifiability and convergence. It is shown that by a suitable selection of the algorithm’s weighting matrix, the estimates become asymptotically efficient. The strong and mean-square convergence rates are explicitly obtained. Optimal input design is discussed. Numerical examples are included to illustrate the main results of this paper

This paper considers the problem of identifiability and parameter estimation of single-input-single-output, linear, time-invariant, stable, continuous-time systems under irregular and random sampling schemes. Conditions for system identifiability are established under inputs of exponential polynomial types and a tight bound on sampling density. Identification algorithms of Gauss-Newton iterative types are developed to generate convergent estimate sequences. When the sampled output is corrupted by observation noises, input design, sampling times, and convergent algorithms are intertwined. Persistent excitation (PE) conditions for strongly convergent algorithms are derived. Unlike the traditional identification, the PE conditions under irregular and random sampling involve both sampling times and input values. Under the given PE conditions, iterative and recursive algorithms are developed to estimate the original continuous-time system parameters. The corresponding convergence results are obtained. Several simulation examples are displayed to verify the theoretical results.

This paper presents a frequency multiplication technique using travelling wave. The proposed technique improves conversion efficiency, cancels byproduct harmonics, and achieves wide bandwidth. citive pressure sensor by means of frequency shifts from its nominal operation frequency. It takes advantage of standing-wave formation and loss cancellation in a distributed structure to generate high amplitude signals resulting in high harmonic power. Wide bandwidth operation and odd harmonic cancellation around the center frequency are the inherent properties of this frequency multiplier.

Model identification and fault detection are of importance for reliable motor control. For fault diagnosis, motor parameters must be estimated accurately and reliably during operation. Typical measurement noises for power electronics encounter unknown and drifting non-zero means. This causes identification bias and has detrimental effects on reliability of fault detection. Based on enhanced model structures of electric motors that accommodate both normal and faulty modes, this paper introduces algorithms that correct this bias and restore diagnosis accuracy. BLDC motors are used as a benchmark type for concrete algorithm development and evaluation. Algorithms are presented, their properties are established, and their accuracy and robustness are evaluated by case studies.

This paper studies identification of systems that can be decomposed into cascaded subsystems. The benefits of using additional sensors for identifying subsystems are investigated in terms of identification accuracy and time complexity. Identification algorithms, input design, and time complexity are first developed for subsystems, under various sensor types and locations. Overall reduction in estimation errors and time complexity is then analyzed to understand optimal selection of sensor locations and impact of sensor types on identification accuracy and time complexity. Copyright © 2009 John Wiley and Sons Asia Pte Ltd and Chinese Automatic Control Society

The relationship between information acquisition (identification) and information processing (control) in their capability of dealing with uncertainty is studied. It is revealed that such a relationship can be established rigorously from the viewpoint of complexity. A notion of information-based complexity is hence introduced, first in its generality, and then in its special applications to metric spaces in feedback control systems. Copyright © 2000 John Wiley & Sons, Ltd.

The relationship between information acquisition (identification) and information processing (control) in their capability of dealing with uncertainty is studied. It is revealed that such a relationship can be established rigorously from the viewpoint of complexity. A notion of information‐based complexity is hence introduced, first in its generality, and then in its special applications to metric spaces in feedback control systems. Copyright © 2000 John Wiley & Sons, Ltd.

The problem of robust stabilization of nonlinear systems with feedback linearizable nominal part and norm-bounded nonlinear uncertainties is investigated. Necessary and sufficient conditions are obtained for robust stabilization of such systems. A design procedure is developed which combines feedback linearization technique and quadratic stabilization via linear feedback to achieve robust global asymptotic stability. © 1997 by John Wiley & Sons, Ltd.

In this paper we attempt to address a number of open issues in closed-loop identification of linear time-varying, possibly unstable systems. The problem is motivated by application of identification in adaptive systems, which inherently requires that identification be conducted on line and in a closed loop configuration. A unique difficulty is found to be the requirement that the identification must be persistent with time, or in other words the identification error must be kept persistently small for all time. It has been found that periodic signals may be used as probing inputs to achieve persistency, and explicit bounds on identification errors, uniform with respect to time, have been derived to show that this is indeed possible. These results further demonstrate that when periodic signals are in use, a simple least-squares estimation scheme constitutes a good identification algorithm.

The problem of robust stabilization of nonlinear systems with feedback linearizable nominal part and norm‐bounded nonlinear uncertainties is investigated. Necessary and sufficient conditions are obtained for robust stabilization of such systems. A design procedure is developed which combines feedback linearization technique and quadratic stabilization via linear feedback to achieve robust global asymptotic stability. © 1997 by John Wiley & Sons, Ltd.

•Real-time multi-forward-step prediction for battery state of charge is investigated.•Various battery states under complex real-world operation conditions can be learned.•A yearlong dataset of an electric taxi is employed for training a full-climate and full-state model.•Prediction accuracy and horizon can be collaboratively controlled based on accuracy benchmark.•Predictable and sufficient predicted time can be available to eliminate driving mileage anxiety.

When lithium-ion batteries age with cycling, the loss of lithium inventory and active material may result in battery aging. The aging characteristics and mechanism of batteries under different cycle conditions differ. Therefore, battery cycle life tests were performed to explore the aging characteristics and mechanism of Li[Ni0.5Co0.2Mn0.3]O2batteries. The results of the one-way analysis of variance show that the temperature has a slight effect on battery aging in the range of 25 °C–55 °C, but the discharge current has a great impact on battery aging in the range of 0.5C–2C. Battery aging mechanism can be analyzed by reconstructing charging voltage curves. But the hypothesis of the nondestructive method had not been verified through experiments. The hypothesis was verified through half-cell experiment, and the nondestructive diagnostic method was used to analyze the aging mechanisms of Li[Ni0.5Co0.2Mn0.3]O2battery. The aging mechanism analysis results revealed that the main aging mechanism will change under different cycle conditions, therefore, the aging mechanism identification is necessary before establishing the battery life prediction model.

In this summary paper, conditions are provided under which sensitivity minimization in slowly time-varying systems can be approximated by solving a sequence of H∞problems for the related frozen-time, time-invariant systems. Each H∞problem employs the Adamjan-Arov-Krein (AAK) parametrization to obtain a δ-suboptimal frozen-time sensitivity whose dependance on the frozen plant and weighting is Lipschitz continuous.