Steven Liu

A variance-constrained approach to recursive state estimation for time-varying complex networks with missing measurements

In this paper, the recursive state estimation problem is investigated for an array of discrete time-varying coupled stochastic complex networks with missing measurements. A set of random variables satisfying certain probabilistic distributions is introduced to characterize the phenomenon of the missing measurements, where each sensor can have individual missing probability. The Taylor series expansion is employed to deal with the nonlinearities and the high-order terms of the linearization errors are estimated. The purpose of the addressed state estimation problem is to design a time-varying state estimator such that, in the presence of the missing measurements and the random disturbances, an upper bound of the estimation error covariance can be guaranteed and the explicit expression of the estimator parameters is given. By using the Riccati-like difference equations approach, the estimator parameter is characterized by the solutions to two Riccati-like difference equations. It is shown that the obtained upper bound is minimized by the designed estimator parameters and the proposed state estimation algorithm is of a recursive form suitable for online computation. Finally, an illustrative example is provided to demonstrate the feasibility and effectiveness of the developed state estimation scheme.

Energy Management for Smart Grids With Electric Vehicles Based on Hierarchical MPC

This paper presents an energy management system for smart grids with electric vehicles based on hierarchical model predictive control (HiMPC). The energy management system realizes load-frequency control (LFC), an economic operation and an electric vehicle integration into the smart grid. The main component is the HiMPC, which allows covering different time scales, regarding constraints (e.g. power ratings) and predictions (e.g. on renewable generation), as well as rejecting disturbances (e.g. due to fluctuating renewable generation) based on a systematic model- and optimization-based design. For the electric vehicle integration, an aggregator is proposed as link between HiMPC and individual vehicle. The aggregator in particular provides predictions to the HiMPC on the availability of electric vehicles for LFC based on the current mobility demand and the statistical mobility behavior of the vehicle users. Throughout the paper, the energy management system is evaluated for the smart grid of an intermediate city.

Integrated current control, energy control and energy balancing of Modular Multilevel Converters

Modular Multilevel Converters (MMCs) are a new circuit topology to realize multilevel converters. Major advantages of MMCs in comparison to other multilevel converters or standard Voltage Source Converters (VSCs) are lower costs and a modular design for high voltage applications, higher reliability and longer maintenance intervals. However, modular multilevel topologies generally require advanced strategies for controlling and balancing energies since many energy storages are distributed all over the converter. This paper presents a novel integrated strategy for current control, energy control and energy balancing of MMCs. The MMC is modeled as a periodic bilinear time-varying system respecting all currents and energies. Furthermore, a control structure is proposed combining a periodic linear quadratic regulator (PLQR) with an extended and least squares (LS) estimator to determine the currents and energies. Finally, simulation results for converter energizing and transmission startup as well as a phase-to-ground fault are given to illustrate the effectiveness of the strategy.

Optimal Control and Scheduling of Switched Systems

This technical note addresses optimal control and scheduling (controlled switching) of discrete-time switched linear systems. A receding-horizon control and scheduling (RHCS) problem is introduced and solved by dynamic programming, leading to a combinatorial optimization problem with exponential complexity. By relaxed dynamic programming, complexity is reduced while relaxing optimality within prespecified bounds. The resulting RHCS strategy is expressed explicitly as a piecewise linear state feedback control law defined over regions implied by quadratic forms. Closed-loop stability is not guaranteed inherently for the RHCS strategy. Therefore, a posteriori stability criteria based on piecewise quadratic Lyapunov functions are proposed. Finally, a region-reachability criterion is presented.

Modeling and current control of modular multilevel converters considering actuator and sensor delays

Voltage source converter (VSC) technology becomes more and more common in high-voltage direct current (HVDC) transmission systems. Compared to conventional VSC technology modular multilevel design offers advantages such as higher voltage levels, modular construction, longer maintenance intervals and improved reliability. A multilevel approach can also help to reduce harmonics due to sinusoidal output voltages so that grid filters become dispensable. Modular multilevel converters (MMC) are a new circuit topology to realize multilevel converters that can be used for high-voltage transmission systems and offer the benefit of an only linearly rising hardware complexity for an increasing number of levels. To control the innovative MMC topology, special requirements exist which exceed classical converter control features. This paper presents a multivariable approach to realize an optimal current control considering actuator and sensor delays that occur in real converter control systems.

On co-design of filter and fault estimator against randomly occurring nonlinearities and randomly occurring deception attacks

In this paper, the co-design problem of filter and fault estimator is studied for a class of time-varying non-linear stochastic systems subject to randomly occurring nonlinearities and randomly occurring deception attacks. Two mutually independent random variables obeying the Bernoulli distribution are employed to characterize the phenomena of the randomly occurring nonlinearities and randomly occurring deception attacks, respectively. By using the augmentation approach, the co-design problem of the robust filter and fault estimator is converted into the recursive filter design problem. A new compensation scheme is proposed such that, for both randomly occurring nonlinearities and randomly occurring deception attacks, an upper bound of the filtering error covariance is obtained and such an upper bound is minimized by properly designing the filter gain at each sampling instant. Moreover, the explicit form of the filter gain is given based on the solution to two Riccati-like difference equations. It is shown that the proposed co-design algorithm is of a recursive form that is suitable for online computation. Finally, a simulation example is given to illustrate the usefulness of the developed filtering approach.

New results on stabilization of networked control systems with packet disordering

In this paper, the stabilization problem is studied for a class of networked control systems (NCSs) with delays, packet disordering and packet dropouts. A new packet reordering method is presented to deal with packet disordering and to choose the newest control input. A relationship between the reordered packet over two consecutive sampling intervals is given for the NCS with both time delays and packet dropouts. A sufficient condition for the NCS to be exponentially stable is presented by using the average dwell-time method. Finally, an illustrative example is given to demonstrate the effectiveness of the proposed method.

Fusion Estimation for Sensor Networks With Nonuniform Estimation Rates

This paper investigates the multi-sensor fusion estimation problem for wireless sensor networks with nonuniform estimation rates. First, each sensor generates local estimates with two rates, namely, a fast rate and a slow rate according to its power situation, where the estimation rates among the sensors are allowed to be different from each other. Second, a fusion rule with matrix weights is designed for each sensor to fuse available local estimates generated at different time scales, and a set of recursive equations are presented to compute estimation error cross-covariances. The fusion algorithm is applicable to both cases where the measurement noises are mutually correlated and are uncorrelated, and is also applicable to the case where the sensors are not time-synchronized. Two types of estimators are designed according to different considerations of design complexity and computation costs, and convergence analysis for the type II estimators is also presented. Finally, two illustrative examples are given to demonstrate the effectiveness of the proposed estimators.

Suboptimal Event-Triggered Control for Time-Delayed Linear Systems

This technical note considers event-triggering conditions and controller synthesis approaches for delayed linear systems. Optimization problems for minimizing the upper bound of quadratic cost functions are formulated in the form of linear matrix inequalities (LMIs). By solving the optimization problems a unique control gain can be obtained. The performance considered in this technical note includes a linear quadratic cost function for quantifying the control performance and average event times at which the control input must be updated for quantifying the transmission reductions. Comparisons with other approaches in the literature are given to demonstrate the advantages with respect to the two performance indices. Furthermore, an experimental implementation of the proposed methods in an inverted pendulum system shows the applicability and effectiveness in real world.

Event-Triggered Control for Linear Systems Subject to Actuator Saturation

Abstract Event-triggered control is developed to reduce the communication load in networked control systems. This means that output or actuator signals are only transmitted over the network when an event-triggering condition is violated which is designed such that a certain control performance can be guaranteed. This paper considers event-triggered control subject to actuator saturation for linear systems. Therefore we present a method to estimate the domain of attraction which represents a contractive invariant set. The contractive invariant set is estimated by an ellipsoid which is determined by solving a linear matrix inequality (LMI) optimization problem. Further a controller synthesis design considering the event-triggering condition in our criterion is given to maximize the contractive ellipsoid. Simulations are given to illustrate the results.