Bayesian network based extreme learning machine for subjectivity detection

Subjectivity detection is a task of natural language processing that aims to remove ‘factual’ or ‘neutral’ content, i.e., objective text that does not contain any opinion, from online product reviews. Such a pre-processing step is crucial to increase the accuracy of sentiment analysis systems, as these are usually optimized for the binary classification task of distinguishing between positive and negative content. In this paper, we extend the extreme learning machine (ELM) paradigm to a novel framework that exploits the features of both Bayesian networks and fuzzy recurrent neural networks to perform subjectivity detection. In particular, Bayesian networks are used to build a network of connections among the hidden neurons of the conventional ELM configuration in order to capture dependencies in high-dimensional data. Next, a fuzzy recurrent neural network inherits the overall structure generated by the Bayesian networks to model temporal features in the predictor. Experimental results confirmed the ability of the proposed framework to deal with standard subjectivity detection problems and also proved its capacity to address portability across languages in translation tasks.     

Adaptive iterative learning protocol design for nonlinear multi-agent systems with unknown control direction

This paper investigates a new adaptive iterative learning control protocol design for uncertain nonlinear multi-agent systems with unknown gain signs. Based on Nussbaum gain, adaptive iterative learning control protocols are designed for each follower agent and the adaptive laws depend on the information available from the agents in the neighbourhood. The proper protocols guarantee each follower agent track the leader perfectly on the finite time interval and the Nussbaum-type item can seek control direction adaptively. Furthermore, the formation problem is studied as an extension. Finally, simulation examples are given to demonstrate the effectiveness of the proposed method in this article.     

Trajectory tracking control for rotary steerable systems using interval type-2 fuzzy logic and reinforcement learning

Rotary steerable system (RSS) is a directional drilling technique which has been applied in oil and gas exploration under complex environment for the requirements of fossil energy and geological prospecting. The nonlinearities and uncertainties which are caused by dynamical device, mechanical structure, extreme downhole environment and requirements of complex trajectory design in the actual drilling work increase the difficulties of accurate trajectory tracking. This paper proposes a model-based dual-loop feedback cooperative control method based on interval type-2 fuzzy logic control (IT2FLC) and actor-critic reinforcement learning (RL) algorithms with one-order digital low-pass filters (LPF) for three-dimensional trajectory tracking of RSS. In the proposed RSS trajectory tracking control architecture, an IT2FLC is utilized to deal with system nonlinearities and uncertainties, and an online iterative actor-critic RL controller structured by radial basis function neural networks (RBFNN) and adaptive dynamic programming (ADP) is exploited to eliminate the stick–slip oscillations relying on its approximate properties both in action function (actor) and value function (critic). The two control effects are fused to constitute cooperative controller to realize accurate trajectory tracking of RSS. The effectiveness of our controller is validated by simulations on designed function tests for angle building hole rate and complete downhole trajectory tracking, and by comparisons with other control methods.     

An update algorithm design using moving Region of Attraction for SDRE based control law

State Dependent Riccati Equation (SDRE) methods have the considerable advantages over other nonlinear control methods. However, stability issues can be arisen in SDRE based control system due to the lack of the global asymptotic stability property. Therefore, the previous studies have usually shown that local asymptotic stability can be ensured by estimating a Region of Attraction (ROA) around the equilibrium point. These estimated regions for stability may become narrow or the condition to keep the states in this region may be very conservative. To resolve these issues, this paper proposes a novel SDRE method employing an update algorithm to re-estimate the ROA when the states tend to move out of the stable region. The tendency is checked using a condition which is developed based on a new theorem. The theorem proves that it is possible to redesign the previous ROA with respect to the current states lying close to its boundary for ensuring the “non-local” stability along the trajectory without the need of solving SDRE at each time instant, unlike the standard SDRE approach. Therefore, the new theorem is now able to enhance the stability of the SDRE based closed-loop control system. The feasibility of the proposed SDRE control method is tested in both simulations and experiments. A validated 3-DOF laboratory helicopter is used for experiments and the control objective for the helicopter is to realise a preplanned movement in both elevation and travel axes. The results reveal that the proposed SDRE approach enables the controlled plant to track the desired trajectory as satisfactorily as the standard SDRE approach, while only solving SDRE when needed. The proposed SDRE method reduces the computational load for practical implementation of the control algorithm whilst ensuring the stability over the operational region.     

A LPV modeling of turbocharged spark-ignition automotive engine oriented to fault detection and isolation purposes

This paper illustrates the derivation of a linear parameter varying (LPV) model approximation of a turbocharged Spark-Ignition (SI) automotive engine and its usage in designing a model-based fault detection and isolation (FDI) scheme. The LPV approximation is derived from a detailed nonlinear mathematical model of the engine on the basis of the well known Jacobian approach. The resulting LPV representation is then exploited for synthesizing a bank of LPV-FDI $H_{\infty }/H_{?}$ Luenberger observers. Each observer is in charge of detecting a particular class of fault and is designed for having low sensitivity to all other exogenous inputs so as to allow an effective fault isolation. The adopted FDI scheme is gain-scheduled and exploits a set of engine variables, assumed to be measurable on-line, as a scheduling parameters. The goodness of the LPV approximation of the engine model and the effectiveness of the LPV-FDI architecture are demonstrated by several numerical simulations.     

A novel performance criterion approach to optimum design of PID controller using cuckoo search algorithm for AVR system

This article presents a novel tuning design of Proportional-Integral-Derivative (PID) controller in the Automatic Voltage Regulator (AVR) system by using Cuckoo Search (CS) algorithm with a new time domain performance criterion. This performance criterion was chosen to minimize the maximum overshoot, rise time, settling time and steady state error of the terminal voltage. In order to compare CS with other evolutionary algorithms, the proposed objective function was used in Particle Swarm Optimization (PSO) and Artificial Bee Colony (ABC) algorithms for PID design of the AVR system. The performance of the proposed CS based PID controller was compared to the PID controllers tuned by the different evolutionary algorithms using various objective functions proposed in the literature. Dynamic response and a frequency response of the proposed CS based PID controller were examined in detail. Moreover, the disturbance rejection and robustness performance of the tuned controller against parametric uncertainties were obtained, separately. Energy consumptions of the proposed PID controller and the PID controllers tuned by the PSO and ABC algorithms were analyzed thoroughly. Extensive simulation results demonstrate that the CS based PID controller has better control performance in comparison with other PID controllers tuned by the PSO and ABC algorithms. Furthermore, the proposed objective function remarkably improves the PID tuning optimization technique.     

Extended dissipative learning of time-delay recurrent neural networks

The paper addresses the issue of extended dissipative learning for a class of delayed recurrent neural networks. Both time-varying delay and time-invariant delay are taken into account. By choosing appropriate Lyapunov–Krasovkii functionals and utilizing some inequalities, several weight learning rules are developed for ensuring the network to be asymptotically stable and extended dissipative. The existence conditions for these learning strategies consist of a few linear matrix inequalities, which are able to be verified readily by Matlab software. Two numerical examples are employed to show the effectiveness and low conservatism of the proposed learning rules.     

Iterative learning control for fractional-order multi-agent systems

In this paper, we apply iterative learning control to both linear and nonlinear fractional-order multi-agent systems to solve consensus tacking problem. Both fixed and iteration-varying communicating graphs are addressed in this paper. For linear systems, a PD^α-type update law with initial state learning mechanism is introduced by virtue of the memory property of fractional-order derivative. For nonlinear systems, a D^α-type update law with forgetting factor and initial state learning is designed. Sufficient conditions for both linear and nonlinear systems are established to guarantee all agents achieving the asymptotic output consensus. Simulation examples are provided to verify the proposed schemes.     

Fault-tolerant safe control design of switched and interconnected nonlinear systems

This paper provides novel fault-tolerant safe control (FTSC) strategies for switched and interconnected nonlinear systems. With several switching and interconnection situations considered, the proposed corresponding strategies ensure that the state never enters the unsafe set and asymptotically converges to the origin in the presence of faults. This relies on a proposed concept named “fault-tolerant control Lyapunov-Barrier functions (FTCLBF)”. Two practical examples are taken to demonstrate the efficiency of the proposed method.     

Sparse subspace clustering with jointly learning representation and affinity matrix

In recent years, sparse subspace clustering (SSC) has been witnessed to its advantages in subspace clustering field. Generally, the SSC first learns the representation matrix of data by self-expressive, and then constructs affinity matrix based on the obtained sparse representation. Finally, the clustering result is achieved by applying spectral clustering to the affinity matrix. As described above, the existing SSC algorithms often learn the sparse representation and affinity matrix in a separate way. As a result, it may not lead to the optimum clustering result because of the independence process. To this end, we proposed a novel clustering algorithm via learning representation and affinity matrix conjointly. By the proposed method, we can learn sparse representation and affinity matrix in a unified framework, where the procedure is conducted by using the graph regularizer derived from the affinity matrix. Experimental results show the proposed method achieves better clustering results compared to other subspace clustering approaches.     

Composite cooperative synchronization and decentralized learning of multi-robot manipulators with heterogeneous nonlinear uncertain dynamics

This paper studies the problem of composite synchronization and learning of multiple coordinated robot manipulators subject to heterogeneous nonlinear uncertain dynamics under the leader-follower framework. A new two-layer distributed adaptive learning control scheme is proposed, which consists of the first-layer distributed cooperative estimator and the second-layer decentralized deterministic learning controller. The first layer aims to enable each robotic agent to estimate the leader’s information. The second layer is responsible for not only controlling each individual robotic agent to track over desired reference trajectory, but also accurately identifying/learning each robot’s nonlinear uncertain dynamics. Design and implementation of this two-layer distributed controller can be carried out in a fully-distributed manner, which do not require any global information including global connectivity of the communication network. The Lyapunov method is applied to rigorously analyze stability and parameter convergence of the resulting closed-loop system. Numerical simulations on a team of two-degree-of-freedom robot manipulators have been conducted to demonstrate the effectiveness of the proposed results.     

Iterative learning control for linear delay systems with deterministic and random impulses

This paper investigates convergence of iterative learning control for linear delay systems with deterministic and random impulses by virtute of the representation of solutions involving a concept of delayed exponential matrix. We address linear delay systems with deterministic impulses by designing a standard P-type learning law via rigorous mathematical analysis. Next, we extend to consider the tracking problem for delay systems with random impulses under randomly varying length circumstances by designing two modified learning laws. We present sufficient conditions for both deterministic and random impulse cases to guarantee the zero-error convergence of tracking error in the sense of Lebesgue-p norm and the expectation of Lebesgue-p norm of stochastic variable, respectively. Finally, numerical examples are given to verify the theoretical results.     

Iterative learning control for boundary tracking of uncertain nonlinear wave equations

This work presents a framework of iterative learning control (ILC) design for a class of nonlinear wave equations. The main contribution lies in that it is the first time to extend the idea of well-established ILC for lumped parameter systems to boundary tracking control of nonlinear hyperbolic distributed parameter systems (DPSs). By fully utilizing the system repetitiveness, the proposed control algorithm is capable of dealing with time-space-varying and even state-dependent uncertainties. The convergence and robustness of the proposed ILC scheme are analyzed rigorously via the contraction mapping methodology and differential/integral constraints without any system dynamics simplification or discretization. In the end, two examples are provided to show the efficacy of the proposed control scheme.     

Distributed cooperative controller design considering guidance loop and impact angle

This paper presents an integrated distributed cooperative guidance and control scheme for multiple missiles to attack a single target simultaneously at desired impact angles. The scheme is divided into two parts: individual part and cooperative part. For the individual part, partial integrated guidance and control method is adopted to generate the elevator deflection (which is a realistic control input) to ensure that the missiles fly along their respective desired line of sight and hit the target; this is in contrast to previous works which analyze only the engagement dynamics and use missile accelerations as the control input, however, the proposed controller also considers the missile dynamics, thus enabling the implementation of an autopilot. For the cooperative part, using only information from adjacent missiles, the proposed distributed cooperative controller can make all missiles hit the target simultaneously. Hence in this scheme, each missile can hit the target at desired angles and at the same time, thus achieving salvo attack. Simulations are performed to verify the effectiveness of the scheme.     

HOG-assisted deep feature learning for pedestrian gender recognition

Pedestrian gender recognition is a very challenging problem, since the viewpoint variations, illumination changes, occlusion, and poor quality are usually encountered in the pedestrian images. To address this problem, an effective HOG-assisted deep feature learning (HDFL) method is proposed in this paper. The key novelty lies in the design of HDFL network to effectively explore both deep-learned feature and weighted histogram of oriented gradient (HOG) feature for the pedestrian gender recognition. Specifically, the deep-learned and weighted HOG feature extraction branches are simultaneously performed on the input pedestrian image. A feature fusion process is subsequently conducted to obtain a more robust and discriminative feature, which is then fed to a softmax classifier for pedestrian gender recognition. Extensive experiments on multiple existing pedestrian image datasets have shown that the proposed HDFL method is able to effectively recognize the pedestrian gender, and consistently outperforms the state-of-the-art methods.     

Distributed cooperative learning for a group of uncertain systems via output feedback and neural networks

This paper studies the problem of adaptive neural network (NN) output-feedback control for a group of uncertain nonlinear multi-agent systems (MASs) from the viewpoint of cooperative learning. It is assumed that all MASs have identical unknown nonlinear dynamic models but carry out different periodic control tasks, i.e., each agent system has its own periodic reference trajectory. By establishing a network topology among systems, we propose a new consensus-based distributed cooperative learning (DCL) law for the unknown weights of radial basis function (RBF) neural networks appearing in output-feedback control laws. The main advantage of such a learning scheme is that all estimated weights converge to a small neighborhood of the optimal value over the union of all system estimated state orbits. Thus, the learned NN weights have better generalization ability than those obtained by traditional NN learning laws. Our control approach also guarantees the convergence of tracking errors and the stability of closed-loop system. Under the assumption that the network topology is undirected and connected, we give a strict proof by verifying the cooperative persisting excitation condition of RBF regression vectors. This condition is defined in our recent work and plays a key role in analyzing the convergence of adaptive parameters. Finally, two simulation examples are provided to verify the effectiveness and advantages of the control scheme proposed in this paper.     

Finite-time and fixed-time synchronization of discontinuous complex networks: A unified control framework design

This paper is concerned with the finite-time and fixed-time synchronization of complex networks with discontinuous nodes dynamics. Firstly, under the framework of Filippov solution, a new theorem of finite-time and fixed-time stability is established for nonlinear systems with discontinuous right-hand sides by using mainly reduction to absurdity. Furthermore, for a class of discontinuous complex networks, a general control law is firstly designed. Under the unified control framework and the same conditions, the considered networks are ensured to achieve finite-time or fixed-time synchronization by only adjusting the value of a key control parameter. Based on the similar discussion, a unified control strategy is also provided to realize respectively asymptotical, exponential and finite-time synchronization of the addressed networks. Finally, the derived theoretical results are supported by an example with numerical simulations.     

Analysis of iterative learning control for one-sided Lipschitz nonlinear singular systems

This paper deals with the problem of iterative learning control for a class of singular systems with one-sided Lipschitz nonlinearity. In order to track the given desired trajectory, a closed-loop D-type learning algorithm is proposed for such nonlinear singular systems. Then the convergence result is derived by utilizing the one-sided Lipschitz and quadratically inner-bounded conditions. In this work, the main contribution is to apply the iterative learning approach to one-sided Lipschitz singular systems, while most researches are focus on the Lipschitz systems. It is shown that the algorithm can guarantee the system output converges to the desired trajectory on the whole time interval. Finally, the effectiveness of the presented algorithm is verified by a numerical example.