No abstract is provided for this article.
Deep neural networks and in particular, deep neural classifiers have become an integral part of many modern applications. Despite their practical success, we still have limited knowledge of how they work and the demand for such an understanding is evergrowing. In this regard, one crucial aspect of deep neural network classifiers that can help us deepen our knowledge about their decision-making behavior is to investigate their decision boundaries. Nevertheless, this is contingent upon having access to samples populating the areas near the decision boundary. To achieve this, we propose a novel approach we call Deep Decision boundary Instance Generation (DeepDIG). DeepDIG utilizes a method based on adversarial example generation as an effective way of generating samples near the decision boundary of a neural network model. Then, we introduce a set of important characteristics that take advantage of the generated borderline instances and provide multifaceted understandings of deep neural networks. We have performed extensive experiments on multiple representative datasets across various deep neural network models and characterized their decision boundaries. The entire code is publicly available. <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup>
The mechatronics systems are widely used in modern society. This paper presents a novel data-driven scheme which can be used for fault diagnose of mechatronics systems. The proposed method is based on the subspace identification of parity vector. By constructing the output observer, critical variables can be acquired by soft sensors. This makes the fault diagnoses free from the limitation of online measurement. A diagnose observer is designed directly from the parity vector. Finally, the proposed scheme is tested by the Simulink benchmark of vehicle suspension and shows its good performance.
The last few decades have witnessed the rapid growth of research & development on large-scale systems (LSSs) due to the increasing complexity and the growing demand of modern engineering systems. The LSSs can normally be viewed as interconnections of multiple subsystems. Many practical systems can be described by the LSSs such as power systems, multi-robot systems, communication networks, transportation networks, and supply chains. In such large-scale systems, the centralized control framework may become impossible for its implementation. The decentralized and distributed controls have emerged as the attractive control methodologies to handle the scale and interactions of large-scale complex systems. However, the interactions among different subsystems introduce many challenges in the analysis and synthesis of such systems. Therefore, it is of significance to address several fundamental problems regarding real-time analysis, estimation and control of these systems.
No abstract is provided for this article.
The paper proposes a performance metric evaluation for a distributed detection wireless sensor network with respect to IEEE 802.15.4 standard. A distributed detection scheme is considered with presence of the fusion node and organized sensors into the clustering and non-clustering networks. Sensors are distributed in clusters uniformly and nonuniformly and network has multilevel fusion centers. Fusion centers act as heads of clusters for decision making based on majority-like received signal strength (RSS) with comparison the optimized value of the common threshold. IEEE 802.15.4 Markov chain model derived the performance metric of proposed network architecture with MAC, PHY cross-layer parameters, and Channel State Information (CSI) specifications while it is including Path-loss, Modulation, Channel coding and Rayleigh fading. Simulation results represent significant enhancement on performance of network in terms of reliability, packet failure, average delay, power consumption, and throughput.
The guaranteed cost control problem is investigated for a class of nonlinear discrete-time systems with Markovian jumping parameters and mixed time delays. The mixed time delays involved consist of both the mode-dependent discrete delay and the distributed delay with mode-dependent lower bound. The associated cost function is of a quadratic summation form over the infinite horizon. The nonlinear functions are assumed to satisfy sector-bounded conditions. By introducing new Lyapunov-Krasovskii functionals and developing some new analysis techniques, sufficient conditions for the existence of guaranteed cost controllers are derived with respect to the given cost function. Moreover, a convex optimization approach is applied to search for the optimal guaranteed cost controller by minimizing the guaranteed cost of the closed-loop system. Numerical simulation is further carried out to demonstrate the effectiveness of the proposed methods.
This paper addresses the problem of fault detection for switched systems with time-varying delay using delta operator approach. By means of the delta operator systems, a fault detection observer is used as the residual generator. Based on the switched Lyapunov function approach, sufficient conditions for the existence of the fault detection observer are given in terms of linear matrix inequalities (LMIs). An example is provided to show the effectiveness and applicability of the proposed method.
This paper presents a novel controller design strategy, which allows computing static output-feedback controllers by solving two successive linear matrix inequality (LMI) optimization problems. This strategy is conceptually simple, easy to implement, computationally efficient, and it can be potentially applied to a wide variety of control problems, in which effective state-feedback controllers can be synthesized by using standard LMI optimization tools. To illustrate the application of the new approach in the field of structural vibration control of large structures, a static velocity-feedback H1 controller for the seismic protection of a five-story building is designed with excellent results.
State estimation problem is considered for a kind of wireless network control system with stochastic uncertainty and time delay. A sliding mode observer is designed for the system under the situation that no missing measurement occurs and system uncertainty happens in a stochastic way. The observer designed for the system can guarantee the system states will be driven onto the sliding surface under control law, and the sliding motion of system states on sliding surface will be stable. By constructing proper Lyapunov-Krasovskii functional, sufficient conditions are acquired via linear matrix inequality. Finally, simulation result is employed to show the effectiveness of the proposed method.
No abstract is provided for this article.
No abstract is provided for this article.
In this study, a cooperative game model is presented to schedule the day-ahead operation of multi-microgrid (MMG) systems. In the proposed model, microgrids are scheduled to achieve a global optimum for the cost of the multi-microgrid system. The minimum cost is achieved by transactions of microgrids with each other. Also, price-based demand response is implemented in the model to build a cost-reducing opportunity for consumers. Applying Shapley value, the optimum cost of the MMG system is fairly allocated between microgrids. To enhance the confidence level of results, data uncertainties are incorporated into the model. The uncertainties of renewable outputs, demand, and prices of trading with the main grid are applied in the model. The presented model is developed as a mixed-integer linear programming problem, and its efficiency is evaluated on a standard test system containing three microgrids. The cost of the MMG system when microgrids form a cooperative game is compared with the isolated status that microgrids do not transact energy with each other. The results indicate that the cost of the MMG system is declined using the proposed cooperative model in comparison with the isolated mode. Also, the cost of microgrid 1, microgrid 2, and microgrid 3 are improved by 2.4, 2.7, and 11.8%, respectively. Therefore, all the microgrids have an incentive to participate in the cooperative game, and both the total cost and each microgrid cost are improved in the cooperative game.
In marine guard, patrol, and racing scenarios, it is of great importance to autonomously helm an underactuated surface vehicle (USV) to accurately achieve successive waypoints tracking (SWT) with prescribed velocities and courses. In this paper, in the presence of completely unknown dynamics and environmental forces, the emerging SWT problem is innovatively solved by creating a novel model-free guidance-control integrated framework. In lieu of direct guidance to the waypoint which inevitably suffers from singularity, a new tool called bridge trajectory (BT) exactly passing through the generalized waypoint (GW) is first developed by defining marching and ahead points, i.e., a marching point (MP) and an ahead point (AP). Combining with pursuit guidance and finite-time unknown observer (FUO), successive BTs are switched ON and OFF, with the aid of MP and AP, respectively. By virtue of the FUO, cascade analysis, filtered backstepping, and Lyapunov approach, BT tracking control laws for surge and yaw motions are further synthesized to ensure successive GWs with desired positions, velocities, and courses can be tracked accurately, and thereby eventually contributing to a BT-guided model-free solution to the SWT problem. Simulation studies on a benchmark USV demonstrate remarkable performance of the proposed method.
