IEEE Transactions on Cybernetics最新文献

In this article, the attitude maneuver control with the arbitrary convergence time is investigated for rigid spacecraft. First, a time-varying sliding mode function expressed by a piecewise function is designed by using an exponential function. This designed sliding mode function contains two equilibria of the attitude control systems. Furthermore, an attitude control law is designed with the aid of this new sliding mode function such that the states of the closed-loop attitude system remain on the sliding mode surface from the initial time instant, and converge to the origin at an arbitrarily preset time. In addition, the unwinding phenomenon can also be avoided when the proposed control law is used.

In this article, the practical fixed-time (FxT) control of switched neutral Filippov systems (SNFSs) on networks is considered. The perturbation functions that can be discontinuous, and the neutral logics expressed by the difference operators, are addressed by new approaches. Several novel Lyapunov inequalities with indefinite functions are proposed, where detailed estimations of the settling-time (ST) are obtained by discussing the different values of the exponents of the Lyapunov functions, which can include the existing results. Considering that the states generally cannot converge to the origin accurately under finite-time control in real applications, practical FxT stability lemmas with indefinite functions are established for the first time, in which the bounded condition imposed on the indefinite function is more practical than the previously unbounded ones. By designing the adaptive control strategies, the FxT and practical FxT synchronization control are investigated based on the Lyapunov-Krasovskii functionals (LKFs), which show the delay characteristic via the adaptive update law containing delay values. Notably, the theoretical deficiency arising from the Lyapunov function when studying the FxT stability of real systems with delays by using the FxT stability lemmas with indefinite function is solved in a successful way. Finally, the validity of the main results is verified by numerical simulations on an electrical device containing an LC transmission line.

Rotating machinery often operates under varying working conditions, which poses significant challenges to achieving reliable bearing fault diagnosis using traditional deep learning-based models. To enhance the diagnostic performance for rolling bearings across diverse operational conditions and noisy environments, a data-selective multiscale dual transfer network (DSMDTN) with a data selector (DS), multiscale concatenation U-Net (MCU-Net), and dual classifier (DC) is proposed. The DS module employs a comprehensive scoring mechanism that integrates math, entropy, and anomaly scores to selectively identify high-quality source samples for model training. Meanwhile, the MCU-Net module incorporates gated convolutional (gated-conv) blocks and convolutional blocks to extract multiscale domain-invariant features and dynamically adjust feature importance. In addition, the DC module comprises separate source and target classifiers that jointly minimize distribution discrepancy and classification loss. The effectiveness of the proposed DSMDTN is validated through experiments on thepublic Case Western Reserve University (CWRU) dataset and the proprietary PT dataset collected from a PT500mini test bed. The experimental results demonstrate that DSMDTN achieves higher accuracy and exhibits stronger transfer capability compared to several state-of-the-art intelligent models across various transfer tasks and under different noise levels.

Linear discriminant analysis (LDA) is a widely used dimensionality reduction (DR) technique that is effective in extracting discriminative features across various fields. Ratio sum LDA (RSLDA), a variant of LDA, was developed to address the shortcoming of the LDA method, which tends to obtain features with weak discriminative information. However, the traditional ratio sum formulation is dominated by the maximum ratio, which makes it difficult to select highly discriminative features and contradicts the original goal of the ratio sum. In this article, we analyzed the underlying causes of the dominance problem in RSLDA. A novel discriminant feature learning method via balance ratio sum discriminant analysis (BRSDA) is proposed. BRSDA effectively balances the ratios of the model formulation, thereby mitigating the domination problem. It focuses on optimizing low-quality projection directions, thereby yielding a well-balanced solution with consistently strong projection quality. First, a minimization ratio sum (Min-RS) criterion is adopted, which leverages the balance property of the harmonic mean to balance the gaps between the ratios. Second, Min-RS is integrated with the $ell _{p}$ -norm to further balance gaps between ratios. By amplifying the differences across projection directions, the $ell _{p}$ -norm drives BRSDA to emphasize the optimization of low-quality directions, thus raising the lower bound of direction quality. Finally, since obtaining a closed-form solution for the BRSDA problem is challenging, the gradient descent method is employed to solve its optimization. Sufficient experimental results verify the effectiveness of BRSDA, and BRSDA can effectively solve the domination problem and extract discriminative features.

This article focuses on the low-conservative stability criteria of delayed neural networks (DNNs). To achieve this goal, new techniques are developed to effectively utilize more system-related information. To use the time-varying delay information, some delay-product terms are introduced into the Lyapunov-Krasovskii functional (LKF), and an extended matrix-injection-based transformation method, which introduces delay-derivative-dependent slack matrices while obtaining the negative definite condition, is proposed. With respect totheuse of activation function information, the terms related to the activation function are fully augmented in the LKF. In particular, by considering the sector-constraint information of the activation function, a new nonlinear-function-dependent functional term is established, and a sector-constraint-dependent matrix-separation-based inequality is developed. By applying the above techniques, several improved stability criteria are derived, and two typical examples are provided to illustrate the advantages of the proposed methods.

This article presents an adaptive neural network (AdNN)-based fault detection framework for the thermal processes of lithium-ion (Li-ion) batteries governed by 2-D semilinear partial differential equations (PDEs) with partially known dynamics. To address the challenges of unknown nonlinear heat generation and limited sensor measurements, a two-stage approach combining reduced-order modeling with adaptive neural observation is proposed. First, a computationally tractable reduced-order model is derived through spectral approximation techniques. An adaptive neural observer is then designed to simultaneously estimate battery states and unknown nonlinear dynamics using only available surface temperature measurements. For robust fault detection, a hybrid scheme is developed that integrates model-based residual generation with data-driven threshold generation. Experimental validation on a pouch-type battery demonstrates the effectiveness of the proposed method in reliably detecting thermal abnormalities.

The existence of stochastic sampling phenomena in wastewater treatment processes (WWTPs) breaks the assumption that the existing control strategies use periodic data, and the operational constraints of equipment and the requirements for effluent water quality impose constraints on the system's input and output. These factors collectively increase the difficulty of achieving stable control of dissolved oxygen concentration (DOC). To solve these problems, a data-driven model predictive control (DDMPC) strategy is proposed to achieve stable control of constrained WWTPs with stochastic sampling intervals. First, a DDMPC framework is designed, which involves designing the objective function based on the mathematical expectation of the predicted output and considering system input and output constraints. In this framework, the problem of stochastic data acquisition caused by stochastic sampling can be solved, and the stable operation of the system can be ensured under constraints. Second, a data-driven multimodel prediction structure is constructed based on the stochastic characteristics of the sampling intervals. Specifically, fuzzy neural networks (FNNs) that match possible sampling intervals are established, thereby providing predictive outputs for the control process at the corresponding sampling instants. Third, a controller solving algorithm based on the generalized multiplier method is proposed, in which the constrained optimization problem within the model-predictive control (MPC) framework is reformulated by incorporating system constraints into the objective function as penalty functions to obtain the optimal control input that satisfies the constraints. Finally, the stability of the proposed DDMPC strategy is demonstrated, and its effectiveness is verified through the simulations on the benchmark simulation model No. 1 (BSM1). The results show that the proposed DDMPC strategy can achieve stable control of DOC in constrained WWTPs with stochastic sampling intervals.

Most research on flexible job shop scheduling assumes constant processing speeds. However, in real production, machines need to operate at variable speeds to achieve energy-efficient scheduling, which requires balancing multiobjective between production efficiency and green development. Such tradeoffs thus trigger the phenomenon in which massive solutions converge to identical objective values (i.e., the multimodal property), which is often neglected in scheduling problems. To address the above challenges, this work introduces a knowledge-enhanced evolutionary multitasking memetic algorithm (KEMMA) to solve the multimodal multiobjective flexible job shop scheduling problem considering speed (MMFJSP-S). First, self-paced learning motivated us to construct a simple auxiliary task and employ an evolutionary multitasking (EMT) framework to tackle the complex MMFJSP-S. Moreover, a knowledge enhancement and explicit transfer strategy is designed to reduce the effects of negative transfer by reinforcing and sharing beneficial knowledge across tasks. Finally, a mapping transformation mechanism is proposed to handle the multimodal property of the MMFJSP-S in the decision space. By comparing with ten advanced algorithms, the experimental results verify the remarkable superiority ofthe proposed KEMMA in solving MMFJSP-S and reveal the significance of studying the multimodal property.

In this article, we investigate a decentralized constrained optimization problem over time-varying directed networks. The nodes in the network aim to collaboratively minimize the aggregate of all locally known convex cost functions, subject to local nonidentical and multiple constraint sets, inequality constraints, and equality constraints. Problems of this nature arise in a number of applications in real networks, such as facility location in wireless sensor networks and image deblurring in machine learning. To address these types of problems, we propose an efficient subgradient-rescaling-based decentralized fixed-random projection (SR-DFRP) algorithm, named the SR-DFRP algorithm. In particular, the SR-DFRP algorithm employs Polyak's random projection to handle nonidentical and multiple constraints, which reduces computational load by avoiding the formulation of complex subproblems. Furthermore, by utilizing dynamically constructed row-stochastic matrices, the algorithm employs a subgradient rescaling strategy to mitigate the imbalance induced by the time-varying directed networks. Rigorous theoretical analyses are provided to establish that the SR-DFRP algorithm converges almost surely to the optimal solution. Extensive simulations on facility location and image deblurring problems are presented to validate the efficacy of the algorithm and the validity of the theoretical results.

This article proposes two types of unscented particle filters (UPFs) that leverage unscented transformation (UT) from a geometric perspective to compute the proposal distribution. An UPF on Lie groups is first developed. Specifically, both the propagation of the sigma points and the computation of the mean and covariance are performed on the Lie groups, while the weight update and resampling are conducted on the Lie algebra. Second, we introduce the log-linear property of group elements to streamline particle propagation by reducing redundant operations, thereby optimizing the proposed UPF framework. In the update process, intermittent measurements that are caused by factors such as packet dropouts and stochastic sensor scheduling are considered. While lowering computational demands, these measurements pose challenges to filter stability. To this end, the introduced property is used to prove that the estimation error remains bounded under certain assumptions. We further establish a critical threshold for the arrival rate of intermittent measurements and derive an upper bound for the expected state error covariance. Moreover, a detailed computational complexity analysis is conducted to evaluate the efficiency of the proposed method. Finally, with the original method serving as a benchmark, simulation and real-world GNSS/INS integrated navigation experiments confirm that the redesigned approach delivers comparable performance and significantly improved computational efficiency.