Permanent Magnet Synchronous Motor (PMSM) is widely used due to its advantages of high power density, high efficiency and so on. In order to ensure the reliability of a PMSM system, it is extremely vital to accurately diagnose the incipient faults. In this paper, a variety of optimization algorithms are utilized to realize the diagnosis of the faulty position and severity of the inter-turn short-circuit (ITSC) fault, which is one of the most destructive and frequent faults in PMSM. Compared with the existing research results gained by particle swarm optimization algorithms, in this paper, the methods using other optimization algorithms incorporating genetic algorithm, whale optimization algorithm and stochastic parallel gradient descent algorithm (SPGD) can acquire more stable and precise results. In particular, the method based on SPGD can obtain the most desirable performance among the methods mentioned above; that is, the relative error of short-circuit turns ratio is approximately as low as 0.03%. In addition, in the case of asymmetric input three-phase voltage and with the adverse impact of high-order harmonics at different load moments, the fault diagnosis method based on SPGD still maintains relatively satisfactory properties. Finally, the verification on the actual PMSM platform demonstrates that the SPGD can still diagnose the faulty severity.

In this paper, a novel fractional-order active disturbance rejection control with fuzzy self-tuning method (FSFOADRC) is proposed for photoelectric tracking system (PTS). Firstly, aiming at the internal uncertainty of PTS and external disturbance, a fraction-order extended state observer (FOESO) is designed, and the FOESO can transform the plant into a simple form, which greatly simplifies the mathematical model. Secondly, a fuzzy regulator is applied to the proportion–differentiation controller (PD), increasing the flexibility and adaptivity of the controller. In addition, the stability of the whole control system can be guaranteed. Eventually, numerical comparative simulations are implemented to verify the feasibility and superiority of the proposed method. Compared with the integral-order active disturbance rejection control (IOADRC), fractional-order active disturbance rejection control (FOADRC) without the fuzzy regulator and proportion–integration–differentiation (PID) controller, the proposed method performs better with faster response, smaller overshoot, and stronger disturbance suppression capability.

In recent years, as a vital branch of artificial intelligence, image recognition (IR) has garnered significant attention across various industries. Although deep convolutional neural networks (DCNNs) and attention-based models have achieved remarkable advancements in this field, they each exhibit inherent limitations and struggle to effectively handle uncertainty. To address these challenges, this article proposes a multiple attentions based multi-level hybrid-guided deep fuzzy convolutional neural network (MAMH-DFCNN), aimed at enhancing recognition accuracy and uncertainty processing ability while simultaneously expanding the application scope of deep fuzzy systems (DFSs). Specifically, by introducing a novel squeeze-and-excitation (SE) attention mechanism based interval type-2 generalized fuzzy hyperbolic tangent autoencoder (SE-IT2GFHTA), a multi-level convolutional fuzzy feature extraction framework (MCFFEF) is developed to capture features across different spatial and granularity levels, enriching feature representations and improving the model's robustness against uncertainty. A spatial pyramid-type fuzzy feature attention fusion module (FFAFM) is then employed to fully integrate the multi-level fuzzy features, acquiring more abstract semantic information while reducing information redundancy. Subsequently, a hybrid-driven cross-attention classifier (HCAC) is designed to facilitate the deep interaction of features with different granularities, ultimately producing the final recognition outcome. Extensive comparative experiments and analytical evaluations are conducted. The results demonstrate that MAMH-DFCNN achieves desirable performance across both binary and multi-class classification tasks, exhibiting certain competitiveness against baseline models and state-of-the-art (SOTA) models. These findings highlight the feasibility and effectiveness of MAMH-DFCNN in improving recognition accuracy and robustness against uncertainty in IR.

In this paper, a novel control approach, namely fuzzy dynamic self-tuning-based linear active disturbance rejection control (FDS-LADRC), is proposed for the speed loop system of permanent magnet synchronous motors (PMSMs). Specifically, a control framework based on the linear active disturbance rejection control (LADRC) is presented. Fuzzy dynamic self-regulators are developed to enable simultaneous adaptive adjustments of both the controller and observer parameters. Additionally, the stability analysis is provided. A series of numerical simulations demonstrates that FDS-LADRC achieves superior adaptivity, transient performance, disturbance rejection capability, and anti-noise ability under various operating conditions. For instance, during no-load startup, compared with the traditional LADRC, nonlinear active disturbance rejection control (ADRC), a variant of FDS-LADRC named IT2FDS which utilizes interval type-2 fuzzy systems as fuzzy dynamic self-regulators, a state-of-the-art fractional-order ADRC with fuzzy self-tuning (FSFOADRC), and sliding mode control (SMC), FDS-LADRC reduces overshoot by 10.82 %, 13.55 %, 7.36 %, 5.53 %, and 3.94 %, respectively, and shortens settling time by 0.0132 s, 0.0076 s, 0.0139 s, 0.0009 s, and 0.0156 s, respectively. Finally, corresponding real-world experiments are conducted to validate the effectiveness and superiority of FDS-LADRC.

Autonomous navigation is a core function of Internet of Things (IoT)-enabled robotic systems, particularly in applications such as smart warehousing, intelligent inspection, and collaborative logistics. Reinforcement learning (RL) has achieved remarkable progress in robotic navigation, but conventional RL methods typically rely on fixed task distributions, making it difficult to adapt to incremental multiscene navigation strategies that are common in dynamic IoT environments. Learning navigation across multiple scenes faces major challenges: catastrophic forgetting in sequential learning, where new scene learning overwrites prior knowledge, and sample inefficiency due to random exploration in complex environments. To address these challenges, this article proposes a knowledge-guided continual RL (KGCRL) approach. Built upon the deep deterministic policy gradient (DDPG) framework, KGCRL integrates elastic weight consolidation (EWC) to mitigate catastrophic forgetting by preserving critical parameters from previously learned tasks. Furthermore, a knowledge-guided learning mechanism is introduced to incorporate external knowledge, thereby improving exploration efficiency and stability. Experimental results in incremental scene navigation demonstrate that KGCRL significantly improves scene adaptability while reducing cross-scene performance degradation. These results validate the effectiveness and practicality of KGCRL, providing new insights for robot incremental scene navigation.

Hierarchical fuzzy systems (HFSs), as a type of deep-structured fuzzy model, exhibit strong expressive power, excellent scalability, high flexibility, and desirable interpretability, making them highly promising for tackling a variety of complex problems. Despite their widespread applications in various fields such as control, classification, and modeling, very little research has been conducted to analyze how inputs influence the system's behavior. Consequently, this paper presents an in-depth analysis of a class of incremental hierarchical fuzzy systems (IHFSs) from two perspectives of input importance. First, an enhanced mutual information-based input importance ranking method (EMI-IIR) is proposed, enabling a more effective and concise evaluation of the degree to which inputs influence or depend on the ground truth. Next, a comprehensive sensitivity analysis, both at the subsystem level and the entire system level, is performed to illustrate the impact of inputs on the system. Specifically, a novel sensitivity metric, termed average sensitivity (AS), is proposed to simplify sensitivity calculations, thereby significantly reducing computational costs. Several related theorems are provided to theoretically analyze the sensitivity of the system to inputs under specific conditions. Furthermore, an evaluation framework based on the AS, called ASEF, is proposed for assessing the rationality of the constructed IHFS structure, which facilitates the effective design of high-performance systems for modeling tasks. A series of experiments on both synthetic and real-world datasets demonstrate the effectiveness and superiority of the proposed methods. Finally, several potential research directions are suggested to drive further progress.