Students’ actual learning engagement in class, which we call learning attention, is a major indicator used to measure learning outcomes. Obtaining and analyzing students’ attention accurately in offline classes is important empirical research that can improve teachers’ teaching methods. This paper proposes a method to obtain and measure students’ attention in class by applying a variety of deep-learning models and initiatively divides a whole class into a series of time durations, which are categorized into four states: lecturing, interaction, practice, and transcription. After video and audio information is taken with Internet of Things (IoT) technology in class, Retinaface and the Vision Transformer (ViT) model is used to detect faces and extract students’ head-pose parameters. Automatic speech recognition (ASR) models are used to divide a class into a series of four states. Combining the class-state sequence and each student’s head-pose parameters, the learning attention of each student can be accurately calculated. Finally, individual and statistical learning attention analyses are conducted that can help teachers to improve their teaching methods. This method shows potential application value and can be deployed in schools and applied in different smart education programs.

Automated high-density storage systems (AHDSS) have attracted widespread attention in recent years owing to their advantages of high throughput and space utilisation. However, owing to the characteristics of large-scale, multi-disturbance, and short-period task scenarios, a system is required to make instant and efficient decisions. To this end, this paper proposes a data-driven real-time decision-making method to solve the real-time equipment scheduling and dynamic location assignment problem in AHDSS. The proposed method comprises two phases: decision scheme learning and real-time decision-making. The operation state attribute features of the AHDSS were constructed to generate training data for equipment scheduling and location assignment scheme learning. Thereafter, a hierarchical learning and decision-making mechanism based on the deep belief network (DBN) is proposed. The integrated learning of better scheduling solutions was realised by establishing three-stage models of lift selection, shuttle selection, and location priority. Additionally, the Taguchi method was adopted to determine the best performance parameters for DBNs at different learning stages. Compared with other well-known machine learning algorithms, DBNs have a higher learning accuracy. Finally, a real-world AHDSS problem is studied, and the results demonstrate that the proposed approach outperforms existing dispatching rules.

As the core component of permanent magnet motor, the magnetic tile defects seriously affect the quality of industrial motor. Automatic recognition of the surface defects of the magnetic tile is a difficult job since the patterns of the defects are complex and diverse. The existing defect recognition methods result in difficulty in practical application due to the complicated system structure and the low accuracy of the image segmentation and the target detection for the diversity of the defect patterns. A self-supervised learning (SSL) method, which benefits from its nonlinear feature extraction performance, is proposed in this study to improve the existing approaches. We proposed an efficient multihead self-attention method, which can automatically locate single or multiple defect areas of magnetic tile and extract features of the magnetic tile defects. We also designed an accurate full-connection classifier, which can accurately classify different defects of magnetic tile defects. A knowledge distillation process without labeling is proposed, which simplifies the self-supervised training process. The process of our method is as follows. A feature extraction model consists of standard vision transformer (ViT) backbone, which is trained by contrast learning without labeled dataset that is used to extract global and local features from the input magnetic tile images. Then, we use a full-connection neural network, which is trained by using labeled dataset to classify the known defect types. Finally, we combined the feature extraction model and defect classification model together to form a relatively simple integrated system. The public magnetic tile surface defect dataset, which holds 5 defect categories and 1 nondefect category, is used in the process of training, validating, and testing. We also use online data augmentation techs to increase training samples to make the model converge and achieve high classification accuracy. The experimental results show that the features extracted by the SSL method can get richer and more detailed features than the supervised learning model gets. The composite model reaches to a high testing accuracy of 98.3%, and gains relatively strong robustness and good generalization ability.

This paper investigates a consensus problem for a class of T–S fuzzy multiple-agent systems (MASs) with unknown input (UI). To begin with, an unknown input observer (UIO) is able to asymptotically estimate the system state and the UI is designed for each agent. In order to construct the UIO, the state interval estimation is obtained by first using zonotope theory. Next, using the interval estimation of the state, a correlation of the state and the UI is built. Subsequently, a UIO is constructed, which is proposed by building upon the algebraic relationship. Moreover, by using the estimations of the state and the UI, a distributed control protocol is developed based on the proposed UIO. And, with the proposed distributed control protocol, the T–S fuzzy MAS can achieve consensus, in that all the states of the agents can converge to the leader’s state asymptotically. Finally, the effectiveness of the proposed method is demonstrated through two simulation examples.

This paper develops a novel fault reconstruction (FR) method and an FR-based fault-tolerant control (FTC) scheme for systems suffering from both sensor and actuator faults based on the combination of a Luenberger-like reduced-order observer and an interval observer. Firstly, by introducing an output transformation, an auxiliary output that is able to decouple the sensor fault is obtained. Secondly, for addressing the external disturbance and actuator fault, a multiple unknown input (MUI) is formed, and a reduced-order observer that is able to decouple the MUI is constructed. Consequently, asymptotic convergence estimations of the state and the sensor fault can be accomplished. Thirdly, in order to obtain the asymptotic convergence actual FR (AFR), an interval observer is designed. After this, an algebraic connection of the MUI and the state error estimation is given, and, based on the algebraic relationship, an algebraic MUI reconstruction (MUIR) method is proposed. Finally, an FTC scheme is developed by using the state estimation and MUIR. Under the FTC, the closed-loop system is asymptotically stable even if it suffers from sensor and actuator faults simultaneously. Theoretical analysis demonstrates that the observer-based FTC mechanism satisfies the separation principle. At last, two simulation examples are given to verify the effectiveness of the proposed methods.