Publications

This paper investigates the problems of state observer design and observer-based integral sliding-mode control (SMC) for a class of Itô stochastic systems subject to simultaneous input and output disturbances. A new type of sliding-mode-based descriptor observer method is developed to approximate the system state and disturbance vectors. An integral-type SMC scheme is proposed based on the state estimation to stabilize the overall system. The main contributions of this approach are as follows: 1) The desired estimations of state and disturbance vectors can be obtained simultaneously, and 2) in the designed sliding-mode observer, the integral term of the Itô stochastic noise is eliminated in the proposed sliding-mode surface by a matrix design; thus, the reachability of the sliding-mode surface is strictly guaranteed. Finally, the proposed approach is applied to the disturbance reconstruction problem of an F-18 aircraft model to illustrate the effectiveness and applicability.

This paper deals with new finite-time estimation algorithms for Linear Parameter Varying (LPV) discrete-time systems and their application to output feedback stabilization. Two exact finite-time estimation schemes are proposed. The first scheme provides a direct and explicit estimation algorithm based on the use of delayed outputs, while the second scheme uses two combined asymptotic observers, connected by a condition of invertibility of a certain time-varying matrix, to recover solution of the LPV system in a finite-time. Furthermore, two stabilization strategies are proposed. The first strategy, called Delayed Inputs/Outputs Feedback (DIOF) stabilization method, is based on the use of the explicit estimation algorithm. The second technique, called Two Connected Observers Feedback (2-COF) stabilization method, is based on the use of two combined observers providing exact finite-time estimation. A numerical example is given to show the validity and effectiveness of the proposed algorithms by simulation.

Large data sets classification is widely used in many industrial applications. It is a challenging task to classify large data sets efficiently, accurately, and robustly, as large data sets always contain numerous instances with high dimensional feature space. In order to deal with this problem, in this paper we present an online Logdet divergence based metric learning (LDML) model by making use of the powerfulness of metric learning. We firstly generate a Mahalanobis matrix via learning the training data with LDML model. Meanwhile, we propose a compressed representation for high dimensional Mahalanobis matrix to reduce the computation complexity in each iteration. The final Mahalanobis matrix obtained this way measures the distances between instances accurately and serves as the basis of classifiers, for example, the<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M1"><mml:mrow><mml:mi>k</mml:mi></mml:mrow></mml:math>-nearest neighbors classifier. Experiments on benchmark data sets demonstrate that the proposed algorithm compares favorably with the state-of-the-art methods.

The lack of an efficient compression model remains a challenge for the wireless transmission of gearbox data in non-contact gear fault diagnosis problems. In this paper, we present a signal-adaptive asymmetrical autoencoder with a transform domain layer to compress sensor signals. First, a new discrete cosine Stockwell transform (DCST) layer is introduced to replace linear layers in a multi-layer autoencoder. A trainable filter is implemented in the DCST domain by utilizing the multiplication property of the convolution. A trainable hard-thresholding layer is applied to reduce redundant data in the DCST layer to make the feature map sparse. In comparison to the linear layer, the DCST layer reduces the number of trainable parameters and improves the accuracy of data reconstruction. Second, training the autoencoder with a sparsifying DCST layer only requires a small number of datasets. The proposed method is superior to other autoencoder-based methods on the University of Connecticut (UoC) and Southeast University (SEU) gearbox datasets, as the average quality score is improved by 2.00% at the lowest and 32.35% at the highest with a limited number of training samples

This paper designs a resilient stubborn observer for switched autonomous systems (SASs) with an energy harvesting sensor (EHS). In order to extend the lifespan of the sensor, a sensor with the ability of harvesting energy is used. For restraining the effect of measurement outliers, a stubborn observer is designed which includes a saturation function with the varying saturation levels. The observer gain fluctuation is considered which is dealt with by the resilience filtering scheme. The aim of this paper is, considering observer gain fluctuation, the measurement outliers and an EHS for SASs, to design a resilient stubborn observer which can guarantee the error dynamics to meet the given performance requirement. Specifically, the existence conditions are derived firstly and then the resilient stubborn observer is designed by linear matrix inequality (LMI) technology. At the end, the performance of the designed observer is evaluated through a numerical simulation and a set of comparative experiments are implemented to further signify the merits of the designed observer.

No abstract is provided for this article.

Anomaly detection algorithms typically utilize generative models to reconstruct anomaly regions. Post-processing is used to pinpoint the anomalies. However, the paucity of real-world anomaly samples and the complex image backgrounds pose significant challenges for anomaly detection. The work innovatively proposed a self-supervised anomaly detection method. An efficient channel attention mechanism in the autoencoder was introduced to improve the reconstruction performance. Besides, a foreground enhancement strategy was designed to distinguish the foreground from the background by maximizing the inter-class variance. The strategy reduced the effect of background noises and simulated various anomalies that were rare in real samples. The MVTecAD and BTAD datasets were used to experiment with anomaly detection and location. Experimental results demonstrated that our method achieved higher AUC and AP scores at both the image level and pixel level compared to other advanced methods. In particular, the average AP score increased by 12.5% at the pixel level.

The electrical machine core is subjected to mechanical stresses during manufacturing processes. These stresses include radial, circumferential and axial components that may have significant influence on the magnetic properties and it further leads to increase in iron loss and permeability in the stator core. In this research work, analysis of magnetic core iron loss under axial mechanical stress is investigated. The magnetic core is designed with Magnetic Flux Density (MF) ranging from 1.0 T to 1.5 T with estimated dimensions under various input voltages from 5 V to 85 V. Iron losses are predicted by the axial pressure created manually wherever required and is further applied to the designed magnetic core in the range of 5 MPa to 50 MPa. Finite element analysis is employed to estimate the magnetic core parameters and the magnetic core dimensions. A ring core is designed with the selected dimensions for the experimental evaluation. The analysis of iron loss at 50 Hz frequency for non-oriented electrical steel of M400-50A is tested experimentally using the Epstein frame test and force-fit setup test. Experimental evaluation concludes that the magnetic core saturates when it reaches its knee point of the B-H curve of the chosen material and also reveals that the axial pressure has a high impact on the magnetic properties of the material.

In this note, we study the problems of stability analysis and H <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">∞</sub> controller synthesis of discrete-time switched systems with time-varying delay. The system under consideration is firstly transformed into an interconnection system. Based on the system transformation and the scaled small gain theorem, the asymptotic stability of the original system is examined via the version of the bounded realness of the transformed forward system. The aim of the proposed approach is to reduce conservatism, which is made possible by a precise approximation of the time-varying delay and the input-output approach. The proposed stability condition is demonstrated to be much less conservative than most existing results. Moreover, the problem of H <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">∞</sub> controller synthesis involving convex optimization is further solved based on the stability condition, whose effectiveness are also illustrated via numerical examples.