Fabric image retrieval is crucial for textile mills to manage their inventory and samples, but it is challenging due to the diverse appearance and fine-grained texture of fabrics. This paper proposes an algorithm based on finegrained features to deal with this issue. The algorithm uses the coordinate attention(CA) module to extract precise location information of the fabric images and scales the overall network structure of Mobile Net V3 to reduce the training time and model parameters. The optimized model is selected based on the scaling factor method, and fabric retrieval experiments are conducted on the fabric image dataset(FID). The results show that the algorithm effectively improves the accuracy of fabric image feature extraction, with a retrieval accuracy(Acc) of 91. 82% and floating point operations(FLOPs) of 175. 34 MB. The Acc is improved by 13. 49 percentage points compared with that of the original Mobile Net V3 model, while the training time is reduced, and the inference speed is improved by 25. 14%. The algorithm has practical application value.

Two-dimensional(2D) semiconducting transition metal dichalcogenides(TMDs) have unique electrical, optical and mechanical properties, and hold great potential for diverse applications such as digital circuits, light harvesting and energy storage. Controlling the electrical properties of TMDs through doping provides an effective approach for sensitive sensing. This paper presents the experimental study of the doping effect of oxygen plasma on molybdenum disulfide(MoS₂). Firstly, the transport characteristics of the MoS₂ field-effect transistor(FET) were investigated and the MoS₂ FET exhibited p-type doping through plasma treatment. Then, the cause of the doping effect was further studied, and the doping effect was attributed to the formation of MoO₃-like defects on the surface of the channel, confirmed by Raman spectroscopy. Finally, the humidity-sensing behavior of the plasma-treated MoS₂ FET was studied. The MoS₂ FET exhibited high sensitivity to humidity because of the increased adsorption centers for water molecules, with the source-drain current change of approximately 54% in humid environment. The work would provide a simple method to modify the electrical properties of TMDs and show potential for low-dimensional chemical sensors.

Developing pure organic room-temperature phosphorescence(RTP) materials with visible light activation has drawn widespread attention. In this work, a visible light-activated RTP design strategy was developed by incorporating the phenanthroline-based donor-acceptor(D-A) phosphor into a rigid polymer matrix polyvinyl alcohol(PVA). Phenanthroline with rich heteroatom N can promote the generation of triplet excitons and form abundant hydrogen bonds with PVA, inhibiting the non-radiative relaxation and thereby leading to phosphorescence. Upon irradiation with 420 nm visible light, the phosphorescence color of these doped PVA films can be shifted from green to yellow by regulating the molecular conjugated structure and D-A interaction. Based on the phosphorescence properties, these doped PVA films can be used for information encryption. This work offers a simple and feasible approach for constructing visible light-activated phosphorescence materials with excellent application prospects in information encryption, sensors and other fields.

For developing an efficient solar cooling technology, a novel coupled system comprising a photovoltaic(PV) module and a van der Waals heterostructure(vdWH)-based thermionic refrigerator(TIR) is established. With full consideration of internal and external irreversibility, the theoretical model of the coupled system is constructed, and mathematical expressions for the key performance indicators are derived. On this basis, the general properties of the coupled system are investigated, and the voltage region permitting the system to operate is determined. According to the calculations, the maximum refrigerating capacity and the maximum coefficient of performance(COP) are 75.88 W and 0.49, respectively. Furthermore, sensitivity analyses are conducted to derive the regularity and magnitude of the impacts of critical parameters on the overall performance of the coupled system, including solar irradiance, effective Schottky barrier height, inter-layer thermal resistance, external thermal resistance, heat leakage thermal resistance and hot reservoir temperature. The obtained outcomes may contribute to the design and operation of practical coupled systems.

Indigo is one of the most important vat dyes in the textile industry, and it must be reduced to a water-soluble leuco form before dyeing. This study aims to seek a complex green reducing agent as a possible substitute for the environmentally unfavorable sodium dithionite(SD) used in indigo dyeing. Firstly, the stability of three reducing agents, SD, thiourea dioxide(TD) and glucose(GS), is compared. The reduction system of indigo dyeing with TD can maintain good stability after 2 h vigorous stirring. However, SD and GS cannot reach the reduction potential required by indigo after 1 h vigorous stirring, and the dyeing performance of indigo decreased. Considering that GS is more eco-friendly than SD, the complex of TD and GS is selected as the green reducing agent of indigo dyeing. Secondly, the reduction potential, pH and K/S values for indigo dyeing on cotton fabrics with different mass ratios of the complex reducing agents are analyzed. The results show that the optimum mass ratio of TD to GS is 7∶3, and under this condition, a stable reduction potential and a high dyeing ability are obtained.

The complex and changeable environment in the process of bearing operation may lead to inconsistent distribution of training data and test data, and decrease the diagnosis performance of the model. Thus a bearing fault diagnosis model based on the Shuffle-CANet is proposed, and realizes bearing cross-domain fault diagnosis by improving the ShuffleNet V2 and introducing asymmetric convolution. A domain loss function is added to the model based on the idea of domain adaptation in transfer learning so that the common features of the source domain and the target domain can be extracted occasionally and the cross-domain fault diagnosis can be realized. Compared with the traditional deep learning model, this model is friendlier to mobile and embedded devices. The Shuffle-CANet is validated by different transfer tasks on two different datasets. The results show that when the source domain and the target domain are derived from the same dataset, the fault diagnostic accuracy of the model can be more than 99%. When the target domain and the source domain are derived from different datasets, the fault diagnostic accuracy of the model can be more than 95%.

Based on the ciphertext-only attack(COA) assumption, the statistical fault analysis(SFA) is proposed to break all versions of QARMA in the Internet of Everything(IoE), where suitable strategies are taken into consideration for the uncertainty of tweaks to cover more rounds of fault injections. It also presents the novel double distinguishers of Cramér-von Mises test-Hamming weight(CM-HW) and Kuiper's test-maximum likelihood estimation(KT-MLE) to improve the attacking efficiency. According to the experimental results, the attackers can inject 374 and 726 random faults into the deeper antepenultimate round to recover 128-bit and 256-bit secret keys of QARMA with a reliability of at least 99%, respectively. Hence, QARMA is vulnerable to the SFA in the IoE. The results offer a valuable reference for the lightweight tweakable cryptosystems with the reflection structure and the protection of the cryptographic devices.

Aiming at the problems of a large difficulty coefficient and tedious process in the clinical fitting of the orthokeratology(OK) lens, a stacking ensemble learning model is proposed to predict the parameters of the OK lens and realize its intelligent fitting. The feature set that is most suitable for the target variables is constructed by feature derivation based on F-test and feature selection under the variance-improved Boruta algorithm. A stacking ensemble learning prediction model is studied. The model uses random forest(RF), gradient boosting decision tree(GBDT) and support vector regression(SVR) as the first layer basic learners and linear regression(LR) as the second layer meta-learner. The experimental results show that the prediction indexes of the model are highly consistent with the clinical diagnosis results, which verifies that the model can be used as an effective auxiliary diagnosis method.

As an emerging sensing paradigm, mobile crowd sensing(MCS) comprises a collection of mobile users that utilize their sensing devices to efficiently execute and send data contributions. However, the integration of privacy and reputation mechanisms(evaluating reliability) is crucial requirements for building secure and reliable MCS applications. Firstly, participants are assured that their privacy is preserved even if they contribute sensitive personal data. Secondly, the reputation mechanism allows the server to monitor participant behaviors and reliability, as biased or inaccurate contributions may demote the system quality, making it essential for the server to validate participants. Integrating a reputation mechanism with privacy is a challenging and contradictory objective. The reputation mechanism measures the participant behavior during the entire sensing activity, while privacy aims to preserve participant credentials. Thus, a novel privacy-preserving and reputation-aware participant selection(PRPS) scheme for MCS has been proposed. The PRPS scheme integrates privacy with a reputation mechanism, preserves the privacy of participant identities and reputation values by employing pseudonyms and cloaking techniques, respectively, and protects the location and data privacy. Extensive simulations have been conducted. Using performance evaluation, we affirm precision, efficacy and scalability of the PRPS scheme by comparing privacy-preserving and utility-aware participant selection(PUPS) and utility-aware participant selection(UPS) schemes, respectively, and demonstrate the impact of privacy and reputation on data contributions. Next, the outcomes of the PRPS scheme are assessed. Finally, we estimate the efficiency and the accuracy of the PRPS scheme in evaluating participant reliability and behavior.

The robust H_∞ bipartite consensus problem is studied for a class of nonlinear time-varying multi-agent systems(MASs) with parameter uncertainties and external disturbances over signed networks. Following the thought of dealing with uncertainties in robust control, the considered system is transformed into a time-invariant dynamical model with norm-bounded parameter uncertainties. The robust bipartite consensus is converted to a reduced-order H_∞ control problem. Based on the Lyapunov stability theory, sufficient conditions in linear matrix inequalities(LMIs) are obtained for the robust bipartite consensus of MASs with desired H_∞ performance. Moreover, the design procedure of a distributed static output feedback controller is shown. Furthermore, an application for two-degree-of-freedom(2-DOF) planar mobile robots is presented to illustrate the effectiveness of the proposed controller.

In order to study the exact number and states of fixed points in the generalized dynamical system with NAND or NOR local functions over directed rooted trees, structural analysis and classification discussion methods are applied. The exact results of the fixed points in such dynamical systems are obtained. It is proved that the fixed points in such dynamical systems are completely determined by the loops in the rooted trees. This work provides a relevant advance in the knowledge of discrete dynamical systems which constitute mathematical tools to model simulation processes.