In treatment of cancers, especially non-small-cell lung cancers such as lung squamous cell carcinoma(LUSC), tumor proportion score(TPS) of a programmed death-ligand 1(PD-L1) slide is essential for selecting tumor therapies. Many parameters of tumor cells(TCs) are vital to cancer diagnosis. Although the indexes can be estimated via the computational analysis, there is seldom a unified system that could acquire different nucleus information simultaneously. To address the issues, multi-objective learning pipeline(MOLP) is proposed to predict TPS, cell counts, nucleus contours and categories altogether from PD-L1 slides of LUSC. The main network comprises two branches, one estimating TPS via the cell analysis and the other directly regressing TPS. It minimizes the difference between these two approximated values of TPS to gain robustness. The cell-analysis branch increases confidence of the estimated TPS by nucleus segmentation, classification and counting. It also enables the system to estimate appearance parameters of TCs for LUSC diagnosis. Experiments on a large image set show that MOLP is feasible and effective. The TPS predicted by MOLP exhibits statistically significant correlation with pathologists' scores, with a mean absolute error(MAE) of 4.97(95% confidence interval(CI):-0.56-10.49) and a Pearson correlation coefficient(PCC) of 0.97(p < 0.001).

With the popularization of smart living and the rapid development of wearable terminal technology in recent years, sensor-based human activity recognition(HAR) has attracted widespread attention and has significant academic research and commercial application value. This paper focuses on enhancing the HAR model's recognition of users' daily simple activities(SAs) and complex activities(CAs), and proposes a deep learning(DL) model. Firstly, two publicly available datasets, UCI HAR and Shoaib CHA, are normalized. Then the characteristics of distinct activities are retrieved by the proposed model for HAR. Given the high association between users' activities and locations, location information is integrated into the dataset by the one-hot encoding technique to boost the model's classification performance. In addition, the proposed DL model is evaluated against eight traditional machine learning(ML) algorithms and six DL algorithms. Finally, the effect of various types of activities on the HAR model's recognition ability is studied. The experimental findings reveal that the proposed model achieves the highest classification accuracy on UCI HAR and Shoaib CHA datasets, with 96.77% and 99.13%, respectively. The classification accuracy of the HAR model is also greatly enhanced for both SAs and CAs by adding location information to the datasets.

When graphene prepared by chemical vapor deposition(CVD) method is applied to various high-performance devices, graphene's transfer and patterning process usually reduces its intrinsic properties. A new method combining wet transfer and mechanical exfoliation can be utilized to accomplish high-quality and large-scale transfer of graphene patterns. This method requires the preparation of an aurum(Au)-mesh with pores of a specific form as the peeling belt. The contact area between the Au and graphene can offer sufficient adhesion to the homogeneous monolayer graphene for mechanical exfoliation, whereas the non-contact area between the hole and graphene has a relatively weak force to release the graphene patterns on the substrate. The surface morphology and electrical characterization of graphene show that the graphene surface is clean and uniform, while graphene field-effect transistors have high carrier mobility and small Dirac voltage. The Au-mesh transfer method can effectively improve the quality of the graphene patterns as well as the efficiency of the transfer, which opens up a broader scope for the application of CVD graphene and also provides a reference for the transfer of other CVD two-dimensional(2D) materials.

Developing an efficient photocatalyst for hydrogen production is crucial for the conversion of solar energy to hydrogen. An efficient photocatalytic hydrogen evolution material was prepared by altering the graphite carbon nitride(CN) through nitrogen-doping and employing CoSe_x as a co-catalyst through a photochemical synthesis route. The optimal photocatalytic performance of the material is 285 times higher than that of CN. Photoelectrochemical experiments demonstrate that nitrogen-doping and introducing CoSe_x can effectively promote the separation of photogenerated charge carriers, thus increasing the hydrogen evolution activity of CN. This work may provide new perspectives for the preparation of novel photocatalysts by photochemical deposition methods.

A silicoboron-carbonitride (SiBNC) ceramic precursor, polyborosilazane (PBSZ), was successfully synthesized using boron trichloride (BCl₃), trichlorosilane (HSiCl₃) and hexamethyldisilazane (Me₆Si₂NH) as raw materials through the polymer-derived ceramics (PDCs) method. Fourier transform infrared (FTIR) spectroscopy, nuclear magnetic resonance (NMR), X-ray photoelectron spectroscopy (XPS), thermogravimetric analysis (TGA) and differential scanning calorimeter (DSC) were used to analyze the structure and high-temperature performance of the obtained PBSZ. Results showed that the network of silicon nitrogen boron (Si—N—B) and six-membered boron nitrogen (B—N) rings were presented in the PBSZ structure. The ceramic yield of the synthesized PBSZ at

800 ℃ in a nitrogen atmosphere was 53. 9%.

To achieve the ideal scaffolds for liver tissue regeneration, chitosan(CS) was modified with lactobionic acid(LA) or/and glycyrrhetinic acid(GA) to obtain LA-modified CS(LC), GA-modified CS(GC) and GA/LA-modified CS(GLC), and the composite nanofibrous scaffolds composed of silk fibroin(SF) and the above modified CS were fabricated by green electrospinning. Fourier transform infrared(FTIR) spectroscopy, nuclear magnetic resonance(NMR) spectroscopy and X-ray diffraction(XRD) patterns were used to characterize the chemical components and structures. The water contact angle was measured to evaluate the hydrophilicity, and thermal gravimetric analysis(TGA) was carried out to obtain the thermal properties. These scaffolds were hydrophilic, and their hydrophilicity and thermal stability decreased with the increase of the modified CS content, while their crystallinity increased. The scaffolds showed good performance in promoting the proliferation of the human hepatoma cell line(HepG2 cell) as well as their secretion of both albumin and urea. Furthermore, the scaffolds with LC had a better performance of hepatocellular compatibility than those with GC or GLC.

Textile antennas, critical electronic devices in radio frequency(RF) energy harvesting systems for wearable products, are increasingly preferable in recent years. In order to investigate the collection performance of RF energy harvesting system based on textile antennas, in this study, the textile microstrip single-element antenna and the textile microstrip array antenna were designed and prepared with polyester felts as the substrate. To build up a complete RF energy harvesting system, the combination of the signal generator, the power amplifier, the transmitting antenna and the above-designed textile microstrip antenna as the receiving antenna connected with the rectifier circuit was made. The results show that the maximum gain of the array antenna is 5.35 dB higher than that of the single-element antenna. The final output voltage shows the effectiveness of the RF energy harvesting system. As the input power increases or the receiving distance decreases, the output voltage increases. The highest output voltage through the single-element antenna is 39.2 mV, and the highest output voltage through the array antenna is 72.7 mV due to the higher gain of the array antenna. This study will bring more research ideas to this field.

The interactive textile adapts light and color to enhance the aesthetics and the performance of traditional textiles, and it can be applied in different fields such as fashion, interiors and medical devices. To create interactive textiles with illumination named “Twist”, the innovative methods of incorporating weaving techniques with electronic integration and laser engraving are explored. Weave design is employed to achieve the lightness and sheerness of the interactive textile, as well as seamless electronic integration. The interactive platform between textiles and users is created through designing and developing interactive textiles. The creative potential of interdisciplinary practice and theoretical research is also explored.

Assembly error abnormal quality testing of harmonic reducers is an important part of the pre-delivery process of manufacturers and focuses on abnormality assessment, which can reduce financial losses due to product recalls and further protect the interests of users and the reputation of manufacturers. Sound signals offer the benefit of simple and non-contact measurements for acoustic resonance testing and can facilitate pre-delivery fast factory testing of harmonic reducers. This paper presents an experimental method for sound data acquisition, feature extraction and analysis. Hammered excitation of a harmonic reducer is used to obtain acoustic datasets for both abnormal and normal harmonic reducers. Time and frequency domain features of the sound signals are extracted, and the classification algorithms of support vector machine(SVM), random forest(RF) and K-means are compared. The results show that the accuracy of SVM on the test set is 98. 0%, that of RF is 95. 0%, and that of K-means is only 53. 0%. The SVM classifier's accuracy, recall, and F1 scores are high. Based on the SVM harmonic reducer quality detection model, the national instrument(NI) data acquisition card and Labview are used to design the harmonic reducer fast detection software for the harmonic reducer pre-delivery inspection of manufacturers.

Research and development of mechatronic products generally face the problems of large workload, long cycle and high cost of on-site debugging, and the current mainstream simulation software has a single function. Thus, a set of hardware-in-the-loop simulation(HILS) system with both real time and a three-dimensional(3D) display effect is designed, and can be connected to the mechatronic software. The winding machine is taken as the simulation object, a virtual simulation platform using Unity3D is built, and the real-time interaction of data between the virtual simulation platform and the control system is realized through the circuit board. The feasibility of the process simulation of the electromechanical system in the winding machine is verified by outputting the motion timing and spatial trajectory diagrams of the key structures in the winding process. The results show that the system can replace part of the on-site debugging work and improve the efficiency of research and development.

Active control simulation software is scarce, and most control problems are either computed using MATLAB programming or conducted through experimental setups. The former often yields unreliable simulation results, while the latter requires substantial financial resources. To address this issue, this article presents a new simulation method for comparing the active sound absorption and noise reduction performance of circular and square piezoelectric ceramic plates. The simulation method involves three steps. Firstly, ANSYS dynamic analysis is used to obtain the voltage generated by two polyvinylidene fluoride(PVDF) piezoelectric sensors on the piezoelectric ceramic plates under the current load step and separate the sound pressure and particle vibration velocity. Secondly, the sound pressure and particle velocity signals obtained in the first step are imported into MATLAB to separate the incident and reflected sound waves, and the optimal control algorithm in MATLAB is used to calculate the driving voltage to be applied. Thirdly, the driving voltage calculated by MATLAB is applied to both ends of the piezoelectric ceramic plates in ANSYS, and dynamic analysis is performed again to obtain the voltage generated by two PVDF piezoelectric sensors under the next load step. By repeating the above three steps, the active sound absorption control process of the piezoelectric ceramic plates can be simulated. The simulation experiment reveals the relationship between the applied driving voltage and frequency on piezoelectric ceramic plates of different shapes, and the simulation results closely match the theoretical calculations. The simulation experiment results demonstrate that, with the same surface area, circular piezoelectric ceramic plates exhibit significantly superior sound absorption and noise reduction performance compared to square piezoelectric ceramic plates.

The intelligent operation and maintenance management of sewing equipment needs to solve the problem of information mining and language model construction of unstructured text, which is of great significance to improve the speed and accuracy of the diagnosis of equipment defects and faults, and realize the intelligent decision-making of equipment maintenance. In this paper, firstly, we propose a method based on bidirectional encoder representations from transformers-conditional random fields(BERT-CRF) to extract key entity information, such as device names and attributes. Then, through the relationship extraction model based on bidirectional gated recurrent unit-attention(BiGRU-Attention), the semantic association between entities is captured effectively to provide support for the construction of the sewing equipment knowledge graph(KG). According to the text analysis scenario of sewing equipment, the model is specially trained and optimized in the task scenarios of text entity recognition, information extraction and fault diagnosis of sewing equipment. Compared with existing deep learning algorithms, the proposed method achieves a 20% to 30% performance improvement on the validation and test sets, demonstrating significant advantages in the recall rate and the accuracy. To facilitate the mining of unstructured text information on sewing equipment, this study provides a reference for constructing a KG that integrates data on flat sewing equipment, including aspects of equipment fault operation, maintenance and flat sewing process route design.

In computer image processing, owing to the influence of lighting conditions and camera installation locations, incomplete ellipse contour extraction often occurs after the edge extraction of the image. By fitting this residual contour, the result deviates from the original elliptical shape and the fitting error is large, which affects the fitting accuracy. The degree of influence of the characteristics of the incomplete contour on the error was studied, and a numerical simulation method in MATLAB was used to perform incomplete elliptical arc segments at different positions, edge extraction on the arc segment, and ellipse fitting on the arc segment based on the least squares method. The influence of multiple factors, such as the phase angle, the arc length integrity, and the axis ratio of the ellipse on the ellipse fitting error was analyzed, which was significant in understanding the causes of error generation and improving the fitting accuracy. The curves of the fitting error with the three factors yield that all three factors have a significant effect on the fitting error. The effect of contour fitting at different phase angles varies greatly, and the greater the arc length integrity and the smaller the axis ratio of the ellipse, the smaller the fitting error.