Fluorescent aerosol leakage quantification for protective clothing with an entropy-based image processor for industrial and medical workers

Correct procedures for donning and removing protective clothing are critical for preventing transmission of hazardous particulates in industrial and medical settings. The purpose of this study is to examine the levels of airborne particulate leakage of personal protective clothing after a series of exercises performed by volunteers who simulated industrial and medical workers. The fluorescent aerosols were employed as airborne particulates in a controlled chamber with ultraviolet (UV) light-detectable stickers. After an exposure-and-leakage test, the protective clothing was removed and photographed with UV-scanning to evaluate areas where fluorescent aerosols had adhered to the body through the clothing. An image processor installed with an entropy-based algorithm was developed to segment the fluorescent area and calculate its relative leakage ratio (L_r) in real time. This study addresses an optimal technique to confirm the safety of protective clothing removal and decontamination policy formulated for aerosol-transmitted situations.     

Interfacial reactions of Sn–58Bi and Sn–0.7Cu lead-free solders with Alloy 42 substrate

Eutectic Sn–58 wt%Bi (SB) and Sn–0.7 wt%Cu (SC) lead-free solders reacting with Alloy 42 substrates were investigated at 240, 270, 300, and 330 °C for various reaction times. The FeSn₂ phase was the only phase formed at the solder/Alloy 42 interface for all couples. However, this FeSn₂ phase had two different microstructures. The intermetallic compound (IMC) morphology near the solder/Alloy 42 interface was dense with a tiny microstructure after reflowing at 240 °C. Massive spalling FeSn₂ phases with large and platy grain were observed in the solder matrix. When the reflow temperature was increased the IMC morphology assumed bulk or cylindrical shape with no spalling IMCs observed at the solder/Alloy 42 interface. The IMC thickness increased with increasing reaction temperature and time. Both SB/Alloy 42 and SC/Alloy 42 couples presented a diffusion controlled growth mechanism in. The activation energies for SB/Alloy 42 and SC/Alloy 42 couples were 86.2 and 103.4 kJ/mol, respectively.     

Seismic Assessment of Bridge Diagnostic in Taiwan Using the Evolutionary Support Vector Machine Inference Model (ESIM)

Because earthquakes cause bridge damage resulting in incidents such as traffic gridlock and casualties, it is necessary to assess possible bridge damage under different seismic intensity levels in order to reduce the incidence of disasters. However, because there are many bridges in Taiwan, the time and budget will be restricted to conduct traditional structural analysis (preliminary assessment, detailed analysis) of each bridge to obtain its yield acceleration (Ay) and collapse acceleration (Ac). Hence, this study developed an inference model by integrating two AI techniques: support vector machines (SVM) and fast messy genetic algorithms (fmGA). The study applied historical cases to infer Ay and Ac values by the mapping relation between preliminary assessment factors (input) of historical cases and the detailed assessment of Ay and Ac values (output). According to the above inference model to predict Ay and Ac values, the probability of possible bridge damage by earthquakes can be predicted as a suggestion for bridge management personnel in a short period of time. This study adopted 121 RC bridges in Taiwan and selected 109 bridges for training cases and 12 bridges for testing cases to calculate the root mean square error (RMSE). The results indicate that the RMSE of training is 0.087 and of testing is 0.0869.     

Portfolio management system in equity market neutral using reinforcement learning

Portfolio management involves position sizing and resource allocation. Traditional and generic portfolio strategies require forecasting of future stock prices as model inputs, which is not a trivial task since those values are difficult to obtain in the real-world applications. To overcome the above limitations and provide a better solution for portfolio management, we developed a Portfolio Management System (PMS) using reinforcement learning with two neural networks (CNN and RNN). A novel reward function involving Sharpe ratios is also proposed to evaluate the performance of the developed systems. Experimental results indicate that the PMS with the Sharpe ratio reward function exhibits outstanding performance, increasing return by 39.0% and decreasing drawdown by 13.7% on average compared to the reward function of trading return. In addition, the proposed PMS_CNN model is more suitable for the construction of a reinforcement learning portfolio, but has 1.98 times more drawdown risk than the PMS_RNN. Among the conducted datasets, the PMS outperforms the benchmark strategies in TW50 and traditional stocks, but is inferior to a benchmark strategy in the financial dataset. The PMS is profitable, effective, and offers lower investment risk among almost all datasets. The novel reward function involving the Sharpe ratio enhances performance, and well supports resource-allocation for empirical stock trading.

     

10 GHz dual-conversion low-IF downconverter with microwave and analogue quadrature generators

A 10 GHz dual-conversion low-IF downconverter using 0.18-mum CMOS technology is demonstrated. The high-frequency quadrature RF and LO₁ signals are generated by broadside-coupled quadrature couplers while a two-section polyphase filter is utilised for the low-frequency LO₂ quadrature signal generation. As a result, the demonstrated downconverter achieves a conversion gain of 7 dB, IP_{1 dB} of -16 dBm, IIP₃ of -5 dBm and noise figure of 26 dB at a 1.8 V supply. The image-rejection ratio of the first/second image signal is 33/42 dB for IF frequency ranging from 10 to 60 MHz, respectively.     

Fast segmentation of porcelain images based on texture features

This study aims to segment objects within images of porcelain artifacts to help users retrieve the images in an efficient and convenient manner. Through digital archiving, a tremendous number of porcelain images have been created. To avoid interference due to the image’s background during the retrieval process, it is necessary to segment objects in advance to accommodate high-precision image retrieval. In the proposed segmentation process, four texture features, including coarseness, contrast, directionality, and gradient, are first obtained. The morphological processing, which involves PCA (principal component analysis), Otsu’s method, and object filter for opening and closing operation, is applied. Finally, regarding the objects selected by object filter, boundary extraction and watershed segmentation are performed to segment the porcelain objects from the background. In our image segmentation experiment using images of Chinese porcelain from various dynasties, featuring various shapes and colors, complete and accurate segmentation results are produced. The results can be used as a reference for future identification of the era to which the artifacts belong, and also to lay a foundation for future development of porcelain image retrieval techniques as a benefit to academic research.     

Preparation of transparent nanocomposites from UV‐ and thermo‐curable epoxyacrylate and graphene oxide for the application of corrosion protection coatings

In this study, we report UV‐ and thermo‐curable epoxyacrylate/graphene oxide (EA/GO) nanocomposites that present good transparency, excellent pencil hardness and promising improvement in corrosion protection. A dual‐curable EA oligomer with one terminal epoxide and one double bond at the other end was synthesized by reaction of diglycidyl ether of bisphenol‐A and acrylic acid. After mixing EA and GO with the curing agents and reactive diluent followed by UV cure and thermo‐cure, the resulting EA/GO films on a glass slide with GO loading up to 3 phr exhibited over 86% light transmittance. Furthermore, the pencil hardness was enhanced from 3H for EA to 6H for the EA/GO composite at 2 phr GO loading. The corrosion protection of the EA/GO coatings was evaluated by a potentiodynamic polarization technique and electrochemical impedance spectra. The corrosion potential (E_corr) of the EA/GO‐coated steel increased with increasing GO loading. Meanwhile, Nyquist and Bode plots indicated that the higher the GO content in the EA/GO coating was, the higher was the coating resistance and also the charge transfer resistance after immersion in salt solution. All these results proved that the GO had positive effects on enhancement of the corrosion resistance. The improved corrosion protection by the EA/GO coatings was mainly due to the enhanced hydrophobicity, the deviation of electron transfer and the increased tortuosity of the diffusion path. The improved corrosion protection and hardness together with the useful dual‐curability make the EA/GO nanocomposite a competitive candidate for corrosion protection coatings. © 2019 Society of Chemical Industry