Search

From traditionally handmade items to the ability of people to use machines to process and even to human-robot collaboration, there are many risks. Traditional manual lathes and milling machines, sophisticated robotic arms, and computer numerical control (CNC) operations are quite dangerous. To ensure the safety of workers in automated factories, a novel and efficient warning-range algorithm is proposed to determine whether a person is in the warning range, introducing YOLOv4 tiny-object detection algorithms to improve the accuracy of determining objects. The results are displayed on a stack light and sent through an M-JPEG streaming server so that the detected image can be displayed through the browser. According to the experimental results of this system installed on a robotic arm workstation, it is proved that it can ensure recognition reaches 97%. When a person enters the dangerous range of the working robotic arm, the arm can be stopped within about 50 ms, which will effectively improve the safety of its use.

Anomalies are a set of samples that do not follow the normal behavior of the majority of data. In an industrial dataset, anomalies appear in a very small number of samples. Currently, deep learning-based models have achieved important advances in image anomaly detection. However, with general models, real-world application data consisting of non-ideal images, also known as poison images, become a challenge. When the work environment is not conducive to consistently acquiring a good or ideal sample, an additional adaptive learning model is needed. In this work, we design a potential methodology to tackle poison or non-ideal images that commonly appear in industrial production lines by enhancing the existing training data. We propose Hierarchical Image Transformation and Multi-level Features (HIT-MiLF) modules for an anomaly detection network to adapt to perturbances from novelties in testing images. This approach provides a hierarchical process for image transformation during pre-processing and explores the most efficient layer of extracted features from a CNN backbone. The model generates new transformations of training samples that simulate the non-ideal condition and learn the normality in high-dimensional features before applying a Gaussian mixture model to detect the anomalies from new data that it has never seen before. Our experimental results show that hierarchical transformation and multi-level feature exploration improve the baseline performance on industrial metal datasets.

In the global wave of automation, logistics and manufacturing are indispensable and important industries. Among them, the related automatic warehousing system is even more urgently needed. There are quite a few cases of using robotic arms in the current industry cargo stacking operations. Traditional operations require engineers to plan the stacking path for the robotic arm. If the size of the object changes, it will take extra time to re-plan the work path. Therefore, in recent years, quite a lot of automatic palletizing software has been developed; however, none of it has a detection mechanism for stacking correctness and personnel safety. As a result, in this research, an intelligent robotic palletizer system is developed based on a self-developed symmetrical algorithm to stack the goods in a staggered arrangement to ensure the overall structure. Innovatively, it is proposed to check the arrangement status and warnings during the visual stack inspection to ensure the correctness of the stacking process. Besides, an AI algorithm is imported to ensure that personnel cannot enter the set dangerous area during the work of the robotic arm to improve safety during stacking. In addition to uploading the relevant data to the cloud database in real time, the stacking process combined database and vision system also provide users with real-time monitoring of system information.

The pressure swing adsorption (PSA) process was used to capture carbon dioxide (CO2) from the flue gas of a coal-fired power plant to reduce CO2 emissions. Herein, CO2 was captured from flue gas using the PSA process for at least 85 vol% CO2 purity and with the other exit stream from the process of more than 90 vol% N2 purity. The extended Langmuir–Freundlich isotherm was used for calculating the equilibrium adsorption capacity, and the linear driving force model was used to describe the gas adsorption kinetics. We compared the results of breakthrough curves obtained through experiments and simulations to verify the accuracy of the mass transfer coefficient. The flue gas obtained after desulphurization and water removal (13.5 vol% CO2 and 86.5 vol% N2) from a subcritical 1-kW coal-fired power plant served as the feed for the designed three-bed, nine-step PSA process. To determine optimal operating conditions for the process, the central composite design (CCD) was used. After CCD analysis, optimal operating conditions with a feed pressure of 3.66 atm and a vacuum pressure of 0.05 atm were obtained to produce a bottom product with a CO2 purity of 89.20 vol% and a recovery of 88.20%, and a top product with a N2 purity of 98.49 vol% and a recovery of 93.56%. The mechanical energy consumption was estimated to be 1.17 GJ/t-CO2.

The nature and characteristics of free-burning wildland fires have significant economic, safety, and environmental impacts. Additionally, the increase in global warming has led to an increase in the number and severity of wildfires. Hence, there is an increasing need for accurately calculating the probability of wildfire propagation in certain areas. In this study, we firstly demonstrate that the landscapes of wildfire propagation can be represented as a scale-free network, where the wildfire is modeled as the scale-free network whose degree follows the power law. By establishing the state-related concepts and modifying the Binary-Addition-Tree (BAT) together with the PageRank, we propose a new methodology to serve as a reliable tool in predicting the probability of wildfire propagation in certain areas. Furthermore, we demonstrate that the proposed maximum-state PageRank used in the methodology can be implemented separately as a fast, simple, and effective tool in identifying the areas that require immediate protection. The proposed methodology and maximum-state PageRank are validated in the example generated from the Barabási-Albert model in the study.