The rarity of equipment failures results in a high level of imbalance between failure data and normal operation data, which makes the effective classification and prediction of such data difficult. Furthermore, many failure data are dominated by mixed data, which makes the model unable to adapt to this type of failure problem. Second, the replacement cycle of production equipment increases the difficulty of collecting failure data. In this paper, an equipment failure diagnosis method is proposed to solve the problem of poor prediction accuracy due to limited data. In this method, the synthetic minority oversampling technique is combined with a conditional tabular generative adversarial network. The proposed method can be used to predict limited data with a mixture of numerical and categorical data. Experimental results indicate that the proposed method can improve 6.45% compared to other similar methods when equipment failure data account for less than 1% of the total data.

Traffic conditions are among the issues most concerned with the general public, and the freeway is a large-scale Internet of Things application. In addition to obtaining real-time road usage information, analysis of local road usage habits is crucial in evaluations of government policy implementation. Using road usage data provided by the Electronic Toll Collection (ETC) system, we investigated the data on road usage history on the freeways. The ELK Stack was employed to construct a platform for visualizing real-time road usage information and history in this paper; the platform is named the Local Transportation Knowledge (LTK) platform. By analyzing more than 500 million pieces of data, the LTK platform proposed in this study efficiently visualized road usage data and facilitated acquirement of local road usage knowledge. We verified that residents of other counties and cities commuted to Taichung each day. We also discovered that a considerable number of Taichung City residents were employed in Hsinchu Science Park and commuted between the two places. The LTK platform can present real-time freeway traffic conditions, facilitate in-depth analysis of local road usage data, and provide data to verify the relevant information.

Unlike graded data of common semiconductor test results storing in relational databases, log data in the standard test data format (STDF) contain millions of test data entries. In a semiconductor packaging and testing factory, a semiconductor wafer or integrated circuit tests generate thousands of STDF files each day; therefore, how to store these massive databases is a crucial topic. Different products correspond to different test items and STDF content; if a relational database is used to store all forms of data, the practical operation becomes challenging. This paper used a NoSQL document-oriented database collocated with a Docker container to build a system, named the scalable STDF data (SSD) framework, for storing semiconductor test data. According to semiconductor test operations, the SSD framework first converts STDF files into an open standard format for data transmission and subsequently transfers them to the database. The use of NoSQL databases allows for flexibility of specifications of STDF content, and a Docker container exhibits features such as rapid deployment and high scalability. The SSD framework meets the requirements of semiconductor testing for throughput, latency, and parallel experimental projects; possesses excellent execution efficiency; and provides flexible data storage services in a semiconductor testing environment where processing a large quantity of data is required. From our simulation results, the major performance of the proposed system depends on the hardware properties. The higher hardware distribution degree provides better performance. Docker container provides more connections and the scalability of storage, but higher software distribution contributes limited performance enhancement.

Modularity is widely-used objective function to detect communities and there are lots of algorithms based on modularity maximization. The leading eigenvector method is one of them where modularity is maximized by choosing the first eigenvector as partition result. To analyze in depth the information provided by other eigenvectors, modularity maximization could be transformed to vector partitioning problem. This paper proposes a method to find non-overlapping vertex vector sets so as to maximize the quadratic sum of norms of community vectors. We observe spatial distribution of the vertex vectors of networks and then discover two phenomenons. First, the vertex vectors belong to different communities are separated by an angle. Second, the node with a larger degree would correspond to a vertex vector with a larger norm. Based on these two phenomena, we design a heuristic community detection algorithm. When a network with weaker community structure, the over-partition problem is considered. The experiment results show that the proposed solution provides higher accuracy than other solutions.

Identifying pavement damage is a crucial component in road maintenance and infrastructure management. Prompt detection and corrective action of pavement defects can prevent severe deterioration, maintain safety, and prolong the useful life of road infrastructure. Computer vision is widely applied to vehicle applications, such as driver assistance and self-driving, and that is a good platform to equip the pavement damage recognition service. Therefore, this work applies computer vision to develop the pavement damage detection service named high-performance pavement health recognition (HPPH) to detect common pavement damages, including longitudinal cracks, transverse cracks, alligator cracks, and potholes. The proposed HPPH can be deployed on a vehicle for regular road detection. To fit the requirements of vehicle applications, HPPH considers state-of-the-art techniques to optimize the performance of the recognition service on vehicles, e.g., DeepStream is applied to increase the inference performance. At the same time, YOLOv8s (You Only Look Once, YOLO) provides real-time inference. The experimental results reveal that HPPH provides an accuracy of 90.3% and an average precision 0.5 of 74.3%. Moreover, the optimized HPPH provides three times better than the pure YOLOv8s in terms of the inference speed. In summary, the proposed HPPH provides high recognition accuracy with high efficiency and is feasible to be applied to vehicle applications to realize distributed road maintenance.