The bolt joints on steel structures are exposed to the possibility of damage, and thus, require intensive care. Usually, periodic inspections are conducted at the cost of time and money. However, it is very difficult to check so many bolts carefully. The purpose of this study is to propose a system that can more efficiently monitor the tightness/looseness of these bolts. The proposed bolt fastening monitoring system is comprised of sensors that are attached to nuts and a data receiving terminal, which gathers information. The reed switch consists of two thin, metallic contacts enveloped in a glass tube and is an electrical switching sensor that is triggered ON or OFF by changes in the surrounding magnetic field. The verification tests showed that bolt loosening can be effectively detected, proving the applicability of this system to the maintenance of the bolt joints of steel structures. The newly developed sensor system is expected to solve conventional sensor problems by enabling measurement of structural members which was not previously possible, thus providing a basis for a new technology in the construction industry by applying IT to construction technology.

No abstract

The success of multi-sensor data fusions in deep learning appears to be attributed to the use of complementary information among multiple sensor datasets. Compared to their predictive performance, relatively less attention has been devoted to the adversarial robustness of multi-sensor data fusion models. To achieve adversarial robust multi-sensor data fusion networks, we propose here a novel robust training scheme called Multi-Sensor Cumulative Learning (MSCL). The motivation behind the MSCL method is based on the way human beings learn new skills. The MSCL allows the multi-sensor fusion network to learn robust features from individual sensors, and then learn complex joint features from multiple sensors just as people learn to walk before they run. The step wise framework of MSCL enables the network to incorporate pre-trained knowledge of robustness with new joint information from multiple sensors. Extensive experimental evidence validated that the MSCL outperforms other multi-sensor fusion training in defending against adversarial examples.

No abstract

In a launch environment, all satellites are subjected to severe random vibration and acoustic loads owing to rocket separation, airflow, and injection/combustion of the fuel. Structural vibrations induced by mechanical loads cause the malfunction of vibration-sensitive components in a satellite, leading to failures during the launch process or an on-orbit mission. Therefore, in this study, a shape memory alloy-based vibration isolator was used on the connection between the launch vehicle and satellite to reduce the vibration transmission to a satellite. The vibration isolator exhibited a high performance in the vibration isolation, owing to the dynamic properties of super-elasticity and high damping. The vibration-reduction performance of the vibration isolator was experimentally verified using random vibration and acoustic tests in a structural thermal model of the satellite developed in the synthetic aperture radar technology experimental project. Owing to the super-elasticity and high attenuation characteristics of the vibration isolator, it was possible to significantly reduce the random vibration of the satellite in the launch environment. Although the mechanical load of the acoustic test mainly excited the antenna on the upper side of the satellite rather than the bottom side, the results of the acoustic test showed the same trend as the random vibration test. From this perspective, the vibration isolator can contribute to saving the costs required for satellite development. These advantages have made it possible to develop satellites according to the new space paradigm, which is a trend in the space industry worldwide.