This paper describes recent research and development on disaster prevention technology at RTRI. We developed some prevention techniques for extreme weather events such as tornados and localized heavy rainfall as part of RTRI's long-term research plan "RESEARCH2020". As part of this plan, a countermeasure against unusual or extreme weather events in the form of a real time hazard map system applying predictive meteorological information was developed. RTRI also built a method to evaluate resilience and a system for generating restoration strategies. A new long-term plan, called "RESEARCH2025" has been compiled, and RTRI is now working on developing two areas of research: a high precision operation control system applying observation data, and a method to evaluate slope stability after heavy rain.

The railway sector is facing major changes due to declining birthrates, an aging society, the intensification of meteorological disasters, and the need to further reduce CO₂ emissions. The spread of COVID-19 since 2020, has had a major impact on the railway business, leading to an urgent need to respond to the drastic changes in demand for mobility in society. Until now, the Signal and Transport Information Technology Division has focused on research and development to further improve the safety and convenience of train operations and reduce the cost and burden of operations related to train operation management. Now, it is necessary to extend research and development to take into account the consequences of COVID-19 by, for example, speeding up our R&D. This paper introduces the purpose and goals of R & D undertaken by the Transport Operation Systems Lab and the Transport Planning and Marketing Lab in FY2020. This paper also describes the R&D concepts and policies for responding to a post-COVID-19 environment.

The Human Science Division of the Railway Technical Research Institute in Japan, has been conducting research on prevention of human error and occupational accidents as well as into countermeasures for transport disruptions and ways to improve user environments in order to improve the safety, convenience and comfort of passengers using the railways. This paper outlines the recent accomplishments of the representative research on prevention of human error accidents and improvement of user environments.

To investigate deterioration rate of rocks, the authors conducted measurements of the physical properties of naturally weathered Kimachi sandstone and forcedly deteriorated Kimachi sandstone. The naturally weathered Kimachi sandstone was obtained by boring from a surface after excavation which had not been touched for approximately 100 years. From the Measurement results of the physical properties of the naturally weathered Kimachi sandstone, the relationship between these properties, depth and time was obtained. The forced deterioration experiments were carried out by repeated drying-wetting and drying-wetting-freezing-thawing cycles. The experiment results revealed that the decrease in ultrasonic wave speed with each cycle approximated well with the logarithmic function.

Shinkansen trains operating in snowy areas accumulate snow on their bogies, which can sometimes lead to damage to ground facilities. To prevent this type of damage, snow is removed from trains in stations. We developed a method to estimate the amount of snow accretion on bogies: first, snow density on the railway track is estimated using weather data, then, flying up snow flux is estimated. This is then used to predict the accumulated snow amount under the bogies. Our research confirmed that snow accretion under a bogie upon arrival at a station can be estimated to within 3 cm.

To improve railway resilience against earthquakes, we verify and apply a developed support system to evaluate the effects of seismic countermeasure, focusing on the lost transportation volume of the railway networks. The system calculates the recovery process of the transportation volume which decreases after an earthquake, using optimization calculation by which the lost transportation volume is to be minimized. This makes us evaluate quantitatively the effect of the countermeasures. In addition, a different recovery process can be relatively evaluated. In this study, first we evaluate the performance of the system based on past earthquakes. Then comprehensively compare the effects of several seismic countermeasures such as structural and operational countermeasures. This paper describes that the evaluated recovery process of the lost transportation volume with the developed system is useful to implement strategic seismic countermeasures.

Local and intensive short bursts of heavy rain which may cause sudden hazards such as unexpected flooding are becoming more frequent in Japan. Moreover, these rains may be causing heavy rain to fall only over a limited area which may be narrower than the distance between rain gauges along railway lines. Therefore, current safety measures of train operation based on observation data measured by these gauges may not work well for these hazards. To address this problem, we developed a system that assists train stop and assists passenger evacuation to avoid those hazards. In this paper, first, we mainly describe the calculation algorithm and functions of the system. Next, we carry out simulations to confirm the validity and accuracy of calculations made by the system.

Commuter lines in metropolitan areas in Japan suffer frequent short train delays. Train dispatchers reschedule operations according to these delays and how they evolve. However, because of the complex way in which they evolve, it is difficult to predict delays of up to a few tens of minutes. To build an accurate prediction method, we developed prediction method using a deep learning model called Long Short Term Memory. This paper reports on the prediction performance of this method compared with the conventional method using neural networks.

When train schedules are disrupted, rescheduling is carried out to recover delays. Although we have developed computational methods for generating train rescheduling plans, incorporating the knowledge of experienced dispatchers into the method remains an unresolved issue. In our previous research, we proposed a method for extracting knowhow as a set of association rules from past dispatcher decision data and to reflect it in train rescheduling algorithms using mathematical optimization. This paper presents a method and verification results, where rules are routinely updated and rule extraction efficiency is improved.

This study describes the development of two training tasks to improve hazard perception skills: a "Scenario Drawing Task" and an "Oversight Experience Task". The "Scenario Drawing Task" helps trainees acquire knowledge of hazards through drawing a picture of their worksite and considering hazards from a human factor perspective. The "Oversight Experience Task" improves trainee's attitude to search for hazards by experiencing an oversight. The results of experiments confirm that more hazards were detected from more perspectives in the "Scenario Drawing Task" than in a hazard prediction task without drawing pictures. Results also confirmed that the "Oversight Experience Task" improves attitude to search for hazards.

There are two types of method for cleaning station restrooms. One method uses water for cleaning floors, whereas the second does not. This study describes quantitative and qualitative research into the surface bacteria found on station restroom floors where these two cleaning methods were used. Samples for analysis were collected from five positions on the floor of each restroom in each railway station investigated. Samples were treated using a conventional culture method to measure the concentrations of surface bacteria on each restroom floor. Samples were then analyzed with bacterial 16S rRNA genes to analysis the microbiomes on these restroom floors. Results showed that the restrooms cleaned without water had lower concentrations of bacteria, than the microbiomes from restroom floors cleaned with water.

Deer-train collisions have become a serious problem in Japan. To mitigate these collisions, we have composed a deterrent sound which consists of a deer alarm call and fierce barking of a dog. It has been observed that broadcasting the deterrent sound from a train to ward off deer resulted in a 30% increase in the frequency of deer running away from the tracks. We also developed equipment to enable automatic broadcasting of the deterrent sound from trains when they are in designated sections where collisions happen frequently.