We have developed an artificial intelligence (AI) model and a hydraulic model that can predict the water levels of drainage pump stations and drainage channels in low-lying areas. Based on the latest observations and weather forecast data, the water level can be predicted several hours ahead in real time. Based on this forecast result, the operation management of irrigation facilities can be supported, and it helps to reduce the labor required for flood damage and drainage management.
Overview
In low-lying drainage pump stations, the number of skilled managers involved in the management and operation of irrigation facilities are decreasing and there is an increasing trend in the flood damage countermeasures in recent years. Hence, there is a need for a system that supports sustainable and effective operation management. Hence, we have developed a highly accurate and high-speed prediction model that can visualize the water level of the drainage pump station and the water level of the drainage channel in the controlled area. The former is a type of AI and a Long Short-Term Memory (LSTM) model that can provide water level information at drainage pump stations at high speed by learning observation data. The latter is a hydraulic model that can provide the surface water level information of the drainage channel network based on the basic formula of the physical phenomenon. By driving these models on the support system, it is possible to predict the water level during floods in the basin in real time. Based on this visualized information, effective operation management of the drainage pump station is possible. For example, the facility manager checks the water level forecast displayed on the management monitor and can operate the appropriate drainage pumps, predict the location where there is a risk of flooding, etc., and can perform appropriate management operations of irrigation facilities such as floodgates. This can be expected to reduce flood damage and labor required for drainage management. Currently, a support system that implements the model is being tested on a low-lying farmland where drainage management is carried out.
For Inquiries
Contact: http://www.naro.go.jp/english/inquiry/index.html
