논문실적
| DATA MINING FOR ENERGY MANAGEMENT IN A WATER DISTRIBUTION NETWORK |
|---|
|
학술지명 MCCSIS 2019
저자 홍성택,이호현,신강욱,김대욱
발표일 2019-06-18
|
|
In the water treatment process, it is important to predict the flow used by citizens and control valves for using large reservoir tanks maximally. In this study, flow predictions are proposed by applying several non-linear algorithms with recursive form to minimize additional learning. To control the valves properly, the target levels of each tank are determined based on flow prediction and genetic algorithms. A simple polynomial learning algorithm is also proposed to control the valves less frequently than a conventional feedback controller until the target tank levels can be reached, which also help not to break down the valves. The proposed method to maximize the usage of reservoir tanks is expected to reduce the electric fee and improve water quality. |
