智慧型都市排水抽水站即時操作系統之研發

Real-Time Pumping Station Operation for an Urban Drainage System Using Artificial Intelligence

黃建霖(Chien-Lin Huang) ； 徐年盛(Nien-Sheng Hsu) ； 魏志強(Chih-Chiang Wei)

《農業工程學報》 58卷3期 (2012/09) Pp. 64-79

https://doi.org/10.29974/JTAE.201209.0005

摘要

本研究目的為利用人工智慧技術研發都市排水抽水站即時操作之決策模式，以決定颱風與暴雨事件中排水閘門關閉時期抽水站抽水機組之即時操作方式。本研究研發ANFIS-HR及ANFIS-OPT兩種操作模式，ANFIS-HR操作模式為利用歷史操作紀錄結合調適性網路模糊推論系統（ANFIS）來建構模式：ANFIS-OPT操作模式則為利用禁忌、演算法所優選出歷史事件之最佳操作情況結合ANFIS來建構模式。本研究以新北市中港大排為研究區域，並以操作中前池最高水位以及最終水位與標的水位相差之絕對值來比較兩種操作模式與歷史操作之優劣，結果顯示ANFIS-OPT操作模式較ANFIS-HR操作模式及歷史操作紀錄為佳。

關鍵字

抽水站操作 ； 優選 ； 調適性網路模糊推論系統 ； 禁忌演算法

並列摘要

The purpose of this study is to apply artificial intelligence techniques to propose the operation decision models, which can decide the real-time operational way of pumping machines in flood events during drainage gate closed period of a pumping station for an urban drainage system. This study develops two operation models, namely ANFIS-HR and ANFIS-OPT. ANFIS-HR is constructed by ANFIS (Adaptive Network-based Fuzzy Inference System) with historical operation records. ANFIS-OPT are constructed by ANFIS with the operation circumstances which are optimized by tabu search in historical events. This study uses Chung-Kong-Da-Pai basin as study area. The comparison variables which are simulated from real case by two operation models are maximum water level (WL) and absolute difference between final WL and target WL of pumping front-pool. Results show that ANFIS-OPT are better than ANFIS-HR and historical operation records.

並列關鍵字

Pumping station operation ； Optimization ； Adaptive Network-based Fuzzy Inference System ； Tabu search