| 摘要: | 本研究以實廠A2O系統做為研究對象，其是由厭氧/缺氧/好氧/MBR所串連而成的污水生物處理系統，可有效去除水中碳氮磷物質。由於廢水處理系統處理對象為都市生活污水，故進流水質與水量會隨著國人用水方式、用途及習慣等因素變化，使進流水質、水量及系統內微生物反應情況皆會隨著時間而變化，故本研究為整合具有系統分析方法的規則庫與具有學習能力的系統動力模式配合類神經網路的模式庫，以發展一套污水處理自動監測與控制系統來因應動態進流條件，並配合自動監測設備提高A2O系統操作自動化與最佳化的可行性，接著利用系統分析方法找出各種控制參數隨時間的變化規律來建立規則庫，並以系統動力模式建立污水處理自動監測與控制系統的模式庫，期望未來能達到人工智慧自動監測與控制污水處理系統的願景。本研究第一年為A2O自動監測與控制雛型系統的發展，首先分析系統進流水質資訊，並以系統分析方法建立規則庫，接著根據規則庫透過模式庫預測系統狀態以判定其結果對系統可能造成的影響，並針對不符合預期之處進行檢討與改善，並視情況改良規則庫與調整系統操作條件，再整理系統處理過程中可用的資訊以建立資料庫，而本研究第二年為A2O自動監測與控制系統的發展與驗證，透過整合規則庫、模式庫與資料庫成一個整體的A2O自動監測與控制系統，最後藉由實廠水質項目檢驗資訊與A2O自動監測與控制系統判定結果相互比較並進行實廠驗證與修正，以使實廠污水處理自動監測與控制系統更貼近於真實動態系統。 ;In this study, the A2O system of actual wastewater plant was selected as the research object. It is a wastewater biological treatment system connected in series by anaerobic / anoxic / aerobic / MBR, which can effectively remove carbon, nitrogen and phosphorus in the wastewater. Because the wastewater treatment system is targeted at urban domestic wastewater, the inflow quality and quantity of water will change with variation in household water use, usage and habit, etc., so that influent water quality, water quantity and microbial in the system will change over time. In order to develop a set of sewage treatment automatic monitoring and control system to respond to dynamic inflow conditions, this study is integrated the rule-base with systematic method of analysis, the system dynamics model with learning ability, and the neural network., and combined with automatic monitoring equipment to improve the feasibility of A2O system operation automation and optimization. The first phase of the study is the development of a prototype A2O automatic monitoring and control system. First, we analyze the system influent water quality information, establish the rule-base by using systematic analysis method, and then use the model-base and the rule-base to predict the state of the system to determine influent water impact on the system. When we review and improve the results that do not meet expectations, we improve rule-base and adjust system operating conditions as appropriate. Finally, we organize the information available in the system processing to establish a database. And the second phase of this study is the development and verification of the A2O automatic monitoring and control system. Through integrating the rule base, the model-base and the database into an overall A2O automatic monitoring and control system. Finally, through the sewage water quality inspection information of the actual wastewater plant and automatic monitoring and control system to determine the results compared with each other to verify and correct the automatic monitoring and control system, making the actual wastewater plant sewage treatment automatic monitoring and control system closer to the real dynamic system. We are expected to achieve the foresight of automatic monitoring and control of sewage treatment system by artificial intelligence in the future. |
