Application of Fuzzy-Neural Network Theory to Automatic Control of Wastewater Treatment System
| Autor: | Cheng Hung Lai, 賴建宏 |
|---|---|
| Rok vydání: | 1999 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 87 The activated processing is the most widespread sewage treatment processes. The control strategies of conventional biological wastewater treatment systems are based upon analytical or statistical or mathematical model development have been little practical contribution, due to complexities and uncertainties of the systems the fact that many environment parameters affect the degradation processes and the influent flow and organic substrate concentration (organic loading) normally vary with time. Due to the above mentioned, expert system (ES), extensively called artificial intelligence (AI) has become the new development for operation and control of wastewater treatment plant, because the rule-based systems required to understand the complex behavior of wastewater do not provide the insight mathematical equations. However the ES method is based on expert''s knowledge or operator''s related experiences making it an effective tool for overall consideration of steady state condition. The limitations of ES approach are (1) absence of systematic approach available to derive control rules. The rules occasionally require a substantial amount modification by experts and/or operators. (2) Different experts likely have divergent opinions or experience on phenomena. In recent years, fuzzy control and neural network has been successfully applied to many control yields. Because this two theories also have common characteristics, such as, approximation of nonlinear functions, the abilities of nonlinear interpolation, learning through examples, and the numeric processing. However this two methods there are some problems, we have no good ways to solve it. Such as, for fuzzy control how to select the parameters of fuzzy sets and the weight of each fuzzy rule is an ordinary problem. For neural network how to understand internal operations of neural network is difficult. Therefore, in the thesis we integrate of fuzzy control and neural networks theory with the goal of combing the human inference style and natural language use of fuzzy systems with the learning and parallel processing abilities of neural networks to architect fuzzy-neural for control effluent suspended solid of activated sludge processes automatically without expert knowledge. The results show application of fuzzy-neural network theory to automatic control of wastewater treatment system approach is very effective. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|