Two-Level Hierarchical Control for Wastewater Treatment Utilizing Neural-Network Predictors and Nonlinear optimization
| Autor: | Mahsa Sadeghassadi, David T. Westwick, Chris J. B. Macnab |
|---|---|
| Rok vydání: | 2018 |
| Předmět: |
Artificial neural network
Computer science Flow (psychology) 02 engineering and technology 010501 environmental sciences 01 natural sciences Nonlinear programming Data modeling Setpoint 020401 chemical engineering Autoregressive model Control theory Benchmark (computing) Sewage treatment 0204 chemical engineering 0105 earth and related environmental sciences |
| Zdroj: | ECC |
| DOI: | 10.23919/ecc.2018.8550493 |
| Popis: | This paper addresses the problem of identifying optimal setpoints for a biological wastewater treatment plant that can react to changing influent flow due to weather conditions. We propose and compare three methods that identify an optimal fixed setpoint for dry weather, and a moving direction for weather changes. The optimal fixed setpoint is the solution of a nonlinear optimization problem. Neural Network Autoregressive eXogenous (NNARX) models predict the correct moving direction in real time. The methods involve choosing the dissolved oxygen setpoint, or the nitrate/nitrite setpoint, or both. Simulations use the Benchmark Simulation Model Number One (BSMI) model of a wastewater treatment plant with the supplied influent data for three weather conditions. The proposed methods can improve effluent quality or energy cost or both. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|