Controle baseado em redes neurais artificiais, aplicado a um sistema híbrido de tratamento para remoção do corante reativo azul 5g de solução sintética
The textile complex covers one of the most traditional industries of the world economy and of great importance in the lives of people, processes mainly in the steps of dyeing and finishing, they require a large volume of water. Colorants are widely used and effluents become highly contaminants recep...
| Autor principal: | Pinto, Andre Hoffmann |
|---|---|
| Formato: | Dissertação |
| Idioma: | Português |
| Publicado em: |
Universidade Tecnológica Federal do Paraná
2018
|
| Assuntos: | |
| Acesso em linha: |
http://repositorio.utfpr.edu.br/jspui/handle/1/2924 |
| Tags: |
Adicionar Tag
Sem tags, seja o primeiro a adicionar uma tag!
|
| Resumo: |
The textile complex covers one of the most traditional industries of the world economy and of great importance in the lives of people, processes mainly in the steps of dyeing and finishing, they require a large volume of water. Colorants are widely used and effluents become highly contaminants receptor bodies. Several treatment options are being studied. The electrolytic process known as electroflocculation has been seen as a promising method. Another very promising method of treatment is the organic coagulation which has advantages over chemical coagulation such as low toxicity and biodegradability. The electro flocculation combined with the organic coagulating becomes an advantageous hybrid treatment since controlled. In this context, the objective was to implement a control based on Artificial Neural Networks (ANN) in a hybrid treatment system (electro flocculation and natural coagulation) to remove the dye Reactive Blue 5G of a synthetic solution. The choice of Central Composite Rotational Design (CCRD) was given to cover the entire experimental space using a smaller number of trials. From this design were performed 17 tests for statistical analysis which validated the mathematical model generated. Based on this model were generated banks of training and validation of ANN data that was implemented in Matlab software. Empirical tests defined the ANN architecture based on training performance with setting 3 layers and 2 hidden with 9 neurons in the first, 12 in the second and two in the output layer. The control was the feedforward type, but due to residual iron resulting from treatment requires 24 hours of decanting for later reading the feedback action was provided by the predictive mathematical model. Control assays confirmed the efficiency of the controller to tests with negative perturbations in dye concentration at the entrance of the treatment concentration of ensuring output value always below the set setpoint value. For positive disturbances control was not significant which can be attributed to the mathematical model included error may also be related to the variability of the quality of the seeds of Moringa oleifera Lam. |
|---|