Neuro-fuzzy based hybrid method for modeling and control of pH neutralization / Mohd Fauzi Zanil

Mohd Fauzi , Zanil (2012) Neuro-fuzzy based hybrid method for modeling and control of pH neutralization / Mohd Fauzi Zanil. Masters thesis, University of Malaya.

Preview

PDF (Thesis M.A) Download (5Mb) | Preview

Abstract

The pH neutralization is regarded as one of the fundamental parts of industrial chemical process. In electrochemical industry for example, heavy metals must be recovered (by reducing the solubility of the metals) from waste streams by controlling the pH value to prevent polluting the environment. The pH neutralization shows strong nonlinear characteristics because of feed condition. Theoretically, the nonlinear effects for this process come from negative logarithm of ionic hydrogen, where process dynamic occurs when the hydrogen ions increase or decrease during neutralization process and because of dynamic nonlinearity called the ‘‘S-shape’’ curve which consists of extreme sensitivity and insensitivity regions. This study proposes a hybrid model and a Fuzzy Logic controller for an on-line pH neutralization pilot plant. The model is used to identify the on-line pH neutralization plant’s characteristics and to improve the Fuzzy Logic controller decision output. The hybrid model is between neuro-fuzzy (ANFIS) identification technique and first principle model. The identification technique uses training dataset from experimental data to map the neutralization response curve from pH equal to 3 to 11. The first principle model is based on material balances and chemical equilibrium equation. The objective of the proposed model is to extend the robustness effect in the Fuzzy Logic controller by predicting the control action based on on-line titrations characteristics without having to re-design the model if plant undergoes different conditions. The on-line model validation and controller performance analysis for hybrid model and Fuzzy Logic controller was conducted and compared. The lowest values of RMSE (Root Mean Square of Error) and ISE (Integral Square of Error) are desired to justify the goodness of proposed model and controller respectively. In the experiment, the hybrid model (in nominal plant condition, RSME = 0.1013 and in altered plant condition, RMSE = 0.5616) gives best of fit for the on-line neutralization process. The proposed Fuzzy Logic controller with inverse hybrid model is able to handle the nonlinearity and robustness issues for the on-line pH neutralization. In set point tracking analysis, it shows best performance (ISE = 35.032) compared to normal Fuzzy Logic controller (ISE = 157.652) and PID controller (ISE =195.365). Thus, the proposed hybrid model and the proposed Fuzzy Logic controller can be used effectively in on-line/off-line studies of the dynamic behaviour of the pH neutralization pilot plant.

Actions (For repository staff only : Login required)