Please use this identifier to cite or link to this item:
|Title:||A REGION-SENSITIVE FUZZY LOGIC CONTROLLER FOR pH||Authors:||VENKATARAMAN RAJAGOPALAN||Issue Date:||1999||Citation:||VENKATARAMAN RAJAGOPALAN (1999). A REGION-SENSITIVE FUZZY LOGIC CONTROLLER FOR pH. ScholarBank@NUS Repository.||Abstract:||One of the key process parameters that require constant monitoring and tight control is the pH of a given medium. pH processes, in general, are characterized by severe nonlinearity and extreme sensitivity as well as time varying gain. This has given rise to a host of control strategies. In recent years fuzzy logic has been employed successfully to pH control (e.g., Galluzzo et al., 1991, Parekh et al., 1994 and Tan et al., 1996). The advantages of a controller based on fuzzy logic are that it is nonlinear and does not require an analytical model of the process. While fuzzy controller generally outperforms the conventional controller for the highly nonlinear pH process, its performance is not equally good over the entire operating range when severe variations in the process gain occur. This work hence proposes a Region-Sensitive Fuzzy Logic Control (RSFLC) strategy that incorporates the severe process gain variations into the controller design so that control performance can be improved over a wide operating range. The present work is based on a CSTR in which an influent acid and a base stream get mixed to produce an effluent. The controlled variable chosen was pH in the CSTR and the manipulated variable was flow rate of the base stream. The perfectly mixed model of the CSTR was modified to account for the non-idealities and was validated by open-loop tests. The first step in designing the proposed RSFLC consisted of dividing the process into different gain regions. This was done based on gain of the static titration curves of three common acid-base systems, viz., strong acid-strong base, weak acid- strong base and a mixture of strong and weak acid-strong base. Four regions of high, low and transition gains were identified and PI-like fuzzy logic controllers (FLCs) of varying sensitivities were designed for each of these regions. The outputs of these FLCs were then combined according to the operational need at any given instant to produce an overall RSFLC output. A method of incorporating process information in the design of RSFLC was also developed to enable the controller to be extended to other acid-base systems with different titration curves. The proposed controller was first evaluated for its efficacy by simulation studies. Disturbances of different magnitudes in acid flow, concentration and composition were considered at different set points. The performance of RSFLC was compared to that of an ordinary FLC (called SFLC) in terms of under/overshoot, settling time and ISE. RSFLC was found to outperform SFLC under all the operating conditions studied with the performance becoming progressively better as the set points was shifted from pH 7 to 5 and to 11 to cover the various gain regions within the operating range. In addition, RSFLC was also compared with a model-based controller called the Adaptive Internal Model controller (AIMC) proposed by Lakshmi Narayanan, et. al. (1997). The results obtained, again confirm the superior performance of RSFLC. The proposed RSFLC was then implemented in a real time lab-scale CSTR. Experimental data confirmed the overall trend revealed by the simulation results under varied operating conditions. The control action of RSFLC was found to be stable, while control valve chattering was observed in the case of SFLC under certain disturbance conditions. Besides, SFLC was not successful in restoring the process to its set point in some cases. Regarding the AIMC, in general, it showed significantly higher settling time and deviation from the set point. Thus, the results obtained are quite encouraging and reflect the practical validity of the proposed RSFLC.||URI:||https://scholarbank.nus.edu.sg/handle/10635/153137|
|Appears in Collections:||Master's Theses (Restricted)|
Show full item record
Files in This Item:
|b22384649.pdf||7.13 MB||Adobe PDF|
checked on Jul 10, 2020
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.