Improved fractional augmented control strategies for continuously stirred tank reactors. Issue 3 (14th June 2022)

• Record Type: Journal Article
• Authors: Mukherjee, Deep; Raja, G. Lloyds; Kundu, Palash; Ghosh, Apurba
• Is Part Of: Asian journal of control. Volume 25:Issue 3(2023)
• Page Start: 2165
• Page End: 2182
• Publication Date: 2022-06-14
• Subjects: CSTR -- FOIMC‐FOPD predictor -- FOIMC‐FOPID -- FO‐Lyapunov -- fractional calculus -- modified PSO; Automatic control -- Periodicals; Control theory -- Periodicals; 629.805
• DOI: 10.1002/asjc.2887
• Languages: English
• ISSNs: 1561-8625

Abstract

Abstract: Controlling concentration and temperature of continuously stirred tank reactor (CSTR) is an extremely challenging task in chemical process industries. This is because conventional unity feedback schemes do not guarantee stable operating conditions for the CSTR process. Therefore, fractional calculus is augmented with multi‐loop control to achieve enhanced stability and closed‐loop performance than unity feedback structure in this work. Accordingly, three different fractional‐order‐based novel multi‐loop control structures are proposed based on time domain analysis. In two of the proposed strategies, fractional‐order PID controller (FOPID) and internal model control (IMC)‐based FOPID controller with fractional filter are used in the inner‐loop. Moreover, fractional‐order‐based Lyapunov stability rule of model reference adaptive control (MRAC) is used in outer‐loops for both of the above mentioned methods. Finally, an advanced multi‐loop predictor with FOPD controller in the inner‐loop and FOIMC controller in the outer‐loop is also proposed. In this work, fractional‐order, filter parameters, and FOPID settings are obtained by minimizing multi‐objective functions using modified particle swarm optimization (PSO) algorithm. Closed‐loop responses and control efforts of the proposed control strategies are compared with that of FO‐Lyapunov‐based MRAC scheme. Quantitative performance analysis is also carried out on all proposed methods based on error metrics and total variation of the control signal.