Concentration estimation and fault tolerant control in a CSTR modelled as a quasi linear parameter varying system

A design of a product concentration estimation and a Fault Tolerant Control (FTC) strategy for compensate an actuator fault in a Continuous Stirred Tank Reactor (CSTR) are developed in this paper. Second and third-order CSTR systems are considered to validate the proposed FTC scheme. Furthermore, a comparison between nonlinear, linear and quasi Linear Parameter Varying (qLPV) models for both CSTR systems are presented. The results demonstrate that the qLPV representation reproduce the nonlinear, in the selected segment, better than the linear model. Then, a Proportional-IntegralObserver (PIO) is designed using the qLPV representation in order to estimate the states and the actuator fault presented in the process. These estimations are used by the FTC system for compute a new control law using a Linear Matrix Inequality (LMI) to ensure the stability of both the qLPV PIO and the FTC law. Thus, the main contributions of this work are four: i) to propose a new representation of the second and three-order CSTR nonlinear model by means of a qLPV system, preserving the nonlinear dynamics of the original nonlinear model, ii) to exploit the easiness to extend theoretical results that originally were conceived for linear systems, to qLPV systems, iii) to estimate the product concentration in order to generate a new FTC law, and iv) to validate in simulation the FTC scheme to reduce the effect of an actuator fault in a CSTR process.

Keywords: fault tolerant control, fault estimation, quasi linear parameter varying, Continuous Stirred Tank Reactor.