Identification and control design for a class of non-minimum phase dead-time systems based on fractional-order Smith predictor and genetic algorithm technique

Abstract

The design of Smith predictor (SP) controllers has been largely investigated in the literature to solve fractional-order dead-time systems control problems; this is due to its robustness in presence of modeling uncertainties and measurement noises. In this paper, a new SP control strategy is proposed for a special class of non-minimum phase dead-time systems. This controller can be designed only by imposing a new fractional-order model and new robust controller in the modeling and synthesis stages. A fractional multi-low-order dead-time (FMLODT) model is accordingly proposed in the SP modeling stage, where an actual process behavior is well ensured by high model accuracy. The proposed fractional model parameters are identified by an adequate optimization tool based genetic algorithm. Furthermore, a robust fractional order controller based FMLODT model is synthesized, its structure is analytically determined through a three terms fractional-order reference (TTFOR) model. The proposed fractional controller parameters are tuned according to some guidelines available in the literature. Thus, two main contributions are proposed in this work: the first one consists of introducing a general framework for the SP modeling scheme. This is ensured by the proposed FMLODT model, which enhances significantly the modeling accuracy compared to the existing models in the literature. The second innovation consists of proposing a general framework for the controller design step in the SP scheme. This is guaranteed by the proposed TTFOR model, which improves significantly the trade-off between the two contradictory goals: performance and robustness. To highlight the proposed strategy of good performances, both the new fractional order model and the classical controller are applied on the minimum phase dead-time systems. The obtained simulation results illustrate the performance and improvement of the proposed model comparatively to preceding works.