Fuzzy predictive control of highly nonlinear pH process

https://doi.org/10.1016/S0098-1354(97)87570-9Get rights and content

Abstract

This paper proposes a new approach to predictive control of highly nonlinear processes based on a fuzzy model of the Takagi-Sugeno form. Standard Model Based Predictive Control (MBPC) methods use linear process models and are therefore unable to deal with strong process nonlinearities. But, the advantage of linear MBPC is in implementation due to fast optimization algorithms and guaranteed convergence within each time sample. In our approach, step responses for different operating points are extracted on-line from the nonlinear fuzzy model and a modified linear DMC algorithm is used. In this way, all the advantages of both fuzzy modeling of nonlinear processes and DMC control are accomplished. For performance evaluation of this simple but efficient approach, a nonlinear pH-process is used.

Reference (14)

There are more references available in the full text version of this article.

Cited by (70)

  • A predictive control based on decentralized fuzzy inference for a pH neutralization process

    2022, Journal of Process Control
    Citation Excerpt :

    Among them, the T–S fuzzy model has been proved to be very suitable for nonlinear predictive control due to its more accurate approximation ability for complex nonlinear systems [32,33]. By calculating the unit step response of the T–S model, Kavsek-Biasizzo et al. [34] constructed a DMC algorithm to control a strongly nonlinear pH process. The results showed that the algorithm had better inhibition effect on model mismatching and unmeasurable disturbance than the conventional DMC algorithm.

  • Stochastic data-driven model predictive control using gaussian processes

    2020, Computers and Chemical Engineering
    Citation Excerpt :

    Many mechanistic and empirical models are however often too complex to be used online and in addition have often high development costs. Alternatively, black-box identification models can be exploited instead, such as support vector machines (Xi et al., 2007), fuzzy models (Kavsek-Biasizzo et al., 1997), neural networks (NNs) (Piche et al., 2000), or Gaussian processes (GPs) (Kocijan et al., 2004). For example, recently in Wu et al. (2019c,b) recurrent NNs are utilised for an extensive NMPC approach with proofs on closed-loop state boundedness and convergence applied to a chemical reactor.

  • Stable fuzzy control and observer via LMIs in a fermentation process

    2018, Journal of Computational Science
    Citation Excerpt :

    Thus, a gain-scheduling fuzzy control can be used to develop a fuzzy control in different operating points [19], each control can be obtained from a heuristic design and then copied in each submodel; as these controllers are heuristic in nature because there is not a design procedure, it cannot be guaranteed stability for all applications [20]. One way to obtain a design procedure to have a stable control for a bioprocess is to use the local linear models as consequents of fuzzy rules [11] where the approximate linear submodels are obtained for different operating points; in this way, if it is possible to obtain a common matrix P for all submodels using the Lyapunov theory, it is possible to obtain a stable fuzzy controller [21]. The development of an output-feedback controller using this idea was presented in [22] where they were added degrees of freedom to achieve a proper solution of the LMIs.

  • Shape-independent model predictive control for Takagi–Sugeno fuzzy systems

    2017, Engineering Applications of Artificial Intelligence
    Citation Excerpt :

    For instance, the widely-cited work (Sousa et al., 1997) resorts to clustering but, actually, it carries out nonlinear MPC on the resulting identified fuzzy model, via nonconvex branch-and-bound optimisation. The work Kavsek-Biasizzo et al. (1997) computes a linear MPC by “freezing” the memberships at a particular instant and assuming they will be constant in the future; this might work in practice, but it lacks justification in fast transients. The work Lu and Arkun (2000) presents an interesting approach in which a sequence of quadratic cost bounds and state-feedback gains solves (suboptimally) the MPC problem.

  • Model predictive control of pH neutralization processes: A review

    2015, Control Engineering Practice
    Citation Excerpt :

    As only one model was identified the result are very similar to the linear model case described in Section 5.1.1. The first application of the Takagi–Sugeno approach was done by Kavsek-Biasizzo, Skrjanc, and Matko (1997). They obtained a nonlinear model from the fuzzy modeling, while applying triangular weights, hence just linearly combining two models at the time, to obtain the linear model applied at each time step.

View all citing articles on Scopus
View full text