### Abstract

In this paper the applicability of an adaptive control based on a novel branch of Computational Cybernetics is illustrated for two different, imperfectly and inaccurately modeled particular physical sytems. One of them is a water tank stirring cold and hot water as input and releasing the mixture through a long pipe. The mass flow rate and the temperature are prescribed at the free end of the exit pipe while the taps at the input side can diretly be controlled. Due to the incompressibility of the fluid the variation of the mass flow rate of the output is immediately observableat the pipe's end and is related to the control action at the input taps, while its effect on the temperature becomes measurable at the free end of the pipe only after a delay time needed for the fluid to flow through the pipe. This results in asymmetric and non-constant delay time. The other paradigm is the thermal decay of the molecular nitrogen during a throttling down process. As is well known chemical reactions hav very drastic non-linearities and it is not easy to construct their "exact" or satisfacorily avccurate model. The fundamental principles of this new branch of Computational Cybernetics are briefly presented in the paper. To some extent it is similar to the traditional Soft Computing, but by using a priori known, uniform, lucid structure of reduced size, it can evade the enormous structures so characteristic to the usual approach. Clumsy deterministic, semi-stochastic or stochastic machine learning is replaced by simple, short, explicit algebraic procedures especially fit to real time applications.The costs of these advantages may manifest themselves in the expected limitation of the applicabilityof this new approach. However, the simulation results exemplify the applicability of the new method in the control of systems of strong non-linearities and asymmetric delay time.

Original language | English |
---|---|

Pages (from-to) | 26-43 |

Number of pages | 18 |

Journal | Acta Polytechnica Hungarica |

Volume | 1 |

Issue number | 1 |

Publication status | Published - May 2004 |

### Fingerprint

### ASJC Scopus subject areas

- General
- Engineering(all)

### Cite this

**Simulation based verification of the applicability of a novel branch of computational cybernetics in the adaptive control of imperfectly modeled physical systems of asymmetric delay time and strong non-linearities.** / Tar, J.; Rudas, I.; Bitó, János F.

Research output: Contribution to journal › Article

*Acta Polytechnica Hungarica*, vol. 1, no. 1, pp. 26-43.

}

TY - JOUR

T1 - Simulation based verification of the applicability of a novel branch of computational cybernetics in the adaptive control of imperfectly modeled physical systems of asymmetric delay time and strong non-linearities

AU - Tar, J.

AU - Rudas, I.

AU - Bitó, János F.

PY - 2004/5

Y1 - 2004/5

N2 - In this paper the applicability of an adaptive control based on a novel branch of Computational Cybernetics is illustrated for two different, imperfectly and inaccurately modeled particular physical sytems. One of them is a water tank stirring cold and hot water as input and releasing the mixture through a long pipe. The mass flow rate and the temperature are prescribed at the free end of the exit pipe while the taps at the input side can diretly be controlled. Due to the incompressibility of the fluid the variation of the mass flow rate of the output is immediately observableat the pipe's end and is related to the control action at the input taps, while its effect on the temperature becomes measurable at the free end of the pipe only after a delay time needed for the fluid to flow through the pipe. This results in asymmetric and non-constant delay time. The other paradigm is the thermal decay of the molecular nitrogen during a throttling down process. As is well known chemical reactions hav very drastic non-linearities and it is not easy to construct their "exact" or satisfacorily avccurate model. The fundamental principles of this new branch of Computational Cybernetics are briefly presented in the paper. To some extent it is similar to the traditional Soft Computing, but by using a priori known, uniform, lucid structure of reduced size, it can evade the enormous structures so characteristic to the usual approach. Clumsy deterministic, semi-stochastic or stochastic machine learning is replaced by simple, short, explicit algebraic procedures especially fit to real time applications.The costs of these advantages may manifest themselves in the expected limitation of the applicabilityof this new approach. However, the simulation results exemplify the applicability of the new method in the control of systems of strong non-linearities and asymmetric delay time.

AB - In this paper the applicability of an adaptive control based on a novel branch of Computational Cybernetics is illustrated for two different, imperfectly and inaccurately modeled particular physical sytems. One of them is a water tank stirring cold and hot water as input and releasing the mixture through a long pipe. The mass flow rate and the temperature are prescribed at the free end of the exit pipe while the taps at the input side can diretly be controlled. Due to the incompressibility of the fluid the variation of the mass flow rate of the output is immediately observableat the pipe's end and is related to the control action at the input taps, while its effect on the temperature becomes measurable at the free end of the pipe only after a delay time needed for the fluid to flow through the pipe. This results in asymmetric and non-constant delay time. The other paradigm is the thermal decay of the molecular nitrogen during a throttling down process. As is well known chemical reactions hav very drastic non-linearities and it is not easy to construct their "exact" or satisfacorily avccurate model. The fundamental principles of this new branch of Computational Cybernetics are briefly presented in the paper. To some extent it is similar to the traditional Soft Computing, but by using a priori known, uniform, lucid structure of reduced size, it can evade the enormous structures so characteristic to the usual approach. Clumsy deterministic, semi-stochastic or stochastic machine learning is replaced by simple, short, explicit algebraic procedures especially fit to real time applications.The costs of these advantages may manifest themselves in the expected limitation of the applicabilityof this new approach. However, the simulation results exemplify the applicability of the new method in the control of systems of strong non-linearities and asymmetric delay time.

UR - http://www.scopus.com/inward/record.url?scp=78649572202&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=78649572202&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:78649572202

VL - 1

SP - 26

EP - 43

JO - Acta Polytechnica Hungarica

JF - Acta Polytechnica Hungarica

SN - 1785-8860

IS - 1

ER -