Study of genetic algorithm implementation efficiency while turboprop engine modeling

Аuthors

Ivanov A. V.

Moscow Aviation Institute (National Research University), 4, Volokolamskoe shosse, Moscow, А-80, GSP-3, 125993, Russia

e-mail: TemkaW@mail.ru

Abstract

Propellers design and development for modern coaxial propfans and their automated control systems are impossible without in-line simulation test benches, which allow reduce testing fee, imitate failure situations, work through control laws and algorithms and determine automated control systems stability margins.

Turboprop engine mathematical model plays key role while testing propellers and automatic control systems with in-line simulation test benches. The tests validity depends on accuracy of non-stationary processes reproduction by mathematical model. Due to turboprop dynamic characteristics errors when employing linear methods of modeling, at present, non-linear element-by-element models became widely used. In the course of SV-27 coaxial propfan and RSV-27 hydro- mechanical regulator testing bench, JSC SPE “Aerosila” employs D-27 turboprop non-linear element-by-element model. Implementation of gas turbine engines non-linear models results in significant processing power waste due to the multiple recalculation of the thermodynamic mathematical model while compressors and turbines joint operation point search. To optimize the computational process while using a non-linear turboprop engine mathematical model the authors suggest to use of a genetic algorithm. Genetic algorithm was developed with LabView software, employed with in-line simulation test bench and associated with the engine mathematical model. Genetic algorithm of various configurations and probability values of mutations and number of species in population with in-line simulation test implementation efficiency was studied. The results of the study allowed determine the optimal genetic algorithm configuration and parameters of its optimal operation. In its optimal configuration with a small number of species in population and increased calculating error, this genetic algorithm appeared to be effectiveness comparable to method of successive approximations by bisection. However, the genetic algorithm execution instability, leading to computational resources wasting for some calculated points, makes its implementation in turboprop engine mathematical model, used with in-line simulation test bench for air propellers tests and their automated control systems, impractical.

Keywords:

genetic algorithm, turboprop engine simulation, in-line simulation test-bench

References

1. Gurevich O.S. Sistemy avtomaticheskogo upravleniya aviatsionnymi GTD (Automatic control systems for aircraft GTE), Moscow, TORUS PRESS, 2011, pp. 153-160.

2. Akhmedzyanov D.A., Krivosheev I.A., Sunarchin R.A. Vestnik SGAU, 2006, no. 1, pp. 24-25.

3. Akhmedzyanov D.A., Vlasova E.S., Kishalov A.E. Vestnik SGAU, 2006, no. 2 (10), pp. 41-44.

4. Akhmedzyanov D.A. Vestnik UGATU, 2006, vol. 7, no. 1 (14), pp. 36-46.

5. Panteleev A.V., Metlitskaya D.V. Vestnik Moskovskogo aviatsionnogo instituta, 2011, vol. 18, no. 4, pp. 102-113.

6. Panteleev A.V., Metlitskaya D.V. Vestnik Moskovskogo aviatsionnogo instituta, 2012, vol. 19, no. 3, pp. 149-159.

7. Gusev Yu.M., Danilin O.E., Badamshin B.I. Vestnik UGATU, 2010, vol. 14, no. 2 (37), pp. 136-145.

8. Rutkovskaya D., Pilin'skii M., Rutkovskii L. Neironnye seti, geneticheskie algoritmy i nechetkie sistemy (Neural nets, genetic algorithms and fuzzy systems), Moscow, Goryachaya liniya Telekom, 2006, 452 p.