6837092897

12    Diagnostyka’ 4(48)/2008

KEKEZ, AMBROZIK, RADZISZEWSKI, Modeling of Cylinder Pres surę in Compression Ignition Engine ...

The error of prediction of mean indicated pressure by GFSm model is less than 5% (with regard to experimental data) for all measured crankshaft speeds. Error of prediction of maximum value does not exceed 3.3 %. Pressure curve is calculated with high accuracy - root mean sąuare error (RMSE) belongs to rangę of 0.07 to 0.13 MPa.

4. CONCLUDING REMARKS

During experimental research the cylinder pressure curv'es in function of crankshaft angle were recorded. The acąuired indicator diagrams were analyzed in the rangę of 180 to 540 °ĆA, because then we can assume that processes in engine cylinder take place in thennodynamic closed system. The results obtained in tliis work can be used for regulation. evaluation and control of engine working cycle. They also give opportunity to evaluate the usefulness and relevance of fueling the engine with certain fuel. The results of experimental research acquired on engine test bench were used to build analytical-empirical model of engine work, based on fuzzy sets theoiy. In order to achieve this, the new genetic-fuzzy system GFSm with advanced infonnation encoding and adjustable number of mles, being a modification of Pittsburgh approach, was designed. By using this system, three models of cylinder pressure curves were built. The best accuracy (indicated work error less than 5% for each examined crankshaft speed) was achieved by Mamdani-type GFSm model consisting of 12 rules. Tlie model allows simulation of cylinder pressure curves with high accuracy, for all allowable cranksliaft speeds. There is also a possibility to model selected fragments of pressure curves morę precisely (e.g. places where pressure changes are fastest), by means of separate model. Time of computation and saving the results to disk equals 0.04 s. Proposed model can be used to evaluate the quality of working cycles of intemal combustion engines, with the accuracy required in practical teclmical applications.

REFERENCES

[1]    Amsden A. A.: KIVA-3V: A Block-Structured KIVA Program for Engines with Vertical or Canted Valves. Los Alamos National Laboratory, LA-13313-MS, 1997.

[2]    Brzozowski K., Nowakowski J.: An application ofartificial neural network to exhaust emission modelling from diesel engine. Journal of KONES, Intemal Combustion Engines, vol. 12, 1-2, 2005, pp. 51-58.

[3]    Jakubek S., Keuth N.: A local neuro-fuzzy network for high-dimensional models and optimization. Engineering Applications of Artificial Intelligence, vol. 19, 2006. pp. 705-717.

[4]    Kalogirou S. A.: Artificial intelligence for the modeling and control of combustion processes: a review. Progress in Energy and Combustion Science, vol. 29, 2003, pp. 515-566.

[5]    Kekez M.: Modeling of work of compression ignition intemal combustion engine with use of artificial intelligence methods. PhD dissertation (in Polish), Kielce, 2008.

[6]    Kimmich F.. Schwarte A., Isermann R.: Fault detection for modern Diesel engines using signal- and process model-based methods. Control Engineering Practice vol. 13, 2005, pp. 189-203.

[7]    Kurczyński D.: Influence of vegetable fuels and its blends with diesel oil on parameters of work of compression ignition engine. PhD dissertation (in Polish), Kielce, 2007.

[8]    Lee S. H„ Howlett R. J., Walters S.D., Crua C.: Fuzzy logie and neuro-fuzzy modelling of diesel spray penetration: A comparative study. Journal of Intelligent and Fuzzy Systems, vol. 18, no. 1. 2007, pp. 43-56.

[9]    Rychter T., Teodorczyk A.: Theory of pis ton engines (in Polish). Wydawnictwa Komunikacji i Łączności, Warszaw a, 2006.

dr inż. Michał KEKEZ jest adiunktem, pracownikiem Politechniki Świętokrzyskiej. Obronił doktorat w roku 2008 na Wydziale Mechatroniki i Budowy Maszyn. Zajmuje się systemami sztucznej inteligencji oraz silnikami spalinowymi.

dr hab. inż. Andrzej AMBROZIK, prof. PŚk w Katedrze Maszyn Cieplnych Wydziału Mechatroniki i    Budowy Maszyn

Politechniki Świętokrzyskiej. W    swojej    działalności

naukowej zajmuje się teorią procesów⁷ spalania w silnikach cieplnych.

dr hab. inż. Leszek RADZISZEWSKI, prof PŚk, jest kierownikiem Katedry Maszyn Cieplnych Wydziału Mechatroniki i Budowy Maszyn Politechniki Świętokrzyskiej. Zainteresowania naukowe obejmują: metody wibroaku-styczne badań materiałów i urządzeń, metrologia.