- Tytuł:
- IC engine valve fault detection using energy distribution of different resolution levels of dwt as a input data to PNN classifier
- Autorzy:
-
Madej, H.
Czech, P. - Powiązania:
- https://bibliotekanauki.pl/articles/245388.pdf
- Data publikacji:
- 2009
- Wydawca:
- Instytut Techniczny Wojsk Lotniczych
- Tematy:
-
diagnostics
combustion engines
artificial neural networks
vibration - Opis:
- This article presents the attempt to detect the valve faults in the engine by using the vibroacoustic signal registered on the si engine block. the object of the research was 4-cylinder 4-stroke with eight valves 1.3 l SI engine. the vibration energy casedby combustion process depends on the average rotation speed and the crankshaft position. Mechanical faults which are having an impact on combustion pressure and misfire cause temporary changes of the rotational speed and instantaneous energy spectral density. Form the research analyzed it shows that there is apossibility of using artificial neural networks to assess the condition of the valves in the combustion engines. As part of the study, the descriptors calculated on the basis of the vibration acceleration signal registered on the engine block were proposed to serve as the source of information on the engine condition. The results have corroborated effectiveness of using the signal approximation and detail energy, acquired from the discrete wavelet decomposition, as the base for building models of engine operation. the use of a probabilistic neural network with a correctly selected value of coefficient gamma enables obtaining a faultless classification.
- Źródło:
-
Journal of KONES; 2009, 16, 4; 307-313
1231-4005
2354-0133 - Pojawia się w:
- Journal of KONES
- Dostawca treści:
- Biblioteka Nauki