This article describes methods of diagnosing of internal combustion engines (ICE). The conclusion is drawn that the majority of modern methods and ICE diagnostic devices don't solve fully a problem of determination of technical condition of the engine, often are labor-consuming and expensive. The choice of a method and mode of diagnosing of ICE on the basis of external speed characteristics is carried out for what the list of sensors and executive mechanisms of a control system of the engine is defined. The choice of a method of training of fuzzy Sugeno systems on the basis of hybrid neural networks is reasonable. The possibility of identification of difficult dependences by the systems of fuzzy sets on the basis of hybrid networks is proved. Possibilities of systems for fuzzy conclusion on identification of dependences are the basis for algorithms. Assessment of influence of external factors on the accuracy of measurements therefore it is established that the maximum error doesn't exceed 5% is carried out. Experimental studies of metrological characteristics of the diagnostic system have been carried out, which showed that the relative errors do not exceed the estimated errors. In this case, speed characteristic was determined in the entire range of engine speed.
Volume 11 | 08-Special Issue
Pages: 2273-2279