Application of the discrete wavelet transform and probabilistic neural networks in IC engine fault diagnostics

Around the world are continued attempts to use the vibroacoustic phenomena for purposes of diagnosis of machine condition. Particularly important becomes non-invasive methods including methods based on vibration and acoustic signals. Vibroacoustic phenomena, which relates to the working of technical objects, includes all necessary information connected with the technical condition. The biggest difficulty is the transformation of registered vibroacoustic signals and creation on their basic measures, which will be non-sensitive to any type of interference occurring during the operation of objects and recording signals. To the group of technical objects, for which are already conducted numerous studies all over the world, connected with use of vibroacoustic phenomena for diagnostic purposes which relates to the automotive drive systems, including combustion engines. Combustion engines during its working generate a whole range of vibroacoustic phenomena bringing information on the proper operation of the engine, as well as on condition of each of its elements. In a combustion engine, there are many sources of vibroacoustic phenomena, which contributes to the disruption of diagnostic information. The development of appropriate methods for vibroacoustic signal processing and complete diagnostic systems may allow future extension of the on-board diagnostics OBD system - currently in used cars. The most interesting would be the development of complex system for diagnosing the condition of the individual elements of the car engine operating by basing on information from vibroacoustic signals. In this article are shown results of research, which aim is to diagnose damages of mechanical elements of car combustion engine using vibration signals and artificial neural networks.