Fault Diagnosis Of The Automobile Engine’s Valve Clearance By Rough Set And Neuron Network

Four-cylinder automobile engine as a typical reciprocating machinery and equipment, like the rotating machinery and equipment, plays an important role in all the machinery and equipment, but the research to reciprocating machinery like this is not as advanced as it to rotating machinery. For the automobile, the engine is one of the most important unit. With the development of science, the structure of the four-cylinder automobile engine has become increasingly sophisticated. This paper we will do some research to a typical four-cylinder automobile engine fault--abnormity of the valve clearance. I hope we can to reduce the cost of automobile repair, save the energy, improve the automobile’s running condition and so many things like these which can bring economic benefits and improve the work efficiency.According to the structure and the characteristic of four-cylinder automobile engine, we establish a vibration signal model of engine cylinder cover. Then we analyze the kinematics and dynamics of the admission gear. By researching the vibration mechanism we can analyze the characteristic of fault.Plenty of state information is included in the vibration signals when the automobile engine is running, Now the analysis of the vibration signals is widely used. In this paper, I design a method of feature extracting is based on time domain analysis and wavelet analysis and the features are from four-cylinder automobile engine. Then we research the selection principle of wavelet base function in processing vibration signal of four-cylinder automobile engine, based on this the appropriate wavelet base is determined. Then use the wavelet base function to do wavelet packet decomposition, so we can obtain the characteristic parameters from fault information.Based on the rough set theory, we can realize the attribute discretization and reduction for the characteristic parameters. And put forward a diagnostic model which based on the analysis of the integrated technology of rough set theory and BP neural network. Then we understand the structure and learning method of the rough-neural. In the networks, replace the neural unit in traditional with a pair of rough neural units, so the input of the network become a continuous interval, at the same time, the information understandability will be strength. Using this method to diagnose we get a result and we can compare it with the traditional BP neural network. This paper offers a feasible way to solve the problem of four-cylinder automobile engine fault which is abnormity of the valve clearance.