The Research Of Intelligent Methods On Fault Diagnosis For Reciprocating Compressor Valves

In recent years machinery fault diagnosis technology has been greatly developed all over the world and the technology is very important in national production.It is rather difficult to monitor the state of reciprocating compressor owing to its complex construction and so many stimulation sources. Although the researches have been made and some achievements have been gained, the total diagnosis level isn't very high and does not match with practical application. The research on reciprocating compressor fault diagnosis is of great significance owing to the particularity of its application. The research work which is combined with practice in this thesis is based on predecessors'achievements, intelligent diagnosis methods are adopted and fault diagnosis is effectively solved for reciprocating compressor valves. In this thesis the denoising method and the feature extraction method are researched according to valves signals characteristic that periodic signals, impact signals and random signals etc are intermixed, signals feature is diffcultly extracted and so on. The rule of choosing wavelet basis function and wavelet denosing soft-thresholding valve is proposed after making further research on wavelet transform technique using for signals disposing of valves. Denoising disposing and feature extraction of signals are carried on through using wavelet transform technique. Feature vectors of signals are constructed by way of "energy"element. Directly gained information is very finite from original data of vibrational signals. Therefore, the feature extraction method of information fusion is proposed. Wavelet packet is used to extract "energy"feature of pressure signals and vibrational signals in the method. Information fusion is carried out to extract valves feature. Feature vectors of fusion is used for fault diagnosis.Its feasibility and validity are verified by the diagnosis result on pressure signals and vibrational signals of valves signals. Fault diagnosis method based on BP neural network is researched for reciprocating compressor valves. Artificial neural network has the capacities of disposing non-line problems, self-learning and collateral computing and on-line diagnosis, which makes it widely used in fault diagnosis of reciprocating compressor. However, BP neural network has some shortcomings which confine it to be used. So improved BP neural network is effectly used in fault diagnosis for reciprocating compressor valves, which gains excellent result. Fault diagnosis method based on resource limited artificial immune system(RLAIS) is researched for reciprocating compressor valves. More and more information is needed for compressor fault diagnosis, much information rubbish is produced, so diagnosis method should have better capacity of reducing data. And new fault swatches are brought with time going, so diagnosis method should have continual learning capacity. Therefore, fault diagnosis method based on RLAIS is proposed for reciprocating compressor valves.This method has merits of expressing knowledge definitely, better capacity of reducing information, good robustness and so on. Based on deeply analyzing the fault diagnosis process of reciprocating compressor, the system software of fault diagnosis prototype for reciprocating compressor is designed and developed by the way of MATLAB language which has toolbox functions and Visual Basic senior program language. The validity of fault diagnosis result is well tested through gathering signals , which testifies this system has usability.