Vibration Signal Analytic Techniques Of Reciprocating Machinery Based On Fractal Theory

Reciprocating machine is used in all kinds of industry field widely, its condition monitoringand fault diagnosis is one of focus of the fault diagnosis studies. Because of the complicatedstructure of the reciprocating machinery and the variety of the movement forms of thecomponents, developing a secure and accurate diagnosis system becomes an urgent issue inpractical production. Foreign technology can't adapt to the domestic company.As a new study, fractal theory applied to mechanical fault diagnosis field is a new trend inacademia in recent years. Aim at the difficult problems of fault diagnosis of reciprocatingmachinery, combine the speciality of fractal theory ,this paper analyses and studies the vibrationsignal of reciprocating compressor based on the nonlinear processing method of fractal theory.First, we lucubrate some problems related to fault diagnosis, such as the technology offractal dimension extraction, geometric fractal dimension, natural fractal dimension and so on.We calculated Correlation Dimension in different working conditions by a series of mathematicmethods, such as program parameters analysis, signal de-noising, curve fitting, using time delayembedding method and other two kinds of typical fractal dimension computing theory. Regardthis one as the basis of diagnosis of reciprocating compressor.Secondly, we analyses the no stationary characteristic of the vibration signal ofreciprocating compressor, and introduced the wavelet de-noising method based on Mallatdecomposition and reconfiguration, according to the rules of time delay embedding method. Weanalyses the de-noise effect according to the known signal and get the satisfactory result by themethod. This method can ensure to extract the fractal dimension exactly.Next, combine several kinds of typical chaos as well as branching mapping, make use ofprogram compiled by time delay embedding method to operate simulation research to knownsignal and discuss the affection that parameter effects to Correlation Dimension, put forward thecertain methods and value area of several significant parameters that affect CorrelationDimension .simplify the program effectively, improve computing efficiency. The last, aim at different working conditions, according to the characteristic that thevibration wave form under the diagnosis state is more complex and more irregular, calculate theCorrelation Dimension, Spectrum Dimension as well as Box Counting Dimension of everymeasuring point respectively. Analysis indicates dimension values of measuring points varyremarkably;thereby realize many kinds of fault diagnosis to compressors.In addition, aim at several computing methods of fractal dimension, and combinereciprocating compressors vibration signal characteristic, develop mechanical fault diagnosissoftware to reciprocating compressors.