Research And Experiment On Extraction Of Engine Fault Components Based On Abnormal Sound Analysis

Engine is an important part of a car.With the increasing level of modern technology,the car engine is more intelligent and multi-functional,its structure is more complicated.However,traditional methods relying on artificial experience have gradually failed to meet higher diagnostic needs.Therefore,research and test of engine fault component extraction based on abnormal sound analysis is carried out for online diagnosis of the engine to eliminate the fault in time,which can avoid larger damage and improve engine reliability and service life.Diagnosing engine faults based on acoustic signals is a hot research direction in modern fault diagnosis,with non-contact advantages.Extraction of fault components as a core step in fault diagnosis process directly affects the accuracy and reliability of the diagnostic results.Therefore,this paper develops an adaptive misalignment superposition method(ADSM)to extract fault components in abnormal sound based on summarization of status in domestic and foreign countries.The ADSM is divided into three steps.Firstly,an automatic search algorithm for initial superposition point in abnormal sound is established.Secondly,the abnormal sound signal is intercepted based on the initial superimposed point,and then the different intercepted segments are superimposed to improve the signal to noise ratio.Finally,part with main information and energy of superimposed signal is intercepted to obtain the fault component.Taking the four-stroke engine model EA211 as the experimental object,we have built a data acquisition and processing platform,which is mainly composed of an engine,an industrial computer,an acoustic sensor,an encoder and a data acquisition card.The acoustic sensor is used to collect abnormal sound signal,and the encoder is responsible for intercepting abnormal sound signal by control of angular displacement.A mathematical model for judging rising edge of the encoder and a interception method of the abnormal signal is proposed.In addition,an abnormal noise signal processing system based on Lab VIEW and Matlab is built for processing the abnormal noise signal in time,to determine whether the engine has failed and to separate fault components.Common knocking faults and connecting rod bearing faults have been artificially set.The abnormal sound is processed by ADSM and the extraction results are compared to high signal to noise ratio fault component.Experimental results show that ADSM can effectively extract the fault components in abnormal sound signal.To further illustrate advantage of ADSM,the processing results were compared with those of wavelet denoising and empirical mode decomposition(EMD),and the comparison results show that ADSM has better extraction effect on quasi-periodic impact fault components.ADSM not only extracts faulty components of engine,but is also effective for separation of fault components of other types of rotating machinery.The innovations of this paper are as follows:The ADSM is proposed.The method can process the abnormal signal in real time and automatically separate the fault components contained therein.