Fault Diagnosis Of Centrifugal Pump Based On Principal Component Analysis And Sequential Probability Ratio Test

Pump is a kind of fluid mechanical machinery which transfers liquids by converting mechanical energy into hydraulic energy.The structure of pump is simple and it is durable.It has been widely used in water conservancy projects,farmland irrigation,petroleum,chemical,metallurgy,mining,shipbuilding and other departments.In a variety of pumps,centrifugal pump is most widely used because of it has advantages of large flow,high-lift,wide performance ranges,small size,light weight and stable operation.In the running process of centrifugal pump,impeller,shaft and other parts will inevitably appear faults,so it is important to analyze the faults of centrifugal pump and take corresponding measures to deal with them.The studying object of this paper is centrifugal pump.On the basis of sequential probability ratio test,the principal component analysis is combined with it to study the fault diagnosis of the centrifugal pump.The main research contents are as follows:Firstly,the key step of centrifugal pump fault diagnosis is feature extraction.There are many commonly used feature parameters in the signal,so it is unreasonable to directly select one of them as testing parameter.Principal component analysis(PCA)can transform more original variables into a few independent new variables with enough original information,and it has the advantages of simple concept and convenient calculation.Therefore,the principal component analysis method is used to reduce the dimension of the extracted feature parameters,and the influence of each parameter is comprehensively considered while reducing the parameters.Secondly,sequential probability ratio test(SPRT)is introduced in detail.The use of sequential probability ratio test algorithm in fault diagnosis is discussed,and the steps of SPRT in practical application are summarized.Sequential probability ratio test does not require a fixed sample size.It compares the hypothesis test data with a threshold.The fault diagnosis method based on parameter sequential probability ratio test should put forward the hypothesis according to the specific problem.Then,the SPRT model is established according to the fault information.Finally,the test results are obtained according to thespecific fault information and the diagnostic capability of the method can be evaluated by root mean square error.Thirdly,a novel centrifugal pump fault diagnosis method based on principal component analysis and sequential probability ratio test is proposed.First,the signal is de-noised by wavelet packet transform and the characteristic parameters of the signal are extracted by time domain analysis.Then,the dimension of the extracted characteristic parameters is reduced by PCA,and the principal component with the largest contribution rate is selected as the testing sequence.Finally,the sequential probability ratio test algorithm is used to analyze the running status of the centrifugal pump and combine the root mean square algorithm to classify the fault.After that,the fault diagnosis experiments and analysis of centrifugal pumps under are carried out.The sequential probability ratio test is used to identify different faults and combined with the root mean square error to further give the identification criterion.The results indicate that this method is accurate and reliable for the fault diagnosis of centrifugal pumps.