Prediction Of The Remaining Service Life Of Mechanical Seal Based On Stochastic Process

For the process industry,the volume of the seal is small,but it determines the safety,reliability and durability of the industrial equipment.Studying the remaining using life(RUL)of mechanical seals can effectively improve the service life of industrial equipment,and help to make reasonable production plans,greatly reducing the economic loss and accidental injury caused by sealing failure to the enterprise.In view of the key and difficult points in the maintenance and management of the actual rotating machinery,taking the diamond-coated mechanical seal ring as the research object,a condition monitoring system based on Acoustic Emission(AE)technology was set up to analyze and process the acoustic emission signal of its full-life cycle so as to achieve the purpose of predicting the remaining service life Taking the diamond-coated mechanical seal ring as the research object,a condition monitoring system based on Acoustic Emission(AE)technology was set up to analyze and process the acoustic emission signal of its full-life cycle so as to achieve the purpose of predicting the remaining using life.In signal analysis and processing,the collected acoustic emission signals need to be preprocessed to reduce the interference of noise on useful signals.In view of the shortcomings of the traditional wavelet packet threshold selection,a method of optimizing the wavelet packet threshold using the Fruit Fly Optimization Algorithm(FOA)optimization algorithm is proposed.Compared with the classic statistical estimation threshold method,the algorithm has the advantages of fast computing speed,global search,and easy control of the algorithm.Through the simulation experiment,it is found that the wavelet packet denoising after optimization has a greater signal to noise ratio and lower RMS than the other five threshold estimation methods,and the effect is better.The signal after preprocessing needs to be extracted by features.The commonly used time-domain features and frequency-domain features are selected as parameters to characterize the signal characteristics.Correspondence coefficient(CC)was used to determine the characteristics of this paper.Kernel Principal Component Analysis(KPCA)was used to reduce the dimension of the feature.Finally,the single parameter degradation index of the remaining using life of mechanical seal is established by means of the Mahalanobis distance(Mahalanobis Distance,MD).The remaining using life prediction method for the mechanical seal of diamond coating based on the Wiener process is studied.It is assumed that the mechanical seal degradation obeys the exponential degradation model.In order to describe the randomness in the degradation process,the Wiener process is introduced here.Assuming that the parameters of the model are subject to normal distribution,the parameter estimation is carried out by the Expectation Maximization Algorithm(EM),and the explicit expression of the remaining using life is obtained.The established model is simpler,and fewer parameters need to be estimated.Besides,the algorithm of Bayesian update combined with the EM estimation can make full use of the degraded data.Through continuous updating,the estimation accuracy of the model can be improved.When the RUL is calculated,the iteration is not needed repeatedly,which greatly speeds up the calculation speed.Experimental results show that this method can predict the RUL of products effectively.At the same time,the accuracy of prediction is improving with the accumulation of degraded data.