Research Of Key Technologies On Mechanical Seal End Face Condition Monitoring And Life Prediction

As the important components of rotating machines such as reaction kettle, centrifugal pump, compressor and turbine, mechanical seals are mainly used to prevent leakage, save energy and protect environment and so on. If the above machines can be supposed as hearts of power cycle systems, then the mechanical seals are equivalent to the heart valves.The purpose of Mechanical Seal Condition Monitoring (MSCM) is to obtain the abnormal symptom and degradation information before the seal failure, which can make engineers prevent seal faults as far as possible by some corresponding measures. Thus the downtime and maintenance cost produced by seal failure are reduced, and the availability of the equipment is improved. By taking non-contact mechanical seal as research objects, the seal opening rotational speed is detected through Multi-sensors seal monitoring, then two evaluation models about seal face contact and its wear degree are constructed, and last, the remaing working life of the seal is successfully predicted. The main innovation and research results are as follow:(1)Generally speaking, the running condition of the mechanical seal is mainly determined by the contact condition of the seal end faces. In order to ensure mechanical seal keep the balance of low friction and low leakage, therefore, first of all must be take measurement of opening speed. The Laplace wavelet is used to detect the end face opening speed based on the AE signal.(2)It is significant to monitor the friction states of the seal end faces for evaluating the seal healthy. Since fixed friction has had effect on features clustering of the AE signal, HSSVM is used to identify the contact of the seal end faces at high shaft speed. Data shows that, this model can effectly detect the seal face-contact.(3)It has a great practice value that establishing a model to preliminary make more accurately assessment of the running state of mechanical seal in the absence of all history state data. In this paper, the method of Bias Factor Hidden Markov Model(BFHMM) is put forward. Based on BFHMM, a mechanical seal wear condition assessment technologies is proposed, and a accurate evaluate performance indicators to the running state of the mechanical seal is given. Because this model doesn’t need strict requirement of the time series to meet homogeneity compared with the past hidden markov model (HMM), it has the stronger ability of modeling and analysis. In addition to, this model have a high recognition rate, it enhances the universality and effectiveness of its application in industry site.(4) The final purpose of seal health evaluation is to predict the remaining working life of the seal. Since grey theory can effectly predict overall trend of a signal, but Partical filter is more sensitive to its local change, a new method, grey partical filter, combined with these two theory is presented. First, the grey theory is used to predict the total life curve of the seal, the particle filter is used to correct the curve inflextion. Data shows that this method can effectively predict the remaining work life of the seal.The research content of this paper can provide a theoretical basis and technical support for the condition monitoring of mechanical seal in the industrial site.