Research On Wear Condition Identification Method Of Hydraulic Reciprocating Seal Based On Information Fusion

Hydraulic reciprocating seal is an important part of the hydraulic system,it is an important guarantee for normal operation of the whole equipment.In the actual work of hydraulic reciprocating seal,the friction in long time reciprocating process and the pollutant particles in the oil may cause the hydraulic reciprocating seal wear.As the result,it will cause oil leakage and more serious consequences.Therefore,accurate monitoring and identification of hydraulic reciprocating seal wear state is very important.However,the reciprocating seal works inside the hydraulic system,the state of the wear is concealed,and the information of the sensor is often affected by various factors,so that the information obtained is uncertain and fuzzy.In order to solve this problem,the paper took the reciprocating seal of piston as the research object,focused on the following contents:(1)The wear and seal mechanism of hydraulic reciprocating seal were studied,the key parameters of hydraulic reciprocating seal monitoring was clarified,the influence of hydraulic reciprocating seal leakage on fluid pressure and displacement were analyzed,the effectiveness and necessity of these parameters to monitor the wear state of reciprocating seal was clarified.(2)The test bench of hydraulic reciprocating was specially designed for monitoring experiment of hydraulic reciprocating seal,and the seal contact strain,fluid pressure and piston rod displacement signals were obtained;The seal leakage characteristic values of different wear conditions were calculated based on the test data by FBG;The wavelet packet method was used to extract three kinds of state features of pressure signal and displacement signal,such as the ratio of bottom energy,wavelet packet entropy and wavelet packet variance,the dimensionless amplitude index statistical method was used to extract five kinds of state features of the strain signals,such as the waveform index,the peak index,the pulse index,the margin index and the kurtosis index.(3)The method of BP neural network and D-S evidence theory were used to carry out local recognition and multi-information fusion recognition respectively for wear state of hydraulic reciprocating seal.In the 18 groups of selected samples,the results show that the multi-information fusion recognition has improved the accuracy of recognition and reduced the recognition uncertainty.(4)From the point of engineering application,Visual Basic 6.0,MATLAB R2014 a and SQL Server R2 were used to develop " Recognition System of Hydraulic Reciprocating Seal Wear State " preliminarily.It provided an effective tool for wear condition information fusion identification of hydraulic reciprocating seal.