Research On Friction Vibration Feature Extraction And Friction State Recognition Of Reciprocating Sliding Friction Pair

Friction vibration is a phenomenon caused by the friction and wear of the friction pair during the operation of mechanical equipment,and contains information reflecting the friction and wear state of the friction pair.The vibration caused by friction is one of the important information of the tribological system output.Compared with the output information such as the friction coefficient,the surface morphology of the friction pair,the condition of the lubricating medium,the friction torque,and the wear particles,the friction vibration signal can be collected in real time during normal operation of the device.Therefore,the analysis of friction vibration signal is the best means to monitor the friction and wear state of the friction pair in real time during the operation of the equipment.The research on friction vibration denoising and feature extraction in the friction and wear process of friction pair can provide theoretical basis and method for friction and wear state monitoring and fault diagnosis of mechanical equipment.During the friction wear test of the friction pair,the background noise is also collected by the friction vibration acquisition system.The collected friction vibration signals are denoised by the effective noise reduction method for subsequent analysis,which can effectively improve the accuracy of the analysis result.This paper establishes a friction vibration noise reduction method based on spectral subtraction and ensemble empirical mode decomposition(EEMD)algorithm.The composite signal obtained by superimposing the noise signal and the structured friction vibration simulation signal is denoised by applying the established noise reduction method.The signal-to-noise ratio of the signal has been greatly improved,which verifies the effectiveness of the noise reduction method and realizes the noise reduction of the friction vibration signal.In order to find out the evolution law of the friction vibration during friction and wear,the chaos theory and multifractal theory were used to extract the characteristics of friction vibrations in friction and wear process.The friction vibration signals obtained from the friction and wear tests were reconstructed in phase space,and the change of friction vibration was qualitatively analyzed according to the evolution of the chaotic attractor trajectory convergence and divergence.Correlation dimension is sensitive to the time process of the system.The correlation dimension was used to characterize the chaotic characteristics of the friction and wear process,and the chaotic characteristics of the friction vibrations of different friction states of the friction pair were quantitatively described.The friction vibration signals were analyzed by using the multifractal detrended fluctuation analysis(MFDFA)algorithm,The obtained multifractal spectrums ??f(?)of the friction vibrations have different opening sizes,and the shapes of the multifractal spectrums are changing as the test time progresses.The shapes of the ??f(?)curves of different friction states are similar,with different parameter ranges and showing certain regularity.In the mixed friction test,the friction vibration multifractal spectrum parameters ??,?f,fmax,?fmax,?max and I? show a decreasing trend,while The parameters of the multifractal spectrum of friction vibration in the boundary friction and dry friction tests show an increasing trend.The range and variation trend of the multifractal spectrum parameters of the friction vibration in different friction states reflect the variation of the friction vibration in different friction states,and realize the feature extraction and quantitative characterization of friction vibration in different friction states.At the same time,the friction coefficient signals during the test were analyzed simultaneously.The multifractal spectrum parameters of friction vibration and friction coefficient under the conditions of mixed friction,boundary friction and dry friction show basically the same increasing and decreasing trend,indicating the effectiveness of the friction vibration feature extraction method.On the basis of the research on the feature extraction of friction vibration,the multifractal spectrum parameters of friction vibration time domain and frequency domain signals were analyzed based on the principal component analysis algorithm.The original feature space composed of the multifractal spectrum parameters of the friction vibration was a high-dimensional data space,and each friction vibration sample point cannot be visually represented.After principal component analysis,the characteristic parameters of each sample point were converted into three comprehensive parameters,which creates a condition for the direct representation of friction vibration characteristics.Principal component analysis was performed on the multifractal spectrum parameters of the friction vibration signals of the three sets of experiments.The analysis results show that the first three components of the friction vibration multifractal spectrum parameters of mixed friction,boundary friction and dry friction have their respective positions in a three-dimensional space.The first three principal components of the same friction states are close to each other with different states in different spatial zones,with good differentiation degree.The centroids of the three-dimensional spatial distribution of friction vibration multi-fractal spectrum parameters of different friction states were used for friction state recognition.The friction vibration samples of the verification test were used to identify the friction state of the friction pair.The accuracy of the recognition result is 91.7%.Furthermore,the applied friction state identification method was applied and the integrated centroid was obtained according to the centroids of the main component of the friction vibration multifractal spectrum parameters of the three sets of experiments.The integrated centroid can be used for the identification of the friction state within the parameters of the three sets of test conditions,and the accuracy of the friction state recognition for the verification test is 87.5%.It is indicated that the established friction state identification method of friction pair has higher reliability,and provides a method for identifying the friction state of friction pair by friction vibration signal.