Evaluation Method of Slipper Pair Performance Degradation State Based on Multi-parameter Intelligent Fusion of Deep Belief Network
The performance degradation state of the wear process of the slipper pair is affected by many aspects such as surface morphology and friction characteristics.It is inaccurate to extract any one of the characteristic parameters as indicators to evaluate its performance degradation state.In order to improve the evaluation accuracy,a multi-parameter intelligent fusion method based on deep belief network is proposed to evaluate the performance degradation state of the slipper pair.Fractal parameters such as fractal dimension,scale coefficient and characteristic roughness are extracted from the surface topography of slipper by fractal theory,and friction coefficient is used as the evaluation index of friction characteristics to construct the evaluation index system of performance degradation state.The gray correlation degree between the friction coefficient signal and the completely stationary Gaussian white noise sequence signal is calculated,and the degradation state is divided into regions according to the gray correlation degree.Deep belief network theory is applied to perform intelligent fusion and feature extraction of multiple index parameters,and a performance degradation state assessment model is established.The simulation test of slipper wear process is carried out to analyze the influence rule of index parameters and gray correlation degree on performance degradation state.The evaluation model is trained and tested through the constructed sample data set.The results show that the evaluation accuracy of the model for performance degradation state can reach more than 97％,which verifies the effectiveness and high accuracy of the method for performance degradation state evaluation of slipper pair.
slipper pairperformance degradation status assessmentfractal theorydeep belief network
NETL
NSTL
万方数据
