Research On Running State Analysis And Prognosis Of Vertical Roller Mill Based On Complex Network

As a key equipment in the process of cement production,vertical roller mill is a typical complex major equipment with large size,complex structure and many working parameters and constraints.The research on the analysis and prediction of the running state of the complex equipment like vertical roller mill is important to improve the reliability,safety and maintainability of the complex and large-scale equipment.At present,most of the research is focused on a single component of vertical roller mill(such as shell,rocker arm,disc and so on),and the research on the analysis of the whole equipment is seldom,which cannot reveal the coupling relationship and evolution law of the running state of the whole machine.Aimed at the above problems,this paper starts from the analysis of single components based on the complex networks theory.The time series are mapped to construct the complex networks.By analyzing the dynamic behavior of the complex networks,the operating status analysis and prediction of single component can be drawn in this paper.On this basis,the multi parameter complex networks model of equipment system-level running state is established so as to realize the system-level analysis and prediction of vertical roller mill running states and provide the systemlevel support to ensure the reliability of the equipment.The main research contents are as follows:(1)Analysis of single component running states.In order to analyze the running state of single component,the data collected by the Fiber Bragg grating displacement sensor on the rocker arm of vertical roller mill is taken as the research object.Firstly,adopt the SAX method to achieve the dimensionality reduction and symbolization of the data.On this basis,take a few adjacent symbols as nodes in complex networks.The sequential connection between nodes constitutes a directed weighted edge in the networks,forming a weighted directed networks model.Finally,take the obtained average degree,average path length and average clustering coefficient of the complex networks as the characteristic value of the classification so as to realize the recognition of the running state of the single component.(2)Trend prediction for single component running states.In this paper,the data collected by the Fiber Bragg grating displacement sensor on the rocker arm of vertical roller mill is taken as the research object.In order to eliminate the long-time correlation between sequences,a local visual map is used to map the time series of a single component to complex networks,which forms an undirected network.Based on the superposition of random walk to determine the points to be predicted and the connection of the nodes in the networks,the nodes connected with the prediction points are determined.And then map the complex networks back to time series so as to get the predicted value.By comparing the experimental results with the time series prediction model(ARIMA),the effectiveness of the proposed algorithm is verified.(3)Analysis of system-level running states.Taking the roller mill system of vertical roller mill as the research object,a multi parameter complex networks model is established based on the data from pin axis force sensors,rocker arm proximity displacement sensors and the roller support hydraulic cylinder pressure sensors.In order to use the nonlinear analysis method,the data are firstly validated by nonlinear method.Then,the Recurrence Plot is adopted to obtain the recurrence matrix of multidimensional time series.By recurrence matrix transformation,the adjacency matrix of complex networks can be obtained so as to get the complex networks of multidimensional time series.Through the analysis of the spectral density of complex networks,it is found that the spectral density of the complex networks is different under different running states.On this basis,the distinction between different running states can be obtained so as to achieve the identification of the running states.(4)Design and Realization of the intelligent monitoring and fault diagnosis system for the running state of the vertical roller mill.Load the running state analysis and prediction algorithms proposed in this paper to the system so as to verify the practicality of the algorithms.