Nonlinear Analysis And Prediction On Operating Characteristics Of Vibrating Screen

Vibrating screen is an important equipment of coal preparation plant, which meets production requirement by vibration performance. The equipment contains many nonlinear links, such as the large deformation of structure, motivation of eccentric blocks, material impact on screensurface, nonlinear support and so on, and there will be nonlinear phenomenon in operation for above-mentioned reasons. Using the current linear analysis method based on spectrum analysis not completely reflect operation characteristics of the system, therefore, the author introduces chaos and fractal theory to extract nonlinear characteristics to achieve condition monitoring and fault diagnosis of the linear vibrating screen.In the actual production, forecasting equipment state and analysing trend have an great significance. If the reasonable forecast can be achieved, maintenance personnel can take measures in time to avoid fault and reduce economic losses. Accordingly, the paper studies time series prediction methods based on phase space reconstruction.In order to extract feature information from the collected signal, the author firstly studies wavelet de-noising theory, and then adopts moving average method to smoothing signal and eliminate its trend. Subsequently, the paper uses time-domain integral method of discrete data to compute first and second integral about measured acceleration signal and then obtains corresponding velocity and displacement signal. Finally, the author regards acceleration-velocity-displacement signal as coordinate axes to get phase space chart, meanwhile, combines amplitude spectrum analysis method to study motion of vibrating screen model in the different frequencies.Based on chaos characteristics of vibration signal of vertical on vibrating screen sides, the paper introduces fractal and chaos theory firstly, presents selection methods of the parameters in phase space reconstruction, and then studies extraction methods of correlation dimension and largest Lyapunov exponent and does simulation experiments for an example of the Lorenz signal. Finally, these characteristics are carried out to achieve crack diagnosis and position of linear vibrating screen. Through the analysis, it shows the two features can be used as a quantify index for depicting vibrating screen state and can also provide an effective basis for crack position.In paper, weighted first-order local prediction model, BP network prediction model based on phase space reconstruction and RBF network prediction model are established. Simulation experiment indicates that RBF network based on phase space reconstruction has better prediction precision, subsquently, it is applied to predict vibrating screen time series and proved feasible. Finally, the author does a preliminary analysis for the trend of characteristics of vibrating screen sides.