Research On Linear Vibration Screening Mechanism And Parameters Optimization Based On Gradient Boosting Algorithm

As an equipment for sorting granular materials,vibrating screens are used in various scenarios of industrial production.At the same time,improving the screening efficiency and processing capacity has always been the purpose of researchers in the field of vibrating screens.However,under the premise of not increasing the overall size and structural strength of the screen machine,the screening efficiency and the volume of processing are often two contradictory performance indicators,that is,unilaterally improving the screening efficiency tends to slow down the screening process,resulting in the volume of processing is reduced.To this end,many scholars at home and abroad have conducted detailed researches on the screening mechanism of vibrating screens.In order to completely analyze the influence of the screening intermediate state on the screening results,six features describing the particles distribution in the intermediate state of screening were defined(the average of stratification index,the mean of small particle touching sieve,the variance of small particle touching sieve,the average of small particle penetrating sieve,the average of large particle penetrating sieve and the screening time).The numerical simulation technology was used to simulate the screening process of the linear vibrating screens.The gradient boosting decision tree algorithm was employed to model and analyze the simulation data.The main research contents are listed as follows:(1)The vibrating screen's parameters were selected in a large range,and simulations were set up and ran based on these parameter combinations.The screening intermediate state features and the unit time screening efficiency could be calculated according to these simulations.(2)Based on the observation of the screening process,the screening intermediate state features were defined,and the influence of each features on the unit time screening efficiency was analyzed;(3)The gradient boosting decision tree algorithm was adopted to model the screening intermediate state features and unit time screening efficiency,and analyze the weight of each features,and the joint influence;(4)The gradient boosting decision tree algorithm was employed to model the vibrating screen's parameters and the screening intermediate state features,and analyze the influence of each vibrating screen's parameter on the screening intermediate state features;(5)The particle swarm optimization algorithm was used to optimize parameters of the vibrating screen's data model.Simulations and physical experiments were implemented to verify the accuracy,stability and effectiveness of the data model.Finally,according to the analysis of verifying simulations and physical experiments,each of vibrating screen's parameter was relatively changed in a large range,and the screening intermediate state features only changed in a small scale,and the final unit time screening efficiency values were similar,that means,there were many optimal parameter combinations that can produce similar screening intermediate state and the final screening results.