Flotation Height Prediction Under Stable And Vibration States In Air Cushion Furnace Based On Division Method

China has become a major country in the production and use of automobiles.Society is facing problems of environmental pollution and consumption of non-renewable resources.Among them,the increase in car ownership has become an important factor in environmental pollution and resource consumption.The continuous increase in the production and ownership of automobiles has brought convenience to people’s travel.At the same time,three issues of public concern such as energy consumption,environmental pollution and passenger safety have become increasingly prominent.For environmental protection and reduction of resource consumption,it is of great significance to reduce the weight of automobiles.This makes more and more high-end aluminum alloy sheets for automobiles as the main material for lightweight vehicles.As the key equipment for producing high-end aluminum alloy sheet,air-cushion furnace is playing an increasingly important role.However,there are still some problems in the localization process of the air cushion furnace.In the actual operation of the air-cushion furnace,there are problems in the strip floating process,such as complex working conditions,coexistence of steady-state and vibration conditions,and strong interference in the production process.In view of this,this paper will divide the floating state of the strip in the air cushion furnace into means,and combine the advantages of the mechanism model and the data-driven model to carry out research to solve the problem of the floating state and the height of the floating strip in the air cushion furnace.The main research contents of this article are summarized as follows:(1)This paper proposes an active learning framework to classify the floating state of the strip in an air-cushion furnace.It mainly designs a selection strategy based on uncertainty measures,local density measures and local weight measures,and stops based on the latter term interval strategy.(2)A hard partitioning method based on stacked noise reduction encoder and knowledge clustering method is proposed,and a clustering algorithm that can fuse the hovercraft process knowledge and suitable for time continuous problems can be built.The knowledge divides the steady state and vibration state of the strip.(3)Based on the theory of thick-walled jet impingement,a mechanism prediction model for strip floating height is constructed.A parallel hybrid prediction model for strip floating height in a stable floating state is proposed,and an unmodeled part of the prediction model of the strip floating height mechanism is compensated by using the LSSVR data error compensation model.Finally,according to the characteristics of the vibration state in the air-cushion furnace,the maximum and minimum values of the floating height of the strip under the vibration state were predicted using the LSSVR model.The vibration prediction model has achieved good prediction results in predicting the maximum and minimum values of the strip floating height in the vibration state.The related research content carried out in this thesis lays an important theoretical basis for the prediction research of the strip floating height during the hovering furnace’s floating process.