| In industrial production,as a chute to save labor,material resources and other production costs,it is widely used to complete the material transportation between different equipment.Due to the impact between the material and the chute,it is very easy to have problems such as material blocking,material crushing,and even material falling.When these problems occur,workers need to carry out high-altitude operation maintenance to ensure the normal progress of production activities.The risk factor of high-altitude operation is relatively high,which is prone to accidents.If materials fall,it will threaten the personal safety of workers.On the other hand,the chute is damaged due to the impact of long-term use of materials.Therefore,how to reduce the crushing rate of materials and improve the service life of chute has important research significance.According to the actual situation of the factory,this paper establishes the three-dimensional model of the initial structure of the chute,and uses EDEM software to simulate the transportation process of materials in the chute,so as to obtain the maximum impact force and maximum material conveying speed data at different structural parameters,so as to provide data support for the optimization of chute structure design.Next,taking the structural size of the impact part of the chute as the design variable and the optimal speed range of material transportation as the constraint condition,the chute structural parameter optimization model to minimize the impact force is constructed.In order to reduce the computational complexity of the optimization model,an surrogate model is used to fit the implicit relationship between impact force,transportation speed and chute structure design variables.Among them,support vector regression(SVR)model can better solve the practical problems such as small samples,nonlinearity,high dimension,over fitting,multiple local minima and so on,and has strong generalization ability.Firstly,this paper solves the chute structural parameter optimization model based on polynomial(Poly)kernel function SVR model and radial basis function(RBF)kernel function SVR model.By studying the classification performance of common kernel functions,aiming at the problems of generalization and learning ability of single kernel function,an surrogate model aggregation method is used to aggregate the SVR models of different kernel functions to obtain the aggregate kernel support vector regression(E-SVR)surrogate model.The effectiveness of the aggregated E-SVR model is verified by three numerical examples.Finally,the sequential quadratic programming(SQP)algorithm calls the e-svr model to solve the chute structure parameter optimization model,and compares it with the SVR model based on single kernel function.The results show the effectiveness of E-SVR model and the feasibility of chute structure optimization design based on E-SVR model. |
