Performance Modeling And Multi-objective Particle Swarm Optimization Of Cyclone Separator

Cyclone separator is the most widely used equipment in gas-solid separator.It is a separator that separates solid particles from air flow by centrifugal force.Because of its simple structure and convenient maintenance,it is widely used in the fields of chemical industry,energy and environmental protection.Efficiency and pressure drop are two contradictory performance parameters of cyclone separator.The design of cyclone separator with high efficiency and low resistance is a complex multi-objective optimization problem.In the multi-objective optimization problem of cyclone separator,the objective function generally has three forms:theoretical and semi-empirical model,computational fluid dynamics model and meta-model.Among them,the calculation results of different theoretical and semi-empirical models will be greatly deviated due to different assumptions and simplified conditions.Computational fluid dynamics model is a time-consuming and unsuitable work and is not suitable for application in iterative optimization tasks.Therefore,it is very meaningful to construct a simple and effective meta-model that is modeled by experimental data or simulated computation data.In order to clearly describe the complex nonlinear relationship between efficiency,pressure drop and structural parameters,operating conditions,a support vector regression(SVR)algorithm based on principal component analysis(PCA)and particle swarm optimization(PSO)is applied to the performance modeling of cyclone separator.The experimental data provided in the literature are preprocessed by principal component analysis,random sampling technique and normalization method.Then the particle swarm optimization algorithm is used to determine the penalty coefficient C,the parameter g of the kernel function and the insensitive loss?in the support vector regression model.Finally,the 80%of the pre-processed experimental data are used to train the SVR model with the optimized parameters,and the generalization ability of the model is tested with the remaining 20%data.The mean square error of PCA-PSO-SVR model coupled with PCA,PSO and SVR in predicting the test set of grade efficiency and pressure drop is 6.948e-4 and 8.982e-4,and the correlation coefficient is 0.982 and 0.990,respectively.The model has higher accuracy,generalization ability and robustness than the existing theoretical and semi-empirical models and meta-models of cyclone efficiency and pressure drop.Applying these two meta-models as objective functions to the multi-objective optimization design of the multi-stage cyclone separator will give better results.Cyclone separator is an important equipment for gas-solid separation.However,in most industries,the single-stage cyclone separator cannot meet the needs of production,and three or more cyclone separators need to be used in series to work together.However,there are few studies on the multi-stage series cyclone separator,and most of the studies on the structure optimization of cyclone separator only focus on the two goals of improving efficiency and reducing pressure drop,without considering how to reduce the cost.Multi-objective particle swarm optimization(MOPSO)is used to optimize the ratio of cyclone cross-sectional area to inlet cross-sectional area K_a,ratio of diameter of vortex finder to that of cyclone (?) and cylinder diameter D with pressure drop,efficiency and cost as objective functions.The result is a set of Pareto optimal solutions that weigh the relationship between pressure drop,efficiency and cost.