Research On Dynamic Many-objective Model Of Coal Green Production And Its Optimization Algorithm

As the basic energy,coal plays an important role in the development of industrialization in China.However,there are two problems in the previous coal production research.The first is most of the papers focuses on coal production models with two or three objectives,without studying that with four or more objectives.The second is all works showed static problems and lacking the research on the dynamic problems of coal production.Therefore,this paper proposes a dynamic many-objective model of coal green production,and proposes the corresponding algorithm for optimization.Because of the lack of dynamic many-objective optimization models in coal production research,this paper constructs a dynamic many-objective model of coal green production.The model not only considers five objectives: economy,energy,environment,recyclable resources and safety,but also adds dynamic parameters changing with time in coal production process.It provides a systematic and comprehensive guidance for coal production and allocation.In order to verify the feasibility of the model,this paper takes the production data of a coal mine in Shanxi Province as an example,and uses the classic DNSGA-II dynamic algorithm to optimize the model.The experimental results show that the model is effective and feasible.In order to solve the dynamic many-objective model of coal green production,this paper proposes a dynamic algorithm based on group center prediction.The algorithm uses reference point strategy to cluster population,and uses historical information to predict the group center through gradient prediction model in the new environment.After the new cluster center is generated,the method of based on individual Gaussian disturbance and group-center uniform distribution is proposed to reinitialize population,that enhances the ability of tracking Pareto front.Finally,we combine 11 classic static many-objective evolutionary algorithms with dynamic algorithm based on cluster center prediction to solve proposed dynamic manyobjective model of coal green production.Experimental results show that One-by-one evolutionary algorithm combined with dynamic algorithm based on cluster center prediction has optimal convergence and diversity,and its performance is better than DNSGA-II.Because of the limitation that dynamic algorithm based on cluster center prediction needs to build reference points,this paper proposes a new algorithm,which is the dynamic algorithm based on K-means clustering center prediction.When the environment changes,K-means clustering method is used to cluster the population,and historical information is used to predict the clustering center of the new environment.In the reinitialization,the accuracy of prediction is enhanced by introducing prediction error.11 classic static many-objective evolutionary algorithms are combined with dynamic algorithm based on K-means clustering center prediction to solve proposed dynamic many-objective model of coal green production.Experimental results show that One-by-one evolutionary algorithm combined with dynamic algorithm based on K-means clustering center prediction has best performance.Finally,through comparison,the dynamic algorithm based on K-means clustering center prediction is superior to the dynamic algorithm based on cluster center prediction in terms of convergence and diversity.