Research On Multi-Population Optimization Algorithom And Its Application In Cement Rotary Kiln

Energy consumption,atmosphere pollution and other issues in cement production have gradually become the focus of industry.As the key equipment in cement production process,how to effectively realize energy saving and emission reduction and improve clinker quality in rotary kiln is of great significance to the development of cement industry.However,the calcination process of rotary kiln is complexity,which make it difficult for traditional methods to establish its system model and optimize its working conditions.Artificial Intelligence technology has been widely concerned because of its good application in system modeling and optimization.Therefore,two improved algorithms are proposed: symbiosis multi-population particle swarm optimization algorithm based on velocity communication(SMPSO)and multi-group grey wolf optimization algorithm based on K-means clustering(KMGWO).Parameters of extreme learning machine are optimized by KMGWO algorithm and KMGWO-ELM model is established.SMPSO algorithm and KMGWO-ELM algorithm are applied to the parameter optimization model in order to control rotary kiln effectively.The specific research work is described as follows:Firstly,SMPSO particle swarm optimization algorithm is constructed to solve the problems of poor population diversity and low precision.The whole population is divided into two parts: the dominant population and the slave population.The slave population is responsible for global search of solution space by using speed exchange mechanism,and the optimal information is shared to the dominant population.The dominant population is responsible for local search of solution space by using adaptive mutation mechanism and the experience of the slave population,and the optimal information is shared to the slave population.Then the symbiotic relationship between the dominant population and the slave population is established.According to the different search performance of each sub-population,different combinations of learning factors and inertia weight are introduced.The performance of SMPSO algorithm is tested by simulation experiments.Secondly,KMGWO gray wolf algorithm is constructed under the inspiration of symbiosis principle and multi-group idea of SMPSO algorithm to solve the limitation of the search performance.K-means clustering strategy is used to divide the initial population into several small populations,which increases the diversity of the population.Competitive relationship is established among decision-making levels of multiple sub-populations,which realizes information exchange among sub-populations and updates decision-making levels of each sub-population dynamically.Parameters of ELM are optimized by KMGWO algorithm and KMGWO-ELM model algorithm is established.The performance of KMGWO-ELM network is validated by using sample data sets.Finally,the prediction model of free calcium oxide content in cement rotary kiln is established by KMGWO-ELM model algorithm,and the functional relationship between fCaO content and operation parameters is obtained.Based on the functional relationship,SMPSO algorithm is used to optimize the operation parameters,and the optimal adjustment values of the operation parameters are obtained when the fCaO content of cement reaches the ideal value.The feasibility of parameter modeling and optimization modeling of cement rotary kiln based on KMGWO-ELM algorithm and SMPSO algorithm is verified by simulation experiments.