Optimization Design And Application Of Extreme Learning Machine Based On Particle Swarm Algorithm

With the development of China's mining industry and the continuous improvement of the technological level,the air pollution caused by the emission of a large number of harmful substances has become increasingly prominent,seriously affecting the health and normal life of the majority of residents.Air Quality Index(AQI),as an important indicator that can indirectly reflect the state of Air pollution,its forecast results can provide basis and guidance for environmental protection departments,and quickly attract the attention of experts in related fields.However,AQI is affected by many factors,such as CO,NO₂and O₃,which makes the prediction of AQI more difficult.Therefore,promptly and accurate analysis and prediction of the AQI can not only help people understand the future air conditions in advance and make reasonable travel plans,but also provide some help for air pollution control.In this paper,the optimized Extreme Learning Machine(ELM)is used to predict AQI,and the main research contents are as follows:(1)Based on the standard PSO algorithm,an improved PSO algorithm based on inertia weight and learning factor is proposed in this paper.At first,this algorithm adjusts the inertia weight by the way of nonlinear decline,then the learning factors are linearly changed,and the mutation operation of genetic algorithm is added to improve the population diversity,so as to improve its global searching ability and local searching ability to a certain extent.In addition,in order to evaluate the performance of the improved PSO calculation method,this paper selects four commonly used test functions to evaluate and verify.(2)Aiming at the sensitive problem of parameter selection of extreme learning machine network model,the improved particle swarm optimization algorithm(IPSO)was used to optimize the input weights and bias of extreme learning machine,and an optimized particle swarm optimization model of extreme learning machine(IPSO-ELM)was constructed,which was applied to the prediction of air quality index.In addition,in order to evaluate the performance of the optimized IPSO-ELM model,this paper also compares it with BP,ELM and PSO-ELM models.The experimental results show that IPSO-ELM has better generalization performance and higher prediction accuracy than the other three models,which provides a certain theoretical basis for future air quality monitoring.(3)On the basis of the above theory,two aspects of optimization are further made in this paper:one is the introduction of quantum thought on the original PSO algorithm,the other is the addition of kernel function in the extreme learning machine model.Finally,we use quantum behaved particle swarm optimization(QPSO)to select the parameters of Kernel Extreme Learning Machine(KELM),and then the AQI prediction model based on QPSO-KELM is obtained.At the same time,this paper also introduced PSO-KELM model and CPSO-KELM model to compare with the QPSO-KELM model mentioned above,and the experimental results prove that QPSO-KELM model has stronger prediction ability compared with the other two models,which shows the superiority and reliability of QPSO-KELM algorithm,and provides a new idea for air quality research.