Research On Application Of Improved QPSO_RBF Neural Network In Air Quality Forecast

With the development of social economy,the air pollution caused by industrialization and urbanization is becoming more and more serious.PM_2.5is an important component of air pollutants,and its increasing concentration value has brought extremely serious problems to people’s daily working life.Related research shows that the incidence of cardiovascular and lung diseases in people who inhale PM_2.5particles for a long time will increase significantly.In addition,the extreme smog caused by excessive PM_2.5concentration has continuously appeared in recent years,which has brought about a negative impact on transportation and industrial production that cannot be ignored.In the management of PM_2.5pollution,accurate and timely prediction of PM_2.5concentration value is a necessary prerequisite for effective control and prevention of PM_2.5hazards.Therefore,the establishment of a scientific and reliable PM_2.5concentration prediction model for effective management of PM_2.5pollution It has important practical value.This paper presents an air quality prediction model based on improved QPSO＿RBF neural network,and uses the improved QPSO＿RBF neural network algorithm for air quality prediction.The model is based on the RBF neural network.It selects 25 influence factors related to PM_2.5concentration,such as air temperature,wind speed,and SO₂concentration value,and uses random forest algorithm to select the characteristics of the influence factors and screen out PM_2.5.Some factors with high correlation of concentration values are used as input factors for the final prediction model.In optimizing the three important parameters of the RBF neural network,such as the weight,center,and base width,the optimization algorithm used is the QPSO quantum particle swarm algorithm with extremely high performance evaluation.In order to further improve the performance of the optimization algorithm and solve the problems of premature particles in the QPSO algorithm and particles falling into local extrema,this paper proposes a quantum particle swarm algorithm with position optimization and weight adaptation,referred to as EA＿QPSO algorithm,to further optimize QPSO To further improve the accuracy of the prediction model.Use this model to predict the time value of PM_2.5concentration in a certain area of?Nanjing City,and compare the prediction results with 6 models including RBF model,PSO＿RBF model,QPSO＿RBF model,SVM model,and GA＿BP model.The model proposed in this study is analyzed in many aspects,such as comparison with predicted values.Predictive analysis shows that the performance of the model is greatly improved during the gradual improvement process,and it performs well in terms of accuracy and convergence ability,and can be applied to the actual PM_2.5concentration prediction.In the extended comparison test,the model is used to predict other air pollutants such as NO₂,PM₁₀etc.The predicted value is also close to the actual value,indicating that the model has good applicability.