Research On Modeling Technology Of Malodorous Gas Prediction In Landfill Based On Historical Data

With the development of China’s economy,residents’ demands for living environment have gradually increased,and the adverse effects of odorous gases from large-scale landfills on the normal life of surrounding residents have deepened.In order to solve the problem of people and garbage competing in environmental resources in the process of urban development,this paper proposes a landfill gas time series fusion model and landfill gas component prediction model by analyzing the landfill gas related data of Hangzhou Tianziling landfill.The gas correlation prediction model can prevent the odor from affecting the physical and mental health of the surrounding residents by accurately predicting the future trend of landfill gas and timely processing the landfill gas.The main contents and research results of this paper are as follows:1.Explain the different composition characteristics of landfill gas.Analyze the potential relationship of landfill gas via the data mining technology;obtain the correlation between CO2 and H2S,NH3 and PID and PM2.5.Summarize the daily and weekly variation of different components of landfill gas.Analyze the relationship between gas and environmental factors.2.Propose an outlier detection algorithm based on bidirectional gradient(Bidirectional gradient anomaly detection algorithm,BADA).By investigating the related outlier detection algorithm,the algorithm of bidirectional gradient detection outliers is proposed to distinguish the abnormal points and the mutation points of the landfill gas data.3.Propose a time series prediction model(WT-LSTM-Prophet,WLP)based on frequency domain fusion model.Based on the analysis of the relevant time series model,this paper proposes a frequency domain fusion model(WLP)that transforms time domain data into frequency domain data and fuses LSTM and prophet models by wavelet transform.The model has a superior effect on the overall fitness and the accuracy of the mutation point prediction.4.Propose ameliorated particle swarm BP network(Ameliorated PSO-BP,APSO-BP).Based on the analysis results of different components of landfill gas,the correlation model of different components of landfill gas is divided into three categories.An optimized particle swarm BP network model(APSO-BP)is proposed by investigating the relevant prediction algorithms.The experimental results show that the model can control the prediction error between different components of landfill gas within 5%.