Research On Forecast Of Pollutant Emission In Actual Working Condition Of Tractor And Uncertainty Of Emission Inventory

With the continuous improvement of the level of agricultural mechanization in China,the rapid growth of agricultural machinery ownership,and the impact of non-road mobile source pollutant emissions such as tractors on the atmospheric environment have attracted more and more attention,and effective control of the emission of tractor pollutants has become an urgent and important issue.In recent years,with the continuous optimization of neural network algorithm models in the field of data mining,the release of new emission inventory establishment guidelines has provided methods and methods for emission prediction and emission inventory establishment for specific regions.Accurately predicting the actual operating conditions of diesel machinery such as tractors and constructing emission inventories is of great significance in analyzing the emission trends of agricultural mechanized production,formulating governance measures,and regional pollutant emission control.In view of this,the research in this paper includes the following 5 aspects:（1）Based on the portable emission measurement system（PEMS）,the actual working condition emission test of the tractor was carried out,and the emission factors of the fuel consumption was measured and the measured data such as engine speed,fuel consumption,combustion ratio,CO,HC,NO_X and PM under different operating conditions was measured,and then the deep extreme learning machine（DELM）was established.and predict pollutant emissions under basic conditions such as tractor idle,walking and rotary tillage.The results showed that the DELM model had certain advantages in predicting emission time series,and its average rms error of NO_X,HC,CO and PM emissions of tractors in idle,walking and rotary tillage states was 5.269×10^-5,5.195×10^-5,5.135×10^-5 and 2.795×10^-5,respectively.（2）To further evaluate the adaptability of the DELM predictive model,it was compared with the support vector machine（SVM）and back propagation neural network（BPNN）models.Comparing the DELM model with SVM and BP,it was found that the DELM model had significant advantages in terms of robustness and adaptability.（3）Based on the measured emission factors,combined with the"Estimation Method for Emission Inventory of Non-road Mobile Sources",the inventory of PM,HC,NO_X and CO emissions of non-road mobile agricultural tractors in 66 counties in Xinjiang from 2008 to 2019 was established and the evolution trend was analyzed,and the results showed that the average annual growth rate of total county tractor pollutant emissions from 2008 to 2019 was 2.60%,experiencing three changes of sharp rise,steady rise and fluctuation decline.（4）Using emission inventories as clustering samples,66 counties were zoned using the K-means clustering method.In terms of spatial dimension,66 counties in Xinjiang were divided into mild,moderate and severe emission control areas,of which the pollutant emissions of each emission control area accounted for 19.87%,39.92%and 40.21%respectively in 2019.Arc GIS software was used to draw a spatial distribution map of pollutant emissions at the county level,in which the distribution pattern of heavy emission control areas was relatively scattered,and the spatial distribution of mild and moderate emission control areas showed more concentrated characteristics.（5）Based on the Monte Carlo method（Monte Carlo）to analyze the uncertainty of the 2019 Xinjiang county tractor emission inventory,among them,NOx emission uncertainty was the largest,the uncertainty range was between-34.1%and 168.4%,and the uncertainty range of HC was the smallest,-41.9%～89.3%,which showed that the results of the Xinjiang county emission inventory established by this institute were more reliable to a certain extent.It can provide a reference for the formulation of pollution control policies for agricultural machinery such as tractors in Xinjiang county.This study provides methods and ideas for estimating the emission inventory of tractors at the county level,dividing the emission control areas of agricultural mechanized production,subsequently formulating differentiated pollutant emission control measures,and predicting mobile source exhaust emissions based on engine condition data.