Construction of automatic air monitoring point siting model based on convolutional neural network and K-means clustering

To enhance the level of air monitoring in urban areas and accurately assess the air quality across the entire urban region, a model for automatic air monitoring point placement based on a convolutional neural network (CNN) and K-means clustering is studied. A CNN is employed to extract features from air monitoring data, and the K-means clustering algorithm is used to partition the extracted features. Based on the air monitoring data features, an air quality model is established to maximize the similarity of pollutant concentrations between the optimal grid p oints and all candidate grid points. Using this air quality model, the feature information entropy of the air monitoring data is computed to determine the recommendation priority of unlabeled nodes in an urban spatiotemporal map, and automatic air monitoring points with the highest average recommendation priority are selected. The model test results demonstrate that the established model can effectively identify the optimal locations for automatic air monitoring, with an air quality monitoring error of less than 1%. When the number of hidden nodes is 256, the CNN used in this model exhibits strong generalization capability and high stability in the output results.