Abstract:Earthwork-related locations (ERLs), such as construction sites, earth dumping ground, and concrete mixing stations, are major sources of urban dust pollution (particulate matters). The effective management of ERLs is crucial and requires timely and efficient tracking of these locations throughout the city. This work aims to identify and classify urban ERLs using GPS trajectory data of over 16,000 construction waste hauling trucks (CWHTs), as well as 58 urban features encompassing geographic, land cover, POI and transport dimensions. We compare several machine learning models and examine the impact of various spatial-temporal features on classification performance using real-world data in Chengdu, China. The results demonstrate that 77.8% classification accuracy can be achieved with a limited number of features. This classification framework was implemented in the Alpha MAPS system in Chengdu, which has successfully identified 724 construction cites/earth dumping ground, 48 concrete mixing stations, and 80 truck parking locations in the city during December 2023, which has enabled local authority to effectively manage urban dust pollution at low personnel costs.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is the effective management and classification of earth - moving - related locations (ERLs) in the city. Specifically, the paper focuses on how to use the GPS trajectory data of more than 16,000 construction waste hauling trucks (CWHTs), combined with 58 urban features (including geographical, land cover, points of interest, and traffic dimensions), to identify and classify earth - moving - related locations in the city, such as construction sites, earth - dumping sites, and concrete mixing plants. These locations are the main sources of urban dust pollution, so their effective management is crucial for improving urban air quality.
### Background and Objectives of the Paper
Urban dust pollution, especially particulate matter (PM), is one of the main sources of air pollution. With the acceleration of the urbanization process and the development of infrastructure, the impact of dust pollution on air quality is becoming more and more serious. Earth - moving - related locations (ERLs) are particularly important in urban dust management because they are closely related to the entire life cycle of earth - moving operations, from production (construction sites) to processing (concrete mixing plants), then to transportation (truck parking lots), and disposal (construction sites and earth - dumping sites). Traditional ERLs management methods rely on manual inspection and recording, which have problems such as information delay, inaccuracy, and incompleteness, and it is difficult to achieve effective management.
### Research Methods
In order to improve the management ability of urban ERLs, the paper proposes a machine - learning - based framework to identify and classify ERLs using urban big data. Specific methods include:
1. **Data Collection**: Use the GPS trajectory data of more than 16,000 construction waste hauling trucks, as well as 58 urban features including geographical, land cover, points of interest (POI), and traffic - related features.
2. **Feature Extraction**: Extract stop points from GPS trajectory data, and identify regularly - occurring stop - point clusters through a multi - layer clustering method. These clusters are mapped to small grids of 200m×200m to form a pixelated representation of ERLs.
3. **Model Selection**: Compare four machine - learning models (logistic regression, random forest, gradient - boosted decision trees, and multi - layer perceptrons), and evaluate the impact of various spatio - temporal features on classification performance.
4. **Performance Evaluation**: Use metrics such as accuracy, precision, recall, F1 - score, and AUROC to evaluate the performance of the model.
### Main Contributions
1. **High - Precision Classification**: The research results show that a classification accuracy of 77.8% can be achieved using a limited number of features.
2. **Feature Importance Analysis**: Extract the importance of features through the SHAP method and feature distribution, revealing the feature relationships between different ERLs categories and enhancing the interpretability of machine - learning results.
3. **Practical Application**: This classification algorithm has been successfully deployed in the Alpha MAPS system in Chengdu for intelligent air pollution monitoring and management. In December 2023, the system successfully identified 724 construction sites/earth - dumping sites, 48 concrete mixing plants, and 80 truck parking lots, significantly improving the local authorities' ability to monitor and manage locations with high dust - pollution risks.
### Conclusion
The paper provides an efficient and intelligent urban ERLs management method by combining big data and machine - learning technologies, which helps to improve urban air quality and reduce management costs. This research fills the research gap in the field of urban ERLs classification and has important practical application value.