Abstract
This study proposes an integrated framework to create a more comfortable, safe, and healthy urban public health environment. By leveraging intelligent multi-objective optimization, the system provides theoretical and practical support for environmental health management in smart cities.
Introduction
As urban populations expand, traditional public health and sanitation management methods struggle to keep pace. Current systems face significant bottlenecks: slow, manual environmental monitoring; labor-intensive waste management; and static traffic control that exacerbates air pollution. This research introduces a multi-objective system that integrates real-time IoT monitoring, AI-driven waste sorting, and intelligent traffic management to improve urban resident well-being.
System Architecture
The proposed framework utilizes a three-tiered architecture to ensure seamless data flow and actionable insights, as illustrated in the overall design framework.
- Perception Layer: A network of IoT sensors collecting data on water quality (pH, turbidity, dissolved oxygen), air quality (PM2.5, CO, CO2), and noise levels.
- Transport Layer: Utilizes 5G modules and high-performance processors to transmit and process data in real-time.
- Application Layer: Provides an interface for end-users via the Internet, accessible through computers and mobile devices.
Environmental Monitoring and Node Design
The core of the Environmental Monitoring System (MS) lies in its sensor nodes. Each node is designed to handle multiple data streams simultaneously, integrating various collection circuits with a central processing module.
To ensure data reliability, the system employs the Kalman Filter (KF) algorithm. By recursively fusing previous state estimates with new measurements, the KF effectively denoises sensor data. This algorithmic approach proved critical; for instance, water turbidity detection error dropped from over 10% to less than 1% when the Kalman Filter was applied.
AI-Driven Waste Sorting and Traffic Management
Beyond monitoring, the system actively intervenes in urban sanitation and logistics:
- Waste Recycling: A Single Shot MultiBox Detector (SSD) model, trained on the SpotGarbage dataset, identifies recyclables like paper, metal, and plastic. This allows for automated robotic sorting, significantly increasing efficiency over manual methods.
- Intelligent Traffic Control: By analyzing real-time flow data, the system optimizes signal timing to reduce "stop-and-go" idling. This intervention prioritizes public transport and emergency vehicles, viewing traffic management as a direct public health tool to reduce tailpipe emissions.
Key Results and Impact
The integrated system demonstrated high performance across all objectives:
- Detection Accuracy: Achieved over 95% accuracy for environmental data, with relative errors as low as 1.2% for air quality and 1.4% for water quality.
- Waste Identification: The SSD model reached 95.3% accuracy for metal and 93.4% for paper. While plastic was more challenging (83.6%) due to irregular shapes, it remained viable for automated processes.
- Urban Health: Intelligent traffic management led to a 20% reduction in congestion, translating to an estimated annual reduction of 20 tonnes of tailpipe emissions in the study area.
Conclusion
The proposed intelligent multi-objective system offers a robust solution for modern urban health management. While future work will focus on refining algorithms for complex materials and improving sensor durability, this framework establishes a scalable model for healthier, smarter cities.
Summary updated