"We initially relied on phone calls to handle citizen complaints about air pollution cases. However, air pollution incidents tend to dissipate quickly, and by the time inspection personnel arrive at the reported site, it can be challenging to gather relevant evidence. With the high volume of air pollution complaints from the public, we started exploring how to conduct environmental inspections more swiftly and effectively," explained Chang Chan-Chuan, Technical Specialist of the Environmental Inspection Division, the Department of Environmental Protection, Taoyuan City Government, highlighting the practical challenges of investigating air pollution cases.
In the early days, citizens primarily reported air pollution incidents via phone calls. These reports were then relayed by the monitoring center to inspection teams via phone notifications. However, the transient nature of such pollution events often made it difficult for inspection personnel to collect evidence upon arrival. In response, the Department of Environmental Protection began implementing an electronic dispatch system in 2014 to enhance inspection efficiency, enabling inspection teams nearest to the reported location to respond immediately. With advancements in Internet of Things (IoT) technology in recent years, since 2014, the Department has introduced various technological devices and software under an IoT framework to improve and integrate operations. These include handheld tablets for inspectors, GPS-enabled electronic dispatch systems, air quality sensors, meteorological instruments, a smart air quality management platform, and an AI smoke detection system. Together, these components form an IoT platform designed to strengthen air quality governance strategies in industrial zones.
AI Smoke Detection System Enhances Inspection Efficiency
“Through our collaboration with the Ministry of Environment, we have deployed 1,000 air quality sensors across 13 industrial zones in Taoyuan City and two neighboring community areas, significantly improving our ability to monitor and manage air quality. Last year, we also introduced AI technology, integrating surveillance cameras with sensors to enable AI-based smoke detection,” explained Hou Guan-Yu, Technical Specialist of the Department of Environmental Protection.
Since 2018, the Department of Environmental Protection has installed 1,000 air quality sensors across 13 industrial zones, two nearby community areas, and one hotspot identified by public complaints. These sensors provide real-time monitoring of air quality fluctuations. When sensors or the AI smoke detection system detect anomalies, inspection teams are dispatched immediately. Furthermore, proactive analysis of sensor data helps identify pollution hotspots, narrowing the scope for tracing pollution sources and improving inspection efficiency.
Between 2021 and 2023,Taoyuan's Environmental Protection Department facilitated 3,319 air pollution inspection cases with the support of these technologies, resulting in 137 businesses being fined a total of NT$21.752 million. Additionally, one business was fined NT$160,000 through AI smoke detection system findings. Complaints about air pollution in the 15 sensor-deployed areas decreased by approximately 33% from 2021 to 2023. Moreover, the integration of the IoT Air Quality Platform, AI smoke detection, and digital dispatch systems has enabled inspectors to reach pollution sites more efficiently, significantly reducing travel times and cutting carbon dioxide emissions.
Optimizing Monitoring Accuracy
“For instance, if a large vehicle idles near a sensor or someone smokes nearby, the sensor might immediately trigger an alert. In such cases, it becomes challenging to determine whether the pollution is caused by nearby factories or other sources, requiring manual efforts to trace the pollution origin,” explained Hou Guan-Yu, Technical Specialist, while discussing areas for improvement in the sensors’ practical applications.
Air quality sensors are often influenced by obvious nearby pollution sources such as vehicle idling, cooking fumes, or smoking. When confirmed as such cases, adjustments are made to optimize sensor placement through a rolling review process, ensuring their effectiveness in identifying pollution sources. AI-based image recognition also faces challenges from steam or background lighting, which can result in false alerts due to misidentified smoke. To improve accuracy, the system captures images for professional review, and feedback on actual pollution patterns is integrated into the system for further learning, enabling proactive alerts and real-time inspections.
Curtailing Pollution and Building a Livable City
“In September 2022, the southern Taoyuan region experienced an unidentified odor event affecting areas such as Yangmei, Longtan, Zhongli, and Pingzhen. Using air quality sensor data, we analyzed and narrowed down potential pollution zones. We then deployed drones to collect evidence and ultimately located the pollution source in the Longtan Rugu Mountain area. This success demonstrates how technological tools are instrumental in identifying pollution sources,” said Zhang Zhan-Chuan, Technical Specialist, sharing a case study on air pollution management and the broader vision of improving urban air quality.
With the rapid development of AI and IoT technologies, technological and smart management approaches have become mainstream compared to traditional inspection methods. Real-time monitoring of various environmental data in industrial zones, coupled with automated computation and analysis, allows a proactive rather than reactive response. This approach enables the early detection and mitigation of pollution incidents, encourages manufacturers to enhance their self-regulation mechanisms, prevents arbitrary pollutant emissions, and effectively addresses public complaints about pollution. Such strategies strengthen environmental governance, improve living conditions, and contribute to creating a progressive and livable city.