The project includes data quality management and maintenance of 300 air quality sensors in Chiayi County. The project includes understanding the status of sensor equipment through clustering, decay, dynamic calibration, and operations, also using air quality sensor data to integrate with air quality information on the Internet of Things to analyze the county's air pollution hotspots and identify suspected polluters. The results of the work performed from February 24, 2021, to December 31, 2022 are as follows:
1. Optimal point planning for air quality sensors
There are a total of 300 air quality sensors in Chiayi county, and according to the application situation, the 2021 EPA deployment principles, and other factors, the proposed sites would be used and relocated for planning operations, which could be divided into industrial areas, auxiliary areas and communities, with 112 units in industrial areas, 55 units in auxiliary areas, and 133 units in communities, to effectively grasp the situation of suspected pollution sources.
2. Work plan and plan for "Refinement of sensor data management and inspection and maintenance operations”
The plan was submitted in 12 copies on March 22, 2021. The plan for "Refining Sensing Data Management and Inspection and Maintenance Operations" was submitted to the Environmental Protection Bureau for review and completion within one month after the approval of the work plan.
3. 300 sensors inspection (on-site visual inspection)
The project team should conduct an on-site visual inspection of each sensor at least once every 6 months, including the appearance of the sensor, internal structure, power supply, the environment around the sensor and whether there were new sources of pollution emissions, etc., and established a video and text record sheet for reference. Currently, we had completed the annual inspection in 2021 and 2022, all equipment components were in normal operation, and there were no new sources of pollution emissions around the sensor.
4. Sensor inspection and comparison operation (more than 10% of all sensors sampled)
The project team sampled at least 10% of all sensors every three months and conducted on-site inspections and compare them with the national monitoring station in parallel or high-level reference equipment to further strengthen the sensor performance level. As of June 30, 2022, the second quarter of 2021 (29 out of 30 sampled sensors passed 100%, 1 unit failed to reach 90%) and the third quarter of 2021 (25 out of 30 sampled sensors passed 100%, 5 sensors failed to reach 90%) had been completed, and the sensors with integrity rate less than 90% will be calibrated to restore their integrity rate to more than 90%. From the fourth quarter of 2021 to the fourth quarter of 2022, the comparison operation achieved a completeness rate of 98% or more, indicating that the quality of data completeness had been improved after sensor calibration.
5. Sensor cluster analysis
The project conducts monthly sensor cluster analysis for 8 clusters: Minxiong industrial park, Touqiao industrial park, Puzi industrial park, Jiatai industrial park, Yi-jhu industrial park, Daipumei industrial park, community, and auxiliary areas to screen out sensors with suspected equipment or sensor deviations, and conducted on-site inspection and comparison to evaluate and clarify the deviation behavior of the suspected abnormal sensor equipment. From October 2021 to December 31, 2022, a total of 44 outlier sensors were verified, 6 were abnormal and were replaced.
6. Sensor decay analysis operation
The project implemented the monthly sensor decay analysis, and the long-term trend analysis of the relative error (Error) and the coefficient of determination (R2) would be applied to the sensors installed on the national monitoring station within 1 km. The error of Bias (%) was larger in 2021 and 2022 due to good air quality from June to August.
7. Dynamic calibration of sensors
The project team established a multivariate calibration model through a long-term parallel comparison between the sensor and the reference instrument (PM2.5 monitoring equipment at the national monitoring station) so that the amount of deviation between the sensor and the monitoring station data meets the application classification performance requirements. The deviation of the concentration was determined based on the three environmental concentration ranges from June 2021 to December 2022, and the passing rate was 99% on average, indicating the accuracy of the project with dynamic calibration operations.
8. Sensor failure repair and removal of abnormal conditions
The project team of this project would set up a standard inspection and repair process for abnormal conditions of sensor equipment. Firstly, the initial abnormality would be determined by the remote processing mechanism of the air quality sensor, and if abnormal conditions still occurred, staff would be sent to the site for inspection and a trouble-shooting record sheet would be established for reference. The total number of abnormalities from October 2021 to December 2022 was 129, and the highest number of abnormalities was 87 for the power supply, instead of other factors (no electricity for the lamp pole and external damage to the sensor), all sensors were functioning normally.
9. Data analysis of sensors
This project performed a quarterly analysis of sensor data and provided the EPA with the results of the analysis of pollution hotspots, pollution periods (including 24-hour, weekly frequency, etc.), or related air quality model simulation applications. The analysis report had been completed from the first quarter of 2021 to the fourth quarter of 2022. The Air Quality Center analyzed the sensor hotspots, pollution hotspots, and the number of alarms. Arranged a visit to Minxiong industrial park, Touqiao industrial park, and Jiatai industrial park and used handheld direct reading instruments to measure and locate several pollution sources on site.
10. Promotion of air quality and air pollution
The public was increasingly concerned about the air quality and the sources of pollution. The promotions could increase the awareness of sensors. From August 2021 to September 2022, the project held 8 air quality and pollution promotions with a total of 386 participants. Through an easy-to-understand approach to enhance technical knowledge and conveyed the knowledge of technology applications and the government used the smart grid to combat pollution crimes and improve environmental quality.
11. Equipment maintenance and data management information
By following the EPA's regulations, we have submitted one copy each of the "2021 first-half equipment maintenance and data management information", "2021 second-half equipment maintenance and data management information", "2022 first-half equipment maintenance and data management information" and "2022 second-half equipment maintenance and data management information" by July 30, 2022, and updated the reported data to comply with the subsequent EPA operations.
12. Education training
To enhance the professional knowledge of the staff, and to propose a further improvement for each work content according to the actual implementation of the current situation. This project could be smoothly promoted to achieve the relevant business assigned by the EPA as scheduled, the project team had completed a total of 4 training sessions for programmers in 2021 and 2022 as of September 30, 2022.
13. Air quality micro-sensors pass safety standards
According to the EPA standard, the air quality micro-sensing device had passed the Taiwan Association of Information and Communication Standard Safety Class II standard by December 22, 2021.
14. Effective data completeness of IoT platforms on sensors (over 90%)
The project team was working on the sensor data uploading completeness rate, due to the air sensor network transmission outage for about 15 days from February to March, so the sensor data in the first quarter of 2021 did not reach 90%. The valid data completeness rate for 2021 was 90.3%, and the valid data completeness rate for January to December 2022 was 97.1%, which was in line with the contract specifications.
15. Air Quality Sensor IoT Inspection and sanctions
The project team of this project used the list of suspected pollution sources provided by the sensor data analysis report or found the compilation and notification of pollution emissions by on-site inspection. After the relevant control programs inspected the on-site situation, they calculated a total of 31 cases of sanctions, including open burning 16 cases, 9 cases of factory violations, and 6 cases of construction site dust, the total amount of sanctions was NT$ 3,259,540 until December 31, 2022.