| 英文摘要 |
This project assisted the Environmental Protection Administration (EPA) in planning a new cloud audit and certification system for recycling waste vehicles, as well as completed the third-generation system, CAR 3.0, with significant adjustments. To implement the smart system for waste vehicle recycling and improve the cloud audit and certification function, this project completed several tasks, namely providing on-site counseling for recyclers, organizing system introduction seminars, adjusting the manual of cloud audit function, integrating new and existed functions, planning one-stop services for waste vehicles, developing a system for automatically recognizing vehicle identification numbers (VINs), providing system maintenance services, and promoting the new system.
To help users adapt to the changes, this project introduced the system by using supporting materials such as promotion leaflets, system demo clips, and manuals. Six seminars were held for recyclers to introduce and demonstrate the system. In addition, 148 on-site counseling sessions were organized (for a total of 202 recyclers) to show new system functions and provide answers to frequently asked questions and concerns to ensure that users could adapt to the new system.
In response to the revision of the Cloud Audit and Certification Handbook, this project simultaneously adjusted and integrated the new and existed system functions and took the initiative to propose system expansion plans to enhance system efficiency and friendliness. In addition, in line with the Smart Government 2.0 Program proposed by the Executive Yuan, this project assisted the EPA in planning and improving inter-ministerial communication. MyData was introduced to the current waste vehicle recycling in advance to plan and implement the digital one-stop waste vehicle recycling service.
The current method of recognizing VINs of waste vehicles by eye is labor-intensive and time-consuming. Therefore, this project tested various optical character recognition (OCR) software and algorithms (both commercial and open-sourced) to introduce appropriate AI auto-recognition software. This project also conducted automated training to identify the VINs in the photos stored in the cloud audit system and compared them with those in the EPA regulatory system. After collecting nearly 10,500 engine photos for self-built model training and learning, the identification precision reached 89.1%. In the future, this project will continue to provide training materials to improve the model and recognition rate to implement smart cloud audit.
The official launch of CAR 3.0 is expected to be in 2021. To smoothly convert the system and database, reduce the influence on users, and maintain system stability, this project voluntarily maintained the previous system. This project adopted numerous measures to improve the system, including stress testing, information security strategies, sustainable operation plans, system service standards, and customer service human resources. In addition, the new system will affect people’s habit of car disposal. This project will promote the new system through multiple media channels, such as posters, articles, e-newsletters, and videos, to increase public recognition and then enhance their cooperation with the new system.
Significant reduction in operating time and labor costs is expected after the new system implementation. For the recycling industry, the new system will save 23% labor for audit and certification, increase 80% of the utilization rate of engine storage space, and save 2.26 million sheets of paper per year. For the audit and certification groups, the system will save 35 hours of travel time per month and complete the audit and certification one month in advance. For the public, the system will shorten the application time for subsidies by three to five days and provide instant notification and information disclosure through instant message systems. For the EPA, the system will streamline audit and certification costs and enhance project quality monitoring.
|