Short-term traffic flow prediction at isolated intersections based on parallel multi-task learning
This paper proposes a novel phase-based short-term traffic flow prediction method based on parallel multi-task learning for isolated intersections. Different from traditional short-term traffic flow prediction methods, we take the traffic flow of each phase as the minimum prediction unit, instead of...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Taylor & Francis Group
2024-12-01
|
| Series: | Systems Science & Control Engineering |
| Subjects: | |
| Online Access: | https://www.tandfonline.com/doi/10.1080/21642583.2024.2316160 |
| _version_ | 1797300140179980288 |
|---|---|
| author | Bao-Lin Ye Shiwei Zhu Lingxi Li Weimin Wu |
| author_facet | Bao-Lin Ye Shiwei Zhu Lingxi Li Weimin Wu |
| author_sort | Bao-Lin Ye |
| collection | DOAJ |
| description | This paper proposes a novel phase-based short-term traffic flow prediction method based on parallel multi-task learning for isolated intersections. Different from traditional short-term traffic flow prediction methods, we take the traffic flow of each phase as the minimum prediction unit, instead of directly utilising the traffic flow of a single lane with large random fluctuations. Meanwhile, we design a novel encoding and decoding structure whereby external influencing factors have been incorporated both into encoding and decoding operations. Furthermore, a fusion strategy is proposed to predict the traffic flow of each phase by integrating the traffic flows of lanes whose right of way are provided by the phase. In the fusion strategy, we develop a parallel multi-task prediction framework whereby a new loss function is defined to improve the prediction accuracy. Finally, the proposed method is tested with the traffic flow data collected from an intersection of South Changsheng Road located in the city of Jiaxing. The findings illustrate that the proposed method can achieve better prediction results at different sampling time scales, compared to the existing short-term traffic flow prediction methods. |
| first_indexed | 2024-03-07T23:01:54Z |
| format | Article |
| id | doaj.art-26e88d86c3b840cb82fd1d90d88dfed7 |
| institution | Directory Open Access Journal |
| issn | 2164-2583 |
| language | English |
| last_indexed | 2024-03-07T23:01:54Z |
| publishDate | 2024-12-01 |
| publisher | Taylor & Francis Group |
| record_format | Article |
| series | Systems Science & Control Engineering |
| spelling | doaj.art-26e88d86c3b840cb82fd1d90d88dfed72024-02-22T12:26:57ZengTaylor & Francis GroupSystems Science & Control Engineering2164-25832024-12-0112110.1080/21642583.2024.2316160Short-term traffic flow prediction at isolated intersections based on parallel multi-task learningBao-Lin Ye0Shiwei Zhu1Lingxi Li2Weimin Wu3School of Information Science and Engineering, Jiaxing University, Jiaxing, Zhejiang, People's Republic of ChinaSchool of Information Science and Engineering, Jiaxing University, Jiaxing, Zhejiang, People's Republic of ChinaDepartment of Electrical and Computer Engineering, Purdue School of Engineering and Technology, Indiana University-Purdue University Indianapolis, Indianapolis, IN, USAInstitute of Cyber-Systems and Control, Zhejiang University, Hangzhou, People's Republic of ChinaThis paper proposes a novel phase-based short-term traffic flow prediction method based on parallel multi-task learning for isolated intersections. Different from traditional short-term traffic flow prediction methods, we take the traffic flow of each phase as the minimum prediction unit, instead of directly utilising the traffic flow of a single lane with large random fluctuations. Meanwhile, we design a novel encoding and decoding structure whereby external influencing factors have been incorporated both into encoding and decoding operations. Furthermore, a fusion strategy is proposed to predict the traffic flow of each phase by integrating the traffic flows of lanes whose right of way are provided by the phase. In the fusion strategy, we develop a parallel multi-task prediction framework whereby a new loss function is defined to improve the prediction accuracy. Finally, the proposed method is tested with the traffic flow data collected from an intersection of South Changsheng Road located in the city of Jiaxing. The findings illustrate that the proposed method can achieve better prediction results at different sampling time scales, compared to the existing short-term traffic flow prediction methods.https://www.tandfonline.com/doi/10.1080/21642583.2024.2316160Short-term traffic flow predictionencoding and decoding structuremulti-step predictionmulti-task learning |
| spellingShingle | Bao-Lin Ye Shiwei Zhu Lingxi Li Weimin Wu Short-term traffic flow prediction at isolated intersections based on parallel multi-task learning Systems Science & Control Engineering Short-term traffic flow prediction encoding and decoding structure multi-step prediction multi-task learning |
| title | Short-term traffic flow prediction at isolated intersections based on parallel multi-task learning |
| title_full | Short-term traffic flow prediction at isolated intersections based on parallel multi-task learning |
| title_fullStr | Short-term traffic flow prediction at isolated intersections based on parallel multi-task learning |
| title_full_unstemmed | Short-term traffic flow prediction at isolated intersections based on parallel multi-task learning |
| title_short | Short-term traffic flow prediction at isolated intersections based on parallel multi-task learning |
| title_sort | short term traffic flow prediction at isolated intersections based on parallel multi task learning |
| topic | Short-term traffic flow prediction encoding and decoding structure multi-step prediction multi-task learning |
| url | https://www.tandfonline.com/doi/10.1080/21642583.2024.2316160 |
| work_keys_str_mv | AT baolinye shorttermtrafficflowpredictionatisolatedintersectionsbasedonparallelmultitasklearning AT shiweizhu shorttermtrafficflowpredictionatisolatedintersectionsbasedonparallelmultitasklearning AT lingxili shorttermtrafficflowpredictionatisolatedintersectionsbasedonparallelmultitasklearning AT weiminwu shorttermtrafficflowpredictionatisolatedintersectionsbasedonparallelmultitasklearning |