文档介绍：该【基于FCM-SVM方法的时空航路网交通状态识别研究 】是由【niuwk】上传分享，文档一共【3】页，该文档可以免费在线阅读，需要了解更多关于【基于FCM-SVM方法的时空航路网交通状态识别研究 】的内容，可以使用淘豆网的站内搜索功能，选择自己适合的文档，以下文字是截取该文章内的部分文字，如需要获得完整电子版，请下载此文档到您的设备，方便您编辑和打印。基于FCM-SVM方法的时空航路网交通状态识别研究
Abstract
Traffic state identification is an important task in transportation management and control. Traditional methods mainly rely on manual inspection or sensor-based data acquisition, which are not cost-effective and efficient for large-scale road networks. In recent years, artificial intelligence and machine learning techniques have been utilized to improve the accuracy and efficiency of traffic state identification. In this paper, we propose a FCM-SVM method for traffic state identification in spatiotemporal air route networks. The proposed method is evaluated using real-world air traffic data, and the results demonstrate its effectiveness and robustness.
1. Introduction
Traffic state identification is a crucial task in transportation management and control. The accurate and timely identification of traffic states can help transportation authorities make informed decisions, such as rerouting traffic, adjusting signal timings, and managing incidents. Traditionally, traffic state identification relies on manual inspection or sensor-based data acquisition. However, these methods are costly and inefficient for large-scale road networks. In recent years, artificial intelligence and machine learning techniques have been utilized to improve the accuracy and efficiency of traffic state identification.
Air transportation is an important mode of transportation that has a significant impact on the global economy. The spatiotemporal air route network is complex and dynamic, which makes traffic state identification a challenging task. In this paper, we propose a FCM-SVM method for traffic state identification in spatiotemporal air route networks. The proposed method combines fuzzy c-means (FCM) clustering and support vector machine (SVM) classification to identify traffic states.
2. Related Work
Several studies have proposed machine learning methods for traffic state identification in road networks. Dey et al. (2019) proposed a method based on long short-term memory (LSTM) networks for traffic state identification. Wu et al. (2019) proposed a cardiovascular neural network for traffic state identification. Ren et al. (2020) proposed an adversarial spatial-temporal multi-task learning method for traffic state prediction. These methods have achieved promising results in various road networks.
However, few studies have focused on traffic state identification in spatiotemporal air route networks. Sun et al. (2016) proposed a Bayesian network-based approach for air traffic flow state prediction. Liu et al. (2019) proposed a deep embedding model for air traffic flow state prediction. These methods mainly focus on the prediction of future traffic states in air route networks, but do not directly identify the current traffic state. Therefore, a method for traffic state identification in spatiotemporal air route networks is still needed.
3. Methodology
The proposed FCM-SVM method for traffic state identification consists of two components: FCM clustering and SVM classification. The FCM algorithm is used to cluster the air routes based on the current traffic state, while the SVM algorithm is applied to classify the traffic state of each air route.
The FCM algorithm is a widely used fuzzy clustering method that assigns data points to clusters based on fuzzy membership functions. The algorithm iteratively updates the membership functions and the cluster centers until convergence. In the proposed method, the FCM algorithm is applied to cluster the air routes based on the current traffic state, which is represented by the following features:
- Traffic flow: the number of planes passing through the air route within a certain time interval.
- Airspace capacity: the maximum number of planes that can safely pass through the air route within the same time interval.
- Weather condition: the weather condition in the air route, which affects the safety and efficiency of air traffic.
After the air routes are clustered by FCM, the SVM algorithm is applied to classify the traffic state of each air route. The SVM algorithm is a widely used machine learning method for classification tasks. It constructs a hyperplane to separate the data points of different classes in feature space. In the proposed method, the SVM algorithm is applied to classify each air route into one of the following traffic states:
- Congestion: the traffic flow exceeds the airspace capacity, which may cause delays and safety issues.
- Moderately busy: the traffic flow is close to the airspace capacity, but within a safe range.
- Lightly busy: the traffic flow is less than the airspace capacity, indicating a safe and efficient traffic state.
4. Experimental Evaluation
The proposed FCM-SVM method is evaluated using real-world air traffic data from the Federal Aviation Administration (FAA). The data consists of air traffic flow information and weather conditions in the United States from January 1st, 2020 to December 31st, 2020.
To evaluate the performance of the proposed method, we compare it with two baseline methods: FCM clustering only, and SVM classification only. For the FCM clustering only method, the air routes are clustered into three groups based on traffic flow, airspace capacity, and weather conditions. For the SVM classification only method, the traffic state of each air route is classified into one of the three categories directly based on the same features.
The results show that the proposed FCM-SVM method achieves the highest accuracy, precision, and recall in traffic state identification among the three methods. The accuracy of the FCM-SVM method is %, while the FCM clustering only method and SVM classification only method achieve accuracies of % and %, respectively. The FCM-SVM method also achieves higher precision and recall values for each traffic state category.
5. Conclusion
In this paper, we propose a FCM-SVM method for traffic state identification in spatiotemporal air route networks. The proposed method combines fuzzy c-means (FCM) clustering and support vector machine (SVM) classification to identify traffic states based on traffic flow, airspace capacity, and weather conditions. The experimental results show that the proposed method achieves higher accuracy, precision, and recall in traffic state identification compared to the baseline methods. Future work will focus on the integration of real-time data acquisition and the deployment of the proposed method in actual air traffic management systems.