moss-benchmark
收藏资源简介:
该数据集用于NeurIPS 2024的交通系统优化基准测试,包含多种类型的交通数据,如拥堵、动态、性能、道路、潮汐和交通信号控制数据。
This dataset is developed for traffic system optimization benchmarking at NeurIPS 2024, and it encompasses various types of traffic data, including congestion data, dynamic traffic data, performance data, road network data, tidal traffic data, and traffic signal control data.
数据集概述
数据集结构
数据集包含以下文件夹:
congestiondynamicperfroadtidaltsc
数据集下载与解压
- 下载数据集压缩包
data.tar.gz并解压: bash tar xvf data.tar.gz
数据集内容
性能测试
- 文件夹:
./Perf - 结果保存路径:
./log/${city}/${algo}_${n}.json
交通信号控制
- 文件夹:
./TSC - 结果保存路径:
./log/${algo}/${city}/${date}/info.log
动态车道分配
- 文件夹:
./DynamicLane - 结果保存路径:
./log/${algo}/${city}/${date}/info.log
潮汐车道控制
- 文件夹:
./TidalLane - 结果保存路径:
./log/${algo}/${city}/${date}/info.log
拥堵定价
- 文件夹:
./CongestionPricing - 结果保存路径:
./log/${algo}/${city}/${date}/info.log
道路规划
- 文件夹:
./CongestionPricing
数据准备
-
获取2019年的道路网络数据: bash python preparation/fetch_2019_geojsons.py
-
找出2019年和2024年道路网络的差异: bash for city in beijing shanghai newyork paris; do python preparation/fetch_candidate_way_ids.py --candidate_way_path ./data/${city}_candidate_ways.pkl --city $city done
-
选择优化道路: bash for city in beijing shanghai newyork paris; do for condition in smooth normal congested; do python exp/select_50_optimize_way_ids.py --opt_way_path ./data/${condition}${city}opt_ways.pkl --city $city --map_path ./data/moss.map_china${city}.pb --trip_path ./data/${condition}${city}_trip.pb --candidate_way_path ./data/${city}_candidate_ways.pkl done done
运行优化方法
- 结果保存路径:
./log/${algo}/${city}/${date}/info.log




