Traffic State Clustering for More Accurate Road Link Speed Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing traffic prediction technologies rely on historical speed patterns, which are inaccurate due to variations in weather and season, and lack precision in micro-level speed prediction for road links, leading to degraded real-time traffic information accuracy.
Innovation Solution
An apparatus and method using a K-means clustering algorithm to classify traffic states into stability-maintained, added congestion, and smoothly recovered states, correcting representative speeds based on probe vehicle data to enhance prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If traffic information is predicted based on historical speed patterns for the same period, then the prediction process is simple, but the accuracy is degraded due to weather and seasonal variations
Solution Approach 1:
The patent segments traffic prediction into two distinct components: (1) baseline prediction using historical speed patterns for simplicity, and (2) correction values generated by machine learning models for accuracy. This segmentation allows each component to optimize for its specific strength while combining to resolve the overall contradiction.
Solution Approach 2:
The patent changes the parameter of traffic information from raw historical speeds to corrected speeds that incorporate multiple factors including weather conditions, seasonal variations, and real-time probe data. This parameter transformation maintains the simplicity of historical baseline while improving accuracy through additional corrective parameters.
2Ease of operation
If representative speed is provided without correction, then the system operation is simple, but the real-time traffic information accuracy is significantly degraded
Solution Approach 1:
The patent introduces correction values as an intermediary element between historical speed data and final traffic information delivery. These correction values act as a mediator that adjusts the baseline prediction without requiring complete system redesign, thus maintaining operational simplicity while improving accuracy.
Solution Approach 2:
The patent performs preliminary correction of traffic information by pre-calculating correction values based on probe vehicle data and storing them for later application. This preliminary action ensures accuracy is improved before final delivery without adding complexity to the real-time operation phase.
3Area of stationary object
If probe vehicle data is used to predict traffic congestion time macroscopically, then the prediction scope is broad, but the micro-level speed prediction for each link is limited due to insufficient probe samples
Solution Approach 1:
The patent merges macroscopic congestion time prediction with micro-level link speed prediction by combining probe vehicle data with correction models for each road link. This merging allows the system to leverage the broad coverage of macro prediction while enhancing it with precise micro-level corrections where probe data is available.
Data Source
AI summary
An apparatus and a method for predicting traffic information are provided. The apparatus includes a storage to store a model for correcting traffic information for each traffic state in a road section, and a controller that determines the traffic state in the road section to be predicted based on K-means clustering algorithm, obtains a correcting value by using a model for correcting traffic information corresponding to the traffic state in the road section to be predicted, and corrects traffic information based on the obtained correcting value to predict real-time traffic information with higher accuracy.