Traffic Flow Prediction Using Dynamic Data Input Ranges
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing traffic information prediction systems using deep learning models face performance limitations due to the heuristic determination of data ranges, which are not optimized for specific prediction time points or road sections, leading to suboptimal prediction accuracy.
Innovation Solution
An apparatus and method that utilize a deep learning-based prediction model to determine an appropriate data input range for predicting traffic flow by analyzing past travel speed data, calculating relative errors, and subdividing road sections based on spatial characteristics, thereby optimizing data input ranges for improved prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a fixed heuristic data range is used for prediction, then the system operation is simple, but the prediction accuracy deteriorates
Solution Approach 1:
The patent implements dynamic data range determination by training a deep learning model to automatically select the optimal historical data range (e.g., 30 minutes, 60 minutes, 90 minutes, or 120 minutes) based on the specific prediction time point and traffic conditions, replacing the static heuristic approach with an adaptive system that adjusts the data range dynamically to maximize prediction accuracy
Solution Approach 2:
The patent changes the parameter of data range selection from a fixed heuristic value to a variable determined by the deep learning model, which evaluates multiple time range options and selects the optimal parameter (data range) for each prediction task based on learned patterns from historical traffic data
2Device complexity
If the same data range is used for all prediction time points, then the data processing is consistent, but the prediction performance is suboptimal
Solution Approach 1:
The patent applies local quality by allowing different data ranges to be used for different prediction time points and road sections. The deep learning model learns that certain time ranges are more appropriate for specific locations and times (e.g., morning rush hour vs. nighttime), enabling locally optimized prediction performance rather than a uniform approach across all conditions
3Measurement precision
If a deep learning model dynamically determines data range, then the prediction accuracy is improved, but the system complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-training the deep learning model offline using historical traffic data to learn the optimal data range selection strategy. Once trained, the model contains the learned knowledge and can quickly determine appropriate data ranges during operation without requiring complex real-time computations, thus improving accuracy while managing system complexity through advance preparation
Data Source
AI summary
An apparatus and a method for providing traffic information are provided. The apparatus includes a traffic data database (DB) that stores traffic data and a processor connected to the traffic data DB. The processor generates a prediction model to predict a traffic flow and determines an appropriate data input range for a target time point in the future using the prediction model. Additionally, the processor extracts past traffic data from the traffic data DB based on the determined appropriate data input range, predicts a traffic flow at the target time point based on the extracted past traffic data and provides the predicted traffic flow as traffic information.