Trend Based Predictive Traffic Using Weighted Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing navigation systems fail to accurately predict traffic speeds in the short term future due to their reliance on historical data that does not account for unanticipated, unique, and temporary events or conditions, such as traffic accidents or construction, which can significantly impact road conditions.
Innovation Solution
A trend-based extrapolation methodology that combines real-time and historical traffic speed data over a specific evaluation window to predict future traffic speeds, accounting for the impact and dissipative nature of such events, using weighted averages and exponential decay functions to adjust predictions accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If historical traffic pattern data is used to predict future traffic speeds, then the prediction covers a broad range of recurring conditions, but it fails to account for unanticipated, unique, and temporary events such as traffic accidents or construction
Solution Approach 1:
The patent merges historical traffic pattern data with real-time traffic speed data to create a hybrid prediction system. The historical data provides the baseline expected speed for recurring conditions, while real-time data captures unexpected events. By combining these two data sources through a blending mechanism, the system achieves both broad coverage of normal conditions and accuracy during unexpected events.
Solution Approach 2:
The patent introduces real-time traffic speed data as an intermediary element between the historical traffic patterns and the final prediction. This intermediary captures current unexpected conditions (accidents, construction, weather) and mediates the prediction process by adjusting the historical baseline to reflect current reality, thereby resolving the contradiction between relying on historical patterns and adapting to unexpected events.
2Reliability
If real-time traffic data is used alone for prediction, then unexpected events are captured, but the prediction lacks the context of normal traffic patterns and cyclical variations
Solution Approach 1:
The system merges real-time traffic data with historical traffic patterns, where each data source compensates for the other's weaknesses. Real-time data provides responsiveness to current conditions, while historical patterns provide context for cyclical variations. The combination ensures that predictions are both responsive to unexpected events and adaptable to normal traffic cycles.
Solution Approach 2:
The patent uses historical traffic patterns to establish a baseline prediction before real-time data is fully processed. This preliminary action based on historical patterns provides the contextual framework of normal traffic behavior, which then serves as a reference point for adjusting predictions based on real-time unexpected events.
3Ease of manufacture
If a simple average of historical speeds is used, then the calculation is straightforward, but it does not account for the dissipative nature of traffic events over time
Solution Approach 1:
The patent changes the parameter of the averaging calculation by introducing time-based weighting. Instead of a simple uniform average, the system applies exponential decay weights that reflect the dissipative nature of traffic events over time. More recent observations are weighted more heavily, while older observations are weighted less, accurately capturing how traffic conditions evolve and return to normal after disturbances.
Solution Approach 2:
The patent transforms the static simple average into a dynamic weighted average that adapts to the temporal characteristics of traffic events. The weighting scheme dynamically adjusts the contribution of each historical observation based on its age, reflecting the reality that traffic events have a dissipative effect over time. This dynamic approach maintains computational simplicity while significantly improving prediction accuracy.
Data Source
AI summary
A method for predicting traffic wherein the method is a trend based extrapolation method that uses real time traffic data and historic traffic data to generate a predictive traffic product. The predictive traffic product provides expected traffic speeds for the short term future, for example, between two to twelve hours into the future.