Machine Learning Model for Road Network Vehicle Flow Prediction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for determining vehicle traffic flow in road networks are costly, require extensive data collection, and are not suitable for predicting rapid changes, especially on sections without measurement coverage, limiting their accuracy and applicability to large networks.

Innovation Solution

A method using macroscopic data and machine learning techniques to build models for maximum and daily vehicle flow, allowing for accurate prediction on any road section, including those without sensors, by constructing models from measurements and macroscopic data, and applying them to the entire network.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If travel modeling tools are implemented to plan and simulate traffic regulations, then traffic flow prediction capability is improved, but implementation complexity and cost increase significantly

Engineering Contradiction:
Improvetraffic flow prediction accuracyVSAvoidmodel implementation complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent replaces expensive, complex travel modeling tools with a simpler machine learning model that uses readily available macroscopic data. The model is trained once and then applied repeatedly to different road segments without requiring continuous expensive data collection or complex recalibration, effectively using a 'cheap' alternative to replace the 'expensive' traditional approach.

Inventive Principle:
Principle #27Cheap short-living objects (Disposable)

Solution Approach 2:

The patent changes the input parameters from requiring detailed population survey data and extensive calibration parameters to using only macroscopic data such as traffic counts and basic road characteristics. This parameter simplification dramatically reduces implementation complexity while maintaining prediction capability.

Inventive Principle:
Principle #35Parameter changes

2Reliability

If population survey data is collected to improve model accuracy, then prediction reliability is improved, but data collection cost and time increase

Engineering Contradiction:
Improveflow rate determination accuracyVSAvoiddata collection time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent extracts only the essential macroscopic data elements needed for flow rate prediction, specifically traffic count data and basic road characteristics, while eliminating the need for comprehensive population survey data. This extraction approach maintains prediction accuracy by focusing on the most relevant parameters while reducing data collection burden.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent uses pre-existing macroscopic data that is already collected for other purposes (traffic monitoring, road management) rather than conducting new population surveys. The machine learning model is trained in advance on this readily available data, eliminating the need for time-consuming data collection campaigns.

Inventive Principle:
Principle #10Preliminary action

3Measurement precision

If measurements are taken on all road segments to improve coverage accuracy, then measurement completeness is improved, but sensor deployment cost increases

Engineering Contradiction:
Improveflow rate measurement accuracyVSAvoidsensor quantity required
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent creates a universal machine learning model that can predict flow rates on any road segment using only macroscopic data, making the measurement system universally applicable across the entire road network without requiring segment-specific sensors. The model performs the function of multiple sensors simultaneously by inferring unmeasured flows from measured ones through spatial correlations.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent introduces a machine learning model as an intermediary that translates available macroscopic measurements into flow rate estimates for unmeasured segments. This intermediary enables the system to achieve complete network coverage using only a subset of actual sensors, mediating between limited measurements and comprehensive prediction needs.

Inventive Principle:
Principle #24Intermediary (Mediator)

4Reliability

If continuous measurement updating is performed to maintain data currentness, then data up-to-date status is improved, but computational load increases

Engineering Contradiction:
Improvedata currentnessVSAvoidcomputational energy consumption
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

The patent performs the computationally intensive model training in advance using historical data, then applies the trained model to new data with minimal computational effort. This preliminary action separates the heavy computational burden from the continuous updating process, allowing frequent predictions with low energy consumption.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent implements a dynamic prediction approach where the model adapts to changing traffic patterns by incorporating recent measurements, but only updates predictions when necessary rather than continuously recalculating. This dynamic updating maintains data currentness while avoiding unnecessary computational energy expenditure.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP4181100A1Method for determining a maximum throughput and/or a daily throughput of vehicles on a road network
Publication Date: 2023.05.17 IFP ENERGIES NOUVELLES
  • EP4181100A1 patent drawingFigure 1~3
  • EP4181100A1 patent drawingFigure 4~5
  • EP4181100A1 patent drawingFigure 6~7

AI summary

The present invention relates to a method for determining the maximum vehicle flow rate on at least one segment of a road network. For this method, flow rate measurements are taken, using at least one fixed sensor, at at least one measurement point on a training road network. A maximum vehicle flow rate model is then built, using machine learning, based on macroscopic data from the training road network and the measurements. This model is then applied to the segment of the road network under consideration.