Graph-Based Transit Time Prediction for Multi-Modal Shipments
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional multimodal transportation systems are inadequate for predicting shipment arrival times due to complexity, high costs, data quality issues, and inflexibility, especially when dealing with missing data and unexpected disruptions in multi-modal shipping scenarios involving multiple modes of transportation like trucks, trains, and ships.
Innovation Solution
A computer-implemented method using machine learning to construct and train a graph that predicts target outcomes for cargo shipments by processing historical transit data, incorporating location nodes with mode inputs and outputs, and receiving origin and destination parameters to determine estimated time of arrival and transit times across all possible shipment modes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional TMS and route optimization algorithms are used to plan multi-modal shipments, then real-time visibility and routing decisions can be made, but the systems cannot accurately predict transit times when route information is missing or incomplete
Solution Approach 1:
The system performs preliminary action by pre-computing and storing transit time estimates for various route combinations in a database before actual shipments occur. When a new shipment is processed, the system queries this pre-computed data to quickly determine transit times without needing to analyze complete route information in real-time, thus handling missing data scenarios effectively.
Solution Approach 2:
The patent introduces an intermediary approach by using a database of pre-computed transit time estimates as a mediator between the shipment data and the final transit time determination. This intermediary layer allows the system to make accurate predictions even when complete route information is missing, as the pre-computed database fills in the gaps with historical estimates.
2Productivity
If multiple modes of transportation are coordinated to optimize cost and transit time, then shipping efficiency improves, but the complexity of coordinating different modes increases
Solution Approach 1:
The patent applies segmentation by dividing the multi-modal shipping problem into separate, manageable segments. Each transportation mode (truck, train, ship, air) is handled as an independent segment with its own transit time estimates stored in the database. The system then combines these segmented estimates to determine overall transit times, reducing the complexity of coordinating multiple modes while maintaining optimization benefits.
3Measurement precision
If historical transit data is processed to train machine learning models, then transit time predictions can be made for complex multi-modal scenarios, but the cost and complexity of implementing such systems increases
Solution Approach 1:
The patent uses copying by creating a simplified representation of complex multi-modal shipping scenarios through pre-computed transit time estimates stored in a database. Instead of implementing complex machine learning models that process historical data in real-time, the system copies and stores key transit time information from historical data, allowing accurate predictions without the full complexity of advanced ML infrastructure.
Data Source
AI summary
A computer-implemented method includes constructing a plurality of nodes; adding the nodes to a graph; receiving origin, destination and shipment parameters; and processing the parameters to determine target outcomes corresponding to a cargo shipment. A computing system includes a processor; and a memory having stored thereon computer-executable instructions that, when executed by the processor, cause the computing system to: construct a plurality of nodes; add the nodes to a graph; receive origin, destination and shipment parameters; and process the parameters to determine target outcomes corresponding to a cargo shipment. A computer-readable medium includes instructions that, when executed by a processor, cause a computer to: constructing a plurality of nodes construct a plurality of nodes; add the nodes to a graph; receive origin, destination and shipment parameters; and process the parameters to determine target outcomes corresponding to a cargo shipment. The target outcomes may include time, emissions and/or costs.