Fleet Arrival Prediction Using Global and Local Learning Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for predicting vehicle arrival times and managing fleets are not adequately accurate due to reliance on generalized assumptions, even when using real-time data, which fail to account for individual vehicle and environmental variations.
Innovation Solution
A system utilizing a combination of global and local self-learning models, based on reinforcement learning, to estimate schedule parameters like arrival times by integrating vehicle and environmental data, with continuous model updating and validation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If static methods based on basic estimations are used, then the system is simple to operate, but the prediction accuracy of arrival times is insufficient
Solution Approach 1:
The prediction system is segmented into multiple specialized models: a first predictive model for estimating duration of vehicle activities, a second predictive model for estimating arrival times at locations, and a third predictive model for estimating energy consumption. Each model focuses on specific parameters and can be independently optimized and updated based on relevant data subsets, improving overall prediction accuracy without requiring a single monolithic complex system.
Solution Approach 2:
The system dynamically adapts by continuously learning from new data. The models are trained on historical data and can be retrained as new data becomes available, allowing the system to evolve and improve its predictions over time. This dynamic adaptation enables the system to handle varying conditions and maintain high accuracy without requiring manual reconfiguration.
2Measurement precision
If generalized assumptions are used for prediction, then the system is easy to implement, but the accuracy deteriorates when vehicles or routes deviate from generalized models
Solution Approach 1:
The system applies local quality by using vehicle-specific and route-specific parameters in the predictive models. Instead of applying a single generalized model to all vehicles and routes, the system incorporates individual vehicle characteristics, specific route conditions, and location-specific factors into the predictions. This allows each prediction to be tailored to the specific context, improving accuracy for diverse scenarios.
Solution Approach 2:
The system changes parameters based on specific conditions by adjusting the input variables and model parameters according to the particular vehicle, route, and environmental conditions. The predictive models accept various parameters including vehicle type, cargo weight, route geometry, weather conditions, and traffic patterns, allowing the system to adapt its predictions to match the specific characteristics of each situation rather than relying on fixed generalized assumptions.
3Measurement precision
If real-time data based methods are used, then the accuracy of arrival time estimation is improved, but the system still relies on generalized assumptions that are not applicable for all situations
Solution Approach 1:
The system incorporates feedback mechanisms where actual vehicle data, arrival times, and energy consumption measurements are fed back into the predictive models. This feedback loop allows the models to learn from real-world outcomes and continuously improve their predictions. The system compares predicted values with actual measurements and adjusts its parameters accordingly, ensuring both high accuracy and reliability across diverse situations by adapting to actual performance rather than relying solely on generalized assumptions.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for operating a plurality of vehicles is disclosed. The method comprises receiving (101) a first set of vehicle data and a second set of vehicle data, the vehicle data comprising information about each vehicle of the plurality of vehicles, each vehicle operating along at least one fixed route, receiving (102) a first set of environmental data and a second set of environmental data, the environmental data comprising information about each fixed route, and estimating (103), by means of the global self-learning model and each local-self learning model, a schedule parameter for each vehicle of the plurality of vehicles based on the received first set of vehicle data, the received first set of environmental data, the received second set of vehicle data, the received second set of environmental data, and a predefined interaction model between the global self-learning model and each local-self learning model. The method further comprises receiving (104) a measured schedule parameter for each vehicle, comparing (105) the estimated schedule parameter with the received measured schedule parameter, and updating (106) the global self-learning model and each local self-learning model based on the comparison of the estimated schedule parameter with the received measured schedule parameter.