EV Driving Range Indication Using an Autoencoder Threshold
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Range anxiety in electrified vehicles due to uncertainty about the remaining driving range, which can occur even in hybrid vehicles not designed as plug-in hybrids, is a significant concern.
Innovation Solution
An electrified vehicle system utilizing an autoencoder, trained with data indicative of remaining battery range, processes vehicle operating data to generate a binary indication of whether the driving range is adequate for the current trip, reducing the need for direct battery capacity or range calculations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If numerical range calculation methods are used to determine driving range, then measurement precision of remaining range is improved, but device complexity and computational energy demands increase
Solution Approach 1:
The patent replaces complex numerical calculation systems with a trained autoencoder neural network model. Instead of using traditional computational methods to calculate remaining driving range based on multiple parameters, the system uses a pre-trained autoencoder that has learned the mapping from operating conditions to adequate range determination, thereby reducing computational complexity while maintaining accuracy
Solution Approach 2:
The patent creates a simplified copy of the complex range calculation problem through the autoencoder model. The autoencoder is trained to replicate the behavior of complex numerical range calculations, allowing the system to use the trained model for rapid inference without performing the full computational sequence again
2Reliability
If intensive computational methods are used to determine adequate driving range, then reliability of range assessment is improved, but use of energy increases
Solution Approach 1:
The patent performs the computationally intensive work in advance by training the autoencoder model offline using historical operating data. Once trained, the model can rapidly assess whether driving range is adequate with minimal computational energy during actual vehicle operation, as the complex learning has already been completed beforehand
Solution Approach 2:
The patent substitutes energy-intensive numerical calculation systems with a trained neural network model that performs inference with significantly lower computational requirements, thereby reducing real-time energy consumption while maintaining reliable range assessment
3Measurement precision
If direct battery capacity measurement methods are used, then measurement precision of remaining capacity is improved, but device complexity increases
Solution Approach 1:
The patent introduces the autoencoder model as an intermediary between raw operating data and battery capacity assessment. Instead of directly measuring or calculating battery capacity through complex sensor arrays and computation, the autoencoder processes operating conditions and indirectly determines adequate range, serving as a mediator that simplifies the measurement process
Data Source
AI summary
An electrified vehicle and associated method for controlling an electrified vehicle having an electric machine powered by a traction battery include an autoencoder trained with training data indicative of a remaining driving range of the traction battery. The trained autoencoder processes vehicle operating data to generate a reference data record and determines a value indicative of a similarity between the vehicle operating data and the reference data record. The autoencoder generates an output data record if the value indicative of the similarity is below a predetermined threshold value. The output data record may be used to display an alert or message to a vehicle occupant and/or control the vehicle to reduce power consumption to increase vehicle driving range.