Vehicle Data Dimension Alignment for Neural Network Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies for predicting the remaining life of a vehicle battery fail to effectively align the number of dimensions in vehicle data, leading to inefficient training of neural networks and issues with missing data, especially when the number of trips per day varies.
Innovation Solution
The method involves using principal component analysis to generate training data by reducing the number of dimensions in vehicle data, aligning them for neural network training, and filling in missing values with averages or zeros, allowing for efficient data processing and model training regardless of the number of trips.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If vehicle data is collected for each trip unit with varying numbers of trips per day, then the data reflects actual user usage patterns, but the number of dimensions of the data becomes different for each day making model training difficult
Solution Approach 1:
The patent applies dimensionality change by transforming the variable number of trip units into a fixed-dimensional format. It introduces a dummy column to pad shorter days and reorders columns so that trip units are consistently positioned across all days. This converts the problem from variable dimensionality to fixed dimensionality, enabling standard neural network training while preserving the adaptive collection method.
2Productivity
If the number of trip units is reduced to align data dimensions, then model training becomes feasible, but data properties are lost and missing data issues arise
Solution Approach 1:
The patent applies preliminary action by performing dimension alignment and dummy column insertion before the actual model training process. It pre-processes the data to ensure all days have the same number of columns by adding dummy columns to shorter days. This preliminary structuring allows the neural network to train efficiently on aligned data while preserving information through the dummy column approach rather than simply reducing or dropping data.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables the generation of comprehensive training data without missing values, allowing for accurate neural network model training and efficient battery life estimation, even with varying trip frequencies and missing data acquisition.
Implementation Method 1
generating, by using principal components obtained by principal component analysis of the vehicle data, training data by reducing the number of units relating to the predetermined operation in the vehicle data
Data Source
AI summary
An information processing method including, by a computer: receiving vehicle data, which is data relating to a vehicle and is accumulated for each predetermined period of time and for each unit relating to a predetermined operation, a number of units in the data relating to the predetermined operation being different for each predetermined period of time in accordance with use by a user: generating, by using principal components obtained by principal component analysis of the vehicle data, training data by reducing the number of units relating to the predetermined operation in the vehicle data, and aligning a number of dimensions of the vehicle data; and training a model of a neural network by using the training data.