ML Training Data Partitioning for Rail Vehicle Prediction Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for building machine learning models from limited test data in rail vehicle operation fail to ensure that the sum of characteristic quantities after data partitioning remains within a tolerance value, affecting generalization performance.
Innovation Solution
A machine learning model building device that divides teaching data sets to maintain a specified error tolerance, reduces similar data, and adjusts data partitions to ensure the sum of characteristic quantities remains unchanged, using error function thresholds to refine the model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If data partitioning is performed to increase the number of training data, then the quantity of training data is improved, but the manufacturing precision of characteristic quantities deteriorates because the sum of characteristic quantities after partitioning may differ from the original
Solution Approach 1:
The teaching data set is divided into multiple partitioned data sets, where each partition contains a portion of the original teaching data. This segmentation increases the number of available training data sets while maintaining the integrity of characteristic quantities through controlled division processes.
Solution Approach 2:
The patent introduces a tolerance value parameter that defines the acceptable range for the sum of characteristic quantities after partitioning. By changing the parameter control from exact equality to tolerance-based acceptance, the system allows data partitioning while ensuring characteristic quantities remain within acceptable precision bounds.
2Quantity of substance
If data augmentation is performed to increase training data, then the quantity of training data is improved, but the reliability of characteristic quantities deteriorates because physically meaningful results cannot be guaranteed
Solution Approach 1:
Before performing data partitioning, the system calculates the characteristic quantities of the original teaching data set. This preliminary calculation establishes a reference baseline that is used to verify the reliability of characteristic quantities after partitioning, ensuring physically meaningful results are maintained.
Solution Approach 2:
The patent implements a feedback mechanism where the sum of characteristic quantities from partitioned data sets is compared against the original characteristic quantities within a tolerance range. This feedback loop ensures that only partitioned data maintaining reliability thresholds are used for training.
3Productivity
If the teaching data set is divided into multiple partitions, then the productivity of model training is improved, but the measurement precision of characteristic quantities deteriorates due to accumulation of errors
Solution Approach 1:
The system dynamically adjusts the tolerance value based on the number of partitions and their individual error characteristics. As more partitions are created to improve training productivity, the tolerance threshold is adaptively modified to maintain overall measurement precision while accommodating the increased number of data divisions.
Data Source
Figure 1
Figure 2~3
Figure 4A
AI summary
A machine learning model building device comprises an actual operation database that holds actual operation data. The machine learning model building device creates a teaching data set including one or more pieces of teaching data based on the actual operation data obtained from the actual operation database. The machine learning model building device creates a post-division teaching data set containing a plurality of pieces of teaching data after dividing the teaching data contained in the teaching data set by dividing the teaching data so that an error between a characteristic quantity of the teaching data before division and the sum of characteristic quantities of the plurality of pieces of teaching data after division becomes less than a tolerance value; and creates the machine learning model using the post-division teaching data set.