Automated AI Diagnostic Model Generation for Vehicle Sensors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current AI-based diagnostic technologies for vehicles rely heavily on human expertise and are time-consuming, requiring significant resources and trial-and-error processes to develop effective machine learning models for sensor data analysis, which limits their efficiency and accuracy.
Innovation Solution
An automation method that automatically extracts features from vibration, noise, and CAN signal data, using both machine learning and deep learning architectures, and generates ensemble prediction models to optimize model accuracy, reducing the need for human intervention in hyper-parameter setting and model structure design.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automated machine learning is used to reduce model development time, then productivity is improved, but barriers remain in data collection, algorithm selection, and training adjustment
Solution Approach 1:
The patent segments the complex model development process into distinct automated modules: data collection module, preprocessing module, algorithm selection module, training module, and evaluation module. Each module handles specific tasks independently, reducing overall complexity while maintaining productivity gains from automation.
Solution Approach 2:
The system implements self-service automation where the automated machine learning platform independently performs data collection, preprocessing, algorithm selection, hyperparameter tuning, and model training without requiring manual intervention. The system automatically evaluates multiple algorithms and selects the optimal one based on performance metrics.
2Reliability
If open analysis method is used to develop diagnostic models, then diagnostic performance can be improved, but a lot of time and efforts are required through trial and error processes
Solution Approach 1:
The patent performs preliminary actions by automatically collecting and preprocessing data before model development begins. The system pre-processes sensor data, performs feature extraction, and prepares datasets in advance, eliminating the need for repeated trial-and-error data preparation and reducing overall development time while maintaining diagnostic performance.
Solution Approach 2:
The system implements automated feedback loops where model performance is continuously evaluated against diagnostic criteria, and results are fed back to automatically adjust hyperparameters and select optimal algorithms. This automated feedback mechanism eliminates manual trial-and-error processes while maintaining high diagnostic performance.
3Manufacturing precision
If tens of machine learning models are generated and compared to find the best model, then manufacturing precision of diagnostic accuracy is improved, but significant domain knowledge and time are required
Solution Approach 1:
The automated machine learning platform performs self-service by automatically generating, training, and evaluating multiple machine learning models without requiring domain expert intervention. The system autonomously compares model performances, selects the best-performing model, and deploys it for diagnostic purposes, making the process accessible without significant domain knowledge.
Solution Approach 2:
The system automatically changes and optimizes multiple parameters including algorithm selection, hyperparameter values, and model architectures to achieve high diagnostic accuracy. The automated parameter tuning and optimization processes eliminate the need for domain experts to manually adjust these parameters while maintaining precision in diagnostic accuracy.
Data Source
AI summary
An automation method of an artificial intelligence (AI)-based diagnostic technology for equipment application includes receiving one or more pieces of data among vibration data, noise data, and controller area network (CAN) data, a data input processing operation of trimming the input data, an operation of extracting features from the trimmed data, setting a setting value of a hyper-parameter with respect to the one or more pieces of data thereamong, and generating a total of N models to include both of machine learning (ML) and deep learning (DL) as N individual models and generating ensemble prediction model structures for the N individual models. As a parameter updating is being proceeded due to the hyper-parameter so as to minimize values of cost functions of the N individual models, a reward for model accuracy performance is optimized and the ensemble prediction model structures of the N individual models change.