Sequence Modeling of Manufacturing Sensor Data for Virtual Measurement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Industrial machine failures in manufacturing processes lead to production disruptions and resource losses due to inadequate predictive monitoring capabilities.
Innovation Solution
A computer-implemented method using a machine learning system with sequence modeling, employing a first neural network to generate parameter data from observed measurements and a second neural network to predict future measurements based on history data, leveraging latent variables to model the dynamics of manufacturing time series and predict future trends.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional measurement monitoring is used, then measurement accuracy is maintained, but production time is lost and costs increase due to time-consuming measurements
Solution Approach 1:
The patent creates a virtual copy of the manufacturing process through digital twins and neural network models. Instead of performing physical measurements on actual parts, the system generates predicted measurement data from virtual models trained on historical measurement data, thereby maintaining measurement accuracy while eliminating time-consuming physical measurements
Solution Approach 2:
The system performs preliminary training of neural networks using historical measurement data before actual production. This preliminary action creates predictive models that can rapidly generate measurement predictions during production, avoiding the need for time-consuming real-time physical measurements while maintaining accuracy
2Measurement precision
If physical measurements are performed at each station, then accurate measurement data is obtained, but manufacturing efficiency decreases due to time-consuming measurements
Solution Approach 1:
The patent replaces physical measurements with virtual measurements generated by neural networks. The system trains models on historical measurement data and uses them to predict measurement outcomes, eliminating the need for time-consuming physical measurements at each manufacturing station while maintaining data accuracy
Solution Approach 2:
The patent substitutes mechanical measurement systems with computational models. Instead of using physical sensors and measurement devices at each station, the system employs neural networks that process historical data to generate predicted measurements, thereby improving manufacturing efficiency while maintaining measurement quality
3Reliability
If predictive monitoring is implemented using machine learning, then production disruptions are reduced, but system complexity increases
Solution Approach 1:
The patent divides the predictive monitoring system into modular components: separate neural networks for different manufacturing stations, independent digital twins for different parts, and distinct training/inference modules. This segmentation makes the complex system more manageable and easier to implement while maintaining production reliability
Data Source
AI summary
A computer-implemented system and method includes establishing a station sequence that a given part traverses. A first neural network generates a set of parameter data based on observed measurement data of the given part at each station of a station subsequence. The set of parameter data is associated with a latent variable subsequence corresponding to the station subsequence. A second neural network generates next parameter data based on history measurement data and the set of parameter data. The history measurement data relates to another part processed before the given part and is associated with each station of the station sequence. The next parameter data is associated with a next latent variable that follows the latent variable subsequence. The next latent variable corresponds to a next station that follows the station subsequence in the station sequence. The second neural network generates predicted measurement data of the given part at the next station based on the next latent variable and the next parameter data.