Kernel-Based Data Screening for Industrial Machine Model Building
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current approaches for selecting data for industrial machine model building are labor-intensive, require domain knowledge, lack accuracy, and are non-repeatable and non-standardized, leading to user dissatisfaction.
Innovation Solution
The use of recursive machine algorithms to automatically select a subset of reference data from raw historical data, employing unsupervised kernel-based methods like one-class SVM to identify healthy operational data without explicit labeling, reducing dependency on domain knowledge and enabling adaptive data selection strategies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual data selection by human experts is used, then domain knowledge can be applied to identify healthy operation data, but the process becomes labor-intensive and non-repeatable
Solution Approach 1:
The system performs automated data screening using unsupervised kernel-based algorithms that operate independently without requiring human intervention. The algorithm recursively processes raw historical data to identify healthy operation segments, eliminating the need for manual expert analysis while maintaining consistent, repeatable results across different datasets and users.
Solution Approach 2:
The patent replaces the mechanical process of manual data selection by human experts with an automated computational system. The unsupervised kernel-based algorithm acts as a substitute for human cognitive processes, objectively identifying healthy operation data through mathematical operations on multi-variate sensor data without subjective judgment variations.
2Adaptability or versatility
If manual data selection is used, then domain knowledge can guide data labeling, but the approach lacks capability to process high-dimensional multi-variate data
Solution Approach 1:
The patent applies kernel-based methods that implicitly map multi-variate sensor data into high-dimensional feature spaces where healthy and unhealthy operation patterns become separable. This dimensional transformation enables the algorithm to process complex multi-variate data effectively, capturing non-linear relationships that would be difficult for manual methods to identify across multiple dimensions simultaneously.
Solution Approach 2:
The system changes the parameters of data processing by transitioning from manual threshold-based selection to automated algorithmic processing with configurable parameters such as noise tolerance levels and recursion depth. These parameter adjustments enable the system to adapt to different industrial applications and data characteristics while maintaining automated high-dimensional processing capability.
3Reliability
If manual data selection approaches are used, then some data quality assessment can be performed, but the results are non-standardized and vary between different persons
Solution Approach 1:
The patent creates a universal data screening system that functions consistently across different industrial applications, machines, and datasets. The unsupervised kernel-based algorithm provides a standardized methodology that eliminates variability between different operators, delivering repeatable results regardless of who implements or uses the system, thereby establishing industry-wide applicability and consistency.
4Measurement precision
If comprehensive data screening is performed to ensure high data quality, then model building accuracy improves, but the complexity of the selection process increases
Solution Approach 1:
The system performs preliminary data screening and segmentation before model building, recursively identifying and isolating healthy operation data segments in advance. This preliminary action prepares clean, standardized training data that improves subsequent model building accuracy while the automated nature of the process prevents complexity from escalating, as the screening is handled algorithmically rather than requiring complex manual procedures.
Data Source
AI summary
Raw data is received from an industrial machine. The industrial machine includes one or more sensors that obtain the data, and the sensors transmit the raw data to a central processing center. The raw data is received at the central processing center and an unsupervised kernel-based algorithm is recursively applied to the raw data. The application of the unsupervised kernel-based algorithm is effective to learn characteristics of the raw data and to determine from the raw data a class of acceptable data. The class of acceptable data is data having a degree of confidence above a predetermined level that the data was obtained during a healthy operation of the machine. The acceptable data is successively determined and refined upon each application of the unsupervised kernel-based algorithm. The unsupervised kernel-based algorithm is executed until a condition is met.