Decision Tree Diagnosis for Systems With Unknown Equations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for diagnosing complex industrial systems require knowledge of the mathematical equations governing their operation, which is often unavailable, especially as systems become increasingly complex, limiting the applicability of structural analysis methods.
Innovation Solution
A method for constructing a decision tree using symbolic classification that discriminates between nominal and failure states of a system based on observable variables, without needing to know the underlying equations, by iteratively splitting nodes and applying classification functions derived from training data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If structural analysis method is used to diagnose complex industrial systems, then diagnosis precision is improved, but the method becomes unusable when mathematical equations are unknown
Solution Approach 1:
The patent replaces the traditional structural analysis method (which relies on mechanical/mathematical equations) with a machine learning-based approach. Instead of using analytical models and equations to diagnose system failures, the invention trains neural networks on operational data to learn failure patterns, enabling diagnosis of systems where equations are unknown or too complex.
Solution Approach 2:
The patent transforms the diagnosis approach from equation-based parameters to data-driven parameters. By changing from using mathematical relationships between variables to using statistical patterns in operational data, the method becomes applicable to systems with unknown equations while maintaining diagnostic capability.
2Reliability
If traditional structural analysis is applied, then diagnostic capability is improved, but device complexity increases due to numerous equations
Solution Approach 1:
The patent substitutes complex mathematical equation systems with machine learning models. Instead of managing numerous analytical equations that increase complexity, the invention uses trained neural networks that automatically learn diagnostic patterns from data, reducing the burden of equation management while improving diagnostic reliability.
Solution Approach 2:
The patent creates a virtual model of the system through machine learning instead of using the actual complex equations. By training on operational data, the system learns a simplified representation of failure patterns that captures diagnostic essence without requiring the full complexity of the underlying equations.
3Measurement precision
If machine learning is combined with structural analysis, then ARR identification performance is improved, but knowledge of system variables is still required
Solution Approach 1:
The patent replaces the requirement for knowledge of system variables and equations with a data-driven approach. Instead of needing to identify analytical redundancy relations through variable knowledge, the invention directly learns failure patterns from operational data using machine learning, eliminating the information barrier.
Data Source
AI summary
A method and device for constructing a decision tree for diagnosing a system with components. The method includes obtaining a training data set comprising pairs with a vector of measured values of observable variables representing a system operation and a label. The label represents a nominal operating state of the system or a failure state of one component. The method includes processing a current node created by splitting a previous node and associated with a subset of the training data set as a current data set. When a splitting criterion is satisfied, splitting the current node into a first and a second child node by applying a classification function obtained from the current data set and defined to a nominal class representing the nominal operating state of the system or a failure class representing the failure states of the system. The method includes providing a decision tree for system diagnosis.