Aggregate Diagnostic Model Using Bayesian Networks

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional diagnostic techniques, such as expert systems and knowledge bases, face challenges in accurately identifying problems due to incomplete failure data and subjective interpretation, leading to incorrect fault isolation and false faults in complex environments.

Innovation Solution

A method is developed to create an aggregate diagnostic model based on a topological relationship between systems and applications, using causal networks represented as Bayesian networks, which identifies causal relationships between faults and observations, and links models with probability values to represent influence, enabling effective diagnostics.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If expert systems with rule-based deterministic approaches are used for diagnostics, then the diagnostic process is structured and systematic, but the system fails to accurately identify problems in generic environments and produces false faults due to incomplete failure data

Engineering Contradiction:

Improvestructured diagnostic processVSAvoidproblem identification accuracy

Core Design Contradiction:

Ease of operationVSMeasurement precision

Solution Approach 1:

The patent transforms the diagnostic approach from deterministic rule-based parameters to probabilistic parameters using Bayesian networks. This allows the system to handle incomplete data by computing posterior probabilities of faults given observed symptoms, thereby improving measurement precision while maintaining systematic analysis through the structured network model.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent introduces Bayesian probability theory as an intermediary between observed symptoms and fault identification. This intermediary layer allows for reasoned inference under uncertainty, enabling accurate problem identification even when failure data is incomplete, by calculating the likelihood of different fault causes based on observed evidence.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Quantity of substance

If knowledge bases with unstructured tribal knowledge are used for diagnostics, then the system can capture extensive operational experience, but users must tediously read and interpret incomplete knowledge, leading to incorrect problem isolation and false faults

Engineering Contradiction:

Improvevolume of diagnostic knowledgeVSAvoidtime for problem diagnosis

Core Design Contradiction:

Quantity of substanceVSLoss of time

Solution Approach 1:

The patent segments the unstructured tribal knowledge into structured causal networks that model specific fault scenarios. By dividing the extensive knowledge base into modular causal relationships between components, symptoms, and faults, the system enables targeted diagnostic queries without requiring users to read through all available knowledge, thereby reducing diagnosis time while preserving comprehensive knowledge coverage.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent replaces the manual mechanical process of reading and interpreting unstructured knowledge with an automated computational system. The Bayesian network automatically processes the structured knowledge and performs probabilistic reasoning, substituting human interpretation effort with algorithmic computation, thus eliminating time loss while maintaining access to extensive diagnostic knowledge.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Ease of manufacture

If rule-based expert systems are used for diagnostics, then the diagnostic approach is deterministic and easy to implement, but the system cannot handle incomplete failure data and produces incorrect fault isolation

Engineering Contradiction:

Improveease of system implementationVSAvoidfault isolation accuracy

Core Design Contradiction:

Ease of manufactureVSReliability

Solution Approach 1:

The patent changes the parameter representation from deterministic binary states (fault present/absent) to continuous probabilistic parameters (probability of fault given evidence). This parameter transformation enables the system to handle incomplete data by representing uncertainty explicitly, improving reliability while maintaining computational tractability through standard Bayesian inference algorithms.

Inventive Principle:

Principle #35Parameter changes

4Productivity

If conventional diagnostic techniques are used, then the diagnostic process is simple and quick, but the system fails to consider system-level interactions and produces incomplete problem signatures

Engineering Contradiction:

Improvediagnostic speedVSAvoidcompleteness of problem signature

Core Design Contradiction:

ProductivityVSLoss of information

Solution Approach 1:

The patent merges multiple individual component models into a unified system-level causal network. By combining the causal relationships of individual components with their interactions, the system maintains diagnostic speed through modular model composition while capturing complete problem signatures that reflect system-level interactions, thereby preventing information loss.

Inventive Principle:

Principle #5Merging (Combining)

Data Source

PatentUS8417656B2Techniques for building an aggregate model for performing diagnostics

Publication Date: 2013.04.09 ORACLE INT CORP

AI summary

Techniques for building a model for performing diagnostics. In one embodiment, a set of models is determined based upon a topological relationship created upon receiving an alert or a request for which diagnostics are to be performed. An aggregate model is then generated based upon the set of models and the topological relationship. The aggregate model is then used for performing the diagnostics.