Manufacturing Causal Graph Pruning for Root Cause Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern manufacturing systems with interconnected machines face challenges in identifying production bottlenecks and determining their impact on specific fault modes and overall downtime, as existing methods struggle with noise and irrelevant data in manufacturing log data, leading to computationally intractable root cause analysis.
Innovation Solution
A system and method using additive noise models to preprocess manufacturing log data, generate a directed acyclic graph representing machine connectivity, cluster machines based on processing similarities, and apply an additive noise model to prune the graph and determine causal pathways, facilitating efficient causal inference and root cause analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional root cause analysis methods are applied to manufacturing log data, then comprehensive fault analysis can be performed, but computational complexity becomes intractable due to noise and irrelevant data
Solution Approach 1:
The patent segments the manufacturing system into interconnected machine clusters represented as nodes in a graph structure. By dividing the complex system into manageable components and analyzing causal relationships between clusters rather than individual machines, the computational complexity is reduced while maintaining comprehensive fault analysis capability.
Solution Approach 2:
The patent extracts and removes noise and irrelevant data from manufacturing logs through preprocessing steps. By taking out only the relevant causal signals and filtering out distracting information, the system achieves accurate fault analysis without being overwhelmed by the full complexity of raw manufacturing data.
2Loss of information
If detailed analysis of all machines and pathways is performed, then complete root cause identification is achieved, but analysis time and computational resources increase significantly
Solution Approach 1:
The patent applies local quality by focusing analysis on specific clusters of interconnected machines that are relevant to the fault being investigated. Rather than uniformly analyzing all machines in the system, the method concentrates computational resources on local areas where causal relationships are most likely to exist, achieving complete root cause identification with reduced analysis time.
Solution Approach 2:
The patent performs preliminary actions by pre-processing manufacturing logs to extract relevant features and pre-identifying machine connectivity patterns before fault analysis begins. This preliminary preparation reduces the computational burden during actual fault analysis, enabling complete root cause identification without excessive analysis time.
3Productivity
If additive noise models are applied to prune candidate clusters, then computational efficiency is improved, but risk of removing relevant causal pathways increases
Solution Approach 1:
The patent employs feedback mechanisms where the additive noise model's pruning decisions are continuously validated against the graph structure and causal relationships. The system monitors the pruning process and can adjust or revert decisions if potential causal pathways are at risk, ensuring computational efficiency is improved without compromising the reliability of causal pathway identification.
Data Source
AI summary
A system and method are provided for determining a causal inference in a manufacturing process. During operation, the system can receive data associated with a processing system which includes a set of interconnected machines and an associated set of processes. The system can generate, based on the data, a graph indicating flows of outputs between the machines as part of the processes. The system can determine, based on a set of variables, one or more candidate clusters in the graph. The system can perform, based on one or more variables of interest, root cause analysis on the one or more candidate clusters by: applying an additive noise model to prune the one or more candidate clusters from the graph; and determining, based on the pruned graph, a candidate pathway likely to cause an issue in at least one process, thereby facilitating improved efficiency in the processing system.