Additive Manufacturing Fault Detection Using Multiple Process Models

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional additive manufacturing (AM) process monitoring struggles to detect process deviations and defects in real-time, leading to potential cost and quality issues due to delayed diagnostics and measurement variability.

Innovation Solution

The implementation of time-dependent and spatially-dependent models for laser-on transient behavior, combined with multiple model hypothesis testing (MMHT) algorithms, allows for enhanced in-process monitoring. This system compares sensor data to outputs from multiple process models to detect deviations and improve part quality control.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If conventional between-build diagnostics are used, then device complexity is reduced, but reliability deteriorates due to delayed detection of process deviations

Engineering Contradiction:

Improveprocess monitoring reliabilityVSAvoidmonitoring system complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The system performs preliminary actions by establishing multiple process models representing different fault conditions before monitoring begins. These models are prepared in advance with associated probability density functions, enabling immediate comparison with incoming sensor data without requiring complex real-time computation of fault probabilities from scratch.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system implements feedback by continuously comparing sensor measurements against predictions from multiple process models and updating fault probability assessments in real-time. This closed-loop feedback mechanism allows the system to adapt to process deviations dynamically while maintaining computational efficiency through the use of pre-established model frameworks.

Inventive Principle:

Principle #23Feedback

2Measurement precision

If steady state averaged measurement is used, then measurement precision is improved, but difficulty of detecting and measuring worsens due to inability to resolve individual parameter shifts

Engineering Contradiction:

Improvemelt pool measurement precisionVSAvoidparameter shift detection difficulty

Core Design Contradiction:

Measurement precisionVSDifficulty of detecting and measuring

Solution Approach 1:

The system segments the monitoring approach by dividing the analysis into multiple distinct process models, each representing a specific fault condition or operational state. This segmentation allows the system to compare measurements against multiple specialized models rather than a single averaged model, enabling detection of specific parameter shifts while maintaining measurement precision through targeted model comparisons.

Inventive Principle:

Principle #1Segmentation

3Reliability

If measurement variability from known sources is not accounted for, then device complexity is reduced, but reliability deteriorates due to inability to distinguish faulty conditions from normal variations

Engineering Contradiction:

Improvefault detection reliabilityVSAvoidmodel-based analysis complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The system applies local quality by creating spatially-dependent models that account for location-specific characteristics of the additive manufacturing process. Each process model incorporates local variations in part geometry, scan path history, and sensor viewing angles, allowing the system to distinguish between normal location-dependent variations and actual fault conditions without requiring overly complex global models.

Inventive Principle:

Principle #3Local quality

4Reliability

If in-process monitoring with multiple models is implemented, then reliability is improved, but loss of time increases due to computational requirements

Engineering Contradiction:

Improveprocess deviation detection reliabilityVSAvoidcomputation time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system performs preliminary actions by pre-establishing multiple process models and their associated probability density functions before monitoring begins. This advance preparation eliminates the need for complex real-time computation of model parameters and probability distributions, significantly reducing computational time during actual fault detection while maintaining high reliability through comprehensive multi-model comparison.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS20250147861A1System and methods for generating fault indications for an additive manufacturing process based on a probabilistic comparison of the outputs of multiple process models to measured sensor data

Publication Date: 2025.05.08 GENERAL ELECTRIC CO

AI summary

Generating fault indications for an additive manufacturing machine based on a comparison of the outputs of multiple process models to measured sensor data. The method includes receiving sensor data from the additive manufacturing machine during manufacture of at least one part. Models are selected from a model database, each model generating expected sensor values for a defined condition. Difference values are computed between the received sensor data and an output of each of the models. A probability density function is computed, which defines, for each of the models, a likelihood that a given difference value corresponds to each respective model. A probabilistic rule is applied to determine, for each of the models, a probability that the corresponding model output matches the received sensor data. An indicator is output of a defined condition corresponding to a model having the highest match probability.