Additive Manufacturing Sensor Data Compression With Process Context
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional additive manufacturing sensor configurations produce 'context-less' data, leading to incomplete and inaccurate predictive models for DMLM processes, resulting in defects such as subsurface porosity and thermal shrinkage, and generate unmanageable quantities of data that hinder analysis and optimization.
Innovation Solution
The system contextualizes sensor data by linking it to manufacturing process data, allowing for adaptive compression and analysis, enabling the development of higher fidelity digital twin models and real-time anomaly detection, and optimizing machine performance through informed design changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sensor data is collected at high acquisition rates to capture complete process information, then measurement precision and process monitoring capability are improved, but data quantity becomes unmanageable and transmission requirements increase
Solution Approach 1:
The patent extracts only the essential and relevant sensor data from the complete data stream. By identifying and extracting only the critical process parameters and anomalies, the system maintains high measurement precision for important features while significantly reducing the overall data quantity that needs to be managed and transmitted.
Solution Approach 2:
The patent segments the continuous sensor data stream into meaningful process phases and identifies key events within each phase. By dividing the data into relevant segments and focusing on critical transitions and anomalies, the system preserves essential process information while eliminating redundant data, thereby reducing total data volume without sacrificing monitoring precision.
2Loss of information
If all sensor data is transmitted and stored without compression to maintain data完整性, then data completeness is improved, but data transmission requirements and storage needs become unmanageable
Solution Approach 1:
The patent applies different data quality levels to different portions of the sensor data based on their importance. Critical process parameters and anomaly data are maintained with high fidelity and transmitted completely, while routine or less important data is compressed or summarized. This local differentiation preserves essential information while reducing overall transmission requirements.
3Device complexity
If conventional sensor configurations are used to monitor the DMLM process, then device complexity is kept simple, but the produced data lacks contextual information needed for accurate predictive modeling
Solution Approach 1:
The patent merges sensor data with process control data and contextual information from the build file. By combining these different data sources and correlating them through process events and timestamps, the system enriches the sensor data with essential context without significantly increasing the physical complexity of the sensor configuration itself.
Data Source
AI summary
Method, and corresponding system, for receiving and adaptively compressing sensor data for an operating manufacturing machine. The method includes determining the sensor data values at the working tool positions based on a time correlation of the values of the sensor data relative to time and the working tool positions relative to time. Tool control magnitude values relative to the working tool positions are determined based on the process data. The method further includes determining a magnitude differential, relative to the working tool positions, between the sensor data values and the tool control magnitude values. Scoring data is determined by applying a scoring function to the magnitude differential. The magnitude differential data is compressed based at least in part on the scoring data. The method further includes decompressing the magnitude differential data and determining the sensor data values versus the working tool positions based on the magnitude differential data.