Semiconductor Run Classification Using Compressed Sensor Signatures
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manufacturing environments face challenges in detecting equipment failures and inefficiencies in real-time, leading to substantial waste and costs due to the complexity of processing vast amounts of sensor data and the inflexibility of custom-built planning systems.
Innovation Solution
A method is introduced that compresses sensor data by generating representations of its shape and magnitude, allowing for efficient querying and classification of runs, using techniques like dynamic time warping and Symbolic Aggregate Approximation, to facilitate fault detection and automated maintenance in manufacturing environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If custom-built planning systems are used to monitor and analyze sensor data, then fault detection capability is improved, but system complexity and maintenance difficulty increase significantly
Solution Approach 1:
The patent creates a virtual copy of the manufacturing process by generating a digital twin that mirrors the physical system's behavior. This digital model captures the essential patterns and relationships in sensor data without requiring complex custom-built analysis systems, thereby improving fault detection while reducing system complexity
Solution Approach 2:
The patent introduces an intermediary layer that translates complex sensor data into simplified representations. This intermediary processing layer enables effective fault detection by converting raw data into meaningful patterns without requiring the end system to handle the full complexity of原始传感器数据
2Measurement precision
If comprehensive sensor data is collected from multiple sources, then measurement precision is improved, but data processing time increases
Solution Approach 1:
The patent extracts only the essential features and patterns from comprehensive sensor data that are most relevant for anomaly detection. By selectively extracting key information rather than processing all raw data, the system maintains high measurement precision while significantly reducing data processing time
Solution Approach 2:
The patent segments the comprehensive sensor data into distinct components and time intervals, analyzing each segment independently. This segmentation allows the system to process large volumes of data from multiple sensors efficiently by breaking down the complex dataset into manageable, analyzable units
3Speed
If real-time monitoring of all production runs is performed, then fault detection speed is improved, but computational resources required increase
Solution Approach 1:
The patent applies partial monitoring by focusing computational resources on critical parameters and high-risk time periods rather than continuously analyzing all sensor data at full resolution. This approach maintains fast fault detection speed for critical events while reducing overall computational resource consumption
Solution Approach 2:
The patent performs preliminary analysis and filtering of sensor data before full processing, identifying and flagging only those data points that warrant detailed examination. This preliminary action enables fast fault detection for anomalies while avoiding unnecessary computational expenditure on normal operating conditions
Data Source
AI summary
Embodiments provide techniques for compressing sensor data collected within a manufacturing environment. One embodiment monitors a plurality of runs of a recipe for fabricating one or more semiconductor devices within a manufacturing environment to collect runtime data from a plurality of sensors within the manufacturing environment. The collected runtime data is compressed by generating, for each of the plurality of sensors and for each of the plurality of runs, a respective representation of the corresponding runtime data that describes a shape of the corresponding runtime data and a magnitude of the corresponding runtime data. A query specifying one or more runtime data attributes is received and executed against the compressed runtime data to generate query results, by comparing the one or more runtime data attributes to at least one of the generated representations of runtime data.