Machine Availability Prediction Using Event Logs and Process Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for predicting machine downtime in manufacturing processes are unreliable and lack precision, relying mainly on machine sensor data and failing to accurately anticipate production delays.
Innovation Solution
A computer-implemented method that analyzes event logs and associated manufacturing process data using machine learning algorithms to predict temporal parameters such as machine availability, downtime probability, and duration, integrating IoT data and cloud-based platforms for real-time analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine sensor data is used for predicting machine downtime, then the prediction can be made, but the prediction precision and reliability are insufficient
Solution Approach 1:
The patent combines multiple data sources including machine sensor data, event log data, and manufacturing process data into a unified prediction model. This merging of diverse data types enables more precise and reliable predictions of machine downtime by considering both technical machine states and contextual manufacturing factors simultaneously.
Solution Approach 2:
The patent introduces machine learning models as intermediaries that process and analyze the collected data from multiple sources. These models act as mediators between raw data and prediction outcomes, transforming diverse data types into actionable insights about machine availability and downtime probability.
2Loss of information
If complex analysis processes are used to identify reasons for machine downtime, then the analysis completeness improves, but the analysis complexity increases
Solution Approach 1:
The patent implements automated analysis systems that self-process the complex analysis tasks without requiring manual intervention. The machine learning models automatically identify patterns, correlate events with downtime, and generate predictions, enabling the system to serve itself in performing complex analytical work that would otherwise require extensive human expertise.
Solution Approach 2:
The patent transforms complex qualitative analysis problems into quantitative parameter measurements by defining specific metrics such as downtime probability, expected downtime duration, and machine availability scores. This parameterization enables complex analysis to be performed through mathematical computations rather than manual qualitative assessment.
Data Source
Figure 1~3
Figure 4
AI summary
A method for determining a prediction of a temporal parameter, e.g., a temporal parameter of machine availability, associated with an event is provided. The event is associated with a machine deployed in a manufacturing process. An occurrence of the event is determined based on an event log associated with the machine. The prediction of the temporal parameter is determined based on the event and on data of at least one parameter associated with the manufacturing process.