Production Machine Downtime Tracking With ML Cause Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods do not provide an easy way to gain insights into the reasons for production machine downtimes, which significantly impact overall productivity.
Innovation Solution
A method using a machine learning module, such as an ensemble of Random Forest and XGBoost, processes sensor and event data to identify and characterize downtime periods, providing reasons and component identifiers for machine downtimes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If manual classification of downtime reasons is used, then operators can understand downtime causes, but the process is time-consuming and complex
Solution Approach 1:
The patent replaces the manual mechanical classification process with an automated machine learning system. The control unit automatically classifies downtime reasons by processing sensor data through trained machine learning models, eliminating the need for manual operator intervention in the classification process while maintaining accurate downtime reason identification
Solution Approach 2:
The system enables self-service automation where the machine learning module autonomously performs downtime classification without human intervention. The control unit automatically receives sensor data, processes it through the machine learning model, and generates classified downtime reasons, allowing the system to serve itself in the classification task
2Ease of operation
If simple downtime tracking is used, then the system is easy to operate, but it cannot provide deep insights into downtime reasons
Solution Approach 1:
The patent introduces a machine learning module as an intermediary between raw sensor data and downtime classification. This intermediary automatically processes complex sensor data patterns and translates them into meaningful downtime reasons, providing deep insights without requiring operators to manually analyze complex data, thus maintaining ease of operation
Solution Approach 2:
The system segments the complex downtime analysis task into automated components: sensor data collection, data processing, machine learning classification, and result presentation. This segmentation allows the complex analytical work to be automated while presenting simplified information to operators, maintaining ease of operation while providing deep insights
3Measurement precision
If multiple sensors and complex processing are used, then downtime analysis precision is improved, but device complexity increases
Solution Approach 1:
The patent implements a universal machine learning model that can process multiple types of sensor data (temperature, pressure, vibration, etc.) and classify various downtime reasons using a single integrated system. This multi-functional approach improves analysis precision across different sensor types while avoiding the complexity of separate processing systems for each sensor
Solution Approach 2:
The system merges multiple sensor inputs and processing functions into a unified machine learning-based classification system. By combining diverse sensor data streams and processing them through a single trained model, the patent achieves high precision downtime analysis while reducing overall system complexity compared to multiple separate analysis systems
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
A method for tracking the downtime of a production machine (12) comprises the steps of: - Receiving sensor data (SD) from the production machine (12) and production target data (PTD), - Combining the sensor data (SD) and the production target data over a certain period (40) providing combined data (D) and calculating characteristic data (CD) of the combined data (D), - Determining if the combined data (D) is from a downtime period (44) of the production machine (12) based on the characteristic data (CD), and - Characterizing the downtime period (44) using a machine learning module (48) implemented in the control unit (32), the machine learning module (48) providing a reason (R) for the downtime period (44) as an output value. Further, a system (19) for tracking the downtime of a production machine (12) is shown.