Predictive Alarm Generation for Early Process Abnormality Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing paperless recorders often fail to provide timely alerts for abnormalities in manufacturing processes, leading to potential defective product creation, as alarms are triggered too late or become less accurate when alarm conditions are made less strict.
Innovation Solution
An alarm generation system that acquires sensor data, generates a learning model to predict future measured values, and triggers an alarm when these values meet predetermined conditions, including displaying predicted values and their corresponding times, along with probability information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the alarm condition is made stricter to improve alarm accuracy, then false alarms are reduced, but the alarm is triggered too late to prevent defective product creation
Solution Approach 1:
The system performs preliminary actions by predicting future measured values before the actual abnormality occurs. The prediction unit uses the learning model to forecast future sensor readings, and the alarm generation unit triggers an alarm based on these predicted values, allowing operators to take preventive action before the abnormality actually manifests in the manufacturing process.
Solution Approach 2:
The system applies preliminary anti-action by generating an alarm in advance of the actual abnormality. By predicting future measured values that would satisfy alarm conditions, the system triggers an alarm that counteracts the potential harmful effect (defective product creation) before it occurs, enabling preventive rather than reactive response.
2Loss of time
If the alarm condition is made less strict to enable earlier alarm triggering, then response time is improved, but alarm accuracy decreases and false alarms increase
Solution Approach 1:
The system performs preliminary actions by predicting future measured values before the actual abnormality occurs. The prediction unit uses the learning model to forecast future sensor readings, and the alarm generation unit triggers an alarm based on these predicted values, allowing operators to take preventive action before the abnormality actually manifests in the manufacturing process.
3Ease of operation
If traditional alarm methods are used to maintain simple system operation, then ease of operation is maintained, but the system cannot provide proactive abnormality notification
Solution Approach 1:
The system applies self-service by automatically learning from historical measured values and generating prediction models without requiring manual intervention. The learning unit continuously processes sensor data to build and update the learning model, which the prediction unit then uses to forecast future values and trigger alarms, enabling the system to improve its own performance autonomously.
Data Source
AI summary
An alarm generation system includes: an acquisition unit that acquires a measured value obtained from a sensor; a learning unit that generates a learning model through learning of the measured value acquired by the acquisition unit; a prediction unit that obtains, by using the learning model generated by the learning unit, a predicted measured value that is a measured value to be obtained in a future from a current time point; and an alarm generation unit that generates an alarm when the predicted measured value obtained by the prediction unit satisfies an alarm generation condition.