Vacuum Pump Abnormality Detection Using Normal Variation Ranges
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vacuum pumps in semiconductor manufacturing processes are prone to stopping due to the deposition of solidified or liquefied products, leading to excessive load and potential damage to the pump and articles, with existing prediction methods limited to specific timings like pump startup, failing to anticipate arbitrary operational stops.
Innovation Solution
An information processing apparatus and method that determines a normal variation range or behavior of state quantities based on past data to compare with current values, allowing detection of abnormalities caused by solidified or liquefied products during operation, thereby predicting pump stops at arbitrary times and reducing article damage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the vacuum pump operates continuously to maintain manufacturing process, then productivity is improved, but solidified products deposit in clearances causing excessive load and pump stop
Solution Approach 1:
The system performs preliminary detection of abnormality trends by continuously monitoring state quantities and comparing them against normal variation ranges before the pump actually stops. This early warning allows maintenance to be scheduled proactively, preventing unexpected stops during manufacturing while maintaining continuous operation benefits.
Solution Approach 2:
The system implements continuous feedback monitoring of pump state quantities (vibration, temperature, pressure) and compares them against dynamically determined normal variation ranges. When deviations exceed thresholds, the system provides feedback warnings to schedule maintenance before failure occurs, resolving the contradiction between continuous operation and reliability.
2Difficulty of detecting and measuring
If existing prediction methods only check at pump startup, then detection complexity is reduced, but abnormality detection precision is insufficient for arbitrary operational stops
Solution Approach 1:
The system transitions from static detection (only at startup) to dynamic continuous monitoring during entire pump operation. State quantities are monitored throughout operation and compared against normal variation ranges, enabling detection of abnormalities that develop during runtime while maintaining manageable complexity through automated comparison logic.
Solution Approach 2:
The system performs preliminary determination of normal variation ranges using past operational data before actual abnormality detection begins. This preparatory step establishes baseline expectations for normal operation, enabling more accurate detection of true abnormalities during subsequent monitoring without requiring complex real-time analysis.
3Adaptability or versatility
If normal variation range is determined using only past data from same pump, then adaptability is improved, but measurement precision for detecting novel abnormalities decreases
Solution Approach 1:
The system merges multiple data sources including past data from the same pump, past data from similar pumps, and theoretical models to determine comprehensive normal variation ranges. This combination leverages pump-specific characteristics while incorporating broader knowledge bases, improving both adaptability to individual pumps and precision in detecting novel or rare abnormalities.
Solution Approach 2:
The system creates a multi-functional determination mechanism that can adapt to different pump types, operating conditions, and abnormality patterns. By incorporating diverse data sources and flexible comparison methods, the system achieves universal applicability across various scenarios while maintaining pump-specific precision through appropriate data weighting and selection.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables early detection of vacuum pump abnormalities due to solidified or liquefied products, preventing unexpected stops and minimizing damage during semiconductor manufacturing by predicting failures at arbitrary operation times.
Implementation Method 1
the pump rotor is configured to rotate while being in non-contact with the casing. As the pair of pump rotors rotates in an opposite direction to each other while synchronizing, the gas in the casing is transferred from a suction side to a discharge side
Implementation Method 2
the vacuum pump described above generates compression heat during the transfer of the gas, the temperature of the operating vacuum pump rises to some extent
Implementation Method 3
the temperature of the vacuum pump is maintained by heating a pump main body from the outside or heating an inflowing gas
Implementation Method 4
the solidified products (reaction products) are deposited in a clearance between the pump rotors and a clearance between the pump rotor and the casing
Data Source
AI summary
An information processing apparatus detecting presence or absence of abnormality of a vacuum pump derived from a product produced within a target vacuum pump, including: a determination unit configured to determine a normal variation range or a normal time variation behavior of a target state quantity which is a state quantity varying depending on a load of gas flowing into the vacuum pump, based on at least one of past target state quantities of the target vacuum pump or another vacuum pump; and a comparison unit configured to compare a current target state quantity of the target vacuum pump with the normal variation range or the normal time variation behavior and output the comparison result.