CN-121980224-A - Vacuumizing detection system based on micro-flow sensor

CN121980224ACN 121980224 ACN121980224 ACN 121980224ACN-121980224-A

Abstract

The invention relates to the technical field of vacuum detection, in particular to a vacuum pumping detection system based on a micro-flow sensor, which comprises a data acquisition module, a reference reconstruction module, a parameter injection module, a differential extraction module and a similarity judgment module, wherein the data acquisition module acquires real-time micro-flow time sequence data and real-time pressure data, the reference reconstruction module constructs an ideal attenuation model and generates an ideal flow reference curve, the parameter injection module injects a physical hard leakage factor and a surface air release factor to generate a theoretical leakage flow curve, the differential extraction module executes differential operation to respectively generate a real residual characteristic and a theoretical residual characteristic, the similarity judgment module calculates waveform similarity and judges a physical leakage state, and the invention accurately locks micro leakage signals under strong background noise, thereby remarkably reducing false alarm rate.

Inventors

YAN WEI
LIANG SHUIGEN
DIAO JIAN
PENG JIE

Assignees

广州市新鲜世界电器有限公司

Dates

Publication Date: 20260505
Application Date: 20260130

Claims (8)

1. The vacuum pumping detection system based on the micro-flow sensor comprises a data acquisition module, an ideal reference reconstruction module, a knowledge parameterization injection module, a double-track differential extraction module and a coupling verification judgment module, wherein the data acquisition module is used for acquiring real-time micro-flow time sequence data and real-time pressure data of a target vacuum cavity, the ideal reference reconstruction module is used for constructing an adiabatic leak-free ideal attenuation model according to preset cavity volume parameters and pumping speed parameters and generating an ideal flow reference curve by utilizing the adiabatic leak-free ideal attenuation model, the knowledge parameterization injection module is used for injecting preset physical hard leakage factors and surface deflation factors into the ideal flow reference curve to generate a theoretical leak-containing flow curve, the double-track differential extraction module is used for executing first differential operation on the real-time micro-flow time sequence data and the ideal flow reference curve to generate a real residual error characteristic and executing second differential operation on the theoretical leak-containing flow curve and the ideal flow reference curve to generate a residual error characteristic, and the coupling verification module is used for calculating topological similarity of the real residual error characteristic and the theoretical residual error characteristic and judging waveform similarity according to the target vacuum waveform state.
2. The micro-flow sensor-based vacuum inspection system of claim 1, wherein the ideal reference reconstruction module generates an ideal flow reference curve by obtaining an effective pumping speed parameter and a volume parameter of the target vacuum chamber, constructing an exponential decay function with time as an independent variable and flow as a dependent variable based on a gas state equation and a conductance theory, and setting a ratio of the effective pumping speed parameter to the volume parameter as a decay coefficient of the exponential decay function, wherein the ideal flow reference curve characterizes a theoretical pumping behavior of the system in a zero fault state.
3. The micro-flow sensor-based vacuum inspection system of claim 2, wherein the knowledge parameterized injection module generates a theoretical leak-containing flow curve by constructing a physical hard leak component that is a quasi-constant flow value proportional to the real-time pressure data and the barometric pressure difference, constructing a surface bleed component that is a flow value that decays with time in a power law, and adding the physical hard leak component and the surface bleed component to the ideal flow reference curve to synthesize the theoretical leak-containing flow curve.
4. The micro-flow sensor-based vacuum inspection system of claim 3, wherein the differential operation performed by the dual-rail differential extraction module comprises subtracting the ideal flow reference curve from the real-time micro-flow time series data, eliminating common-mode background flow, extracting the real residual features containing environmental noise and potential leakage signals, subtracting the ideal flow reference curve from the theoretical leakage flow curve, eliminating common-mode background flow, and extracting the theoretical residual features containing only preset fault patterns.
5. The micro-flow sensor-based vacuum pumping detection system as set forth in claim 4, wherein the process of calculating the waveform topology similarity by the coupling verification decision module comprises setting a sliding time window, intercepting a local sequence of the real residual feature and a template sequence of the theoretical residual feature within the sliding time window, calculating a correlation coefficient between the local sequence and the template sequence by using a cross-correlation algorithm or a cosine similarity algorithm, and taking the correlation coefficient as the waveform topology similarity.
6. The micro-flow sensor-based vacuum detection system according to claim 5, wherein the process of determining the physical leakage state by the coupling verification determination module comprises presetting a similarity determination threshold, determining that a real leakage signal conforming to a physical rule exists in the real residual feature and marking the state of the target vacuum cavity as the physical leakage state if the waveform topological similarity is greater than or equal to the similarity determination threshold, and determining that the real residual feature is sensor thermal drift or random environmental noise and marking the state of the target vacuum cavity as a false abnormal state if the waveform topological similarity is less than the similarity determination threshold.
7. The micro-flow sensor-based vacuum inspection system of claim 6, further comprising an adaptive feedback control module for generating a leakage alarm signal and intercepting a vacuum flow in response to the target vacuum chamber being marked as a physical leakage state, and automatically filtering a fluctuating component in the real-time micro-flow time series data and maintaining a current vacuum flow in response to the target vacuum chamber being marked as a false anomaly state.
8. The micro-flow sensor-based vacuum inspection system of claim 7, wherein the adaptive feedback control module is further configured to perform a model correction operation to obtain a drift slope when the target vacuum chamber is marked as a false anomaly and the real-time micro-flow time series data monotonically drift, and to compensate for zero parameters of the adiabatic leak-free ideal attenuation model based on the drift slope to update the ideal flow reference curve at a next time.

Description

Vacuumizing detection system based on micro-flow sensor Technical Field The invention relates to the technical field of vacuum detection, in particular to a micro-flow sensor-based vacuumizing detection system. Background Along with the rapid iteration of the semiconductor manufacturing and aerospace technology, the requirements of various vacuum process equipment on the cleanliness and the tightness of the vacuum environment are increasingly severe, how to realize accurate and rapid online leakage detection under the working condition of high background noise is a great difficulty facing the current vacuum technical field, and the anti-interference capability and the sensitivity of a detection system are focused on the premise of pursuing high production beats; Conventional vacuum cavity leak detection currently relies primarily on static pressure boosting, helium mass spectrometry leak detection, and absolute threshold determination based on a single sensor reading. However, the static pressure boosting method, the helium mass spectrum leak detection method, the absolute threshold judgment method and the like have certain defects, for example, the static pressure boosting method needs to interrupt the air extraction flow and maintain pressure for a long time to seriously influence the production efficiency, the helium mass spectrum leak detection method relies on expensive special equipment and trace gas and is difficult to integrate for low-cost real-time process monitoring, the absolute threshold judgment method has the problems of insufficient resolution on sudden environmental noise and zero drift of a sensor when facing the micro-flow sensor, and the micro physical leakage signal is very easy to be covered by common mode flow because of huge background flow in the early stage of vacuumizing, and the waveform characteristics of the signal cannot be identified by single numerical comparison, so that the false alarm rate and the false alarm rate under the high-vacuum complex working condition are higher. Disclosure of Invention In order to solve the technical problems, the invention provides a micro-flow sensor-based vacuumizing detection system, which specifically comprises the following technical scheme: The system comprises a data acquisition module, an ideal reference reconstruction module, a knowledge parameterization injection module, a double-track differential extraction module and a coupling verification judgment module, wherein the data acquisition module is used for acquiring real-time micro-flow time sequence data and real-time pressure data of a target vacuum cavity, the ideal reference reconstruction module is used for constructing an adiabatic leak-free ideal attenuation model according to preset cavity volume parameters and pump speed parameters and generating an ideal flow reference curve by utilizing the adiabatic leak-free ideal attenuation model, the knowledge parameterization injection module is used for injecting preset physical hard leakage factors and surface gassing factors into the ideal flow reference curve to generate a theoretical leak-containing flow curve, the double-track differential extraction module is used for executing first differential operation on the real-time micro-flow time sequence data and the ideal flow reference curve to generate a real residual error characteristic and executing second differential operation on the theoretical leak-containing flow curve and the ideal flow reference curve to generate a theoretical residual error characteristic, and the coupling verification judgment module is used for calculating the waveform topological similarity of the real residual error characteristic and the theoretical residual error characteristic and judging the physical leakage state of the target vacuum cavity according to the waveform topological similarity. The ideal reference reconstruction module generates an ideal flow reference curve by acquiring an effective pumping speed parameter and a volume parameter of the target vacuum cavity, constructing an exponential decay function taking time as an independent variable and flow as a dependent variable based on a gas state equation and a conductance theory, and setting the ratio of the effective pumping speed parameter to the volume parameter as a decay coefficient of the exponential decay function to generate the ideal flow reference curve, wherein the ideal flow reference curve represents the theoretical pumping behavior of the system in a zero fault state. Preferably, the process of generating the theoretical leakage-containing flow curve by the knowledge parameterized injection module comprises the steps of constructing a physical hard leakage component, constructing a surface air release component, and combining the physical hard leakage component and the surface air release component to the ideal flow reference curve to synthesize the theoretical leakage-containing flow curve, wherein the physical hard leak