Search

CN-121980244-A - Archival storage health monitoring method based on feature extraction

CN121980244ACN 121980244 ACN121980244 ACN 121980244ACN-121980244-A

Abstract

The invention discloses a method for monitoring the health of archive storage based on feature extraction, which relates to the technical field of archive management, the trace characteristic gas released by acidification and embrittlement of the paper can be accurately identified, so that the non-contact early warning of the file health degree is realized, and the limitation of the traditional pure dependence on the environmental temperature and humidity is broken through. Meanwhile, by combining gradient source searching logic of distributed sensing or mobile inspection, the system can automatically invert and lock three-dimensional space coordinates of abnormal gas and directly position the abnormal gas to a specific damaged file box. In addition, a physical isolation and adjacent archives association monitoring mechanism is innovatively introduced, so that the cross contamination source of the acid gas can be rapidly cut off through the transfer device, the monitoring level of surrounding high-risk suspected objects can be automatically improved, and the robustness and the intellectualization level of the monitoring result are remarkably improved.

Inventors

  • LI GUANGYI

Assignees

  • 银泰智能科技有限公司

Dates

Publication Date
20260505
Application Date
20260128

Claims (10)

  1. 1. The archive storage health monitoring method based on the feature extraction is characterized by comprising the following steps of: Presetting a file damage characteristic odor library in a central control unit, wherein the file damage characteristic odor library comprises gas fingerprint characteristic vectors corresponding to different file damage types; acquiring environmental gas data of a storage space in real time through a gas sensing assembly, and extracting a current gas characteristic value after noise reduction treatment of the gas data; comparing the current gas characteristic value with the file damage characteristic odor library to calculate the matching degree, judging that an abnormal file exists in the storage space when the matching degree exceeds a preset threshold value, and determining the type of the abnormal gas; Triggering a source positioning strategy in response to an abnormal identification result, calculating three-dimensional space coordinates of an abnormal gas release source in a storage space by using concentration space distribution gradients of gas data based on a deployment mode of a gas sensing assembly, and locking a target archive box; the dispatching and transporting device goes to the three-dimensional space coordinates to grab the target file box and carries the target file box to a preset physical isolation area; And acquiring adjacent file information of the original storage position of the target file box in the system, marking the file positioned at the adjacent coordinates as a high-risk suspected object, and improving the monitoring level of the high-risk suspected object.
  2. 2. The profile storage health monitoring method as in claim 1, wherein the gas fingerprint feature vector comprises one or a combination of two gas indices: The acetic acid concentration index is used for representing the hydrolysis acidification degree of hemicellulose in the archival paper; and the furfural concentration index is used for representing the embrittlement degree caused by cellulose dehydration in the archival paper.
  3. 3. The method for monitoring archival storage health based on feature extraction of claim 1, wherein the deployment of the gas sensing component and the corresponding data acquisition logic comprises at least one of the following; the distributed deployment comprises the steps that a plurality of gas sensor nodes are fixedly distributed at different positions of an archive storage layer rack, and a three-dimensional concentration field is constructed through reading differences among the nodes; the movable deployment comprises the steps that the gas sensor is arranged on the moving device or an independent inspection mechanism, and a concentration change track is constructed through continuous sampling data in the moving process of the moving device along a preset path.
  4. 4. The method for monitoring archival storage health based on feature extraction according to claim 3, wherein when the gas sensing component is deployed in a distributed manner, the source localization strategy specifically comprises the following steps: Constructing a three-dimensional discrete concentration field model of the storage space, and mapping real-time concentration readings of each sensor node into model grid points; Estimating the concentration value of the non-grid point area by using a spatial interpolation algorithm to generate a continuous concentration distribution curved surface; Calculating a gradient vector of the concentration distribution curved surface, and searching local maximum points along the gradient ascending direction; and judging the geometric center coordinates corresponding to the local maximum points as three-dimensional space coordinates of the abnormal gas release source.
  5. 5. A profile storage health monitoring method based on feature extraction as in claim 3, wherein said source localization strategy specifically comprises the steps of: The transporting device carries a gas sensor to carry out global inspection on all the archives storage shelves; When the gas characteristic value of the target position is detected to exceed the trigger threshold value, interrupting global inspection, and entering an accurate positioning mode; the moving device takes the target position as an origin, and respectively collects the gas concentrations in different directions in the three-dimensional space where the target position is located; calculating the directional derivative of the gas concentration change rate and the displacement vector of the transporting device, and moving the transporting device to the direction of the fastest concentration increase; repeating the steps of calculating and moving until the detected concentration change rate is changed from positive to negative, and locking the physical coordinates corresponding to the concentration peak point as the position of the target file box.
  6. 6. The method for monitoring the health of archival storage based on feature extraction according to claim 1, wherein the physical isolation area is a closed cabin with an independent exhaust system and a negative pressure environment, and the method further comprises the following steps after the target archival box is transported to the physical isolation area: automatically starting an exhaust system of the physical isolation area, and adjusting the exhaust air quantity to keep the air pressure in the area continuously lower than the air pressure of an external storage space, so as to prevent abnormal air from flowing back; And continuously monitoring the release of the monomer gas of the target file box in the physical isolation area, and recording the concentration rising rate in unit time so as to evaluate the degradation activity of the file.
  7. 7. The method for monitoring the health of archival storage based on feature extraction according to claim 1, wherein the noise reduction process is performed on the gas data, and further comprising the temperature and humidity compensation process is performed on the gas data, and specifically comprising the following steps: Invoking a preset temperature and humidity-sensitivity drift curve, wherein the drift curve describes baseline drift amounts of the gas sensing component under different temperature and humidity environments; determining a corresponding correction coefficient from the drift curve according to the currently acquired temperature and humidity data; and carrying out weighted correction on the acquired gas concentration original data by using the correction coefficient, and eliminating signal errors caused by environmental temperature and humidity fluctuation.
  8. 8. A method for monitoring archival storage health based on feature extraction as set forth in claim 3, wherein the step of enhancing the monitoring level of the high-risk suspected object specifically includes the following execution strategies: If the gas sensing assembly is distributed, the sampling frequency of the sensor nodes in the adjacent areas of the original position of the target file box is increased, and the sampling interval is shortened to a preset emergency sampling period; If the gas sensing component is in mobile deployment, inserting a fixed-point sampling instruction at the adjacent coordinates of the original position of the target file box in the follow-up inspection task planning of the transporting device, and increasing the stay sampling time in the area.
  9. 9. The profile storage health monitoring method based on feature extraction as in claim 1, further comprising the steps of: generating an alarm through a human-computer interaction interface, wherein the alarm content comprises the number of a target archive box, the type of the detected abnormal gas, the concentration peak value and an affected adjacent high-risk archive list; and when the monitoring data in the physical isolation area show that the gas release rate of the target file box exceeds the irreversible damage threshold, generating a paper deacidification or repair suggestion list.
  10. 10. The profile storage health monitoring method based on feature extraction of claim 1, wherein the feature scent library further comprises interfering gas features of non-paper materials; if the matching degree of the current gas characteristic value and the interference gas characteristic of the non-paper material is highest, judging that the non-file damage is abnormal; And triggering no scheduling target file box action and physical isolation step, and generating environment interference early warning information.

Description

Archival storage health monitoring method based on feature extraction Technical Field The invention belongs to the technical field of archive management, and particularly relates to an archive storage health monitoring method based on feature extraction. Background The main components of the paper file are cellulose and hemicellulose, and in the long-term storage process, the paper file is affected by the temperature and humidity, illumination and self acidity of the environment, and slow chemical degradation reaction inevitably occurs to release volatile organic compounds, such as acetic acid generated by hydrolysis of paper, furfural generated by dehydration of cellulose and the like. The existing archive monitoring technology mainly focuses on macroscopic regulation and control of temperature and humidity of a storage space, and the single environmental monitoring means can only reflect external preservation conditions, cannot directly represent the actual aging degree or pathological state of an archive paper body, and is difficult to discover early paper acidification. Moreover, the health examination of the archives often depends on manual spot check or regular turn-over, which not only has low efficiency and narrow coverage, but also can cause secondary physical damage to the fragile ancient books due to frequent manual turn-over. Disclosure of Invention Aiming at the defects in the prior art, the invention provides a archive storage health monitoring method based on feature extraction, so as to solve the technical problems. A archive storage health monitoring method based on feature extraction comprises the following steps: Presetting a file damage characteristic odor library in a central control unit, wherein the file damage characteristic odor library comprises gas fingerprint characteristic vectors corresponding to different file damage types; acquiring environmental gas data of a storage space in real time through a gas sensing assembly, and extracting a current gas characteristic value after noise reduction treatment of the gas data; comparing the current gas characteristic value with the file damage characteristic odor library to calculate the matching degree, judging that an abnormal file exists in the storage space when the matching degree exceeds a preset threshold value, and determining the type of the abnormal gas; Triggering a source positioning strategy in response to an abnormal identification result, calculating three-dimensional space coordinates of an abnormal gas release source in a storage space by using concentration space distribution gradients of gas data based on a deployment mode of a gas sensing assembly, and locking a target archive box; the dispatching and transporting device goes to the three-dimensional space coordinates to grab the target file box and carries the target file box to a preset physical isolation area; And acquiring adjacent file information of the original storage position of the target file box in the system, marking the file positioned at the adjacent coordinates as a high-risk suspected object, and improving the monitoring level of the high-risk suspected object. Preferably, the gas fingerprint feature vector at least comprises one or a combination of the following two gas indexes: The acetic acid concentration index is used for representing the hydrolysis acidification degree of hemicellulose in the archival paper; and the furfural concentration index is used for representing the embrittlement degree caused by cellulose dehydration in the archival paper. Preferably, the deployment mode of the gas sensing component and the corresponding data acquisition logic comprise at least one of the following; the distributed deployment comprises the steps that a plurality of gas sensor nodes are fixedly distributed at different positions of an archive storage layer rack, and a three-dimensional concentration field is constructed through reading differences among the nodes; the movable deployment comprises the steps that the gas sensor is arranged on the moving device or an independent inspection mechanism, and a concentration change track is constructed through continuous sampling data in the moving process of the moving device along a preset path. Preferably, when the gas sensing assembly adopts distributed deployment, the source positioning strategy specifically comprises the following steps: Constructing a three-dimensional discrete concentration field model of the storage space, and mapping real-time concentration readings of each sensor node into model grid points; Estimating the concentration value of the non-grid point area by using a spatial interpolation algorithm to generate a continuous concentration distribution curved surface; Calculating a gradient vector of the concentration distribution curved surface, and searching local maximum points along the gradient ascending direction; and judging the geometric center coordinates corresponding to the local m