Intelligent archival repository management method and system based on multiple modules

Abstract

The invention relates to the technical field of intelligent archival repository management, and discloses an intelligent archival repository management method and system based on multiple modules. The method comprises the following steps: acquiring environment monitoring data such as storehouse temperature and humidity distribution and air quality indexes, and generating an environment state feature vector through feature extraction; inputting the data into an archive storage risk assessment model, outputting a risk level, and generating an environment regulation and control instruction set containing temperature and humidity regulation and air purification parameters according to the risk level; monitoring access request data containing file numbers and access timestamps in real time, and constructing an access frequency matrix to calculate file thermodynamic diagram distribution; a warehouse partition regulation strategy is generated by combining the thermodynamic diagram and the regulation instruction set, and the high-frequency access file is dynamically migrated to an environment stable area; incremental learning is adopted periodically to update model parameters, and risk prediction precision is optimized. The method can improve the intelligent degree of warehouse management and guarantee the safety of file storage.

Description

Technical Field

[0001] This invention relates to the field of intelligent archive storage management technology, specifically to an intelligent archive storage management method and system based on multiple modules. Background Technology

[0002] Archives contain important information such as historical data, academic achievements, and administrative records. Their long-term secure storage is of practical significance for institutional operations, historical preservation, and academic research. Currently, in the management of archive storage facilities, environmental monitoring often suffers from problems such as limited monitoring dimensions and incomplete data coverage. Most management methods only monitor temperature and humidity at a few fixed points within the storage facility, making it difficult to obtain temperature and humidity distribution data across the entire area. Furthermore, there is a lack of systematic collection of air quality indicators that affect archive preservation (such as the concentration of harmful gases and dust content). This results in the failure to detect instances of excessive temperature and humidity or poor air quality in localized areas, leading to slow damage to the paper, writing, or other media materials of the archives and shortening their lifespan.

[0003] In terms of risk assessment and environmental control, traditional management relies heavily on the experience and judgment of managers, lacking standardized assessment systems and intelligent model support. Managers need to judge the risks of file storage based on their personal experience, which is easily influenced by subjective factors. Different personnel have different judgment standards, leading to untimely or misjudgment of risks. Consequently, environmental control measures lack specificity. In most cases, a uniform control model is adopted across the entire area. Regardless of the file storage needs or access frequency in the area, the same temperature and humidity control and air purification parameters are implemented. This not only wastes energy resources but may also fail to meet the file storage needs of specific areas due to excessive or insufficient control in some areas.

[0004] The archival storage and management process also has shortcomings. Current management methods mostly employ a fixed storage location model, failing to dynamically adjust based on the frequency of access. Managers struggle to monitor the access status of each archive in real time and cannot accurately identify the distribution of frequently accessed and infrequently accessed archives. This results in frequently accessed archives potentially being stored in areas with significant environmental fluctuations or remote locations, increasing the time cost for staff to locate archives, reducing access efficiency, and accelerating the wear and tear of frequently used archives due to insufficient environmental stability. Furthermore, the model parameters used for risk assessment are often fixed, failing to consider long-term changes in the storage environment (such as temperature and humidity fluctuations due to seasonal changes, and localized environmental changes caused by equipment aging). The models cannot adapt to dynamic environmental changes, and their risk prediction accuracy gradually decreases over time, making it difficult to effectively support subsequent environmental control and archival management decisions. Summary of the Invention

[0005] The purpose of this invention is to provide a multi-module-based intelligent archive storage management method and system to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides a multi-module-based intelligent archive storage management method, the method comprising:

[0007] Collect environmental monitoring data from the archive storage facility, including temperature and humidity distribution data and air quality index data;

[0008] The environmental monitoring data is subjected to feature extraction processing to generate an environmental state feature vector;

[0009] The environmental state feature vector is input into the archive storage risk assessment model, and the archive storage risk level is output.

[0010] An environmental control instruction set is generated based on the risk level of the archive storage, and the environmental control instruction set includes temperature and humidity adjustment parameters and air purification parameters.

[0011] Real-time monitoring of file access request data, which includes file number and access timestamp;

[0012] Based on the file access request data, a file access frequency matrix is ​​constructed, and the file heatmap distribution is calculated.

[0013] By combining the heat map distribution of the archives and the environmental control instruction set, a warehouse zoning control strategy is generated;

[0014] Dynamically adjust the storage location of files, and migrate frequently accessed files to areas with a stable environment;

[0015] The parameters of the archive storage risk assessment model are periodically updated, and the risk prediction accuracy is optimized by using incremental learning.

[0016] Preferably, the step of performing feature extraction processing on the environmental monitoring data to generate an environmental state feature vector includes:

[0017] The temperature and humidity distribution data are spatiotemporally aligned to generate standardized temperature and humidity time-series data.

[0018] The air quality index data is processed to remove outliers, resulting in a smoothed air quality data sequence.

[0019] The standardized temperature and humidity time series data and the smoothed air quality data sequence are fused with multi-dimensional features to generate the environmental state feature vector.

[0020] Preferably, the step of inputting the environmental state feature vector into the archive storage risk assessment model and outputting the archive storage risk level includes:

[0021] Analyze the temperature and humidity fluctuations and air quality change trends in the environmental state feature vector;

[0022] Based on the preset material degradation threshold, match the degree of deviation between the current environmental data and the material's tolerance range;

[0023] The degree of deviation is classified and quantified using a decision tree classification algorithm to generate the risk level of the archive storage.

[0024] Preferably, the step of generating an environmental control instruction set based on the archive storage risk level includes:

[0025] When the risk level of the archive storage is Level 1, the environmental control instruction set containing only humidity fine-tuning parameters is generated;

[0026] When the risk level of the archive storage is level two, an environmental control instruction set that includes both temperature and humidity adjustment parameters and local ventilation parameters is generated.

[0027] When the risk level of the archive storage is level three, a full warehouse environment reset instruction is generated. The full warehouse environment reset instruction includes forced dehumidification parameters and air circulation and purification parameters.

[0028] Preferably, the step of constructing a file access frequency matrix based on the file access request data and calculating the file heatmap distribution includes:

[0029] The frequency of occurrence of the file number is counted according to a preset time window to generate the file access frequency matrix;

[0030] Spatial interpolation is performed on the file access frequency matrix to generate a continuously distributed access density surface;

[0031] The access density surface is mapped to the three-dimensional coordinate space of the warehouse to generate the archive heat map distribution.

[0032] Preferably, the step of generating a warehouse zoning control strategy by combining the archive heat map distribution and the environmental control instruction set includes:

[0033] Identify high-frequency and low-frequency access regions in the heatmap distribution of the archives;

[0034] The high-frequency access region is overlapped with the priority control region in the environmental control instruction set.

[0035] Based on the overlap analysis results, warehouse control priorities are determined, and warehouse zone control strategies are generated for execution in different time periods.

[0036] Preferably, the dynamic adjustment of file storage location, migrating frequently accessed files to a stable environment, includes:

[0037] Based on the warehouse zoning control strategy, storage areas that meet environmental stability standards are marked;

[0038] Filter the target file numbers whose access frequency exceeds a threshold from the file access frequency matrix;

[0039] The entity files corresponding to the target file number are migrated in batches to the storage area where the environmental stability meets the standards.

[0040] Preferably, the periodic updating of the parameters of the archive storage risk assessment model, using incremental learning to optimize risk prediction accuracy, includes:

[0041] Collect correlation data between historical environmental control records and actual archive damage records;

[0042] Calculate the nonlinear correlation coefficient between changes in environmental parameters and the rate of archive damage;

[0043] The nonlinear correlation coefficient is injected as a weighting factor into the parameter update process of the archive storage risk assessment model.

[0044] Preferably, the method further includes:

[0045] Establish a log for changes in file storage location, recording the source and target coordinates for each migration operation;

[0046] Perform timestamp correlation analysis between the file storage location change log and the environmental control instruction set;

[0047] The regional stability judgment criteria in the warehouse zoning control strategy were revised based on the correlation analysis results.

[0048] Preferably, the present invention also includes an intelligent archive storage management system, the system including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor, when executing the computer program, implements the steps of the intelligent archive storage management method described above.

[0049] Compared with the prior art, the beneficial effects of the present invention are:

[0050] In the environmental monitoring of archive storage facilities, by collecting data on temperature and humidity distribution and air quality indicators, it is possible to achieve a comprehensive perception of the entire environment of the storage facility. This is no longer limited to traditional single-point monitoring. It can promptly detect abnormal temperature and humidity or air quality problems in local areas, avoid damage to archives due to undetected local environmental hazards, ensure the overall stability of the archive storage environment, and provide a good foundation for the long-term preservation of archives.

[0051] The environmental state feature vector generated by feature extraction can transform scattered environmental monitoring data into a structured and analyzable vector form, making the environmental data more in line with the input requirements of the archive storage risk assessment model, reducing the interference of data redundancy on model calculation, enabling the model to more accurately capture key information in the environmental data, accurately identify potential risks under different environmental conditions, and avoid deviations in the model's judgment of environmental conditions due to data format mismatch or missing key information.

[0052] The application of the archive storage risk assessment model can output the archive storage risk level in an objective and standardized way, replacing the traditional judgment method that relies on the subjective experience of managers. It eliminates the impact of differences in judgment standards among different personnel, making the risk assessment results more consistent and reliable. It can clearly classify the security level of archive storage according to the risk level, providing a clear direction for subsequent environmental control and avoiding the lack of targeted control measures due to ambiguity in risk assessment.

[0053] The environmental control instruction set generated based on risk level includes temperature and humidity adjustment parameters and air purification parameters. It can formulate differentiated control plans for different risk levels. Higher-risk areas can adopt more precise temperature and humidity adjustment and stronger air purification measures, while lower-risk areas can appropriately adjust the control intensity. This avoids the energy waste caused by traditional uniform control across the entire region, while ensuring that the control measures in each area match the actual risk situation and improve the effectiveness of environmental control.

[0054] By monitoring archive access request data in real time and obtaining archive numbers and access timestamps, we can keep track of the usage of each archive in real time, clearly understand the access frequency and time patterns of each archive, provide accurate data support for the construction of the archive access frequency matrix, avoid the inability to judge the usage popularity of archives due to lack of access data, and ensure that the understanding of archive usage is more in line with actual management needs.

[0055] The archive access frequency matrix constructed based on access request data and the calculated archive heat map distribution can transform abstract access data into an intuitive and visual distribution form, clearly showing the specific locations of high-frequency and low-frequency access archives in the warehouse. This allows managers to quickly grasp the spatial distribution characteristics of archive usage frequency and identify the concentrated areas of high-frequency access archives without manual statistics and analysis, providing an intuitive basis for subsequent warehouse zoning control and archive storage location adjustment.

[0056] By combining heat map distribution with environmental control instruction sets to generate warehouse zoning control strategies, environmental control can be integrated with the frequency of archive usage. For areas with high-frequency access to archives, more stable temperature and humidity control and air purification measures are prioritized to ensure a more stable environment for high-frequency access archives. At the same time, low-frequency access areas are reasonably controlled according to the actual risk level, further optimizing the allocation of energy resources and avoiding waste caused by excessive control of low-frequency areas, thus achieving a synergistic adaptation between environmental control and archive usage needs.

[0057] Dynamically adjusting the storage location of files and migrating frequently accessed files to a stable environment can place frequently used files in a more suitable preservation environment, reducing file damage caused by environmental fluctuations. At the same time, storing frequently accessed files in a stable environment also makes it easier for staff to quickly find and access them, shortening access time, improving file usage efficiency, and avoiding the problem that improper storage location of frequently accessed files affects usage and exacerbates damage.

[0058] Periodically updating the parameters of the archive storage risk assessment model using incremental learning allows the model to continuously absorb new environmental data and management experience during use, adapting to long-term changes in the warehouse environment. This maintains high risk prediction accuracy without requiring model retraining, reducing model maintenance time and resource costs, ensuring the model operates efficiently in the long term, and providing reliable support for archive storage risk assessment and environmental control. It also avoids a decline in prediction accuracy due to fixed model parameters failing to adapt to environmental changes, thus affecting overall management effectiveness. Attached Figure Description

[0059] Figure 1 This is a schematic diagram illustrating the working principle of a multi-module-based intelligent archive storage management method and system as described in this invention.

[0060] Figure 2 A flowchart for generating environmental state feature vectors;

[0061] Figure 3 This is a diagram illustrating the characteristic vector analysis of environmental states.

[0062] Figure 4 This is a flowchart for calculating the distribution of archive heatmaps. Detailed Implementation

[0063] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0064] Please see Figure 1 This invention provides a multi-module intelligent archive storage management method and system. The method includes: collecting environmental monitoring data, including temperature and humidity distribution data and air quality index data, through a sensor network deployed in various areas of the storage facility. The temperature and humidity distribution data is acquired by multiple temperature and humidity sensors at a preset sampling frequency, forming a spatiotemporal sequence. The air quality index data includes parameters such as PM2.5 and VOCs. Next, feature extraction processing is performed on the environmental monitoring data. Signal processing algorithms are used to eliminate noise and extract key features, generating a high-dimensional environmental state feature vector. This vector integrates the spatiotemporal variation patterns of temperature and humidity and the dynamic trend of air quality. The environmental state feature vector is input into a pre-trained archive storage risk assessment model. This model is built based on machine learning algorithms. By analyzing the environmental fluctuation factors in the feature vector, it outputs a quantified archive storage risk level. The risk level is divided into multiple levels, indicating the potential threat level of the current environment to the archive materials. Based on the risk level, the system automatically generates an environmental control instruction set, which includes specific temperature and humidity adjustment parameters and air purification parameters, and implements environmental adjustments through control actuators. Simultaneously, the system monitors archive access request data in real time. This data, derived from the logs of the warehouse management system, includes archive numbers and access timestamps, used to analyze archive usage patterns. Based on the access request data, an archive access frequency matrix is ​​constructed. This matrix statistically analyzes access frequency in time windows and calculates archive heatmap distribution using spatial interpolation algorithms, visually displaying high-frequency access areas. Combining the archive heatmap distribution and environmental control instruction sets, the system performs overlap analysis to generate a warehouse zoning control strategy. This strategy divides the warehouse into control zones of different priorities and formulates time-segmented execution plans. Dynamically adjusting archive storage locations is based on the heatmap and strategy results, automatically migrating high-frequency access archives to environmentally stable areas to reduce environmental interference during access. The system periodically updates the parameters of the archive storage risk assessment model, employing incremental learning and combining historical control data and actual damage records to optimize risk prediction accuracy and ensure the model adapts to environmental changes.

[0065] Example 1: See Figure 2During feature extraction, environmental monitoring data is standardized and fused, and temperature and humidity distribution data undergoes spatiotemporal alignment. Since the models and sampling cycles of temperature and humidity sensors deployed in the warehouse may differ, the raw data they collect may have discrepancies in timestamps and spatial locations. Spatiotemporal alignment is addressed through a data coordination module. This module has a built-in unified time reference clock, and all sensor data streams are stamped with a standardized timestamp based on this clock upon input. For spatial location, the three-dimensional coordinates of each sensor (e.g., X, Y, Z coordinates relative to the origin of a corner in the warehouse) are already entered into the system database when deployed. When data is input, the system retrieves the corresponding coordinate information based on its device ID, thus binding each data point to a specific physical location within the warehouse. Through this processing, the originally scattered and heterogeneous temperature and humidity readings are integrated into a clearly structured standardized temperature and humidity time-series data matrix. The rows of this matrix represent continuous time points, the columns represent various fixed monitoring locations in the warehouse, and each cell is filled with the temperature and humidity readings at the corresponding time and location, forming a complete dataset describing the spatiotemporal evolution of the warehouse environment. The processing of air quality index data focuses on data smoothness and consistency. This type of data may come from equipment monitoring PM2.5, VOCs (volatile organic compounds), or carbon dioxide concentrations. These readings are easily affected by transient interference (such as brief passage of people or fluctuations in the equipment itself), resulting in outliers. Outlier removal is achieved using an algorithm based on dynamic thresholds. The system maintains a sliding window for each air quality parameter, containing historical readings over a recent period. When real-time data flows in, the algorithm calculates the relationship between the data point and the mean and standard deviation of the data within the window. If a data point deviates from the mean by more than a preset standard deviation multiple (this multiple can be configured according to the characteristics of different sensors), it is considered an outlier. This outlier is not simply discarded but replaced with the moving average of the valid data within the sliding window, thereby preserving the data trend to the maximum extent while eliminating noise. After this processing, a smoothed air quality data sequence is generated. This sequence can more realistically reflect the overall trend of air quality changes in the warehouse and avoid misleading subsequent judgments due to accidental fluctuations.

[0066] The key step in generating the environmental state feature vector is to fuse the two types of processed data. Although the standardized temperature and humidity time series data and the smoothed air quality data series have been cleaned, their data structures and physical meanings are different. Direct splicing cannot effectively utilize their inherent correlation. The multi-dimensional feature fusion process extracts features from the two types of data separately. For the standardized temperature and humidity time series data, in addition to directly using the readings, the statistical characteristics of each sensor location over a period of time are calculated, such as the average, maximum, minimum, and volatility (e.g., standard deviation). These statistics can summarize the temperature and humidity stability at that location. For the smoothed air quality data series, the focus is on extracting its trend features. For example, by calculating the difference between consecutive time points, the air quality is captured to determine whether it is improving, deteriorating, or remaining stable. Subsequently, these features extracted from different dimensions are combined to form a comprehensive environmental state feature vector. Each dimension of this vector represents a specific attribute of a certain aspect of the warehouse environment, such as "average temperature of area A", "humidity volatility of channel B", and "overall VOCs concentration trend". This vector is no longer just a collection of raw data, but carries refined environmental state information.

[0067] The entire feature extraction process runs continuously on a dedicated data preprocessing server. It receives real-time data streams from a distributed sensor network via a high-speed data bus. After processing, the generated environmental state feature vectors are pushed to a message queue for subsequent subscription and use by the archive storage risk assessment model. This pipelined processing method ensures low latency and high throughput in the conversion from raw data to feature vectors, providing a foundation for the real-time response of the entire intelligent management system. The dimensions of the feature vectors are pre-designed according to the warehouse partitions and sensor layout to ensure that all environmental factors that need to be monitored are covered without omission. At the same time, the algorithm parameters for feature extraction can be configured through the management interface, enabling the system to adapt to the specific environment and archive protection requirements of different warehouses.

[0068] Example 2: The operation of the archival storage risk assessment model involves environmental data analysis and hierarchical quantification. The model analyzes key indicators in the environmental state feature vector, including the amplitude of temperature and humidity fluctuations and the trend of air quality changes. The calculation of temperature and humidity fluctuation amplitudes does not simply rely on instantaneous readings, but analyzes the distribution pattern of data within a certain time window. For example, it quantifies instability by calculating the standard deviation or range of temperature and humidity data at each monitoring point within a specific time period. An environment that fluctuates within a small range is considered more stable than an environment that fluctuates frequently and drastically, even if their average values ​​may be similar. The analysis of air quality change trends focuses on identifying the long-term direction of parameter movement. The system uses methods such as linear regression to fit the smoothed air quality data sequence. The slope of the resulting fitted line intuitively reflects whether the parameter is rising, falling, or stable. This trend analysis helps to predict potential environmental deterioration risks, rather than just focusing on the current instantaneous value.

[0069] The parsed data needs to be matched with preset degradation thresholds for archival materials. The archives store a variety of media, from common paper documents and drawings to more environmentally sensitive media such as film and magnetic media. Each material has its unique material properties, corresponding to different temperature and humidity tolerance ranges and safe storage conditions. These tolerance ranges are based on materials science and long-term preservation practices and are pre-stored in the system's knowledge base. The matching process essentially compares the currently monitored environmental data with the ideal environmental range set for the current area's archives in the knowledge base and calculates the degree of deviation. For example, if the archives in a certain area are mainly paper materials, and the set ideal humidity range is 45% to 55%, if the currently monitored humidity is consistently above 60%, the system will calculate a positive deviation value, and the deviation increases with the magnitude and duration of the actual value exceeding the threshold range.

[0070] The decision tree classification algorithm is used to classify and quantify the calculated deviation. The structure of the decision tree model is pre-trained using historical data and contains a series of judgment rules. The model takes various deviation degrees and fluctuation frequencies extracted from the environmental state feature vector as input. Starting from the root node, it makes judgments according to preset conditions (e.g., "Does the current area's average humidity deviate from the ideal upper limit by more than 5% and last for more than 2 hours?"). Based on the judgment results, it traverses downwards along different branches until it reaches a leaf node. Each leaf node corresponds to a preset risk level for file storage. For example, the final output may be a classification result such as "Level 1 Risk", "Level 2 Risk", or "Level 3 Risk". The advantage of this classification method is that its judgment process is relatively clear, easy to understand and trace.

[0071] Based on the risk level output by the archival storage risk assessment model, the system generates a targeted environmental control instruction set. When the model determines the risk level to be Level 1, it indicates that the environmental parameters have only slightly deviated and may not yet pose an immediate threat to the archives, but there is a tendency to worsen. At this time, the generated environmental control instruction set is relatively simple, usually only including humidity fine-tuning parameters. The instructions will specify the storage area that needs fine-tuning, the target humidity value, and the equipment to perform the fine-tuning. The adjustment range will be controlled within a small range to avoid new drastic fluctuations in the environment due to the control action itself. When the model determines the risk level to be Level 2, it indicates that the environmental parameters have deviated moderately, or the deviation of a single parameter has increased, or multiple parameters have slight problems at the same time. At this time, the risk of damage to the archives increases significantly. The generated environmental control instruction set will include both temperature and humidity control parameters and local ventilation parameters. The temperature and humidity control parameters will set a more specific target range and the adjustment rate to achieve that range, while the local ventilation parameters will specify to activate ventilation equipment in a specific area (usually the risk-identified area), set the air exchange rate and running time, in order to improve local air quality by introducing fresh air or internal air circulation, and to assist in temperature and humidity control. When the model determines the risk level to be Level 3, it indicates that the environmental parameters have deviated significantly from the safe range, possibly due to equipment failure, sudden changes in the external environment, or continuous cumulative effects, and the archives face an imminent risk of damage. At this time, the system will generate the highest-level full-warehouse environment reset command. This command set includes powerful measures such as forced dehumidification parameters and air circulation and purification parameters. The forced dehumidification parameters may instruct all dehumidification equipment to operate at maximum power, with the goal of reducing the overall humidity of the warehouse to below the safe threshold in the shortest possible time. The air circulation and purification parameters will instruct the activation of the entire warehouse's fresh air system and air purification system to carry out high-intensity, large-scale air replacement and pollutant removal. This command aims to quickly reverse the environmental deterioration, but it also has the greatest energy consumption and impact on the stability of the warehouse. Therefore, it is only activated at the highest risk level. The entire process, from risk analysis and assessment to instruction generation, is an automated closed loop. The risk assessment model is typically deployed on a dedicated analysis server, continuously receiving environmental state feature vectors from the feature extraction module. The assessment results and generated control instructions are sent to the actuators of the environmental control system via the system bus to control the operation of equipment such as air conditioners, dehumidifiers, and ventilation fans. At the same time, each risk assessment result and corresponding control instruction is recorded in a log. This log data can be used for subsequent model optimization and audit trails, enabling the entire system to continuously improve itself over time.

[0072] See Figure 3This figure illustrates the transformation process from raw environmental monitoring data to structured feature vectors. The chart is divided into two parts: the upper part uses a horizontal bar chart to visually display the scoring results of five key environmental characteristic indicators, including temperature fluctuation, temperature stability, PM2.5 trend, VOCs change, and spatial evenness. These scores reflect the quantitative performance of the environmental monitoring data after feature extraction processing. The high scores for VOCs change and PM2.5 trend indicate good air quality monitoring data quality, while the relatively low score for spatial evenness suggests uneven environmental distribution within the storage facility. The lower part, a time series analysis, shows the decomposition of environmental characteristic values ​​over time, clearly presenting the dynamic characteristics of trend components, seasonal components, and residual components. This multi-dimensional feature fusion is a concrete manifestation of the environmental state feature vector generation process, providing standardized input data for subsequent archival storage risk assessment models and offering a scientific basis for risk assessment and environmental control decisions.

[0073] Example 3: See Figure 4 The construction of the archive access frequency matrix and the calculation of the heatmap rely on the systematic analysis of continuously generated archive access request data. The system counts the frequency of archive numbers according to a preset time window. The length and sliding step of this time window are configurable parameters that can be adjusted according to the warehouse management cycle (such as daily, weekly, or monthly). Within each time window, the system scans all access request logs and calculates the total number of times each unique archive number is accessed within that window. These statistical data are organized into a two-dimensional matrix, where the row index of the matrix uniquely corresponds to each archive number, and the column index of the matrix corresponds to consecutive time windows in chronological order. Each element value in the matrix is ​​the access frequency of a specific archive within a specific time window, thus generating the archive access frequency matrix. This matrix dynamically reflects the utilization status of different archives over time. Spatial interpolation of the generated archive access frequency matrix is ​​crucial for transforming it into spatial distribution information. Since the archive access frequency matrix itself only contains archive number and frequency information, and the archive number and the specific storage location of the archive entity in the warehouse (usually represented by three-dimensional coordinates) are associated through the background database, the access frequency of each archive can be regarded as a discrete sample occurring at its storage coordinate point. In order to obtain a continuous access density distribution of the entire warehouse space, a spatial interpolation algorithm is required. This process treats each archive storage point as a sample point, its access frequency as a sample value, and estimates the access density at any location in the warehouse through the interpolation algorithm, generating a continuously distributed access density surface.

[0074] The density surface is constructed using a kernel function-based estimation method to obtain dimensionlessly consistent physical density. Let the coordinates of any point within the warehouse space be... The estimated access density at this point Given by the following formula:

[0075] ;

[0076] Where: symbol This represents the total number of files accessed within that time window, i.e., the number of sample points with non-zero access frequency. (Symbol) Indicates the first The known coordinate vector of the accessed file in the three-dimensional space of the storage room. (Symbol) It is the first The frequency of access to a file within that time window is an observed value; it is a dimensionless pure quantity. (Symbol) This is a key parameter, called the bandwidth or smoothing parameter, which has the dimension of length [L] and is used to control the smoothing range of the kernel function. (Symbol) It is a standard kernel function whose independent variable is a dimensionless distance ratio. The function itself satisfies the normalization condition, that is, its integral over the entire space is 1. This makes It has the dimension of [L]⁻³. Symbol Indicates the estimated point to sample point The Euclidean distance between them. Therefore, each term in the summation term on the right-hand side of the formula has the dimension of [degree] / [length]³, which is the same as the visit density on the left-hand side. The dimensions are completely consistent. This estimated value Calculations are performed at each point in the warehouse space to ultimately form a continuous access density surface.

[0077] This access density surface is then mapped onto the warehouse's three-dimensional coordinate space. Using visualization technology, the density values ​​are mapped to different colors or brightness levels, generating an intuitive archive heatmap distribution. Warmer colors (e.g., red, orange) indicate more frequent archive access, while cooler colors (e.g., blue, green) indicate less frequent access. Combining the archive heatmap distribution with the previously generated environmental control instruction set, the warehouse zoning control strategy is generated. The system automatically identifies high-frequency and low-frequency access areas in the archive heatmap distribution. This identification process typically sets dynamic thresholds based on calculated access density values. For example, areas with access densities higher than the overall warehouse average density by a certain percentage (e.g., 1.5 times) are designated as high-frequency access areas, while areas with densities lower than the average density by a certain percentage are designated as low-frequency access areas. Simultaneously, the environmental control instruction set itself implicitly contains spatial priority information; for example, the instruction set may specify areas requiring priority temperature and humidity control (usually areas with higher current environmental risks).

[0078] Overlap analysis is performed between high-frequency access areas and priority control areas indicated by centralized environmental control instructions. This analysis is conducted on a digital floor plan or 3D model of the storage facility, calculating the intersection of the two sets of areas through spatial geometric operations. The results reveal which areas are both hotspots for archival access and areas requiring key environmental control, and which areas possess only one of these attributes. Storage facility control priorities are assigned based on the overlap analysis results. Areas with complete overlap are assigned the highest control priority because these areas must meet both the functional needs of frequent personnel access to archives and ensure the environmental security of archive storage, requiring the most control resources and the most timely response. Areas with partial overlap or proximity are assigned medium priority. Areas with only high-frequency access but a stable environment, or areas requiring environmental control but with very low access frequency, are assigned lower priority. The final product is a time-segmented warehouse zoning control strategy. This strategy not only clarifies the control sequence for different areas but also plans the specific intensity of control measures to be taken during different time periods (such as working hours, nights, and weekends). For example, precise temperature and humidity control is implemented in high-priority areas during peak visit periods, while preventative environmental adjustments are made over a wider area during off-peak periods, thereby achieving a balance between energy efficiency and archive preservation. The entire analysis process is executed by the spatial analysis engine, and the generated strategy is then distributed to the environmental equipment control system for scheduling and execution.

[0079] Imagine a large, intelligent archive storage facility divided into six zones, A through F. During a work week, from Monday to Friday, the storage management system records all file access requests, with the system set to perform daily statistics. The system calculates the frequency of each file number daily, generating a file access frequency matrix. The rows of this matrix represent thousands of unique file numbers, and the columns represent the five time windows from Monday to Friday. A data point in the matrix might be "File Number DA-2023-0158" with a value of "5" in the "Wednesday" column, indicating that the file was accessed five times on Wednesday. Conversely, "File Number DA-2020-1047" might have a value of "1" on Friday, indicating that it was accessed only once. Many files have an access frequency of 0 on most days. This massive matrix visually displays the activity level of different files on different days, but it is discrete and does not directly reveal spatial patterns. Spatial interpolation is performed on the access frequency matrix of the archives. Each archive has its exact physical location coordinates in the storage room. For example, archive DA-2023-0158 is located in the 3rd row, 2nd shelf, 1st floor of section B, with coordinates (B,3,2,1). Its access frequency "5" on Wednesday is assigned to this coordinate point. The system needs to estimate a continuous access density for the entire storage room space. It uses a distance-based weighted average method. For any point in the storage room (e.g., coordinates (C,2,1,2)), the system searches for the location and frequency of all accessed archives within a certain distance (e.g., within a radius of 10 meters). The closer the archive is to the point, the greater its influence (i.e., weight) on the estimated density of that point; the farther away, the smaller the influence. Through this calculation, each point in the storage room is assigned a calculated access density value, thus generating a continuously varying access density surface covering the entire storage room plane. This surface may show a significant density peak in the middle of sections B and D on Wednesday. Mapping the access density surface to the warehouse's three-dimensional coordinate space generates a heat map distribution of the archives. On the warehouse's digital twin model or floor plan, access density values ​​are converted into colors, for example, dark blue represents areas with 0 access density, transitioning through green, yellow, and orange, and finally bright red represents the areas with the highest access density. Observing Wednesday's heat map, one can clearly see striking red blocks in the center of areas B and D, while the peripheral areas such as areas A and F are mostly blue or green. This map intuitively reveals the spatial hotspots of archive utilization. Combining this archive heat map distribution with the environmental control instruction set generated by the environmental risk assessment at the same time, the system begins to generate a warehouse zoning control strategy. The environmental control instruction set may indicate that due to continuous rain, the humidity in areas A and B on the east side of the warehouse is too high, requiring the activation of dehumidifiers for intervention. These areas are marked as "priority control areas." The system automatically identifies high-frequency access areas (areas B and D) and low-frequency access areas (areas A, C, E, and F) in the heat map distribution.Overlap analysis of high-frequency access areas and priority control areas revealed that Area B is both a hotspot for access and a key area for environmental control, making it a completely overlapping area; Area D is a hotspot for access, but its current environment is stable, so it is not a priority control area; Area A is a priority control area, but its access frequency is very low. Based on this overlap analysis, warehouse control priorities were assigned as follows: Area B was given the highest priority, requiring immediate and precise environmental control to reduce humidity while ensuring that the control process does not affect frequent storage and retrieval operations by staff; Area A had the second lowest priority and could undergo more intensive dehumidification operations due to less staff activity; Area D had the next lowest priority and only needed to maintain its existing environmental parameters; the other areas were subject to routine monitoring. The final result is a warehouse zone control strategy implemented in different time periods. The strategy may stipulate that during the morning peak visit period from 9:00 to 11:00 on Wednesday, a low-noise, gradual dehumidification mode will be used in Zone B to avoid strong winds disturbing staff, and the air conditioning in Zone B will be set to a more precise constant temperature and humidity mode; during the midday low visit period from 12:00 to 2:00 in the afternoon, a high-power dehumidification mode will be activated in Zone A to quickly reduce the overall humidity in the area; for Zone D, environmental monitoring will be maintained throughout the day, and no active intervention will be carried out for the time being.

[0080] Example 4: The process of dynamically adjusting the storage location of archives is based on the warehouse zoning control strategy and archive access frequency analysis. According to the generated zoning control strategy, the system needs to mark storage areas that meet environmental stability standards. This marking is not a one-time judgment, but rather based on continuous monitoring data over a period of time (e.g., seven consecutive days). Evaluation criteria include, but are not limited to, whether the temperature fluctuation range of the area is consistently controlled within ±2 degrees Celsius, whether the relative humidity fluctuation does not exceed ±5%, and whether the air quality index (e.g., PM2.5, VOCs concentration) is consistently below the set safety threshold. Only areas that simultaneously meet all stability indicators are marked by the system as preferred locations suitable for storing frequently accessed archives. From the continuously updated archive access frequency matrix, the system filters out archive numbers whose access frequency exceeds a dynamic threshold. This threshold is not a fixed value, but is calculated based on the access frequency distribution of all archives in the most recent statistical period (e.g., one month). For example, archives ranking in the top 15% by access frequency are defined as frequently accessed archives, and their corresponding archive numbers are listed as target archive numbers. After identifying the target files and target region, the system will generate a detailed migration task list. This list includes the complete file number list to be migrated, the current source storage coordinates, and the planned target storage coordinates. See Table 1 for a simplified task list.

[0081] Table 1: List of Archives Relocation Tasks

[0082] The migration priority is determined by the system based on the frequency of access to the archives and the environmental risk level of the source area. Automated track or robot systems that perform migration operations will process tasks according to priority. Before physical migration begins, the system will use RFID or QR code scanning technology to perform a pre-outbound integrity check on the archives on the list. Only after confirming that the archives match the system records will the automated equipment be instructed to pick up and transport them. After the migration is completed, the location information in the archive management database will be updated in real time to ensure the accuracy of subsequent access operations. The process of periodically updating the parameters of the archive storage risk assessment model is closely linked to the above dynamic adjustment. The system will continuously collect correlation data between historical environmental control records and actual archive damage records. The environmental control records detail the content of each environmental control instruction set issued, the execution time, the equipment involved, and the environmental parameter change curves after execution. Actual archive damage records come from regular archive inspection reports. Inspectors use professional equipment (such as acid testing pens to test paper acidity and high-resolution scanners to detect fading of handwriting) or visual inspection to record the physical and chemical changes of the archives they find into the system and associate them with specific archive numbers. Calculating the nonlinear correlation coefficient between changes in environmental parameters and the rate of archival damage is a core step in model updating. This process is not a simple linear regression, but rather employs analytical methods capable of capturing complex relationships. For example, mutual information is used to quantify the statistical dependence between temperature and humidity fluctuation sequences and the probability of specific types of damage (such as paper brittleness and mold). The higher the mutual information value, the stronger the predictive power of changes in that environmental parameter on archival damage. This relationship may be lagged and nonlinear. The calculated nonlinear correlation coefficient (or similar indicator) is transformed into weighting factors, which determine which environmental features should be given higher importance in the next incremental learning process of the model.

[0083] The parameter update process of injecting weighting factors into the archival storage risk assessment model is implemented through an incremental learning algorithm. When new correlated data accumulates to a certain scale (e.g., a quarter), the system initiates a model fine-tuning process. This process does not retrain the model using all historical data, as that would be computationally too costly. Instead, it makes minor adjustments to the model's existing parameters based on new data batches and calculated weighting factors. For example, during gradient descent optimization, the gradient brought by new data points is amplified or reduced according to its corresponding weighting factor, enabling the model to learn more quickly those environmental patterns strongly correlated with archival damage, while gradually diminishing the influence of accidental correlations. In this way, the archival storage risk assessment model can continuously adapt to the environmental characteristics of specific storage facilities and changes in archival materials over time, continuously optimizing its risk prediction accuracy and forming a self-improving closed-loop management system.

[0084] Example 5: The establishment and correlation analysis of archive storage location change logs is an important feedback mechanism for the system to achieve self-optimization. The system creates a detailed log record for each archive migration operation performed by automated equipment. Each record contains a unique migration event ID, the timestamp of the operation trigger, the archive number to be migrated, the source storage coordinates of the archive before migration (e.g., represented by a region-row-shelf-layer coding system), the target storage coordinates after migration, and the strategy batch number that initiated the migration task. These log records are persistently stored in a dedicated table in a relational database, forming a complete archive location change history arranged in chronological order. This log table supports efficient query and aggregation analysis, enabling the system to trace back the location information of any archive at any point in time, or to statistically analyze all migration activities that occurred in a certain area within a specific time period. The core of uncovering causal relationships lies in performing timestamp correlation analysis between the file storage location change log and the environmental control command set. The environmental control command set is also fully recorded by the system, including the timestamp of the command set's generation, the command content (such as the target area, equipment action, parameter settings), and the risk assessment event ID that triggered the command set. The correlation analysis is achieved through a data processing workflow that aligns the two events on the timeline. For example, it finds all file migration operations that occur within a specific time window (such as 24 hours or 48 hours) after a certain environmental control command set is issued. The focus of the analysis is to identify whether the migration events densely follow a specific type of control command in time sequence, and whether the target area of ​​the migration highly overlaps with the control area specified in the command set.

[0085] The results of correlation analysis are used to revise the regional stability judgment criteria in the warehouse zoning control strategy. For example, the system may find a pattern through analysis: after issuing high-intensity dehumidification commands to "Zone B" multiple times, a large number of files are always moved to "Zone B" in a short period of time. This indicates that the system frequently judges "Zone B" as an area with satisfactory environmental stability. However, in-depth analysis of the subsequent environmental data of "Zone B" may show that despite the forced control, the humidity stability of this area is still poor, and the humidity value will rebound to a high level quickly after the control ends. This pattern suggests that the current regional stability judgment criteria may be flawed, and may rely too much on the instantaneous environmental readings after control, while ignoring the evaluation of the persistence of the control effect. Based on this insight, the system will automatically initiate a process to revise the stability judgment criteria. The revision may involve adjusting the time scale of the assessment. For example, the judgment criterion for "environmental stability meeting the standard" may be changed from "parameters meeting the standard within one hour after the completion of the control command" to "parameters continuously meeting the standard within twenty-four hours after the completion of the control command". Or, more complex comprehensive evaluation indicators may be introduced, such as incorporating the volatility of environmental parameters into the assessment instead of just the instantaneous value, requiring that the target area not only meets the average value during the observation period, but its standard deviation must also be lower than a certain more stringent threshold. The revised criteria will be written into the configuration file of the strategy generation engine as a new basis for generating warehouse zoning control strategies.

[0086] This log-based correlation analysis-based feedback mechanism enables the management system to learn from historical operations. It no longer mechanically executes preset rules but optimizes its decision-making logic by observing the actual consequences of its decisions. For example, if the system observes that files migrated to a certain area frequently trigger new control commands in their microenvironment within a week, the system may lower the stability rating of that area in the next decision, even if its instantaneous readings appear acceptable. This avoids repeatedly migrating files to an actually unstable area, reducing unnecessary mechanical wear and file disturbance. The entire implementation relies on the database system's efficient storage and retrieval capabilities for massive amounts of operation logs, and the background analysis program's ability to identify time-series event patterns, ultimately allowing the warehouse management strategy to dynamically approximate actual environmental conditions and storage requirements.

[0087] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0088] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A multi-module-based intelligent archive storage management method, characterized in that, Includes the following steps: Collect environmental monitoring data from the archive storage facility, including temperature and humidity distribution data and air quality index data; The environmental monitoring data is subjected to feature extraction processing to generate an environmental state feature vector; The environmental state feature vector is input into the archive storage risk assessment model, and the archive storage risk level is output. An environmental control instruction set is generated based on the risk level of the archive storage, and the environmental control instruction set includes temperature and humidity adjustment parameters and air purification parameters. Real-time monitoring of file access request data, which includes file number and access timestamp; Based on the file access request data, a file access frequency matrix is ​​constructed, and the file heatmap distribution is calculated. By combining the heat map distribution of the archives and the environmental control instruction set, a warehouse zoning control strategy is generated; Dynamically adjust the storage location of files, and migrate frequently accessed files to areas with a stable environment; The parameters of the archive storage risk assessment model are periodically updated, and the risk prediction accuracy is optimized by using incremental learning.

2. The intelligent archive storage management method according to claim 1, characterized in that, The step of performing feature extraction processing on the environmental monitoring data to generate an environmental state feature vector includes: The temperature and humidity distribution data are spatiotemporally aligned to generate standardized temperature and humidity time-series data. The air quality index data is processed to remove outliers, resulting in a smoothed air quality data sequence. The standardized temperature and humidity time series data and the smoothed air quality data sequence are fused with multi-dimensional features to generate the environmental state feature vector.

3. The intelligent archive storage management method according to claim 2, characterized in that, The step of inputting the environmental state feature vector into the archive storage risk assessment model and outputting the archive storage risk level includes: Analyze the temperature and humidity fluctuations and air quality change trends in the environmental state feature vector; Based on the preset material degradation threshold, match the degree of deviation between the current environmental data and the material's tolerance range; The degree of deviation is classified and quantified using a decision tree classification algorithm to generate the risk level of the archive storage.

4. The intelligent archive storage management method according to claim 3, characterized in that, The step of generating an environmental control instruction set based on the archive storage risk level includes: When the risk level of the archive storage is Level 1, the environmental control instruction set containing only humidity fine-tuning parameters is generated; When the risk level of the archive storage is level two, an environmental control instruction set that includes both temperature and humidity adjustment parameters and local ventilation parameters is generated. When the risk level of the archive storage is level three, a full warehouse environment reset instruction is generated. The full warehouse environment reset instruction includes forced dehumidification parameters and air circulation and purification parameters.

5. The intelligent archive storage management method according to claim 1, characterized in that, The step of constructing a file access frequency matrix based on the file access request data and calculating the file heatmap distribution includes: The frequency of occurrence of the file number is counted according to a preset time window to generate the file access frequency matrix; Spatial interpolation is performed on the file access frequency matrix to generate a continuously distributed access density surface; The access density surface is mapped to the three-dimensional coordinate space of the warehouse to generate the archive heat map distribution.

6. The intelligent archive storage management method according to claim 5, characterized in that, The process of generating a warehouse zoning control strategy by combining the archive heatmap distribution and the environmental control instruction set includes: Identify high-frequency and low-frequency access regions in the heatmap distribution of the archives; The high-frequency access region is overlapped with the priority control region in the environmental control instruction set. Based on the overlap analysis results, warehouse control priorities are determined, and warehouse zone control strategies are generated for execution in different time periods.

7. The intelligent archive storage management method according to claim 6, characterized in that, The dynamic adjustment of file storage location, migrating frequently accessed files to a stable environment, includes: Based on the warehouse zoning control strategy, storage areas that meet environmental stability standards are marked; Filter the target file numbers whose access frequency exceeds a threshold from the file access frequency matrix; The entity files corresponding to the target file number are migrated in batches to the storage area where the environmental stability meets the standards.

8. The intelligent archive storage management method according to claim 7, characterized in that, The periodic updating of the parameters of the archive storage risk assessment model, employing incremental learning to optimize risk prediction accuracy, includes: Collect correlation data between historical environmental control records and actual archive damage records; Calculate the nonlinear correlation coefficient between changes in environmental parameters and the rate of archive damage; The nonlinear correlation coefficient is injected as a weighting factor into the parameter update process of the archive storage risk assessment model.

9. The intelligent archive storage management method according to claim 8, characterized in that, The method further includes: Establish a log for changes in file storage location, recording the source and target coordinates for each migration operation; Perform timestamp correlation analysis between the file storage location change log and the environmental control instruction set; The regional stability judgment criteria in the warehouse zoning control strategy were revised based on the correlation analysis results.

10. An intelligent archive storage management system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the intelligent archive storage management method according to any one of claims 1 to 9.

Images