A multi-modal recognition-based classified information intelligent classified pulverization and security management and control system and method

Abstract

The application discloses a kind of based on multi-modal recognition's classified information intelligence classified powdering and safety control system and method, it is related to information security and secrecy technical field, including the following steps: paper medium of entering equipment is transported positioning, and document image information is collected by visible light imaging, while through near infrared spectrum collection document implicit characteristic information, and utilize medium pre-inspection sensor to obtain thickness, humidity and metal content data.The application forms closed-loop control by multi-modal classified, adaptive powdering, particle size verification and degaussing, realizes that destruction result can be measured and verified;Combined with dynamic strategy and abnormal processing improves system intelligence;Black box and blockchain are used to guarantee audit data reliability;Through structure optimization and pre-inspection mechanism improves operating stability;And support networked collaborative management and control, realize centralized security management.

Description

Technical Field

[0001] This invention relates to the field of information security and confidentiality technology, specifically to a system and method for intelligent classification, shredding, and security management of classified documents based on multimodal recognition. Background Technology

[0002] With increasingly stringent requirements for information security and confidentiality management, the secure destruction of classified media, especially paper-based classified media, has become a crucial aspect of information security systems. Current technologies for destroying classified documents are increasingly moving towards automation and intelligence. Some commercial shredders or confidentiality destruction equipment already possess basic identification capabilities, such as scanning barcodes on documents or using Optical Character Recognition (OCR) technology to identify specific fields (like CONFIDENTIAL) and trigger different shredding modes accordingly. These devices have significantly improved the automation level of the destruction process, enabling them to execute corresponding processing strategies based on document identifiers, and have already been applied in some office and confidentiality scenarios.

[0003] However, the aforementioned existing technologies still have significant shortcomings: First, their recognition capabilities are limited and lack robustness, typically relying solely on barcodes or simple OCR recognition. They are prone to failure due to blurred seals, non-standard layouts, document damage, or tampering, making them illegible for documents without barcodes or those requiring complex seals and layout features for security determination. Second, the security processing chain of existing equipment is incomplete, mostly implementing only a simple recognition-shredding process, lacking online detection and verification of shredding results (such as particle size), and failing to consider the elimination of magnetic information residue, resulting in the inability to effectively verify the destruction effect. Third, the systems mostly operate independently with fixed processing strategies, lacking the ability to link with backend security policies and unable to optimize recognition models or process parameters using operational data, resulting in a low overall level of intelligence. Finally, audit information typically only records the time and operation results, lacking correlation with images of the destroyed object and key sensor data, and is mostly stored on local ordinary storage media, posing a risk of tampering or deletion, making it difficult to meet the audit and evidence collection needs under high security levels.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a system and method for intelligent classification, shredding, and security management of classified documents based on multimodal recognition, in order to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for intelligent classification, shredding, and security management of classified materials based on multimodal recognition, comprising the following steps: The paper-based sealed media entering the equipment is transported and positioned, and document image information is acquired through visible light imaging. At the same time, the implicit feature information of the document is acquired through near-infrared spectroscopy, and the thickness, humidity and metal content data are obtained by media pre-inspection sensing. Multimodal fusion analysis is performed based on document image information, document implicit feature information, and thickness, humidity and metal content data to output the corresponding confidentiality level and the target crushing particle size parameter that matches the confidentiality level. The gap between the shredder shafts is controlled according to the classification level and target particle size parameters to perform shredding on paper-based confidential media, and the generated debris is transported during the shredding process. Image acquisition and particle size analysis are performed on the debris during the conveying process to obtain the debris size distribution results. The debris size distribution results are compared with the target crushing particle size parameters. If the target crushing particle size parameters are not met, the debris is crushed again. The document image information, hidden feature information, confidentiality level, target shredded particle size parameters, debris size distribution results and operation information are associated and recorded throughout the entire process of paper-based confidential media processing. After generating verification data from the associated records, they are securely stored.

[0007] The process of acquiring multi-source information from paper-based dense media involves using multispectral and physical detection to obtain multi-dimensional data, thereby improving information integrity and consistency. The steps are as follows: An imaging window is set in the transport path, and a high-resolution visible light CMOS camera is used to continuously scan the paper-based printing medium to acquire image data including text areas, stamp areas and page structure. Near-infrared spectral sensors are placed at the corresponding positions of the imaging window to simultaneously scan the same area, record reflection response information under different spectral bands and form spectral feature data. A laser thickness gauge, a capacitive humidity sensor, and a metal detection device are sequentially arranged in the feeding channel to measure the thickness, detect the moisture content, and identify the metal components of the paper-based dense medium point by point, and to form a corresponding data sequence. Image data, spectral feature data, and thickness, humidity, and metal detection data are labeled in chronological order and aligned according to spatial location to construct a multidimensional data set in a unified format.

[0008] To enhance the accuracy and stability of classification level determination, a joint analysis and processing method is used, combining image features, spectral information, and physical properties. The steps are as follows: The image data is processed to divide it into regions, extract text content features, seal outline information and page layout structure parameters, and form structured description information. Spectral band analysis is performed on the spectral feature data to identify the response curve features corresponding to different materials, and feature parameters used to distinguish ink types and watermark structures are extracted. The thickness, humidity and metal detection data are numerically normalized, and abnormal data are marked to form physical state description information; The structured description information, spectral feature parameters, and physical state description information are combined in a unified manner to form the classification level determination result and generate the corresponding set of crushing particle size control parameters.

[0009] By constructing an adjustable cutter shaft structure and particle size control relationship, the paper-based dense media processing process is equipped with graded pulverization capabilities. The steps are as follows: A relative mounting structure is set up between the main fixed tool shaft and the auxiliary adjustable tool shaft, and a linear drive device is connected to the auxiliary adjustable tool shaft to provide displacement adjustment capability; The auxiliary adjustable cutter shaft is controlled to move radially by the drive device, thereby changing the distance between the two cutter shafts and recording the current distance parameter value. Receive the granularity control parameters corresponding to the classification level, map the granularity requirements to the tool axis spacing parameters, and send the parameters to the drive device for adjustment; During the rotation of the cutter shaft, the paper-based compacted medium is introduced into the blade area, and the pulverization process is completed through continuous shearing and tearing, outputting fragments of the corresponding particle size.

[0010] By combining the debris discharge path and flow state control, the crushed material is kept in a stable conveying form to adapt to the subsequent testing process. The steps are as follows: A flow guide channel is set at the outlet of the crushing area to guide the debris to move downstream from the cutter shaft area; A dispersing structure is set in the flow channel so that the debris forms an intermittent distribution state during the movement; The debris is driven to move along a fixed path by a conveying device, so that the debris keeps flowing continuously and avoids stagnation; The dispersed debris is transported to the detection area and kept in an expanded state within the detection area.

[0011] Based on image acquisition and statistical analysis methods, the size characteristics of debris are quantified to form distribution data that can be used for evaluation. The steps are as follows: An image acquisition device is set up in the detection area to continuously photograph the passing debris and obtain debris projection images; Perform boundary recognition processing on the image, extract the contour lines of individual debris and calculate the corresponding geometric dimension parameters; Statistical processing was performed on multiple debris samples to calculate the size range, distribution frequency, and proportional relationship. Generate a dataset of debris size distribution data, including size distribution ranges and corresponding proportions.

[0012] Based on image acquisition and statistical analysis, the size characteristics of debris are quantified, and the contour extraction accuracy and statistical consistency are further refined. The steps include preprocessing the debris projection image to enhance the boundary clarity, performing denoising and closure processing on the contour data to obtain complete morphological information, standardizing the size parameters and establishing corresponding relationships, and continuously updating the statistical results to form a stable size distribution data set.

[0013] By introducing a feedback judgment and repeated processing mechanism, the pulverization result can gradually approach the target particle size requirement. The steps are as follows: Receive debris size distribution data and extract key statistical parameters for subsequent judgment; The statistical parameters and granularity control parameters are compared item by item to determine whether the preset conditions are met. If the conditions are not met, the debris is reintroduced into the crushing area and crushed again. The process of testing and comparing the re-crushed debris is repeated and continuously cyclical.

[0014] By combining multi-stage data processing for unified recording and verification, the integrity and traceability of information throughout the entire process are ensured. The steps are as follows: Image data, spectral data, granular data, and operational information are collected and processed in the workflow, and then categorized according to task number. Generate a unique identifier for each task and associate and store various types of data with this identifier; Perform unified calculations on the associated data, generate corresponding verification data, and append it to the record; Data and verification information are written into the storage structure in chronological order, and the records are kept in a continuous sequence.

[0015] A smart classification and shredding system for classified documents based on multimodal recognition, comprising an enhanced sensing and feeding module, a core execution module, a secure storage and recycling module, and a distributed control system: The enhanced sensing and feeding module is used to transport and position the paper-based dense media. It acquires visible light image information and near-infrared spectral feature information of the document through a multispectral scanning unit. At the same time, it acquires thickness, humidity and metal content information through a media pre-inspection sensor group. The multispectral scanning unit includes a high-resolution visible light CMOS camera and a near-infrared spectral sensor. The media pre-inspection sensor group includes a laser thickness gauge, a capacitive humidity sensor and a metal detection sensor. The smooth conveying mechanism adopts a servo motor driven synchronous belt conveying structure to achieve precise stopping of the document at the scanning position and the shredding position. The core execution module includes an adaptive crushing and demagnetizing unit, which receives document-related information and performs crushing processing. The adaptive gap crushing head includes a main fixed cutter shaft and a secondary adjustable cutter shaft. The secondary adjustable cutter shaft is driven by a linear motor to adjust the cutter shaft gap, thereby achieving particle size adjustment. An online particle size analyzer is set at the outlet of the crushing chamber to perform real-time image analysis on the crushed debris to obtain the debris size distribution. At the same time, the integrated alternating demagnetizing and air gap cleaning tunnel performs alternating magnetic field treatment on the crushed debris and cleans the crushing channel with gas. The secure storage and recycling module is used to store data during the processing and recycle debris. The black box audit storage unit includes a security chip and a write-protected storage structure, which is used to record document image information, sensor data and control commands in an append-only manner. The intelligent recycling bin includes an electronic lock and a weight sensing structure, and the debris is compressed by a debris compaction structure. The distributed control system controls the enhanced sensing and feeding module, the core execution module for the adaptive crushing and demagnetizing unit, and the safe storage and recycling module. The main control unit is used for process scheduling and human-machine interaction, the vision processing unit is used for multimodal recognition processing, the motion control unit is used for driving the conveyor and cutter shaft movement, and the security encryption unit is used for data encryption, audit data generation, and communication identity authentication, thereby realizing the classification, crushing, result detection, and security management of confidential materials.

[0016] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention employs a four-tiered security mechanism—multimodal precise classification, adaptive particle size crushing with online particle size verification, and demagnetization—to form a complete closed loop from identification and execution to result verification. This ensures that the destruction process of classified materials no longer relies on a single step but achieves a comprehensive improvement in security through multi-dimensional collaborative control. Specifically, multimodal precise classification enhances the accuracy of classification level determination; adaptive particle size crushing ensures different crushing standards correspond to different classification levels; online particle size verification enables real-time detection and feedback control of the destruction results; and demagnetization further eliminates the risk of potential information residue. In particular, the real-time feedback control of particle size makes the destruction effect quantifiable and verifiable, thus meeting the requirements for demonstrable destruction results in high-level confidentiality scenarios.

[0017] This invention possesses capabilities such as dynamic strategy loading, anomaly self-handling, and data feedback optimization, transforming the overall operation mode from fixed rule-driven to dynamic response mechanism-driven. During actual operation, it can automatically update the processing logic according to changes in the external environment or security policies. Furthermore, it can adaptively handle identification anomalies, fragmentation anomalies, or communication anomalies, thereby ensuring the continuity and stability of the processing flow. By continuously accumulating and utilizing historical operational data, the processing strategy can be gradually optimized, transforming the device from a traditional static execution tool into an intelligent terminal with learning and adaptability capabilities. This significantly reduces the need for manual intervention and lessens reliance on operator experience.

[0018] This invention introduces black-box audit storage and blockchain evidence preservation technology to ensure high security and reliability of audit data during generation, storage, and transmission. Throughout the entire process, all elements of document information, processing parameters, detection results, and operational information are recorded, and data integrity is verified through digital hashing, enabling the detection of any tampering. After integrating blockchain evidence preservation services, distributed storage and time-stamping mechanisms further enhance the immutability and temporal reliability of the data, thereby giving audit logs high legal evidentiary value and providing reliable evidence for accountability, compliance review, and security auditing.

[0019] This invention improves overall operational reliability and reduces maintenance complexity through optimized design of the crushing structure and system architecture. The adaptive crushing structure reduces the probability of mechanical failure by decreasing the number of complex mechanical parts and introducing an adjustable gap control method. The media pre-inspection mechanism identifies abnormal carriers before they enter the processing stage, effectively preventing equipment damage caused by non-compliant media. Simultaneously, self-cleaning and self-calibration functions automatically repair some abnormal states during operation, reducing the frequency of manual maintenance. Combined with a distributed control structure and fault information recording mechanism, maintenance personnel can quickly locate the source of problems, thereby improving the overall maintainability of the system.

[0020] This invention supports linkage with a higher-level security management platform, enabling collaborative operation among multiple devices through policy distribution and data feedback. This allows processing devices in different regions and at different nodes to execute tasks under a unified security policy, thereby achieving centralized management and unified scheduling of confidential destruction work. During operation, the centralized aggregation and analysis of data processed by each device allows for real-time monitoring of the overall operational status and unified response to anomalies. By constructing a networked control system, the destruction of classified materials is no longer limited to the level of a single device, but is elevated to a system-wide security protection system, thus forming a more comprehensive and responsive confidentiality defense. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0022] Figure 1 This is a flowchart of a method for intelligent classification, shredding, and security management of classified documents based on multimodal recognition, according to the present invention.

[0023] Figure 2 This is a schematic diagram of a module of a system and method for intelligent classification, shredding, and security management of classified documents based on multimodal recognition, according to the present invention. Detailed Implementation

[0024] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0025] This invention provides, for example Figure 1 The method for intelligent classification, shredding, and security management of classified materials based on multimodal recognition, as shown, includes the following steps: The paper-based sealed media entering the equipment is transported and positioned, and document image information is acquired through visible light imaging. At the same time, the implicit feature information of the document is acquired through near-infrared spectroscopy, and the thickness, humidity and metal content data are obtained by media pre-inspection sensing. In the above process, the paper-based dense medium first enters the equipment through the feed inlet. After entering the initial channel, it is carried and guided by the smooth conveying mechanism. The smooth conveying mechanism adopts a synchronous belt drive structure driven by a servo motor. By precisely controlling the conveying speed and displacement, it achieves continuous and stable conveying of the paper-based dense medium. At the same time, it achieves high-precision stopping at key positions, so that the paper-based dense medium maintains a stable posture and accurate position within the detection area, thus providing a reliable foundation for subsequent multimodal information acquisition.

[0026] At the verification window position of the transport path, a multispectral scanning unit is set up to image the surface of the paper-based printing medium using a high-resolution visible light CMOS camera, acquiring two-dimensional image information including text content, seal outline, and layout structure to support the recognition of text and seal information. Simultaneously, a near-infrared spectral sensor scans the same paper-based printing medium synchronously to acquire implicit feature information invisible to the naked eye, including special ink reflection characteristics, internal watermark structure of the paper, and potential hidden marking features, thus forming spectral data that is distinct from the visible light image. The above two types of data are consistent in time and correspond to the same location area in space, ensuring that the information of different modes is alignable and consistent in subsequent analysis.

[0027] As the paper-based compacted medium continues to be conveyed forward, it passes sequentially through a pre-inspection sensor group installed in the feeding channel. This sensor group includes a laser thickness gauge, a capacitive humidity sensor, and a metal detection sensor. The laser thickness gauge is used to perform non-contact measurement of the overall thickness of the paper-based compacted medium, thereby identifying whether there are multiple pages overlapping or abnormal thickening. The capacitive humidity sensor is used to detect the moisture content of the paper-based compacted medium to determine whether it is too wet or damp, avoiding instability in subsequent processing due to abnormal humidity. The metal detection sensor is used to detect whether there are metal objects such as staples or paper clips embedded in the paper-based compacted medium, thereby avoiding impact or damage to the subsequent shredding structure.

[0028] The aforementioned pre-inspection data is recorded synchronously with visible light image information and near-infrared spectral feature information during the acquisition process and is continuously updated during the transportation process, thereby forming a multi-dimensional information set for the same paper-based dense medium. When the thickness, humidity or metal content is detected to exceed the preset range, the abnormal state can be identified in a timely manner during the transportation process, so that the paper-based dense medium can be distinguished and processed before entering the subsequent processing area, thereby reducing the risk of equipment jamming and improving the overall operational stability.

[0029] Through the synergistic effect of the aforementioned conveying positioning, multispectral scanning, and media pre-inspection, the paper-based confidential media has completed comprehensive data collection from its physical state to surface information and implicit features before entering the subsequent processing stages. This not only ensures the integrity and accuracy of information acquisition but also provides a sufficient data foundation for subsequent classification level determination and crushing parameter determination. At the same time, through high-precision conveying and stable stopping control, a strict spatial correspondence is maintained between various sensor data, thereby improving the reliability and consistency of the overall processing flow.

[0030] Multimodal fusion analysis is performed based on document image information, document implicit feature information, and thickness, humidity and metal content data to output the corresponding confidentiality level and the target crushing particle size parameter that matches the confidentiality level. In this process, the document image information is first analyzed and processed. By identifying and extracting text areas, seal areas, and page layout features, key information reflecting the document's content attributes and security classification is obtained. This includes sensitive fields in the text content, the morphological features of the seal and its positional distribution, and the format features presented by the page layout structure. At the same time, the implicit feature information of the document is analyzed and processed. Near-infrared spectral data is used to identify the special ink response characteristics, watermark structure, and potential hidden marking information in the paper carrier. This supplements the implicit security features that cannot be directly obtained from visible light images, so that the recognition process not only relies on surface information but also uses material-level features for auxiliary judgment.

[0031] Based on the two types of information mentioned above, thickness, humidity, and metal content data are incorporated into the analysis process. Thickness data is used to determine whether there are multiple overlapping pages or abnormal binding in the document, thereby correcting the reliability of the identification results. Humidity data reflects the physical state of the paper carrier to avoid interference with the judgment results due to image feature distortion or abnormal spectral response caused by moisture. Metal content data is used to confirm the presence of staples or other metal attachments, thereby assisting in the judgment of document integrity and providing constraints for subsequent processing. The above multi-source data are synchronized in the time dimension and correspond to the same paper carrier medium area in the spatial dimension. They are organized through a unified data structure to form a multi-dimensional data set containing surface information, implicit features, and physical attributes.

[0032] In the multimodal fusion analysis process, feature extraction and comprehensive processing of data from different sources enable various features to participate in the determination of confidentiality level within a unified framework. Visible light images provide the foundation for text and seal recognition, near-infrared spectroscopy provides the ability to identify concealed features, while thickness, humidity, and metal content data are used to provide physical state constraints, thus jointly participating in the comprehensive analysis of document security attributes. During the analysis process, matching and verification of the correspondence between different features ensures that when a single feature is uncertain, it can be supplemented by other features, thereby improving the stability and robustness of the overall recognition results and avoiding judgment failures due to blurred seals, non-standard formats, or document damage.

[0033] After completing the multimodal fusion analysis, the classification level of the current paper-based confidential media is determined. This classification level reflects the sensitivity of the document in the security management system and serves as the core control parameter for subsequent processing. Based on the classification level, a target pulverization particle size parameter matching the classification level is further generated. This target pulverization particle size parameter is used to limit the size range of the pulverized debris. For example, different debris size requirements are set for different security levels to ensure that the corresponding security standards are met during physical destruction. This target pulverization particle size parameter not only reflects the particle size range but can also include constraints such as the proportion of debris size distribution to enable precise control and verification of the pulverization effect in subsequent processing.

[0034] The entire multimodal fusion analysis process, through the collaborative processing of document image information, document implicit feature information, and thickness, humidity, and metal content data, transforms the determination of confidentiality level from being driven by a single information source to being the result of the combined effect of multidimensional data. This significantly improves the accuracy and adaptability of the determination. At the same time, by directly mapping the confidentiality level to the target shredding particle size parameter, a continuous correlation between information identification and execution control is achieved, providing a clear and quantifiable control basis for subsequent shredding processing and laying the foundation for a refined and verifiable secure destruction process.

[0035] The gap between the shredder shafts is controlled according to the classification level and target particle size parameters to perform shredding on paper-based confidential media, and the generated debris is transported during the shredding process. In this process, after the confidentiality level of the paper-based confidential medium is determined and the target particle size parameters are generated, it is further transported to the crushing area. Before entering the crushing area, the paper-based confidential medium is positioned and controlled in the transport path to ensure that it enters the working range of the crushing blade shaft in a stable posture, thereby ensuring the continuity and consistency of the subsequent crushing process.

[0036] During the shredding process, the shredding blades employ a combination of a main fixed blade shaft and an auxiliary adjustable blade shaft. The main fixed blade shaft remains in a fixed position, while the auxiliary adjustable blade shaft undergoes precise radial displacement adjustment under the action of the drive device, thereby changing the gap between the two blade shafts. This gap size directly corresponds to the target shredding particle size parameter. By adjusting the gap, the paper-based confidential medium entering between the blade shafts is broken down into fragments of different sizes during the shearing process, thus achieving a shredding effect with controllable particle size. When the confidentiality level is high, corresponding to a smaller target shredding particle size parameter, the auxiliary adjustable blade shaft moves closer to the main fixed blade shaft to reduce the gap, allowing the paper-based confidential medium to be sheared into finer fragments. When the confidentiality level is relatively low, corresponding to a larger target shredding particle size parameter, a relatively larger gap is maintained between the auxiliary adjustable blade shaft and the main fixed blade shaft to improve processing efficiency and reduce energy consumption.

[0037] During the shredding process, the paper-based confidential media is continuously drawn into the blade area under the rotation of the cutter shaft. Through continuous shearing, tearing, and compression, it is gradually decomposed into multiple fragments of varying sizes. Because the gap adjustment has the ability to be continuous or graded, the shredded particle size can be flexibly switched between different processing requirements, thereby adapting to the requirements of different confidentiality levels for destruction precision. At the same time, the relative motion state between the cutter shafts is coordinated with the conveying speed to avoid blockage or jamming caused by uneven feeding or speed mismatch, thereby improving the overall operational stability.

[0038] After the debris is generated, the crushing area is equipped with a conveying path corresponding to the discharge direction of the cutter shaft. The debris is discharged from the cutter shaft area by gravity or auxiliary conveying structure and continuously conveyed to the subsequent processing area along the predetermined path. During the conveying process, the debris remains dispersed to avoid large-area accumulation, thus providing good observation conditions for subsequent particle size detection. At the same time, the continuous design of the conveying path forms a tightly connected processing link between the crushing and conveying processes, ensuring that the debris can leave the cutter shaft area in a timely manner and preventing the debris from affecting the subsequent crushing effect or causing an operational burden on the equipment.

[0039] Through the coordinated processing of the aforementioned shredder shaft gap adjustment and debris conveying, paper-based confidential media can achieve a controllable particle size shredding effect during physical destruction in accordance with confidentiality requirements, while ensuring processing efficiency while also taking into account safety and stability. At the same time, the immediate conveying of debris after generation provides continuous and stable input conditions for subsequent particle size analysis and further processing, thus forming a close connection between parameter control and execution results in the overall process.

[0040] Image acquisition and particle size analysis are performed on the debris during the conveying process to obtain the debris size distribution results. The debris size distribution results are compared with the target crushing particle size parameters. If the target crushing particle size parameters are not met, the debris is crushed again. In this process, after being crushed, the debris continues to move along a predetermined conveying path, maintaining a relatively dispersed state during the movement, thus providing stable and unobstructed observation conditions for image acquisition. Image acquisition areas are set at key locations along the conveying path, and the passing debris is continuously acquired by an imaging device to obtain projected image information of the debris. This image information can reflect the shape contour, size boundaries, and distribution of the debris, thus providing basic data for subsequent particle size analysis.

[0041] After image acquisition, the acquired debris image information is processed. By identifying and extracting debris boundaries, the size parameters of individual debris are calculated, and statistical analysis is performed on multiple debris samples to obtain the debris size distribution results. This debris size distribution result includes not only average size information, but also the size distribution range and the proportion of different size intervals, thus comprehensively reflecting the current crushing effect. By summarizing the data collected in multiple time periods during continuous conveying, a dynamically updated size distribution can be formed, making the analysis results real-time and representative.

[0042] After obtaining the debris size distribution results, the results are compared with the target particle size parameters. By comparing the size distribution range and proportion requirements, it is determined whether the current crushing effect meets the preset particle size control requirements. When the debris size distribution results meet the target particle size parameters, it indicates that the current crushing process has reached the destruction standard of the corresponding confidentiality level, and the debris can continue to enter the subsequent processing stage along the conveying path. When the debris size distribution results do not meet the target particle size parameters, it indicates that some debris is too large or the distribution does not meet the requirements, and further processing is needed to improve the crushing accuracy.

[0043] If the target particle size parameters are not met, the debris in the current conveying process is redirected back into the crushing area for re-crushing. During this re-crushing process, the debris is further sheared and decomposed using the same cutter shaft action as in the initial crushing, refining the originally larger debris and gradually approaching the target particle size parameter requirements. After the re-crushing is completed, the debris re-enters the conveying path, and the image acquisition and particle size analysis process is repeated until the debris size distribution meets the target particle size parameters.

[0044] Through the continuous process of image acquisition, particle size analysis, result comparison, and re-grinding, the grinding process is transformed from a single execution into a dynamic process with feedback adjustment capabilities. This allows particle size control to no longer rely on a single parameter setting, but rather to be corrected through real-time detection results, thereby improving the controllability and consistency of the grinding results. At the same time, continuous monitoring of particle size distribution makes the entire process verifiable, providing reliable data support for subsequent safety audits and destruction effect assessments, and forming a closed-loop relationship from execution results to control adjustment throughout the entire processing chain.

[0045] The document image information, hidden feature information, confidentiality level, target shredded particle size parameters, debris size distribution results and operation information are associated and recorded throughout the entire process of paper-based confidential media processing. After generating verification data from the associated records, they are securely stored. In this process, from the moment the paper-based confidential medium enters the processing flow, the data generated at each stage is continuously collected and uniformly managed. This includes document image information and implicit feature information acquired in the early stage, confidentiality level and target particle size parameters formed during the intermediate analysis process, and the particle size distribution results obtained during the crushing and detection stages. It also includes multiple types of data, including operational information. By uniformly identifying the above different types of data, a corresponding relationship is established between them in the same processing flow, thereby forming a complete data link.

[0046] During the association and recording process, a unique identifier is assigned to each processing task and is integrated into document image information, document implicit feature information, confidentiality level, target shredding particle size parameters, debris size distribution results, and operation information. This ensures that all types of data remain consistent in the time dimension and processing flow dimension, thereby achieving effective association between multi-source data. During the recording process, data generated at different stages are written continuously in the processing order to ensure that the information records have a sequential relationship and integrity, while avoiding data fragmentation or missing data, so that the entire processing process can be completely restored.

[0047] After data recording is completed, verification data is generated for the associated records. By calculating and processing the key content in the associated records, a verification result is formed to verify the integrity of the data. This verification data is saved together with the original records for subsequent detection of whether the data has been changed. During the verification data generation process, key fields in document image information, document implicit feature information, confidentiality level, target crushing particle size parameters, debris size distribution results, and operation information are uniformly processed to ensure that the verification result can cover the entire process of data, thereby ensuring that any data changes at any stage can be identified.

[0048] After generating the verification data, the associated records and corresponding verification data are securely stored. During storage, an append-only approach is used, continuously overlaying newly generated data onto existing records to prevent overwriting or modification. This method ensures the irreversibility of data records, preventing historical data from being tampered with or deleted after being written, thereby improving the overall security and reliability of the data. Simultaneously, different types of data are structured during storage, ensuring that document image information, implicit document features, confidentiality levels, target particle size parameters, debris size distribution results, and operational information are saved in a unified format, facilitating subsequent querying and verification.

[0049] Through the aforementioned methods of linking records, generating and securely storing verification data, all critical data on the paper-based confidential media form a complete, continuous, and tamper-proof record chain throughout the entire processing flow. This not only allows for the traceability of each processing step but also verifies the authenticity of the processing results, thereby providing a reliable basis for subsequent security audits and further enhancing the security and controllability of the overall processing flow.

[0050] Dynamic Security Policy Engine: Devices can preset basic policies and simultaneously receive dynamic policy updates from the central security management server (e.g., during flood control, all 'Secret' level data is upgraded to 'Confidential' level), achieving global security posture linkage. In this mechanism, basic policies are pre-stored as default operating rules to guide the processing of classified information under normal circumstances, including processing requirements and control parameters corresponding to different classification levels. During operation, policy information is continuously received from the central security management server via communication links. When external environments or management requirements change, dynamic policy updates can overwrite or adjust existing policy content, enabling processing rules to respond promptly to changes in external security posture. In this way, different devices can operate under a unified policy framework, ensuring consistent processing behavior across multiple devices and forming cross-regional and cross-node collaborative response capabilities. Furthermore, dynamic policy updates can be configured for specific time periods or events, making classified information processing standards time-sensitive and scenario-adaptable, thereby improving the overall security control level.

[0051] Abnormal Autonomy and Security Orientation: 1. Anomaly Identification: Multiple consecutive low-confidence scores trigger the automatic device self-cleaning and camera calibration process, and report the issue. During the identification process, if the analysis results of document image information or document implicit feature information show multiple consecutive low-confidence scores, it indicates a possible deviation in the current acquisition environment or imaging state. In this case, the system automatically enters a self-calibration process, cleaning the optical acquisition components and recalibrating the imaging parameters to restore the acquisition conditions to a stable state. After completing self-cleaning and calibration, data acquisition and analysis are performed again, thereby improving the reliability of the identification results. Simultaneously, the anomaly and its handling process are recorded and reported for subsequent statistical analysis.

[0052] 2. Crushing Abnormalities: If the motor current is abnormal or the particle size detection fails to meet standards, the system automatically attempts to reverse and re-crush. If this fails again, it will process the material once more according to the highest safety level process, and the entire process will be forcibly recorded. During the crushing process, the operating status of the drive motor is monitored. When abnormal changes in current are detected, it indicates possible jamming or abnormal load. Simultaneously, if the particle size detection result does not meet the target crushing particle size parameters, it indicates that the crushing effect has not met the expected requirements. In the event of any of the above abnormalities, a reverse operation is automatically executed, causing the cutter shaft to briefly reverse to release the jammed material, and then crushing is performed again to improve the crushing state. If the requirements are still not met after this second processing, a higher standard processing method is adopted, raising the processing conditions to the process parameters corresponding to the highest safety level, further refining the debris, thereby ensuring that the final processing result meets safety requirements. Throughout the entire abnormality handling process, the operation process and results are fully recorded to ensure traceability.

[0053] 3. Communication Anomaly: When communication with the server is lost, the system automatically switches to local encrypted caching mode. Audit data is then synchronized in batches upon restoration. During operation, if the communication link is interrupted and a connection to the central security management server cannot be maintained, the system automatically enters local operation mode. Data generated during processing is encrypted and cached locally, including key data such as document image information, document hidden feature information, confidentiality level, target particle size parameters, and debris size distribution results. During the disconnection, the processing flow continues to run according to the currently loaded strategy, ensuring uninterrupted processing. Once communication is restored, the locally cached data is uniformly organized and synchronized to the server in batches according to the generation order, thereby ensuring data integrity and continuity, while avoiding data loss or record breaks due to communication interruption, and maintaining consistency throughout the entire audit chain.

[0054] Full-element associated audit: The audit log entries package all elements such as the unique ID of this task, before / after images of the document (de-identified), the basis for confidentiality (such as the screenshot of the identified stamp area), the process parameters used, key sensor data (such as the average size of the final debris), operator ID, and equipment status, calculate digital hashes, and can selectively write them to the blockchain notarization service to ensure data integrity and time authority. During this audit process, key data generated at different stages of a complete processing task were uniformly incorporated into the audit log entries. A unique ID was used to identify the entire task, ensuring a one-to-one correspondence between various data types within the same recording system. Images before and after document processing were recorded in an anonymized format to reflect changes in the carrier's state before and after processing, while preventing the leakage of sensitive information. The basis for classifying confidentiality was demonstrated by recording key area information, such as screenshots of the seal area, providing a traceable basis for determining the level of confidentiality. The process parameters used described the specific control conditions employed during the crushing process, including particle size requirements corresponding to the level of confidentiality. Key sensor data recorded core detection results during processing, such as the average size of the final debris, to reflect whether the processing results met the established requirements. Operator IDs and equipment status information identified the executing entity and the equipment's operating status, supplementing information on human factors and equipment status.

[0055] After the aforementioned multiple elements are collected, all elements are structured and encapsulated using a unified format, logically forming a complete data set. A numerical hash value is then generated by performing calculations on this data set, serving as a verification identifier for data integrity. This hash value corresponds to the original data; any change in any element will alter the hash result, thus enabling effective detection of data tampering. In this way, the audit log not only records the processing procedure but also possesses self-verification capabilities, allowing for the verification of the authenticity of the recorded content.

[0056] After generating the digital hash, users can choose whether to write the relevant data to the blockchain notarization service based on the requirements of the operating environment. During the writing process, the audit log entries and their corresponding digital hashes are used as notarization content and stored through the distributed recording method of the blockchain, so that the data forms a consistent copy among multiple nodes, thereby enhancing the immutability of the data. At the same time, the blockchain recording process usually includes time stamp information, so that each audit record has a clear time attribute, thus providing a reliable time reference for subsequent audits. Even without access to the blockchain notarization service, the audit logs can still be continuously stored locally in an append-only manner, maintaining the continuity and integrity of the records through the sequential relationship of the internal data chain.

[0057] By employing the aforementioned comprehensive auditing approach, all critical steps in a single processing task are incorporated into a unified recording system. This not only achieves full-process coverage from input data to processing results but also enhances data integrity and credibility through digital hashing and blockchain-based evidence storage mechanisms. In subsequent security audits or accountability investigations, the corresponding task record can be quickly located based on a unique ID. By verifying the record content and its validation results, it can be confirmed whether the data remains in its original state, thus providing reliable and authoritative evidence for the entire processing process. Simultaneously, it ensures that audit information remains undistorted or tampered with during long-term storage and cross-system transmission.

[0058] This invention employs a four-tiered security mechanism—multimodal precise classification, adaptive particle size crushing with online particle size verification, and demagnetization—to form a complete closed loop from identification and execution to result verification. This ensures that the destruction process of classified materials no longer relies on a single step but achieves a comprehensive improvement in security through multi-dimensional collaborative control. Specifically, multimodal precise classification enhances the accuracy of classification level determination; adaptive particle size crushing ensures different crushing standards correspond to different classification levels; online particle size verification enables real-time detection and feedback control of the destruction results; and demagnetization further eliminates the risk of potential information residue. In particular, the real-time feedback control of particle size makes the destruction effect quantifiable and verifiable, thus meeting the requirements for demonstrable destruction results in high-level confidentiality scenarios.

[0059] This invention possesses capabilities such as dynamic strategy loading, anomaly self-handling, and data feedback optimization, transforming the overall operation mode from fixed rule-driven to dynamic response mechanism-driven. During actual operation, it can automatically update the processing logic according to changes in the external environment or security policies. Furthermore, it can adaptively handle identification anomalies, fragmentation anomalies, or communication anomalies, thereby ensuring the continuity and stability of the processing flow. By continuously accumulating and utilizing historical operational data, the processing strategy can be gradually optimized, transforming the device from a traditional static execution tool into an intelligent terminal with learning and adaptability capabilities. This significantly reduces the need for manual intervention and lessens reliance on operator experience.

[0060] This invention introduces black-box audit storage and blockchain evidence preservation technology to ensure high security and reliability of audit data during generation, storage, and transmission. Throughout the entire process, all elements of document information, processing parameters, detection results, and operational information are recorded, and data integrity is verified through digital hashing, enabling the detection of any tampering. After integrating blockchain evidence preservation services, distributed storage and time-stamping mechanisms further enhance the immutability and temporal reliability of the data, thereby giving audit logs high legal evidentiary value and providing reliable evidence for accountability, compliance review, and security auditing.

[0061] This invention improves overall operational reliability and reduces maintenance complexity through optimized design of the crushing structure and system architecture. The adaptive crushing structure reduces the probability of mechanical failure by decreasing the number of complex mechanical parts and introducing an adjustable gap control method. The media pre-inspection mechanism identifies abnormal carriers before they enter the processing stage, effectively preventing equipment damage caused by non-compliant media. Simultaneously, self-cleaning and self-calibration functions automatically repair some abnormal states during operation, reducing the frequency of manual maintenance. Combined with a distributed control structure and fault information recording mechanism, maintenance personnel can quickly locate the source of problems, thereby improving the overall maintainability of the system.

[0062] This invention supports linkage with a higher-level security management platform, enabling collaborative operation among multiple devices through policy distribution and data feedback. This allows processing devices in different regions and at different nodes to execute tasks under a unified security policy, thereby achieving centralized management and unified scheduling of confidential destruction work. During operation, the centralized aggregation and analysis of data processed by each device allows for real-time monitoring of the overall operational status and unified response to anomalies. By constructing a networked control system, the destruction of classified materials is no longer limited to the level of a single device, but is elevated to a system-wide security protection system, thus forming a more comprehensive and responsive confidentiality defense.

[0063] This invention provides, for example Figure 2 The system shown is an intelligent classification and shredding system for classified documents based on multimodal recognition, comprising an enhanced sensing and feeding module, a core execution module, a secure storage and recycling module, and a distributed control system. The enhanced sensing and feeding module is used to transport and position the paper-based dense media. It acquires visible light image information and near-infrared spectral feature information of the document through a multispectral scanning unit. At the same time, it acquires thickness, humidity and metal content information through a media pre-inspection sensor group. The multispectral scanning unit includes a high-resolution visible light CMOS camera and a near-infrared spectral sensor. The media pre-inspection sensor group includes a laser thickness gauge, a capacitive humidity sensor and a metal detection sensor. The smooth conveying mechanism adopts a servo motor driven synchronous belt conveying structure to achieve precise stopping of the document at the scanning position and the shredding position. The core execution module includes an adaptive crushing and demagnetizing unit, which receives document-related information and performs crushing processing. The adaptive gap crushing head includes a main fixed cutter shaft and a secondary adjustable cutter shaft. The secondary adjustable cutter shaft is driven by a linear motor to adjust the cutter shaft gap, thereby achieving particle size adjustment. An online particle size analyzer is set at the outlet of the crushing chamber to perform real-time image analysis on the crushed debris to obtain the debris size distribution. At the same time, the integrated alternating demagnetizing and air gap cleaning tunnel performs alternating magnetic field treatment on the crushed debris and cleans the crushing channel with gas. The secure storage and recycling module is used to store data during the processing and recycle debris. The black box audit storage unit includes a security chip and a write-protected storage structure, which is used to record document image information, sensor data and control commands in an append-only manner. The intelligent recycling bin includes an electronic lock and a weight sensing structure, and the debris is compressed by a debris compaction structure. The distributed control system controls the enhanced sensing and feeding module, the core execution module for the adaptive crushing and demagnetizing unit, and the safe storage and recycling module. The main control unit is used for process scheduling and human-machine interaction, the vision processing unit is used for multimodal recognition processing, the motion control unit is used for driving the conveyor and cutter shaft movement, and the security encryption unit is used for data encryption, audit data generation, and communication identity authentication, thereby realizing the classification, crushing, result detection, and security management of confidential materials.

[0064] This invention provides a method for intelligent classification, shredding, and security management of classified documents based on multimodal recognition. This method is implemented through the aforementioned intelligent classification, shredding, and security management system based on multimodal recognition. For details of the specific methods and processes of the intelligent classification, shredding, and security management system based on multimodal recognition, please refer to the embodiment of the above-mentioned method for intelligent classification, shredding, and security management of classified documents based on multimodal recognition, which will not be repeated here.

[0065] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for intelligent classification, shredding, and security management of classified documents based on multimodal recognition, characterized in that, Includes the following steps: The paper-based sealed media entering the equipment is transported and positioned, and document image information is acquired through visible light imaging. At the same time, the implicit feature information of the document is acquired through near-infrared spectroscopy, and the thickness, humidity and metal content data are obtained by media pre-inspection sensing. Multimodal fusion analysis is performed based on document image information, document implicit feature information, and thickness, humidity and metal content data to output the corresponding confidentiality level and the target crushing particle size parameter that matches the confidentiality level. The gap between the shredder shafts is controlled according to the classification level and target particle size parameters to perform shredding on paper-based confidential media, and the generated debris is transported during the shredding process. Image acquisition and particle size analysis are performed on the debris during the conveying process to obtain the debris size distribution results. The debris size distribution results are compared with the target crushing particle size parameters. If the target crushing particle size parameters are not met, the debris is crushed again. The document image information, hidden feature information, confidentiality level, target shredded particle size parameters, debris size distribution results and operation information are associated and recorded throughout the entire process of paper-based confidential media processing. After generating verification data from the associated records, they are securely stored.

2. The intelligent classification, shredding, and security management method for classified documents based on multimodal recognition as described in claim 1, characterized in that, The process of acquiring multi-source information from paper-based dense media involves using multispectral and physical detection to obtain multi-dimensional data, thereby improving information integrity and consistency. The steps are as follows: An imaging window is set in the transport path, and a high-resolution visible light CMOS camera is used to continuously scan the paper-based printing medium to acquire image data including text areas, stamp areas and page structure. Near-infrared spectral sensors are placed at the corresponding positions of the imaging window to simultaneously scan the same area, record reflection response information under different spectral bands and form spectral feature data. A laser thickness gauge, a capacitive humidity sensor, and a metal detection device are sequentially arranged in the feeding channel to measure the thickness, detect the moisture content, and identify the metal components of the paper-based dense medium point by point, and to form a corresponding data sequence. Image data, spectral feature data, and thickness, humidity, and metal detection data are labeled in chronological order and aligned according to spatial location to construct a multidimensional data set in a unified format.

3. The intelligent classification, shredding, and security management method for classified documents based on multimodal recognition as described in claim 1, characterized in that, To enhance the accuracy and stability of classification level determination, a joint analysis and processing method is used, combining image features, spectral information, and physical properties. The steps are as follows: The image data is processed to divide it into regions, extract text content features, seal outline information and page layout structure parameters, and form structured description information. Spectral analysis of spectral feature data is performed to identify the response curve features corresponding to different materials, and feature parameters used to distinguish ink types and watermark structures are extracted. The thickness, humidity and metal detection data are numerically normalized, and abnormal data are marked to form physical state description information; The structured description information, spectral feature parameters, and physical state description information are combined in a unified manner to form the classification level determination result and generate the corresponding set of crushing particle size control parameters.

4. The intelligent classification, shredding, and security management method for classified documents based on multimodal recognition as described in claim 1, characterized in that, By constructing an adjustable cutter shaft structure and particle size control relationship, the paper-based dense media processing process is equipped with graded pulverization capabilities. The steps are as follows: A relative mounting structure is set up between the main fixed tool shaft and the auxiliary adjustable tool shaft, and a linear drive device is connected to the auxiliary adjustable tool shaft to provide displacement adjustment capability; The auxiliary adjustable cutter shaft is controlled to move radially by the drive device, thereby changing the distance between the two cutter shafts and recording the current distance parameter value. Receive the granularity control parameters corresponding to the classification level, map the granularity requirements to the tool axis spacing parameters, and send the parameters to the drive device for adjustment; During the rotation of the cutter shaft, the paper-based medium is introduced into the blade area, and the shredding process is completed through continuous shearing and tearing, and the corresponding particle size is output as debris.

5. The intelligent classification, shredding, and security management method for classified documents based on multimodal recognition according to claim 1, characterized in that, By combining the debris discharge path and flow state control, the crushed material is kept in a stable conveying form to adapt to the subsequent testing process. The steps are as follows: A flow guide channel is set at the outlet of the crushing area to guide the debris to move downstream from the cutter shaft area; A dispersing structure is set in the flow channel so that the debris forms an intermittent distribution state during the movement; The debris is driven to move along a fixed path by a conveying device, so that the debris keeps flowing continuously and avoids stagnation; The dispersed debris is transported to the detection area and kept in an expanded state within the detection area.

6. The intelligent classification, shredding, and security management method for classified documents based on multimodal recognition according to claim 1, characterized in that, Based on image acquisition and statistical analysis methods, the size characteristics of debris are quantified to form distribution data that can be used for evaluation. The steps are as follows: An image acquisition device is set up in the detection area to continuously photograph the passing debris and obtain debris projection images; Perform boundary recognition processing on the image, extract the contour lines of individual debris and calculate the corresponding geometric dimension parameters; Statistical processing was performed on multiple debris samples to calculate the size range, distribution frequency, and proportional relationship. Generate a dataset of debris size distribution data, including size distribution ranges and corresponding proportions.

7. The intelligent classification, shredding, and security management method for classified documents based on multimodal recognition according to claim 6, characterized in that, Based on image acquisition and statistical analysis, the size characteristics of debris are quantified, and the contour extraction accuracy and statistical consistency are further refined. The steps include preprocessing the debris projection image to enhance the boundary clarity, performing denoising and closure processing on the contour data to obtain complete morphological information, standardizing the size parameters and establishing corresponding relationships, and continuously updating the statistical results to form a stable size distribution data set.

8. The intelligent classification, shredding, and security management method for classified documents based on multimodal recognition according to claim 1, characterized in that, By introducing a feedback judgment and repeated processing mechanism, the pulverization result can gradually approach the target particle size requirement. The steps are as follows: Receive debris size distribution data and extract key statistical parameters for subsequent judgment; The statistical parameters and granularity control parameters are compared item by item to determine whether the preset conditions are met. If the conditions are not met, the debris is reintroduced into the crushing area and crushed again. The process of testing and comparing the re-crushed debris is repeated and continuously cyclical.

9. The intelligent classification, shredding, and security management method for classified documents based on multimodal recognition according to claim 1, characterized in that, By combining multi-stage data processing for unified recording and verification, the integrity and traceability of information throughout the entire process are ensured. The steps are as follows: Image data, spectral data, granular data, and operational information are collected and processed in the workflow, and then categorized according to task number. Generate a unique identifier for each task and associate and store various types of data with this identifier; Perform unified calculations on the associated data, generate corresponding verification data, and append it to the record; Data and verification information are written into the storage structure in chronological order, and the records are kept in a continuous sequence.

10. A system for intelligent classification, shredding, and security management of classified documents based on multimodal recognition, used to implement the method for intelligent classification, shredding, and security management of classified documents based on multimodal recognition as described in any one of claims 1-9, characterized in that, It includes an enhanced sensing and feeding module, a core execution module, a secure storage and recycling module, and a distributed control system. The enhanced sensing and feeding module is used to transport and position the paper-based dense media. It acquires visible light image information and near-infrared spectral feature information of the document through a multispectral scanning unit. At the same time, it acquires thickness, humidity and metal content information through a media pre-inspection sensor group. The multispectral scanning unit includes a high-resolution visible light CMOS camera and a near-infrared spectral sensor. The media pre-inspection sensor group includes a laser thickness gauge, a capacitive humidity sensor and a metal detection sensor. The smooth conveying mechanism adopts a servo motor driven synchronous belt conveying structure to achieve precise stopping of the document at the scanning position and the shredding position. The core execution module includes an adaptive crushing and demagnetizing unit, which receives document-related information and performs crushing processing. The adaptive gap crushing head includes a main fixed cutter shaft and a secondary adjustable cutter shaft. The secondary adjustable cutter shaft is driven by a linear motor to adjust the cutter shaft gap, thereby achieving particle size adjustment. An online particle size analyzer is set at the outlet of the crushing chamber to perform real-time image analysis on the crushed debris to obtain the debris size distribution. At the same time, the integrated alternating demagnetizing and air gap cleaning tunnel performs alternating magnetic field treatment on the crushed debris and cleans the crushing channel with gas. The secure storage and recycling module is used to store data during the processing and recycle debris. The black box audit storage unit includes a security chip and a write-protected storage structure, which is used to record document image information, sensor data and control commands in an append-only manner. The intelligent recycling bin includes an electronic lock and a weight sensing structure, and the debris is compressed by a debris compaction structure. The distributed control system controls the enhanced sensing and feeding module, the core execution module for the adaptive crushing and demagnetizing unit, and the safe storage and recycling module. The main control unit is used for process scheduling and human-machine interaction, the vision processing unit is used for multimodal recognition processing, the motion control unit is used for driving the conveyor and cutter shaft movement, and the security encryption unit is used for data encryption, audit data generation, and communication identity authentication, thereby realizing the classification, crushing, result detection, and security management of confidential materials.

Images