A security management personnel performance behavior intelligent recording system and method

Abstract

The application relates to the technical field of building construction safety monitoring, and particularly discloses a safety management personnel job performance behavior intelligent recording system and method, which comprises a wearable collection terminal, which can be worn on different parts of the body and comprises a camera, a microphone, a processor, a storage unit, a communication module and a man-machine interaction unit, is used for on-site audio and video collection, instruction response and data encryption storage, and comprises a scene-based job performance directory library, an intelligent triggering engine, an automatic classification storage module, an anti-tampering encryption module and a job performance statistical analysis module; the application automatically associates audio and video records generated in the job performance process of safety management personnel with specific job performance scenes through post-scene mapping configuration and an intelligent triggering mechanism, automatically constructs a hierarchical storage path after triggering, completes file classification storage, and significantly improves the retrieval efficiency and readability of job performance records.

Description

Technical Field

[0001] This invention relates to the field of construction safety monitoring technology, and in particular to an intelligent recording system and method for recording the performance of duties by safety management personnel. Background Technology

[0002] In construction sites and other high-risk industries (such as chemical, mining, and power), safety management personnel fulfill their safety production responsibilities through on-site inspections and recording. Traditional record-keeping mainly relies on manual paper records, full-process video recording by ordinary law enforcement recorders, or mobile phone photography, resulting in a large number of continuous video streams and paper documents. These recording methods are characterized by decentralized and unstructured data generation, storage, and management. The audio and video materials generated during the performance of duties are mostly continuous, unclassified raw data, lacking a clear correspondence with specific duty scenarios.

[0003] In existing technologies, it is difficult to accurately trigger and automatically classify key performance behaviors in highly dynamic and obstructed construction sites. Traditional fixed perspectives or manual operations are prone to missing records and false triggers. Furthermore, based on behavior classification records, existing systems generate video or photo files that are at risk of being tampered with, deleted, or forged during local operation and transmission, failing to meet the legal requirements of the "Safety Production Law" regarding the authenticity and completeness of performance data. At the same time, when different positions (such as full-time safety officers and project managers) share the same set of equipment, there is a lack of dynamic scene directory configuration capabilities at the position level, resulting in functional redundancy, chaotic records, and the inability to generate structured and traceable performance ledgers. Therefore, how to simultaneously solve the problems of accurate behavior triggering, data anti-tampering, and position adaptive configuration in the same system is the technical problem that this invention aims to solve. Summary of the Invention

[0004] To overcome the shortcomings of the prior art, the present invention provides an intelligent recording system and method for the performance of duties by safety management personnel, which can effectively solve the problems involved in the prior art.

[0005] The objective of this invention can be achieved through the following technical solution: Firstly, this invention provides an intelligent recording system for the performance of duties by safety management personnel, comprising the following modules: Wearable data acquisition terminal is a portable device with audio and video acquisition capabilities that can be worn on different parts of the body. It includes a camera, microphone, processor, storage unit, communication module, and human-computer interaction unit (the human-computer interaction unit includes physical buttons / voice module). It is used for on-site audio and video acquisition, command response, and encrypted data storage. It supports wearing on multiple parts of the body and flexible shooting. The intelligent recording system for job performance is a software system embedded in the data collection terminal. It includes a scenario-based job performance catalog library, an intelligent trigger engine, an automatic classification and storage module, an anti-tampering encryption module, and a job performance statistical analysis module. It is used to achieve job scenario matching, intelligent trigger recording, automatic classification and storage, and anti-tampering encryption. The cloud management platform is used to receive and store encrypted data uploaded by terminals, and supports the generation of performance ledgers, record traceability, and job configuration management.

[0006] Secondly, the present invention also provides a method for intelligently recording the performance of duties by safety management personnel, which is based on the aforementioned intelligent recording system for the performance of duties by safety management personnel, and includes the following steps: Step 1: The cloud management platform dynamically maps allowed job scenario subdirectories according to job type, forming a job-scenario matrix. After the wearable data collection terminal is synchronized, only the current job scenario is activated, and irrelevant functions are restricted. This realizes the preset job responsibilities and directory configuration, avoiding functional redundancy and misoperation. Step 2: Trigger the corresponding duty scenario through voice keywords or physical button mapping. Upon triggering, automatically create a storage folder in the corresponding subdirectory with the naming rule of date + duty scenario, and write the current time, geographical location, and personnel ID. This enables one-click triggering of duty scenarios and automatic construction of storage paths, improving on-site operation efficiency. Step 3: Enter recording mode according to the trigger command type. The recording mode is full-process video recording plus intermittent photo taking, and generates audio and video files that are automatically saved to the corresponding folder according to the preset path. The file name contains the scene code and personnel ID, realizing one-time trigger and automatic classification, realizing automatic classification and recording of the performance process, and solving the problem of low efficiency of manual classification after the event. Step 4: When the file is generated, the solidification operation is automatically performed, a digital watermark is superimposed in real time, the file is marked as read-only, and encrypted video is recorded in segments of fixed duration to prevent local modification, deletion or forgery, ensuring the authenticity and integrity of the performance data and meeting the requirements of the Safety Production Law for the prevention of evidence tampering. Step 5: When there is a network connection, the wearable data collection terminal automatically backs up the encrypted files to the cloud management platform. The cloud management platform forces the data to be retained for at least 6 months and cannot be deleted normally. It supports traceability by position and scenario, forming a structured performance record, realizing the cloud retention and structured traceability of performance data, and supporting the performance assessment and responsibility determination of the position.

[0007] Preferably, step one specifically includes: The cloud management platform has a built-in job-scenario mapping matrix, which pre-associates allowed job scenario subdirectories for the job types of full-time safety officers, project managers and safety directors, forming an explicit correspondence between jobs and scenarios, realizing standardized binding of jobs and job scenarios, and providing a basis for access control; Administrators can select the allowed subdirectories of job performance scenarios by job position through the cloud management platform interface. The subdirectories are selected from the main scenario library. Unselected scenarios are not displayed or are disabled in the terminal interface, enabling flexible configuration of job scenarios. After the wearable data collection terminal synchronizes the configuration information with the cloud, the human-computer interaction unit only presents the job performance scenario options that have been enabled for the current position. The entry points for other scenario functions are logically blocked to achieve functional permission isolation, ensuring that the terminal only responds to operations within the scope of the current job's responsibilities, and avoiding functional redundancy and misuse.

[0008] Preferably, step one further includes: The duty performance scenario subdirectory adopts a hierarchical storage structure. The root directory is fixed as the duty performance status, and its lower level is pre-set with at least ten pre-set duty performance scenario subdirectories, including the whole process control of major and critical projects and the investigation and management of hidden dangers. This realizes the standardized classification and storage of duty performance scenarios, which is convenient for subsequent retrieval and traceability. In the job-scenario mapping matrix, the permitted scenario subset for full-time safety officers includes pre-job inspections, full-process control of critical and major projects, supervision of high-risk operations, and investigation and management of hidden dangers. The permitted scenario subset for project leaders includes leading production and inspections and supervising major hidden dangers. This achieves a precise correspondence between the responsibilities of different jobs and the performance scenarios, avoiding confusion of permissions. The wearable data acquisition terminal automatically loads the corresponding scene configuration table based on the synchronized job identifier. The wake-up word library of the voice recognition module and the mapping table of physical buttons are updated synchronously. It only matches the scenes that have been activated for the current job, so as to realize the automatic adaptation of the terminal interaction logic and job responsibilities and reduce the complexity of operation.

[0009] Preferably, step two specifically includes: Wearable data acquisition terminal collects voice input in real time. The voice recognition module extracts preset keywords and parses the corresponding performance scenarios, or detects single press, long press and continuous press of physical buttons. Based on the pre-configured mapping relationship, it matches scene identifiers to realize multimodal command recognition and meet the rapid triggering needs in different scenarios. After the scene identifier is confirmed by the intelligent trigger engine, the storage management module is automatically invoked to locate or create a subdirectory corresponding to the scene identifier in the root directory of the storage unit's performance status, thereby realizing the automatic association between the scene identifier and the storage directory and ensuring accurate data archiving. In this subdirectory, a folder is generated using the naming rules of the current system date concatenated with the name of the duty performance scenario. At the same time, the real-time timestamp, GPS geographical coordinates, and the ID of the currently logged-in user are written as metadata into the folder attributes, so as to realize automatic folder naming and metadata recording, which facilitates accurate traceability of the duty performance process afterward.

[0010] Preferably, step two further includes: The physical button is configured with multiple trigger modes. A single short press is mapped to the hidden danger investigation and management scenario, two consecutive short presses are mapped to the pre-shift inspection scenario, and a long press exceeding the preset threshold is mapped to the whole process management scenario of major and hazardous projects. This enables quick switching between multiple scenarios with a single button and improves the convenience of on-site operation. Based on the fuzzy matching supported by the speech recognition module, when a sentence containing a combination of start, supervision, inspection and scene name is recognized, it is automatically completed and mapped to the standard duty performance scene identifier, which reduces the trigger threshold, reduces the burden of remembering voice commands, and improves the success rate and ease of use of voice triggering. The naming format of the completed folders is determined to be "YYYY-MM-DD-Duty Scenario Name". Metadata is embedded in the file system extended attributes in a structured manner, which can be directly read by the subsequent retrieval and verification modules, realizing the standardization of folder naming and the structured storage of metadata, which facilitates efficient retrieval and verification.

[0011] Preferably, step three specifically includes: The intelligent trigger engine automatically selects the recording mode based on the output scene identifier. If the scene identifier is the whole process control of critical and major projects or pre-job inspection, the whole process continuous recording mode is started. The recording frame rate and resolution are executed according to the preset parameters to realize the automatic matching of scene and recording mode and meet the recording needs of different job performance scenarios. If the scene is identified as hazard investigation and management, it enters the intermittent photo mode. Each time a physical button trigger signal is received, a high-resolution static image is generated. The image is appended with the current timestamp and scene code, so as to achieve accurate capture of hazard investigation scenes and highlight key evidence in the performance of duties. The generated video or photo files are stored in the corresponding folders according to the naming rules of timestamp-scene code-person ID. This enables fully automated categorization after a single trigger, without manual intervention, thus achieving automated naming and categorization of performance documents and completely solving the inefficiency of manual organization.

[0012] Preferably, step four specifically includes: During the document generation process, the processor in the wearable acquisition terminal reads the current time, GPS location, personnel ID and scene name in real time. The information read is then embedded into a designated area of ​​the video or photo in the form of a digital watermark through a graphics overlay engine, so as to solidify the performance information into the audio and video in real time and prevent the content from being tampered with from the source. After the file is written to the storage unit, the operating system calls to change the file attributes to read-only, prohibiting any local process from performing modification, deletion or renaming operations on the file, and records the file hash value as the integrity verification benchmark, realizing low-level write protection and integrity verification of the file, ensuring that the evidence cannot be tampered with; Continuous recordings are stored in segments of fixed duration. Each segment is independently encrypted and an encryption verification code is stored. Damage or tampering of any segment does not affect the integrity verification of other segments. Stored segments are not lost when the wearable acquisition terminal experiences an abnormal power outage, achieving independent protection of recording segments and improving data resistance to damage and abnormal recovery capabilities.

[0013] Preferably, step five specifically includes: The wearable data collection terminal periodically checks the network connection status. When a mobile network or wireless network is detected to be available, it automatically adds the locally stored encrypted files to the upload queue in the order of their generation time, realizing automatic backup of the performance data and completing the upload without manual intervention. The upload process employs a breakpoint resume mechanism, which resumes the upload from the point of interruption if the file transfer is interrupted. After the upload is completed, the hash value is compared between the local and cloud management platforms. If they match, the upload is marked as synchronized, ensuring that the upload process is reliable and complete. Automatic recovery is also provided after network fluctuations to avoid data loss. After receiving the file, the cloud management platform stores it in the storage area with the corresponding job position and time index, and sets a mandatory retention period of no less than 6 months. It cannot be deleted or overwritten by anyone through the regular user operation interface. It only supports retrieval by job type, job performance scenario and time range and generates structured job performance ledgers, realizing mandatory retention and structured retrieval of cloud data, meeting the requirements of legal traceability and supporting job performance assessment.

[0014] Preferably, step five further includes: The wearable data acquisition terminal periodically detects the signal strength of mobile or wireless networks through a communication module. When a network is detected, it automatically scans the encrypted and hardened file partitions in the storage unit, builds an upload task list according to the order of file generation timestamps, and adds the encrypted files in the list to the upload queue in sequence, so as to realize the automatic and orderly upload of encrypted files and complete data synchronization without manual intervention. When the wearable data collection terminal initiates a file upload request to the cloud management platform, it carries the file size and upload location information. It uses a breakpoint resume mechanism to resume the upload from the breakpoint. After the file is completely transmitted, the terminal sends the locally stored file hash value to the cloud management platform. The cloud management platform calculates the hash value of the received file and compares it with the hash value sent by the terminal to ensure that the file upload is complete and reliable. After the network is interrupted, the transmission is automatically resumed and the data consistency is verified. After receiving the complete encrypted file, the cloud management platform parses the metadata to extract job type, performance scenario, and timestamp information. It stores the data in the storage area according to the hierarchical structure of job type and time index, and forces a retention period. The stored performance records cannot be deleted or overwritten through the regular user operation interface, thus realizing hierarchical storage and forced retention of performance data to prevent manual deletion or overwriting and ensure the integrity of the evidence chain.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. This intelligent recording system and method for the performance of duties by safety management personnel automatically associates audio and video recordings generated during the performance of duties by safety management personnel with specific performance scenarios through job-scenario mapping configuration and intelligent triggering mechanism. After triggering, it automatically constructs a hierarchical storage path and completes file classification and storage, which significantly improves the retrieval efficiency and readability of performance records.

[0016] 2. This intelligent recording system and method for the performance of duties by safety management personnel superimposes digital watermarks in real time during the document generation process, marks the document attributes as read-only, and encrypts continuous video recordings in segments independently. At the same time, it records hash values ​​as integrity verification benchmarks. Any local modification, deletion, or forgery can be identified, effectively meeting the legal requirements of safety production regulations for the authenticity and integrity of performance data.

[0017] 3. This intelligent recording system and method for the performance of duties by safety management personnel uses a job-scenario mapping matrix built into a cloud management platform to dynamically distribute the sub-directories of allowed job performance scenarios according to the job type. After synchronization, the wearable data collection terminal only activates the function corresponding to the current job, and the function entry of the unactivated scenario is logically blocked, which solves the problem of functional redundancy and operation confusion caused by different jobs sharing the same device.

[0018] 4. This intelligent recording system and method for the performance of duties by safety management personnel automatically classifies and stores audio and video files generated during the performance of duties according to job position, scenario and time, and synchronizes them to the cloud management platform. The cloud forcibly retains the performance records and cannot be deleted in a regular manner. It supports retrieval by multiple dimensions and automatically generates a structured performance ledger, which solves the problems of scattered performance data, difficulty in statistics and traceability. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the workflow of an intelligent recording system and method for the performance of duties by safety management personnel according to the present invention. Detailed Implementation

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0021] Example 1, please refer to Figure 1 This invention provides a technical solution: an intelligent recording system for the performance of duties by safety management personnel, comprising the following modules: The wearable data acquisition terminal is a portable device with audio and video acquisition capabilities. It can be worn on different parts of the body and includes a camera, microphone, processor, storage unit, communication module, and human-machine interface unit (the human-machine interface unit includes physical buttons / voice modules). It is used for on-site audio and video acquisition, command response, and encrypted data storage. It supports multiple wearing locations and flexible shooting. During actual work, safety management personnel fix the wearable data acquisition terminal to the brim of their safety helmet, chest, or shoulder epaulettes. The adjustable connection structure allows adjustment of the camera's shooting angle to ensure coverage of the work area. After powering on, the terminal automatically synchronizes with the cloud management platform and loads the personnel ID and job title of the currently logged-in personnel. The system is designed to recognize physical buttons that respond to single presses, long presses, and consecutive presses according to a preset mapping. The voice module continuously listens for preset wake words, and when a valid command is recognized, it sends the command text to the processor for scene matching. The camera captures video streams at a resolution of at least 1080P, and the microphone simultaneously captures ambient audio. The raw data is temporarily stored in the processor cache. The communication module periodically detects the signal strength of the mobile network or wireless network. The storage unit uses a partitioned management mechanism to store the raw data to be encrypted, the encrypted files, and the system configuration files. The terminal's built-in GPS module obtains the geographic location coordinates in real time, and reads the current coordinates as part of the metadata each time a record is triggered. The intelligent performance recording system is a software system embedded in the data acquisition terminal. It includes a scenario-based performance directory library, an intelligent trigger engine, an automatic classification and storage module, an anti-tampering encryption module, and a performance statistics and analysis module. It is used to achieve job scenario matching, intelligent trigger recording, automatic classification and storage, and anti-tampering encryption. In operation, the scenario-based performance directory library reads configuration information synchronized from the cloud when the terminal starts, constructing a sub-directory index table of performance scenarios corresponding to the current job. The intelligent trigger engine continuously receives input signals from the voice recognition module and physical buttons. When a valid trigger signal is received, it queries the scenario-based performance directory library for a matching scenario identifier and outputs this identifier to the automatic classification and storage module. Upon receiving the scenario identifier, the automatic classification and storage module traverses or creates a "Performance Status" root directory in the storage unit, locating or creating a corresponding scenario. The system identifies the corresponding subdirectory, further concatenates the current system date with the performance scenario name to generate a folder for that day, and simultaneously writes the timestamp, GPS coordinates, and personnel ID as metadata into the folder's extended attributes. The anti-tampering encryption module calls the processor's graphics overlay engine in real time during file writing to embed the time, location, personnel ID, and scenario name as digital watermarks into specified pixel areas of video frames or photos. After file writing is complete, it calls the operating system interface to change the file attributes to read-only, and divides continuous recordings into 10-minute segments. Each segment independently uses the AES encryption algorithm to generate an encrypted file and a corresponding hash check value. The performance statistics and analysis module records the scenario identifier, timestamp, and file size of each triggered event locally on the terminal, forming a local performance log for subsequent uploading to the cloud management platform for aggregation and statistics. The cloud-based management platform receives and stores encrypted data uploaded by terminals, supporting the generation of performance logs, record tracking, and job configuration management. In practice, administrators access the job configuration management module through the cloud management platform's web interface. The platform has a built-in job-scenario mapping matrix, displaying job types such as dedicated safety officers, project managers, and safety directors. Administrators select allowed performance scenario subdirectories from the overall scenario library for each job. After configuration, the platform generates a structured configuration file. When a wearable data collection terminal initiates a synchronization request, the cloud management platform verifies the terminal's identity, distributes the corresponding job configuration information, and records the configuration distribution log. Terminal-uploaded encrypted data... When receiving confidential documents, the cloud management platform first compares the hash value to verify the integrity of the document before storing it in the storage area indexed by the corresponding position and time. A minimum retention period of 6 months is enforced, and regular user interfaces cannot delete or overwrite the file. The performance record generation module allows administrators to perform combined searches by position type, performance scenario, and time range. The search results are displayed in a list format, showing the time, location, personnel, and scenario information for each performance record, and provide online preview and download functions for videos or photos. The record traceability module records every file access, download, and export operation, forming an operation log for auditing purposes, ensuring the traceability of performance records.

[0022] Example 2, as Figure 1 As shown, based on Embodiment 1, the present invention also provides a method for intelligent recording of the performance of duties by safety management personnel, which is implemented based on the above-mentioned intelligent recording system for the performance of duties by safety management personnel, and includes the following steps: Step 1: The cloud management platform dynamically maps allowed subdirectories of job performance scenarios according to job type, forming a job-scenario matrix. After the wearable data collection terminal is synchronized, only the current job scenario is activated, and irrelevant functions are restricted. This realizes the preset of job responsibilities and directory configuration, avoiding functional redundancy and misoperation. The administrator enters the job configuration management module through the web interface of the cloud management platform. The platform has a built-in job-scenario mapping matrix, which displays the job types of full-time safety officers, project managers and safety directors. For each job, the administrator selects the allowed subdirectories of job performance scenarios from the main scenario library. After the configuration is completed, the platform generates a structured configuration file.When a wearable data collection terminal initiates a synchronization request, the cloud management platform verifies the terminal's identity and issues the corresponding job configuration information. Upon power-up, the terminal automatically loads the currently logged-in user's ID and job identifier. The scenario-based job performance directory reads the cloud-synchronized configuration information upon terminal startup and constructs an index table of job performance scenarios corresponding to the current job. The human-computer interaction unit only displays the job performance scenario options enabled for the current job; other scenario function entry points are logically blocked. The cloud management platform has a built-in job-scenario mapping matrix, pre-associating allowed job performance scenario subdirectories for the job types of dedicated safety officers, project managers, and safety directors, forming an explicit correspondence between jobs and scenarios. This achieves standardized binding between jobs and job performance scenarios, providing a basis for access control. According to reports, administrators access the job configuration management module via a web interface. The platform displays job types such as full-time safety officers, project managers, and safety directors. Each job type has a pre-defined list of corresponding performance scenarios. Full-time safety officers are pre-associated with sub-directories for pre-job inspections, full-process control of hazardous projects, supervision of high-risk operations, and hazard identification and mitigation. Project managers are pre-associated with sub-directories for on-duty production and inspection, and supervision of major hazard cases. Safety directors are pre-associated with sub-directories for emergency drill organization and personnel performance verification. Administrators can select the allowed performance scenario sub-directories by job type through the cloud management platform interface. Sub-directories are selected from the overall scenario library. Unselected scenarios are not displayed or are disabled on the terminal interface, enabling flexible configuration of job scenarios. (Unauthorized...) The system automatically masks and simplifies the user interface. The overall scenario library contains at least ten preset subdirectories for job performance scenarios, including full-process management of critical and major projects, hazard investigation and control, pre-job inspection, pre-job training supervision, high-risk operation supervision, emergency and accident response, personnel performance verification, shift production and inspection, major hazard supervision, and emergency drill organization. Administrators select the allowed subdirectories from the overall scenario library for each currently selected job type. After selection, the platform generates a structured job configuration file. Unselected subdirectories are marked as disabled in the configuration file. After the wearable data collection terminal synchronizes the configuration information to the cloud, the human-computer interaction unit only displays the currently enabled job performance scenario options; other scenario function entry points are logically blocked, thus implementing functional permission controls. Isolation ensures that the terminal only responds to operations within the scope of its current job responsibilities, avoiding functional redundancy and misuse. The wearable data collection terminal initiates a synchronization request. After the cloud management platform verifies the terminal's identity, it issues the corresponding job configuration information. After the terminal is powered on, it automatically loads the personnel ID and job identifier of the currently logged-in personnel. The scenario-based job performance directory library reads the synchronized configuration information when the terminal starts up and builds a sub-directory index table of job performance scenarios corresponding to the current job. The wake-up word library of the voice recognition module and the mapping table of physical buttons are updated synchronously to match only the scenarios that have been enabled for the current job. The interface menu of the human-computer interaction unit only displays the job performance scenario options that have been enabled. The physical button mapping for scenarios that have not been enabled is disabled, the voice wake-up word is not recognized, and the entry points for other scenario functions are logically blocked. Furthermore, Step One also includes: The performance scenario subdirectories adopt a hierarchical storage structure. The root directory is fixed as "Performance Status," and its subordinate subdirectories include at least ten pre-set performance scenario subdirectories, such as "Full-Process Management of Major and Hazardous Projects" and "Hazard Investigation and Management." This achieves standardized classification and storage of performance scenarios, facilitating subsequent retrieval and traceability. Within the storage unit, a "Performance Status" folder is created as the root directory, and ten subfolders are pre-created under this root directory, named as follows: Pre-job Inspection, Pre-job Education Supervision, Full-Process Management of Major and Hazardous Projects, High-Risk Operation Supervision, Hazard Investigation and Management, Emergency and Accident Response, Personnel Performance Verification, Shift Production and Inspection, and Heavy... The system includes sections for supervising major safety hazards and organizing emergency drills. Each subfolder corresponds to a specific category of job duties. The terminal automatically initializes this directory level upon initial startup. In the job-scenario mapping matrix, the permitted scenarios for full-time safety officers include pre-job inspections, full-process control of critical engineering projects, supervision of high-risk operations, and hazard identification and remediation. The permitted scenarios for project managers include on-duty production and inspection, and supervision of major safety hazards. This ensures precise matching between different job responsibilities and performance scenarios, avoiding permission confusion. The cloud management platform stores the job-scenario mapping matrix table, with job type as rows and scenario subdirectories as columns. In the corresponding row for all personnel, the four columns of pre-job inspection, full-process control of critical and major projects, supervision of high-risk operations, and hazard investigation and management are marked as permitted. In the corresponding row for project leaders, the two columns of production supervision and inspection and supervision of major hazards are marked as permitted, and the remaining columns are marked as prohibited. The platform generates configuration files for each position based on this matrix table. The wearable data acquisition terminal automatically loads the corresponding scene configuration table based on the synchronized position identifier. The wake-up word library of the voice recognition module and the mapping table of physical buttons are updated synchronously, matching only the scenes that are currently enabled for the position. This achieves automatic adaptation of terminal interaction logic and job responsibilities, reducing operational complexity. After the terminal synchronously obtains the position identifier... The system reads the scenario configuration table that matches the job identifier from the local storage, loads the allowed scenario names in the configuration table into the wake-up word library of the voice recognition module, and dynamically binds the allowed scenarios with the pressing modes of the physical buttons. The dedicated safety officer terminal only recognizes the wake-up words corresponding to allowed scenarios such as "start pre-shift inspection" and "start supervision of major hazardous projects". The project manager terminal only recognizes the wake-up words corresponding to allowed scenarios such as "start on-duty inspection" and "supervise major hidden dangers". Voice commands that do not belong to the current job's start scenario are ignored by the voice recognition module, and the pressing modes of the physical buttons that are not bound to allowed scenarios do not generate a trigger response. Step Two: Trigger the corresponding duty scenario via voice keywords or physical button mapping. Upon triggering, automatically create a storage folder in the corresponding subdirectory using the naming rule of date + duty scenario, and write the current time, geographical location, and personnel ID. This achieves one-click triggering of duty scenarios and automatic construction of storage paths, improving on-site operational efficiency. Safety management personnel fix the wearable data acquisition terminal to the brim, chest, or shoulder patch of their safety helmet, and adjust the camera shooting angle through the adjustable connection structure. The terminal's built-in GPS module obtains the geographical coordinates in real time. The physical buttons respond to single press, long press, and continuous press operations according to the preset mapping relationship. The voice module continuously listens for the preset wake-up word, and when a valid command is recognized, the command text is transmitted to the processor for scene matching. The trigger engine continuously receives input signals from the voice recognition module and physical buttons. When a valid trigger signal is received, it queries the scenario identifier in the scenario-based duty performance directory library, automatically calls the storage management module, locates or creates a subdirectory corresponding to the scenario identifier in the "Duty Performance Status" root directory of the storage unit, and generates a folder using the naming rule of concatenating the current system date with the duty performance scenario name. At the same time, it writes the real-time timestamp, GPS geographical coordinates, and the currently logged-in user ID as metadata into the folder attributes. The wearable data acquisition terminal collects voice input in real time, the voice recognition module extracts preset keywords and parses the corresponding duty performance scenario, or detects single press, long press, and continuous press actions of physical buttons, and matches the scenario identifier according to the pre-configured mapping relationship to achieve multi-mode processing. The system features status command recognition to meet the rapid triggering needs of different scenarios. The voice module continuously monitors preset wake words, and when it recognizes a phrase containing "start," "supervise," "inspect," or a combination of scenario names, it automatically completes and maps it to a standard duty performance scenario identifier. The physical buttons are configured with multiple trigger modes: a single short press maps to a hazard investigation and management scenario, two consecutive short presses map to a pre-shift inspection scenario, and a long press exceeding a preset threshold maps to a major engineering project full-process management scenario. The terminal's built-in GPS module obtains real-time geographic coordinates, reading these coordinates as part of the metadata each time a trigger is recorded. The collected raw data is temporarily stored in the processor cache. After the intelligent trigger engine confirms the scenario identifier, it automatically calls the storage management module and stores the data in the storage unit. The system locates or creates a subdirectory corresponding to the scene identifier in the root directory of the performance status, realizing the automatic association between scene identifiers and storage directories, ensuring accurate data archiving. The intelligent trigger engine continuously receives input signals from the voice recognition module and physical buttons. When a valid trigger signal is received, it queries the scene identifier in the scene-based performance directory library. After confirmation, the scene identifier is output to the automatic classification storage module. The automatic classification storage module traverses the "performance status" root directory in the storage unit. If a subdirectory corresponding to the scene identifier already exists in the root directory, it is located directly. If it does not exist, a new subdirectory is created according to the preset subdirectory names of at least ten performance scene subdirectories. In this subdirectory, a folder is generated using the naming rule of concatenating the current system date with the performance scene name.Simultaneously, the real-time timestamp, GPS geographic coordinates, and the currently logged-in user ID are written as metadata into the folder attributes, enabling automatic folder naming and metadata recording. This facilitates accurate tracing of the performance process afterward. The automatic classification and storage module obtains the current system date and converts it to the "YYYY-MM-DD" format. The date is then concatenated with the performance scenario name to generate the folder name, in the format "YYYY-MM-DD-performance scenario name". This folder is generated in the located or created subdirectory. Simultaneously, the processor reads the real-time clock to obtain the current timestamp, calls the GPS module to obtain the current geographic coordinates, and reads the login authentication information to obtain the user ID. These three pieces of data are embedded in the folder's file system extended attributes in a structured manner for direct reading by subsequent retrieval and verification modules. Furthermore, step two also includes configuring multiple trigger modes for physical buttons. A single short press is mapped to a hazard investigation and management scenario, two consecutive short presses are mapped to a pre-shift inspection scenario, and a long press exceeding a preset threshold is mapped to a major engineering project full-process management scenario. This enables rapid switching between multiple scenarios with a single button, improving on-site operational convenience. The processor detects the pressing duration and interval of the physical button in real time. When the pressing time is less than the preset short press duration threshold, it is determined as a single short press; when the interval between two presses is less than the preset interval threshold, it is determined as two consecutive short presses; and when the pressing duration exceeds the preset long press duration threshold, it is determined as a long press. Based on different pressing judgment results, the corresponding duty performance scenario identifier is matched. Based on the fuzzy matching supported by the voice recognition module, when a sentence containing the combination of "start," "supervision," "inspection," and scenario name is recognized, it is automatically completed and mapped to the standard duty performance scenario identifier, lowering the trigger threshold, reducing the burden of remembering voice commands, and improving the success rate and ease of use of voice triggering. The speech recognition module converts the collected voice commands into text strings and iterates through the text to see if it contains preset keyword combinations. When it detects that "start" and words such as "critical and major project" or "supervision" appear simultaneously, it automatically completes the command and maps it to a critical and major project full-process management scenario identifier, without requiring the user to speak a complete standard command statement. The module determines the naming format of the completed folder as "YYYY-MM-DD-performance scenario name". Metadata is embedded in the file system extended attributes in a structured manner, which can be directly read by the subsequent retrieval and verification modules. This achieves standardized folder naming and structured storage of metadata, which facilitates efficient retrieval and verification. The processor calls the file system interface to write the timestamp, GPS coordinates, and personnel ID into the extended attribute area of ​​the folder. This area stores structured data in the form of key-value pairs. The retrieval module can directly read the extended attributes when traversing the folder without opening the file content. The verification module synchronously reads the metadata in the extended attributes for consistency comparison. Step 3: Based on the trigger command type, the system enters recording mode, which involves full-process video recording plus intermittent photo taking. Audio and video files are automatically saved to the corresponding folders according to preset paths. Filenames include the scene code and personnel ID, enabling automatic categorization upon triggering. This solves the problem of low efficiency in manual categorization after the fact. The camera captures video streams at a resolution of at least 1080P, while the microphone simultaneously captures ambient audio. The raw data is temporarily stored in the processor cache. The intelligent trigger engine automatically selects the recording mode based on the output scene identifier. If the scene identifier is "full-process management of critical engineering projects" or "pre-shift inspection," it activates continuous full-process video recording mode. If the scene identifier is "hazard investigation and management," it enters intermittent photo taking mode, generating a high-resolution still image each time a physical button trigger signal is received. The generated video or photo files are stored in the corresponding folder constructed in Step 2 using the naming rule "timestamp-scene code-person ID." The automatic categorization and storage module completes the automatic categorization and storage of the files. Step 4: During file generation, an automatic solidification operation is performed, with real-time overlay of digital watermarks. The file is marked as read-only, and encrypted recordings are segmented into fixed durations to prevent local modification, deletion, or forgery, ensuring the authenticity and integrity of the performance data. This meets the requirements of the Safety Production Law for evidence tamper-proofing. During file writing, the anti-tampering encryption module calls the graphics overlay engine in the processor in real time to embed the time, geographical location, personnel ID, and scene name into designated pixel areas of video frames or photos as digital watermarks. After the file is written to the storage unit, the storage unit adopts a partition management mechanism. The encrypted and solidified file is stored in a designated partition, and the operating system-level call modifies the file attributes to read-only, prohibiting any local process from modifying, deleting, or renaming the file. Continuous recordings are segmented into fixed 10-minute segments, and each segment independently uses the AES encryption algorithm to generate an encrypted file and a corresponding hash check value. Damage or tampering of any segment does not affect the integrity verification of other segments. Step 5: When a network connection is available, the wearable data collection terminal automatically backs up encrypted files to the cloud management platform. The cloud management platform mandates retention for at least 6 months and prohibits regular deletion. It supports traceability by job position and scenario, forming a structured performance record, achieving cloud-based retention and structured traceability of performance data, supporting job performance assessment and responsibility identification. The communication module periodically checks the signal strength of mobile or wireless networks. When a network is detected, it automatically adds locally stored encrypted files to the upload queue in chronological order of generation. During the upload process, a breakpoint resume mechanism is used, resuming upload from the point of interruption if file transmission is interrupted. After receiving the file, the cloud management platform first performs hash value comparison, verifies file integrity, and then stores it in the storage area with the corresponding job position and time index, and mandates a file retention period of no less than 6 months. Regular user operation interfaces cannot delete or overwrite any files. The performance record generation module allows administrators to perform combined searches by job type, performance scenario, and time range. The search results are displayed in a list format, showing the time, location, personnel, and scenario information for each performance record. The record tracing module records every file access, download, and export operation, forming an operation log for auditing purposes.

[0023] Example 3, as Figure 1As shown, based on Embodiments 1 and 2, the present invention provides a technical solution: Step three specifically includes: the intelligent triggering engine automatically selects the recording mode according to the output scene identifier. If the scene identifier is "full-process control of critical and major projects" or "pre-shift inspection," then the continuous recording mode is started, and the recording frame rate and resolution are executed according to preset parameters to achieve automatic matching between the scene and the recording mode, meeting the recording needs of different duty scenarios. The intelligent triggering engine compares the received scene identifier with the preset mode mapping table. When a match is found for "full-process control of critical and major projects" or "pre-shift inspection," a continuous recording start command is sent to the camera control module, and the camera records continuous recording at a constant frequency. Video streaming begins at a resolution lower than 1080P and a preset frame rate, while the microphone simultaneously captures ambient audio. The raw data is temporarily stored in the processor cache. Recording continues until a stop command is received or a new duty scenario is triggered. If the scenario is identified as "hazard investigation and remediation," an intermittent photo-taking mode is entered. Each time a physical button trigger signal is received, a high-resolution still image is generated, with a timestamp and scene code appended to the image, enabling precise capture of hazard investigation scenarios and highlighting key evidence during duty performance. When the intelligent trigger engine matches the "hazard investigation and remediation" identifier, it sends a photo-ready command to the camera control module, initiating video recording. The head enters a low-power standby state. When the processor detects a physical button press, it immediately triggers the camera to capture a still image with a resolution no lower than the preset high-resolution parameter. Simultaneously, the processor reads the real-time timestamp and the current scene code, overlaying the timestamp and scene code as digital watermarks onto a designated pixel area of ​​the still image. Each button press generates only one image; multiple presses generate multiple independent images. The generated video or photo files are stored in the corresponding folder using the naming convention of timestamp-scene code-person ID. This achieves fully automated categorization after a single trigger, requiring no manual intervention, and realizes automated naming of performance documents. The categorization process completely solves the inefficiency of manual sorting. After recording stops or photos are generated, the processor obtains the recording start timestamp or photo generation timestamp, the scene code corresponding to the current scene, and the currently logged-in personnel ID. It then concatenates these elements in the format of "timestamp-scene code-person ID" to generate a filename. The timestamp is accurate to the second, the scene code is a preset English abbreviation or numerical code, and the personnel ID is a unique identifier. The automatic classification and storage module writes the generated video files or photo files to the corresponding folder that has been built in step two. After writing is completed, the local performance log is updated to record the filename, file size, and generation time. The entire process requires no manual intervention. Step four specifically includes: During the file generation process, the processor in the wearable acquisition terminal reads the current time, GPS location, personnel ID, and scene name in real time. The graphics overlay engine embeds this information as a digital watermark into a designated area of ​​the video or photo, enabling real-time solidification of performance information into the audio and video, preventing content tampering at the source. The processor reads the current time (accurate to the second) from the real-time clock module, the current location coordinates from the GPS module, the personnel ID from the login authentication cache, and the current scene name from the intelligent trigger engine. The graphics overlay engine converts this information into text format, using a preset font size and color. The process involves pixel-level overlay of each frame of video or photograph at its top left or bottom edge. The overlaid information is seamlessly integrated with the original image, making it impossible to separate or erase using conventional image editing tools. After the file is written to the storage unit, the operating system modifies the file attributes to read-only, prohibiting any local process from modifying, deleting, or renaming the file. The file hash value is recorded as an integrity verification benchmark, achieving low-level write protection and integrity verification to ensure the evidence is tamper-proof. The storage unit employs a partitioned management mechanism, storing encrypted files in designated partitions. After the file is written, the processor calls the file system interface provided by the operating system to transfer the file... The attribute flag is set to read-only mode, and modification permissions for the current file's directory are blocked. After the system-level call is completed, the processor uses a hash algorithm to calculate a hash value for the file content and stores this hash value in an independent verification database. Each time the file is read subsequently, the hash value is recalculated and compared with the original hash value. If they do not match, the file is determined to have been tampered with. Continuous recordings are stored in segments of fixed duration, each segment is independently encrypted and an encryption verification code is stored. Damage or tampering of any segment does not affect the integrity verification of other segments. Stored segments are not lost in the event of an abnormal power outage of the wearable capture terminal, achieving independent protection for recording segments and improving data resilience. With its abnormal recovery capabilities, the anti-tampering encryption module sets a segment timer when recording starts. Every fixed duration, such as 10 minutes, it automatically stops the current recording file and starts a new recording file. Each segment recording file is independently encrypted using the AES encryption algorithm. After encryption, a corresponding encryption verification code is generated and appended to the end of the file. There is no data dependency between segments. When the terminal is powered off due to battery depletion or unexpected power failure, the stored segments that have been fully written to the storage unit remain intact. The cached data of the current segment that has not reached the fixed duration is automatically discarded. The recording of a new segment will start again the next time the device is started. The stored segments can be independently decrypted and verified when uploaded to the cloud management platform. Step five specifically includes: The wearable data collection terminal periodically checks the network connection status. When a mobile or wireless network is detected as available, it automatically adds locally stored encrypted files to the upload queue in chronological order of their creation time, achieving automatic backup of performance data without manual intervention. The communication module checks the signal strength of the mobile or wireless network every preset time interval, such as 30 seconds. When a network is detected as available and the signal strength exceeds a preset threshold, the terminal automatically scans the encrypted and solidified file partitions in the storage unit, reads the timestamp attributes of each file, generates an upload task list in chronological order of creation time, and adds the files in the task list to the upload queue in sequence. The upload queue supports priority adjustment, and is suitable for critical scenarios such as the supervision of major and hazardous projects. It can be configured to prioritize uploads, employing a resume mechanism during the upload process. If file transfer is interrupted, it resumes from the point of interruption. After upload, the local device and the cloud management platform compare hash values; if they match, it is marked as synchronized, ensuring a reliable and complete upload process. Automatic recovery is implemented after network fluctuations to avoid data loss. When the terminal initiates a file upload request to the cloud management platform, it includes the file size and upload position information. If the connection is interrupted due to network fluctuations during upload, the terminal sends a resume request to the cloud management platform after network recovery. The platform returns the interruption position based on the length of received data, and the terminal continues transmitting the remaining data segment from that position. After the file is completely uploaded, the terminal sends the locally stored file hash value to the cloud management platform, which calculates the received file hash value. The platform compares the hash value of the file with the hash value sent by the terminal. If they match, the platform returns a confirmation message. Upon receiving the confirmation, the terminal marks the local file as synchronized and records the synchronization completion timestamp. If the hash values ​​do not match, a re-upload mechanism is triggered. After receiving the file, the cloud management platform stores it in the storage area corresponding to the job position and time index, forcibly setting a retention period of no less than 6 months. It cannot be deleted or overwritten by anyone through the regular user operation interface. It only supports retrieval by job type, job performance scenario, and time range, and generates a structured job performance ledger. This achieves mandatory retention and structured retrieval of cloud data, meets regulatory traceability requirements, and supports job performance assessment. After receiving the complete encrypted file, the cloud management platform parses the file metadata to extract the job type and job performance. The platform assigns storage partitions based on job type and timestamp information, creating storage paths according to a hierarchical structure of "job type / month / year / job scenario." Files are stored under these paths, and the platform sets the retention deadline in the file attributes to the current date plus 6 months. A background daemon scans the storage area daily and writes to files that have not reached their retention deadline. Any deletion or overwrite requests initiated through the web interface or API are intercepted by the regular user operation interface and return an insufficient permission message. The job performance ledger generation module supports administrators to combine and search by job type dropdown, job scenario multi-select box, and time range selector. The search results are displayed in tabular form, showing the personnel ID, time, location, and scenario information for each job performance record.It automatically calculates the frequency of job performance for each position and scenario, generating a structured performance record that can be exported to PDF or Excel format. Step five also includes: The wearable data acquisition terminal periodically detects the signal strength of mobile or wireless networks through the communication module. When a network is detected, it automatically scans the encrypted and hardened file partition in the storage unit, constructs an upload task list according to the order of file generation timestamps, and adds the encrypted files in the list to the upload queue in sequence, realizing automatic and orderly uploading of encrypted files without manual intervention to complete data synchronization. The communication module reads the current network signal strength value at preset time intervals and compares it with a preset threshold. When the signal strength exceeds the threshold, a scanning operation is triggered. The processor traverses all file attributes in the encrypted and hardened file partition, reads the timestamp field of each file, sorts it in ascending order, and generates an upload queue. The task list is transmitted, and the upload queue manager adds file paths to the upload queue one by one according to the list order. Manual priority adjustment of files in critical scenarios is supported. When the wearable data acquisition terminal initiates a file upload request to the cloud management platform, it carries the file size and upload location information. A breakpoint resume mechanism is used to resume the upload from the point of interruption. After the file is completely transmitted, the terminal sends the locally stored file hash value to the cloud management platform. The cloud management platform calculates the hash value of the received file and compares it with the hash value sent by the terminal to ensure complete and reliable file upload. Transmission is automatically resumed and data consistency is verified after network interruption. The terminal encapsulates the total file length and the starting byte offset of this resumed upload in the request header. After receiving this, the platform... Data is written starting from the offset position. After each data block is transmitted, the terminal records the number of bytes uploaded locally. If the connection is interrupted, the terminal retains the breakpoint information and re-initiates the resume request after the network is restored. After all bytes of the file have been transmitted, the terminal reads the locally stored hash value and encapsulates it in a verification request before sending it to the platform. The platform uses the same hash algorithm to calculate the received complete file and compares the calculation result with the value sent by the terminal bit by bit. After receiving the complete encrypted file, the cloud management platform parses the metadata to extract the job type, job performance scenario, and timestamp information. It stores the information in the storage area according to the hierarchical structure of job type and time index, and forces a retention period to be set so that it cannot be deleted or overwritten through the regular user operation interface. The stored performance records implement hierarchical storage and mandatory retention of performance data to prevent manual deletion or overwriting and ensure the integrity of the evidence chain. After receiving the encrypted file, the platform calls the parsing module to read the file header metadata segment, extracting the job identifier field, scenario identifier field, and timestamp field. The storage management module creates or locates the corresponding top-level directory based on the job identifier, creates second-level subdirectories under the top-level directory according to the year and month, and creates third-level subdirectories under the second-level subdirectories according to the performance scenario. After the file is stored, the expiration timestamp in the file attributes is set to the current time plus the retention period. The file system implements a write protection policy for this storage area. When regular interface calls delete or overwrite operations, a permission denial status code is returned.

[0024] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. The scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An intelligent recording system for the performance of duties by safety management personnel, characterized in that, Includes the following modules: Wearable data acquisition terminals can be worn on different parts of the body, including cameras, microphones, processors, storage units, communication modules, and human-computer interaction units, for on-site audio and video acquisition, command response, and encrypted data storage; The intelligent record system for job performance includes a scenario-based job performance catalog, an intelligent trigger engine, an automatic classification and storage module, an anti-tampering encryption module, and a job performance statistical analysis module. It is used to achieve job scenario matching, intelligent trigger recording, automatic classification and storage, and anti-tampering encryption. The cloud management platform is used to receive and store encrypted data uploaded by terminals, and supports the generation of performance ledgers, record traceability, and job configuration management.

2. A method for intelligently recording the performance of duties by safety management personnel, implemented based on the intelligent recording system for the performance of duties by safety management personnel as described in claim 1, characterized in that, Includes the following steps: Step 1: The cloud management platform dynamically maps allowed job performance scenario subdirectories according to job type, forming a job-scenario matrix. After the wearable data collection terminal is synchronized, only the current job scenario is activated, and irrelevant functions are restricted. Step 2: Trigger the corresponding job performance scenario through voice keywords or physical button mapping. Upon triggering, automatically create a storage folder in the corresponding subdirectory with the naming rule of date + job performance scenario, and write the current time, geographical location, and personnel ID. Step 3: Enter recording mode according to the trigger command type. The recording mode is to record the whole process and take intermittent photos, and generate audio and video files that are automatically saved to the corresponding folder according to the preset path. The file name contains the scene code and personnel ID. Step 4: When the file is generated, the solidification operation is automatically performed, a digital watermark is superimposed in real time, the file is marked as read-only, and the video is encrypted in segments according to a fixed duration to prevent local modification, deletion or forgery; Step 5: When there is a network connection, the wearable data collection terminal automatically backs up the encrypted files to the cloud management platform. The cloud management platform will retain the files for at least 6 months and will not allow them to be deleted regularly, thus forming a structured performance record.

3. The intelligent recording method for the performance of duties by safety management personnel according to claim 2, characterized in that: Step one specifically includes: The cloud management platform has a built-in job-scenario mapping matrix, which pre-associates allowed job scenario subdirectories for the job types of full-time safety officers, project managers and safety directors, forming an explicit correspondence between jobs and scenarios; Administrators can select the allowed subdirectories of job performance scenarios by job position through the cloud management platform interface. The subdirectories are selected from the main scenario library. Unselected scenarios will not be displayed or will be disabled in the terminal interface. After the wearable data collection terminal synchronizes the configuration information with the cloud, the human-computer interaction unit only presents the job performance scenario options that have been enabled for the current position, and the entry points for other scenario functions are logically blocked.

4. The intelligent recording method for the performance of duties by safety management personnel according to claim 3, characterized in that: Step one also includes: The duty performance scenario subdirectory adopts a hierarchical storage structure. The root directory is fixed as the duty performance status, and its lower level is pre-set with at least ten pre-set duty performance scenario subdirectories, including the whole process control of major and critical projects and the investigation and management of hidden dangers. In the job-scenario mapping matrix, the permitted scenario subset for full-time safety officers includes pre-job inspections, full-process control of critical and major projects, supervision of high-risk operations, and hazard investigation and management; the permitted scenario subset for project leaders includes leading production and inspections and supervising major hazards. The wearable data acquisition terminal automatically loads the corresponding scene configuration table based on the synchronized job identifier. The wake-up word library of the voice recognition module and the mapping table of physical buttons are updated synchronously, matching only the scenes that have been activated for the current job.

5. The intelligent recording method for the performance of duties by safety management personnel according to claim 2, characterized in that: Step two specifically includes: Wearable data acquisition terminal collects voice input in real time. The voice recognition module extracts preset keywords and parses the corresponding duty performance scenarios, or detects single press, long press and continuous press actions of physical buttons, and matches scene identifiers according to pre-configured mapping relationships. After confirming the scene identifier using the intelligent trigger engine, the storage management module is automatically invoked to locate or create a subdirectory corresponding to the scene identifier in the root directory of the storage unit's performance status. In this subdirectory, a folder is generated using the naming rules of concatenating the current system date with the name of the duty scenario. At the same time, the real-time timestamp, GPS geographic coordinates, and the ID of the currently logged-in user are written as metadata into the folder attributes.

6. The intelligent recording method for the performance of duties by safety management personnel according to claim 5, characterized in that: Step two also includes: The physical button is configured with multiple trigger modes: a single short press is mapped to the hidden danger investigation and management scenario, two consecutive short presses are mapped to the pre-shift inspection scenario, and a long press exceeding the preset threshold is mapped to the whole process control scenario of major and hazardous projects. Based on the fuzzy matching supported by the speech recognition module, when a sentence containing a combination of "start", "supervision", "inspection" and scene name is recognized, it is automatically completed and mapped to the standard duty performance scene identifier. Determine the naming format of the completed folders, and embed metadata into file system extended attributes in a structured manner.

7. The intelligent recording method for the performance of duties by safety management personnel according to claim 2, characterized in that: Step three specifically includes: The intelligent trigger engine automatically selects the recording mode based on the output scene identifier. If the scene identifier is the whole process control of critical and major projects or pre-shift inspection, the whole process continuous recording mode is started, and the recording frame rate and resolution are executed according to the preset parameters. If the scene is identified as hazard investigation and management, it will enter the intermittent photo mode. Each time a physical button trigger signal is received, a high-resolution static image will be generated, and the image will be appended with the current timestamp and scene code. The generated video or photo files are stored in the corresponding folder created using the naming convention of timestamp-scene code-person ID.

8. The intelligent recording method for the performance of duties by safety management personnel according to claim 2, characterized in that: Step four specifically includes: During the file generation process, the processor in the wearable acquisition terminal reads the current time, GPS location, personnel ID and scene name in real time, and embeds the read information into a designated area of ​​the video or photo in the form of a digital watermark through a graphics overlay engine. After the file is written to the storage unit, the operating system calls to change the file attributes to read-only, prohibiting any local process from performing modification, deletion or renaming operations on the file, and records the file hash value as a basis for integrity verification. Continuous recordings are stored in segments of fixed duration. Each segment is independently encrypted and an encryption verification code is stored. Damage or tampering of any segment will not affect the integrity verification of other segments. Stored segments will not be lost when the wearable acquisition terminal experiences an abnormal power outage.

9. The intelligent recording method for the performance of duties by safety management personnel according to claim 2, characterized in that: Step five specifically includes: The wearable data collection terminal periodically checks the network connection status. When a mobile network or wireless network is detected to be available, it automatically adds the locally stored encrypted files to the upload queue in the order of their generation time. The upload process uses a breakpoint resume mechanism. If the file transfer is interrupted, the upload will resume from the breakpoint. After the upload is completed, the hash value is compared between the local and cloud management platforms. If they match, the upload is marked as synchronized. After receiving the file, the cloud management platform stores it in the storage area indexed by the corresponding position and time, and sets a mandatory retention period of no less than 6 months.

10. The intelligent recording method for the performance of duties by safety management personnel according to claim 9, characterized in that: Step five also includes: The wearable data acquisition terminal periodically detects the signal strength of mobile or wireless networks through the communication module. When a network is detected, it automatically scans the encrypted and hardened file partitions in the storage unit, builds an upload task list according to the order of file generation timestamps, and adds the encrypted files in the list to the upload queue in sequence. When the wearable data collection terminal initiates a file upload request to the cloud management platform, it carries the file size and upload location information. It uses a breakpoint resume mechanism to resume the upload from the breakpoint. After the file is completely transmitted, the terminal sends the locally stored file hash value to the cloud management platform. The cloud management platform calculates the hash value of the received file and compares it with the hash value sent by the terminal. After receiving the complete encrypted file, the cloud management platform parses the metadata to extract job type, job performance scenario and timestamp information, and stores it in the storage area according to the hierarchical structure of job type and time index.

Images