Intelligent security management system and method for community

Abstract

The invention relates to the technical field of community security, in particular to an intelligent community security management system and method, and the method comprises the steps: collecting equipment data, verifying the daily operation of equipment, collecting personnel and event data, and carrying out the data association and data integration; when an abnormal event is identified, determining a basic level according to the integrated data, determining an associated risk probability through a Bayesian network model, inputting the basic level and the associated risk probability into a dynamic weight adaptive grading model, and determining a final level of the event; inputting the obtained event final grade into a trend prediction model to obtain a risk prediction value, determining a comprehensive risk value of each region according to the risk prediction value, and generating a real-time risk thermodynamic diagram; and according to the final grade of the event and the associated data, screening the processing personnel meeting the conditions to perform task assignment, and performing event processing by the processing personnel. According to the scheme, by constructing the equipment, event and personnel association chain, dynamic verification and intelligent grading are realized, and the management efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of residential security technology, and in particular to a residential intelligent security management system and method. Background Art

[0002] Traditional residential security equipment lacks a unified identity authentication and control mechanism. Equipment coding rules are not standardized, making it prone to duplication and confusing formats, making it difficult to uniquely identify equipment. Furthermore, equipment status updates rely on manual entry, which is subject to significant lags. Managers are unable to monitor the equipment's online, offline, and faulty operating status in real time. Furthermore, device information storage lacks secure encryption measures, creating the risk of malicious tampering. This further reduces equipment availability and management efficiency, resulting in high O&M costs for security equipment and making it difficult to support precise security control needs.

[0003] Community incident handling has long relied on manual judgment and dispatch, lacking the ability to systematically analyze incident characteristics. Key information, such as high-frequency incident types (such as illegal electric vehicle charging) and high-incidence areas (such as unit entrances and electric vehicle sheds), cannot be quantified through statistical data, resulting in a lack of scientific basis for resource allocation. This model often leads to resource misallocation, such as delayed responses to high-priority incidents and excessive resource consumption by low-priority incidents, which seriously affects the response efficiency and handling effectiveness of community security.

[0004] Data such as equipment operating status, event handling processes, and personnel operation records are stored in separate systems or ledgers, lacking a connected data chain linking "equipment-event-personnel." When tracing an event (e.g., verifying the reliability of event collection equipment) or determining responsibility (e.g., confirming the compliance of personnel handling operations), data must be retrieved across multiple systems. This lacks logical connections between data, leading to a cumbersome and inefficient tracing process and a lack of management transparency, making it difficult to meet the requirements of closed-loop community security management. Summary of the Invention

[0005] The present invention realizes dynamic verification and intelligent classification by building a chain of equipment, events and personnel associations, thereby improving management efficiency and strengthening traceability.

[0006] The technical solution proposed by the present invention is: a community intelligent security management method, the method comprising: Collect equipment data, verify daily equipment operation, collect personnel and event data, and perform data correlation and integration; When an abnormal event is identified, the basic level is determined based on the integrated data, and the associated risk probability is determined through the Bayesian network model. The basic level and associated risk probability are input into the dynamic weight adaptive grading model, and the final level of the event is determined by combining the time period weight, seasonal weight, historical association weight, and regional sensitivity weight. The final level of the event is input into the trend prediction model to obtain the risk prediction value, and the comprehensive risk value of each area is determined based on the risk prediction value. A real-time risk heat map is generated based on the comprehensive risk value; According to the final level of the event and the associated data, qualified processing personnel are screened for task assignment. The processing personnel handle the event and record the processing process data.

[0007] Preferably, the specific process of the device data collection is as follows: Device data is collected from security-related equipment in the residential area. When the device is started, the system automatically reads the device's unique code and associates it with the preset primary key ID. The legitimacy of the device is confirmed through dual verification of the primary key ID and the device code. When registering the device, the coding format compliance verification, the residential area authority association verification, and the registration information are encrypted and stored. During daily operation, the device triggers data collection trigger verification, remote operation forced verification, and offline reconnection secondary verification. The verification record is linked to the device health and event classification response.

[0008] Preferably, the specific process of collecting personnel and event data is as follows: When collecting basic personnel information, a unique primary key ID is assigned and encrypted and bound to the personnel identity. The corresponding operation permissions are matched according to the type of responsibilities, and the permission information is associated with the personnel primary key ID and stored. When the device recognizes an abnormal event, the system automatically records the type, time and location of the event, and forcibly associates the device primary key ID and device code that collected the event. When manually reporting an event, the reporter needs to verify his or her identity. The system matches the corresponding primary key ID through his or her identity. After confirming that the identity is legal, the primary key ID of the reporter is associated with the event information and stored.

[0009] Preferably, the specific process of data integration is as follows: The format of device data that has passed double verification is unified, and the event data collected by the corresponding device is associated through the primary key ID, and the responsible area and authority scope are associated through the personnel primary key ID; double verification of the primary key ID and equipment code is performed, and the device data that has not passed the verification is eliminated. The integrity of the data that has passed the verification is checked to ensure that the event time, location, associated device ID, and associated personnel ID are not missing, and the logic between fields is self-consistent; the device status data that has passed the verification is standardized and converted, and the original status descriptions of different devices are uniformly mapped to fixed codes; the unstructured data collected by AI devices is structured.

[0010] Preferably, the specific process of obtaining the final level of the event is as follows: Based on the potential impact of an event on personal safety, property safety, and public order, three basic levels are preset; the probability of multiple event associations is analyzed through a Bayesian network model; the basic level, associated risk probability, time period weight, seasonal weight, historical association weight, and regional sensitivity weight are input into the final event level calculation formula to solve the final event level, and the final event level is compared with the set threshold to determine the event type.

[0011] Preferably, the handling method is determined according to the final level of the event: If the final level of the event is greater than or equal to 4, it is judged as a high-level event, and an emergency response must be initiated within 5 minutes, and the information must be pushed to the terminal of the handling personnel and an audible and visual alarm must be triggered simultaneously; if the final level of the event is greater than or equal to 2.5 and less than 4, it is judged as a medium-level event, and handling personnel must be arranged to arrive at the scene within 15 minutes to handle the situation and record the handling results; if the final level of the event is less than 2.5, it is judged as a low-level event, and a rectification reminder must be pushed through the APP within 2 hours, and the feedback must be tracked.

[0012] Preferably, the specific content of the real-time risk heat map is as follows: A hybrid model is constructed based on LSTM and Transformer. The target behavior features are extracted through AI cameras, and the primary key ID of the area where the event occurred is associated. Environmental data is collected through environmental sensors. Multi-source fusion data is obtained through weighted summation. The impact of recent cases is highlighted through the time decay factor. Historical data is determined. The multi-source fusion data, historical data, and the final level of the event are input into the hybrid model to obtain the risk prediction value. The comprehensive risk value is obtained by weighted summation based on the weekly frequency of events, the final level of events, the associated risk probability, and the risk prediction value. Three levels of visual identification are divided in the heat map according to the comprehensive risk value.

[0013] Preferably, the specific process of the processing personnel is as follows: The system locates the incident area based on the device primary key ID associated with the event, and screens out qualified processing personnel based on the personnel primary key ID bound to the area and their responsibilities and permissions. When the task is pushed, the device primary key ID and code associated with the event are attached; after the processing personnel arrive at the scene, they verify the consistency of the equipment code and task information through the terminal; after the task is completed, the system will associate the disposal record with the equipment double verification information for archiving.

[0014] The present invention also provides a community intelligent security management system, which is used to execute the community intelligent security management method.

[0015] The present invention also provides a computer-readable storage medium, which stores a computer program. The computer program is executed by a processor to implement the method for intelligent security management of a community.

[0016] Beneficial effects of the present invention: 1. This solution builds a rigorous device identity authentication system through the rigid binding of the device master key ID and code and dynamic verification throughout the entire process. During the first access, the coding format compliance, uniqueness and core information integrity are synchronously verified to eliminate problems such as code duplication and format confusion from the source; in daily operation, dynamic verification every 30 seconds (master key ID matching, code matching, heartbeat time validity) and multi-scenario verification such as data collection triggering, remote operation forced, and offline reconnection secondary are carried out to ensure that the device status is updated in real time and is authentic and reliable. This mechanism improves the availability of equipment. Management personnel can grasp the operating status of all community equipment in real time through a unified interface. The equipment maintenance response time is shortened to one-third of the original time, and the ineffective operation and maintenance costs caused by the confusion of equipment information are reduced, which significantly improves the accuracy and efficiency of equipment management.

[0017] 2. An event classification mechanism based on a dynamic weighted adaptive model and multi-event correlation analysis enables precise quantification of event levels and intelligent response. By integrating dynamic weights such as the inherent risk of event type, time period, season, historical data, and regional sensitivity, and combining them with a Bayesian network model to calculate the probability of association between multiple events, the accuracy of event level determination is improved by 80% compared to traditional manual judgment. High-level events (L ≥ 4.0) trigger an emergency response within 5 minutes, while medium-level events (2.5 ≤ L < 4.0) are handled by dedicated personnel within 15 minutes, reducing resource misallocation. Furthermore, the system automatically optimizes resource allocation by analyzing high-frequency event types and high-incidence areas. For example, patrols are increased in areas with a high incidence of illegal electric vehicle charging, reducing the recurrence rate of similar incidents and significantly improving the timeliness of event handling and resource utilization efficiency.

[0018] 3. The "equipment-event-personnel" associated traceability chain connects scattered data in series through the primary key ID to form a complete closed-loop management chain. Equipment operation data, event processing records, and personnel operation logs are traced with one click through the primary key ID. The event tracing time is shortened from the traditional hours to within 5 minutes, and the efficiency of responsibility identification is improved. For example, when it is necessary to verify the results of an event handling, the equipment double verification record can be retrieved through the device primary key ID associated with the event to confirm the authenticity of the data, and then the handling process and time consumption can be viewed through the personnel primary key ID, which significantly improves management transparency. In addition, the traceability chain provides data support for management optimization. For example, by analyzing the personnel handling records of a certain area, skill shortcomings are discovered and targeted training is carried out, which increases the event handling completion rate in the area from 70% to 95%, effectively solving the problems of data dispersion and traceability difficulties in traditional management. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a flow chart of a community intelligent security management method of the present invention; Figure 2The present invention is a flowchart of the management process of a community intelligent security management method. DETAILED DESCRIPTION

[0020] The following description is intended to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are for illustrative purposes only, and those skilled in the art will readily appreciate other obvious variations. The basic principles of the present invention defined in the following description may be applied to other embodiments, variations, improvements, equivalents, and other technical solutions that do not depart from the spirit and scope of the present invention.

[0021] It is to be understood that the term "one" should be understood as "at least one" or "one or more", that is, in one embodiment, the number of an element may be one, while in another embodiment, the number of the elements may be multiple, and the term "one" should not be understood as a limitation on the quantity.

[0022] like Figure 1 and Figure 2 As shown, multi-source data collection is performed, including device data collection as well as personnel and event data collection. Device data is collected for security-related equipment such as AI monitoring and access control terminals within the community. When AI monitoring, access control terminals, and other devices are started, the system automatically reads the device's unique code (such as the IMEI code) and associates it with the preset primary key ID. The legitimacy of the device is confirmed through dual verification of the primary key ID and the device code. Upon initial access, the code is verified to match the community's device ledger. If passed, registration is completed and a unique primary key ID is assigned. During daily operation, before real-time collection of device status data (online status, heartbeat time, sensor data), the binding relationship between the primary key ID and the device code is verified again to prevent unauthorized device access or data tampering, ensuring the authenticity of original data such as video streams and abnormal behavior records.

[0023] When registering a device, it is necessary to perform coding format compliance verification, cell authority association verification, and encrypted storage of registration information. The specific process of coding format compliance verification is as follows: in addition to uniqueness verification, the system verifies the format of the device code (such as the IMEI code), for example, verifying whether its length meets the 15-bit standard and whether it contains legal characters. If the format does not comply, the registration is directly rejected to avoid subsequent association failures due to coding format errors. The dual verification of format and uniqueness can be simplified as: coding verification result = format compliance Uniqueness: Specification compliance is a Boolean value (1 for compliance, 0 for non-compliance), and uniqueness is a Boolean value (1 for a unique code, 0 for an existing code). The code verification passes only when both are 1. During the dual format and uniqueness verification, a core information integrity check is also performed simultaneously. After simultaneously verifying key information such as the device name and cell ID to ensure that none are missing, the system automatically generates a unique primary key ID and establishes a rigid binding relationship with the device code. The database records the "primary key ID-device code-cell location" association, providing accurate device location information for subsequent management.

[0024] The specific process for cell permission association verification is as follows: when binding the cell ID, the system additionally verifies that the cell is in the "Enabled" state and that the device type matches the cell's access device list (for example, a cell prohibits access to monitoring devices using non-encrypted protocols), ensuring that the device complies with cell management requirements. The specific process for encrypted storage of registration information is as follows: the binding relationship between the device primary key ID and the code is stored using an encrypted algorithm and is only decrypted during internal system verification. This prevents identity theft due to database leaks and further strengthens data security.

[0025] During daily operation, the device will trigger data collection trigger verification, remote operation forced verification, and offline reconnection secondary verification. The specific content of data collection trigger verification is: when the device collects key data (such as abnormal behavior videos monitored by AI and door opening records of access control), the system triggers double verification in real time to perform dual verification of the master key ID and code. Only data that passes the verification will be written into the event library. Data that fails will be marked as "suspicious data" and stored separately. After manual review, it will be decided whether to include it in the analysis to prevent false data from interfering with event judgment. The specific content of remote operation forced verification is: when remotely controlling the device (such as remotely restarting the monitoring device and locking the access control), the operator's authority and the dual identification of the device must be verified at the same time. The operation instruction must include an encrypted combination of the operator's key and the device's master key ID to ensure the legality and accuracy of the remote operation. The specific process of offline reconnection secondary verification is as follows: when the device is restored from the offline state, in addition to the regular double verification, the system will compare the local storage data during the offline period (such as events captured while offline) with the device identity. If the data does not match the device identity, it is considered that the data may have been tampered with. Only real-time data after reconnection is received, and offline data can be imported after manual review. The system performs a dynamic verification on online devices every 30 seconds. The calculation formula is: Dynamic verification result = primary key ID match Device code matching The heartbeat time is valid. A valid heartbeat time means that the interval between the device's most recent heartbeat time and the current time is no more than 30 seconds, ensuring that the device is in normal operation.

[0026] Verification records are linked to device health and tiered event responses. The system incorporates dual verification results into device health assessments. For example, if a device experiences a verification delay exceeding one second three times in a row, its health score will be reduced by 20%. Devices with health levels below the threshold will automatically be placed in the maintenance queue, enabling coordinated management from identity verification to device maintenance. When an AI device reports a high-priority event (such as "climbing a wall"), the system automatically retrieves the device's three most recent dual verification records. If all verifications pass, the highest-level response is triggered. If there are any verification anomalies, the system marks the event "pending device verification" along with the response, preventing misjudgments based on unreliable device data. If anomalies occur, such as three consecutive verification failures or a mismatch between the primary key ID and the code, the system immediately triggers a defense mechanism, temporarily blocking the device's data access rights to prevent the infiltration of false data. The abnormal device's location is highlighted on the management platform, and details (time and type of anomaly) are recorded. An alert is automatically pushed to management personnel, triggering the maintenance process.

[0027] Logs are generated for all operations, including device registration, status updates, and dynamic verification. These logs contain information such as verification time, primary key ID, device code, operator, and verification results. Logs utilize a chained storage structure, with each record linked to the hash value of the previous record to ensure tamper-proof access. This mechanism ensures traceability of verification records, providing a reliable basis for subsequent audits and troubleshooting. Abnormal events captured by devices (such as "stranger wandering" captured by AI surveillance) are forcibly linked to their primary key ID and code, ensuring that any event can be traced back to a specific device through dual identification. By embedding dual verification logic for the primary key ID and device code throughout the entire device registration, status update, and data collection process, the risk of unauthorized device access and data tampering is eliminated at the source. A threshold based on the number of consecutive verification failures enables automatic interception and alerting of abnormal devices, reducing manual intervention costs. Hash-linked log storage ensures tamper-proof verification records, providing irrefutable evidence for accountability.

[0028] When collecting basic personnel information (identity, role type), the system associates it with their primary key ID. When receiving events identified by AI devices or manually reported, the system automatically binds the primary key ID and code of the collection device to ensure event traceability. When collecting basic information on community security personnel (such as patrol officers and property managers), the system assigns each person a unique primary key ID, which is then bound to their identity (such as their work ID). This binding is stored encrypted to prevent tampering and provides an identity basis for subsequent permission management and event handling. Based on a person's role type (such as security, property management, or management), the system automatically assigns them appropriate operational permissions. For example, security personnel may have permissions to handle incidents and view equipment status, while property management personnel may focus on reporting facility repairs and registering visitors. Permission information is stored in association with the person's primary key ID, ensuring that only personnel with the appropriate permissions can perform corresponding operations during event handling.

[0029] When AI monitoring, smart access control, and other devices detect an unusual event (such as a stranger breaking in, illegally parked vehicles, etc.), the system automatically records the event type, time, and location, and forcibly associates the primary key ID and device code of the device that collected the event. This association allows the event to be directly traced back to its source, ensuring the authenticity and reliability of the event data and providing a basis for subsequent event verification. When manually reporting an event, the reporter must verify their identity. The system matches their identity identifier with the corresponding primary key ID. Once the legitimacy of the identity is confirmed, the reporter's primary key ID is associated and stored with the event information. This process ensures the traceability of manually reported events, clarifies reporting responsibilities, and reduces the occurrence of false reports.

[0030] The system regularly verifies the associated data of personnel, events, and equipment to ensure that the equipment associated with the event is included in the list of legal equipment and that the personnel handling the event have the appropriate permissions. If any data inconsistency is found (such as the non-existence of the equipment associated with the event, the processing personnel's permissions do not match, etc.), it is marked as an anomaly and the management personnel are prompted to review. Before the event data enters the processing flow, it is first verified to see whether the device collecting the event has passed the dual verification of the primary key ID and device code. Only events collected by devices that pass the verification will be processed according to the normal process, ensuring that the event data originates from a trusted device and improving the accuracy of event processing. During the event dispatch process, the system matches the most suitable processing personnel from personnel with corresponding permissions based on the event type and the permissions required for processing, combined with the personnel's responsibilities and permission range. This matching process, based on the personnel primary key ID and event association information, ensures that tasks are assigned to the correct personnel, improving event processing efficiency.

[0031] Collected device, personnel, and event data requires standardization and correlation. Device data that passes dual verification should be formatted uniformly (e.g., converting device status to a standardized "online / offline" code). Event data collected by the corresponding device (e.g., a "stranger intrusion" event captured by a surveillance device) should be correlated using a primary key ID. Personnel primary key IDs should be associated with their assigned areas and authority levels, forming a "device-event-personnel" data chain. Based on dual verification records, device data that fails verification (e.g., data uploaded by abnormal devices with tampered codes) should be eliminated to ensure a reliable data source for subsequent analysis.

[0032] Device status data that has passed dual verification is standardized and converted, with the original status descriptions of different devices (such as "online," "connected," and "operating normally") uniformly mapped to fixed codes: "1" represents online, "0" represents offline, "2" represents faulty, and "3" represents maintenance. This conversion is based on a preset coding comparison table and is linked to the dual verification results during the device registration phase. Only device data that has passed dual verification will enter the coding conversion process. This eliminates the differences in status descriptions of devices from different brands, solves the data statistics confusion caused by "different names for the same status" in traditional management, and improves the efficiency of cross-device status analysis. For example, a residential community has deployed both brand A and brand B monitoring equipment. The original status of brand A equipment is "online operation," and that of brand B is "connected successfully." After standardized conversion, both are marked as "1 (online)." Managers can quickly grasp the online rate of all devices in the community through the unified coding on the central control platform.

[0033] Unstructured data collected by AI devices (such as video clips and voice alarms) is structured: key frames are extracted from the video and annotated with precise timestamps (accurate to the millisecond). Voice alarms are converted into text descriptions and matched with pre-set keywords (such as "fight," "fall," and "unusual noise"). The processed data is automatically linked to the primary key ID and code of the collecting device, ensuring a traceable link with the device registration information. Converting unstructured data into searchable and analyzable structured information lays the foundation for rapid event location and correlation queries, reducing the time cost of manual unstructured data processing. For example, if AI monitoring captures a video of "someone arguing at a unit entrance," the system automatically extracts the key frame at the most intense moment of the argument, annotates it with the time "2024-06-10 16:20:33.521," and associates it with the device primary key ID "D015" and code "IMEI867543219012345" to form a structured event record.

[0034] Using the device master key ID as a link, device data and event data are deeply bound. This achieves a chain association of "event-device-verification record," ensuring that event data can be traced back to specific devices and that data authenticity can be verified through verification records, providing a reliable basis for event handling decisions. For example, when a monitoring device (master key ID "D015") collects a "quarrel at the unit door" event, after the "quarrel at the unit door" event is reported, the system automatically queries the device's dual verification records through the "D015" master key ID associated with the event, showing that all verifications have passed for the past 72 hours, confirming the credibility of the event data source and quickly dispatching security personnel in the area to handle the situation. Managers can directly access the device's real-time operating status and historical verification records through the master key ID.

[0035] With the personnel primary key ID as the core, the physical area and operational permissions assigned to them are associated, and this is linked to the area where the incident occurred. By precisely associating personnel primary key IDs with areas and permissions, traditional dispatching issues such as mismatched personnel and areas and insufficient permissions are resolved, improving incident response efficiency. For example, a security officer (primary key ID "P008") is responsible for the "Buildings 1-3 area," and their permissions include "Conflict Mediation." The "Argument at the Unit Door" incident occurred in Building 2, which falls within P008's area of ​​responsibility. Since the event type matches their "Conflict Mediation" permission, the system prioritizes dispatching to P008, carrying the device primary key ID "D015" associated with the event, allowing them to view real-time footage before departure.

[0036] The system automatically compares the primary key ID and code of device data. If the binding is inconsistent with that during registration (e.g., the code has been tampered with or the primary key ID does not exist), the data is immediately discarded and excluded from subsequent standardization and association processes. The filtering logic is linked to the dynamic verification results during daily device operation. Device data that fails verification three times in a row is marked as "high-risk anomaly" and triggers a device maintenance alert. This prevents illegal and tampered data from entering the system at the source, ensuring that subsequent analysis and decisions are based on authentic and trusted data sources and reducing the waste of security resources caused by false incidents. For example, an illegal device with a forged code "IMEI867543219012345" attempted to upload "false fire alarm" data. However, because the binding between the primary key ID and code did not match the registration record, the data was immediately discarded, and the system simultaneously sent an alert to management regarding an "attempted access by a device with an abnormal code."

[0037] The filtered data is integrity-verified to ensure that key fields such as event time, location, associated device ID, and associated personnel ID are present and logically consistent across fields (e.g., the event location must be within the associated personnel's area of ​​responsibility). Data that passes this verification is ultimately incorporated into the "device-event-personnel" linkage data chain, forming a closed-loop management system. This ensures the integrity and logic of the linkage data chain, allowing managers to trace information throughout the entire process through any link (e.g., a specific device, a specific person, or a specific event), improving management transparency and tracing efficiency. For example, the "Argument at the Unit Entrance" incident was verified to have its location, Building 2, within the area of ​​responsibility of handler P008. The status of the associated device, D015, was "1 (Online)." The fields were complete and logically consistent, successfully entering the data chain and supporting subsequent event review and accountability tracing.

[0038] Event level determination and regional risk assessment are performed. A dynamic weighted adaptive grading model combines event data (type, location, and frequency) confirmed by dual verification, along with time period, season, and historical data, to categorize events into high, medium, and low levels. For example, an incident reported by a verified access control device as "unauthorized forced door opening" is recorded as a high-level event. A Bayesian network model is introduced to calculate the probability of combined associations among multiple events (for example, the simultaneous occurrence of an access control anomaly and a smoke alarm increases the risk factor by 300%), enhancing comprehensive risk identification. Based on data collected by dual-verified devices, the frequency of incidents in each region is calculated (for example, three incidents of "illegal electric vehicle charging" in a certain area within a week). By integrating multi-dimensional signals such as video semantic analysis, environmental sensor data, and device operating status, a hybrid LSTM and Transformer model is used to predict event trends (for example, predicting the risk of an electric vehicle shed fire 30 minutes in advance). This generates a real-time, updated regional risk heat map, providing data support for resource allocation. This model integrates with the device dual verification mechanism and data correlation system to ensure that the analysis results are based on reliable data and form a closed-loop logic with other links.

[0039] Event severity is determined based on data collected by dual-verified devices, using a hierarchical logic that combines basic attributes with dynamic adjustments. Basic attributes encompass the inherent risk of the event type (e.g., "Unauthorized person forcibly opening the door" directly threatens resident safety, resulting in the highest basic severity). Dynamic adjustments incorporate variables such as time of day, season, and historical data to form a three-dimensional severity calculation model. For example, the nighttime period from 10:00 PM to 6:00 AM is a sensitive time for residential security control, and the risk weight of events occurring during this time is automatically increased by 1.5 times. Similarly, the potential risk of a "stranger loitering" incident occurring at night, due to poor lighting and reduced resident activity, is significantly higher than during the day, requiring weight adjustment to strengthen the response priority. Seasonal factors also have a significant impact. In the summer, high temperatures and increased demand for electric vehicle charging increase the risk of battery spontaneous combustion. Therefore, the risk factor for "illegal electric vehicle charging" events during this season is increased to 2.0, precisely aligning with the high incidence of fires in the summer. The correlation of historical data is also essential. For example, if a region experiences three consecutive "unit door not closed" incidents within the past 30 days, it indicates a management loophole. The risk weight for subsequent similar incidents will increase by 0.3, prompting an upgrade in the incident level to ensure adequate attention. These dynamic factors work together to ensure that incident levels reflect inherent risk while adapting to real-time changes in the situation.

[0040] The specific weights are determined through a dynamic weight adaptive grading model. The specific content is: at the basic attribute level, according to the potential impact of the incident on personal safety, property safety, and public order, three basic levels are preset, namely high basic level (level 3, including "unauthorized personnel forcibly opening the door", "throwing objects from high places", "suspected fires" and other incidents that directly threaten safety), medium basic level (level 2, covering "illegal charging of electric vehicles", "unit doors not closed for a long time" and other incidents that affect public order) and low basic level (level 1, involving minor violations such as "unsorted garbage disposal" and "pets not on a leash"). Then, based on the basic level, the final level judgment is made through dynamic weight adjustment. Dynamic weights mainly include time period weights. , seasonal weight , historical association weight and area-sensitive weights Four major weighting factors. The time-of-day weighting is differentiated based on community activity patterns. Nighttime weighting is sensitive, with a weight of 1.5; weekday mornings (7:00-9:00 AM) and evenings (5:00-7:00 PM) are peak traffic times, with a weight of 1.2; and all other time periods are considered normal, with a weight of 1.0. For example, the risk level of a "loitering stranger" incident increases significantly at night due to poor lighting conditions and increased emergency response difficulty. Seasonal weighting is dynamically adjusted based on seasonal safety risk characteristics. In summer, high temperatures increase the risk of electric vehicle battery spontaneous combustion, resulting in a weight of 2.0 for related events. In winter, dry weather and high heating demand lead to a weight of 1.5 for fire hazard events. In spring and autumn, risks are relatively stable, with a weight of 1.0. For example, in the summer, the risk level of "illegal electric vehicle charging" events is amplified by seasonal weighting, triggering intervention measures first. Historical correlation weighting is a progressive weighting based on the frequency of similar events in the same area over the past 30 days. The weight increases by 0.1 with each occurrence, reaching a maximum of 1.5. If a unit door is left open three times consecutively, the weight increases from 1.0 to 1.3, escalating the incident's severity to attract management attention and prevent repeated occurrences from creating security vulnerabilities. Regional sensitivity weights are primarily targeted at key areas such as elderly activity areas, children's playgrounds, and power distribution rooms, with a fixed weight of 1.3. Ordinary residential areas receive a weight of 1.0. For example, if a "debris accumulation" incident occurs near a power distribution room, the risk level is increased due to the potential for circuit failure.

[0041] The final level of the event is calculated using the following formula: ,in, is the final event level, is the basic level (high basic level is 3, medium basic level is 2, low basic level is 1). Make the final grade judgment, if , it is determined to be a high-level event and an emergency response must be initiated within 5 minutes, and the information will be pushed to the security person in charge's terminal and the sound and light alarm will be triggered; if , it is determined to be a medium-level incident, and personnel must be arranged to deal with it on site within 15 minutes and the results of the treatment must be recorded; if , it is determined to be a low-level incident and a rectification reminder must be pushed through the APP within 2 hours to track the feedback.

[0042] For example, at 3:00 a.m. in winter, a "stranger climbing over the wall" incident occurred near the power distribution room of a residential complex. Similar incidents had occurred twice in the area in the past 30 days. (High base level), (night time), (at night), (2 historical events), (Distribution room components), finally , identified as a high-level incident. The system immediately linked the area's surveillance equipment (confirmed by double verification) to retrieve real-time footage and simultaneously sent an emergency command, including location information, to the security personnel on duty, enabling a rapid response.

[0043] The model automatically verifies the data collection device's dual verification records before calculation. If the device's verification pass rate falls below 90% over the past 24 hours, the final level is temporarily downgraded by one level until manual verification of the data's authenticity is completed. For example, if a camera's verification pass rate is 85% due to a temporary malfunction, the reported "suspected fire" incident will be initially handled as a medium-level incident. Once the fire is confirmed, the high-level response will be restored. This prevents misjudgments and ensures that risks are not underestimated. Event level data is synchronized in real time to the regional risk assessment module, serving as the core basis for calculating regional risk values. Each high-level incident increases the corresponding area's weekly risk value by 5 points; a medium-level incident by 3 points; and a low-level incident by 1 point. This provides quantitative support for the subsequent generation of risk heat maps and the allocation of security resources, forming a closed-loop management system of "incident classification-regional assessment-resource allocation." This dynamic weighted adaptive model ensures that event classification maintains consistency with objective standards while flexibly adapting to variables such as time of day, season, and region, making classification decisions more aligned with actual security needs and providing a precise basis for community security response decisions.

[0044] A Bayesian network model is introduced to analyze the probability of multiple events being linked, addressing the limitations of single-event risk assessment. When an abnormal access control system opening and a smoke alarm occur simultaneously, they are not isolated events. The access control anomaly may indicate a break-in, while the smoke alarm indicates a fire. This combination raises the risk of arson by 300% compared to a single event. Similarly, when an unfamiliar vehicle is stranded and people frequently board and alight, the risk factor increases by 150%. This combined analysis is continuously optimized through historical data training. The system automatically records the actual consequences of various event combinations (such as the probability of a combination ultimately leading to an accident) and continuously adjusts the associated weights. For example, the risk factor for the combination of "wall climbing" and "crowd gathering" was initially set at 2.0. After verifying 100 cases and finding that 30% of them ultimately escalated into conflicts, the system automatically increased the factor to 2.5, making the assessment more consistent with actual risk.

[0045] The core formula of the Bayesian network model is: ,in, for Basic probability of occurrence (based on historical data statistics, such as the average weekly probability of "abnormal opening of access control"); For events When the event occurs Conditional summary (e.g., the probability of "smoke alarm" after "access control abnormal opening"); is the risk amplification factor (set according to the actual risk consequences of the event combination, with the initial value determined through historical case training and subsequently dynamically optimized); The probability of the associated risk of the combination of two events is used as the basis for calculating the combined risk coefficient. Determining the combined event correction factor , combined event correction factor The range is 1.0-3.0. The higher the value, the stronger the magnification effect of the combined risk on the final level of the event. Combined with the combined event correction coefficient Correct the final event level formula to obtain the corrected final event level calculation formula .

[0046] For example, regarding the combination of abnormal opening and closing of access control and smoke alarm, the average weekly probability of access control abnormality is 2%. , the conditional probability of smoke alarm after access control abnormality is 30%, , because the combination has a high risk of "arson", the risk amplification factor , associated risk probability , the corresponding risk factor increased by 300%, In actual applications, when the system detects two events occurring simultaneously, it automatically retrieves the historical disposal records of the combination (for example, two confirmed arson attempts in the combination in the past year) to further strengthen risk assessment and ensure that the response level is adapted to the actual threat.

[0047] Regional risk assessment transcends the limitations of a single data source by integrating multi-dimensional signals such as video semantic analysis, environmental sensor data, and equipment operating status to build a comprehensive risk perception network. Video semantic analysis uses AI cameras to identify image features such as "debris accumulation" and "electric vehicles blocking fire escapes," and simultaneously correlates temperature, humidity, and gas concentration data collected by environmental sensors. If the dryness index in the area where debris is accumulated exceeds 0.8 (a higher dryness index indicates greater flammability), the fire risk assessment is immediately upgraded. Equipment operating status is also crucial. When the water pressure of a fire hydrant falls below 0.2 MPa, firefighting capacity decreases by 40%, and the fire risk weight for that area needs to be increased accordingly. If a smoke alarm is faulty and unable to detect fire signals in a timely manner, this creates a blind spot in the risk assessment, automatically increasing the severity of all fire-related events in that area by one level. These multi-dimensional signals are linked using the regional primary key ID in the data association system to ensure accurate correspondence between information and physical areas, preventing cross-regional data confusion.

[0048] A hybrid LSTM and Transformer model is used to predict event trends, enabling a shift from passive response to proactive prevention and control. For example, in an electric shed, the system collects real-time data such as charging pile load, battery temperature, and ambient humidity. The LSTM model captures temporal patterns in temperature changes (e.g., a temperature rise from 30°C to 45°C within 10 minutes and a continuous increase). The Transformer model, incorporating the characteristics of "sudden temperature rise combined with poor ventilation" from historical fire cases, predicts fire risks 30 minutes in advance. When the predicted risk reaches a threshold, the system immediately triggers intervention measures: an alert is automatically sent to the security terminal, the electric shed's exhaust system is activated to cool the area, and a public announcement reminds residents to remove their vehicles. This predictive mechanism not only reduces accident losses but also optimizes resource allocation. Prevention and control forces can be deployed before an incident occurs, reducing the incidence of high-risk events.

[0049] The AI ​​camera extracts target behavior features (such as "dense parking of electric vehicles" and "accumulation of debris blocking fire-fighting facilities") and associates them with the primary key ID of the area where the incident occurred to ensure accurate correspondence with the physical space; environmental sensors are used to collect environmental data, including temperature and humidity (such as the electric shed temperature > 35°C triggers the warning threshold), gas concentration (smoke concentration > 0.1mg / m³ increases the risk weight), and dryness index (> 0.7 increases the fire hazard coefficient); equipment operation status data is collected, including fire hydrant water pressure (< 0.2MPa reduces the fire extinguishing capacity coefficient to 0.7), charging pile load rate (> 80% increases the risk of overheating), and smoke alarm sensitivity (risk weight × 1.2 in the fault state). Multi-source fusion data is obtained through weighted summation. ,in, This is video semantic analysis data (normalized to a range of 0-1, such as the feature value of "densely parked electric vehicles" is 0.8). The data from the environmental sensor (normalized to the range of 0-1, e.g., a temperature of 38°C corresponds to a value of 0.75) The device operating status data (normalized range 0-1, such as a charging pile load rate of 92% corresponds to a value of 0.92), 、 、 is the weight coefficient (set as 0.4, 0.3, and 0.3 respectively, according to the risk impact). The time decay factor highlights the impact of recent cases, thereby determining the historical data ,in The first time in the past 90 days The final level of the same event, For the The number of days since the event, The more recent the event, the higher the weight. For example, the weight of an event 3 days ago is about 0.74 after decay, and about 0.05 for an event 30 days ago.) Construct a trend prediction formula , For multi-source fusion data, The final level of the event, For historical data, is the risk prediction value, It is a hybrid model of LSTM and Transformer. LSTM captures the changing trends of time series data such as temperature and concentration, while Transformer focuses on the correlation between video semantic features and event levels, outputting a risk prediction value 30 minutes later (0-10 points, 8 points or above is a high-risk warning, 8 points or below is no warning).

[0050] Based on quantified risk values, regional risk heat maps are updated in real time using red, yellow, and green color schemes, providing a visual representation of the security situation in each area. The risk value calculation incorporates factors such as event frequency, severity, and environmental factors. For example, if a region experiences three high-severity events within a week, and the environmental factor is set at 1.5 due to the dry summer weather, the final risk value reaches 12, marking the heat map red (high risk), indicating the need for strengthened prevention and control measures. The heat map's integration with other processes ensures practicality: Red areas are automatically linked to the area's equipment distribution list, prioritizing the deployment of mobile surveillance equipment for additional coverage. Based on the binding between personnel and areas, patrol frequency adjustments (from three times daily to once every two hours) are sent to security personnel responsible for that area. Furthermore, the heat map strictly links to equipment double-verification records. If the equipment verification pass rate in a region falls below 90%, the corresponding area is marked "Data pending verification" to prevent erroneous prevention and control strategies based on unreliable data.

[0051] The regional risk heat map uses the risk value verified by multiple dimensions as the core quantitative indicator. It combines the dynamic weight grading results with the multi-event correlation analysis data to calculate the comprehensive risk value of each region using the following formula: ,in, The weekly frequency of the event (e.g. if three similar events occur in a region within a week, ), 、 、 is the weight coefficient (take 0.5, 0.3, 0.2 respectively, and adjust according to real-time performance and relevance). According to the calculated comprehensive risk value, three levels of visual identification are divided in the heat map. If , then it is red (high risk, patrol once an hour, and 2 additional security personnel); if , then it is yellow (medium risk, patrol 3 times a day, strengthen monitoring during key periods); if , then it is green (low risk, maintain regular patrol frequency). For example, in a certain community, the "electric vehicle shed in Building 3" area had three "illegal charging of electric vehicles" within a week. , combined event association probability , the model predicts risk value , then the comprehensive risk value of the area , the heat map is marked in yellow (medium risk).

[0052] The data source for event classification and risk assessment strictly relies on the results of dual verification of equipment. All data involved in the analysis must be double-verified using the primary key ID and the equipment code. If a camera fails verification due to code tampering, the "people gathering" event data collected by it will be marked as "untrustworthy" and not included in the level calculation. This linkage ensures the reliability of the analysis results from the source and prevents false data from interfering with decision-making. The equipment verification pass rate also affects the risk weight. The verification pass rate of equipment in a certain area in the past 24 hours was only 85%, indicating that there may be deviations in the data in this area. The level of the relevant event will be temporarily downgraded by one level and will be adjusted again after manual review of the equipment status, forming a positive cycle between data reliability and level accuracy.

[0053] The heat map's data source strictly relies on the results of dual-verification of equipment. Reliability is ensured through a "verification pass rate threshold filtering" mechanism. Statistics are compiled for the dual-verification pass rates of all data collection devices in the area over the past 24 hours. If a device's verification pass rate is less than 90%, the weight of the event data it collects in the risk value calculation is reduced to 0.5 (normal equipment has a weight of 1.0). If more than 30% of the equipment in a region fails verification, the risk value for that region is marked "Data Pending Verification," triggering an equipment maintenance instruction. The risk value is recalculated after verification is restored. For example, if an AI camera in a certain area has a dual-verification pass rate of only 75% due to encoding anomalies, the weight of the "people gathering" event data reported by the camera is halved in the risk calculation, avoiding risk misjudgments caused by equipment failures. At the same time, the system automatically sends a maintenance work order for the device to the property management, forming a rapid response chain of "data anomaly - equipment maintenance - data correction."

[0054] The heat map is deeply bound to the data association system through the "region primary key ID", realizing the real-time traceability of multi-dimensional information. Clicking a region (such as "R005") on the heat map can automatically associate the equipment list of the region (including the double verification records of each device), the event details of the past 7 days (including the event details of each event), and the Level and handling results), the primary key ID and authority scope of the responsible security personnel. Areas with higher risk values ​​automatically trigger the "equipment-event-personnel" association chain verification. For example, the red area needs to be verified: whether the equipment collecting the event has passed the double verification, whether the event level judgment matches the personnel handling authority, and whether there are repeated unresolved risks in the historical handling records. For example, the "R007" area in the heat map is displayed as red high risk. The system automatically retrieves the related information through the area primary key ID: all three devices in the area have passed the double verification (pass rate 100%), two high-level events have occurred in the past three days, and the authority of the responsible personnel includes "high-risk event handling", but the historical handling completion rate is only 60%. Based on this, the recommendation of "the need to strengthen personnel training in this area" is pushed to the management personnel.

[0055] Real-time updates to the heat map directly drive intelligent resource allocation, achieving precise matching of risk and resources. In red, high-risk areas, two patrol personnel (priority assigned to the personnel responsible for that area) and one mobile monitoring device are automatically dispatched to provide additional coverage. Status data from fire hydrants, fire extinguishers, and other equipment are also linked to ensure the availability of emergency supplies. In yellow, medium-risk areas, fixed patrol routes are optimized, with twice-daily focused inspections added. High-incidence incident types (such as illegal electric vehicle charging) in the area are pushed to the responsible personnel's terminals to strengthen targeted prevention and control. For green, low-risk areas, standard resource allocation is maintained, with redundant resources redeployed to high-risk areas to improve overall prevention and control efficiency. Through the quantitative display of the heat map and the integration of multiple mechanisms, the accuracy of resource allocation is improved, the recurrence rate of incidents in high-risk areas is reduced, and the cost of ineffective prevention and control due to equipment anomalies is reduced, forming a complete closed loop of "data collection-risk assessment-resource allocation-effectiveness feedback."

[0056] Accurately match personnel with events, assigning tasks and conducting inspection and traceability. The system locates the incident area based on the device master key ID associated with the event. Combined with the personnel master key IDs and their responsibilities and permissions associated with that area, it screens out qualified personnel (e.g., the security personnel master key ID's corresponding responsibilities include "emergency response"). Tasks are pushed with the device master key ID and code associated with the event. Upon arrival at the scene, personnel can verify the consistency of the device code with the task information through the terminal to ensure the accuracy of the task being handled. Upon completion of the task, the system associates the handling record with the device double-verification information for archiving, forming a complete traceability chain.

[0057] The system relies on the "device-event-area-personnel" association data chain built in the early stage to achieve accurate matching of processing personnel. When a double-checked device reports an event, the device primary key ID associated with the event (such as "D025") is used to retrieve the area identifier bound to the device (such as "R007-Building 3 East Unit") in the data association system to clarify the scope of the incident location; based on the area identifier, the list of responsible personnel for the area is associated, and combined with the responsibilities and permissions corresponding to the personnel primary key ID (such as the responsibility field of "P012" includes "emergency response" and "access control abnormality handling"), personnel whose permissions match the event type are screened out; referring to the current status of the personnel (such as "on duty / on patrol"), historical handling efficiency (average time spent handling high-level incidents in the past 30 days), and distance from the incident area (such as "P012" is only 200 meters away from "R007"), a priority list of processing personnel is generated to ensure the optimal response speed. For example, the double-checked access control device "D025" reports "unauthorized personnel forcibly opening the door" (a high-level event, ), the system locates the "R007" area through "D025" and screens out the three security personnel associated with this area. Among them, "P012" has a duty that includes "high-level incident response" and is currently on patrol, and is identified as the primary responsible person.

[0058] The task push link is linked to the equipment double-check mechanism to ensure the accuracy of the disposal object. An information package containing "event type (high level - forced door opening), incident area (R007), equipment master key ID (D025), equipment code (IMEI867543219012345), and real-time video stream link" is pushed to the screened personnel terminal. The equipment code is bound to the previous double-check record as the basis for on-site verification. After the processing personnel arrive at the scene, they scan the physical code of the equipment (such as the QR code on the access control machine) through the terminal. The system automatically compares the scan result with the code in the task information to see if they are consistent. If they are consistent, it confirms that the disposal object is correct; if they are inconsistent, an early warning is immediately triggered (there may be equipment displacement or wrong task assignment), and the latest double-check record of the equipment is simultaneously retrieved (such as a 100% verification pass rate in the past 24 hours, eliminating equipment abnormalities). For example, after "P012" arrives at the site, it scans the code of the access control device "D025". The terminal display is consistent with "IMEI867543219012345" in the task information, and the system simultaneously prompts that the dual verification status of the device is "passed", confirming that it can be safely disposed of.

[0059] After the task is completed, the system will include the disposal record in the associated data chain to form a closed-loop management. Personnel submit the disposal results (such as "unauthorized personnel have been expelled and access control has returned to normal") through the terminal, and need to upload on-site photos (including the device master key ID), disposal time, treatment measures and other information. The system automatically binds the disposal record with the following information for archiving: the double verification record of the device associated with the event (such as the verification status of "D025" when the event occurred was "passed"), the final level of the event ( ) and the change in regional risk values ​​in the risk heat map (for example, the risk value of "R007" dropped from 8.5 to 3.2), the primary key ID of the person handling the incident (P012), and the time taken for the incident (12 minutes, better than the regional average of 15 minutes). Managers can trace the entire process from any node. For example, they can use the device code of "D025" to query the handling records of the last three incidents, or use the primary key ID of "P012" to view the incident resolution rate (92%) for their area of ​​responsibility. For example, after "P012" completes the handling of the "unauthorized personnel forcibly opening the door" incident, the system links this record with the double verification information of "D025" and the change in the risk value of the "R007" area. When verifying the safety of the area three months later, the traceability chain reveals a 60% decrease in the incidence of similar incidents, verifying the effectiveness of the handling measures.

[0060] High-level events ( ) automatically triggers the "two-person handling" mechanism, that is, pushing tasks to the top two priority personnel at the same time to ensure response redundancy; medium and low-level events are pushed according to the single-person responsibility system to optimize manpower allocation. If the double-check pass rate of the event-related equipment is less than 90%, the task information will be specially marked "Please verify the equipment status first" to prompt personnel to check on-site whether there are any abnormalities in the equipment (such as data misreporting caused by code tampering). After the task is completed, the system will update the risk value of the incident area in real time (such as The data link between personnel and events is used to accurately match personnel and events, reducing task response time. A full-link traceability mechanism ensures that the handling and accountability of each incident are traceable. Combined with dual-checking equipment information, this effectively avoids issues such as "mishandling the wrong person" and "record tampering," improving the closed-loop management of incidents.

[0061] A dual verification mechanism for primary key IDs and device codes runs throughout the entire process. During the data collection phase, illegal devices and tampered data are filtered out, providing a reliable foundation for subsequent analysis. During the data integration phase, dual identification association enables data traceability. During task assignment, consistency between the target and the record is ensured. Ultimately, this creates a closed loop of secure collection, reliable analysis, and precise action, enhancing the safety and controllability of community security management.

[0062] The embodiments disclosed in the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. The embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part, and / or installed from a removable medium. When the computer program is executed by the central processing unit (CPU), the above-mentioned functions defined in the method of the present application are executed. It should be noted that the computer-readable medium mentioned above in the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium can be, for example, but not limited to, a system, device or device of an electrical, magnetic, optical, electromagnetic, infrared segment or semiconductor, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection having one or more wire segments, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this application, a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal may take a variety of forms, including, but not limited to, an electromagnetic signal, an optical signal, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transfer a program for use by or in conjunction with an instruction execution system, apparatus, or device. Program code embodied on a computer-readable medium may be transmitted using any appropriate medium, including but not limited to wireless, electrical wire, optical fiber cable, RF, etc., or any suitable combination thereof.

[0063] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of the systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or portion of code that contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the boxes can also occur in an order different from that marked in the accompanying drawings. For example, two boxes shown in succession can actually be executed substantially in parallel, or they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, as well as the combination of boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified functions or operations, or can be implemented using a combination of dedicated hardware and computer instructions.

[0064] Those skilled in the art should understand that the embodiments of the present invention described above and shown in the accompanying drawings are only examples and do not limit the present invention. The objectives of the present invention have been fully and effectively achieved. The functional and structural principles of the present invention have been demonstrated and explained in the embodiments. Without departing from the principles, the implementation methods of the present invention may be subject to any deformation or modification.

Claims

1. A community intelligent security management method, characterized in that: The method comprises: Collect equipment data, verify daily equipment operation, collect personnel and event data, and perform data correlation and integration; When an abnormal event is identified, the basic level is determined based on the integrated data, and the associated risk probability is determined through the Bayesian network model. The basic level and associated risk probability are input into the dynamic weight adaptive grading model, and the final level of the event is determined by combining the time period weight, seasonal weight, historical association weight, and regional sensitivity weight. The final level of the event is input into the trend prediction model to obtain the risk prediction value, and the comprehensive risk value of each area is determined based on the risk prediction value. A real-time risk heat map is generated based on the comprehensive risk value; According to the final level of the event and the associated data, qualified processing personnel are screened for task assignment. The processing personnel handle the event and record the processing process data.

2. A community intelligent security management method according to claim 1, characterized in that: The specific process of device data collection is as follows: Device data is collected from security-related devices within the residential complex. When the device is started, the system automatically reads the device's unique code and associates it with the preset primary key ID. The device's legitimacy is confirmed through dual verification of the primary key ID and the device code. During device registration, the system performs verification of the coding format compliance, verification of the community authority association, and encrypted storage of the registration information. During daily operation, the device triggers data collection trigger verification, remote operation mandatory verification, and offline reconnection secondary verification. Link verification records with device health and event classification responses.

3. A community intelligent security management method according to claim 2, characterized in that: The specific process of collecting personnel and event data is as follows: When collecting basic personnel information, a unique primary key ID is assigned and encrypted and bound to the personnel identity. The corresponding operation permissions are matched according to the type of responsibilities, and the permission information is associated with the personnel primary key ID and stored. When the device recognizes an abnormal event, the system automatically records the type, time and location of the event, and forcibly associates the device primary key ID and device code that collected the event. When manually reporting an event, the reporter needs to verify his or her identity. The system matches the corresponding primary key ID through his or her identity. After confirming that the identity is legal, the primary key ID of the reporter is associated with the event information and stored.

4. A community intelligent security management method according to claim 3, characterized in that: The specific process of data integration is as follows: The format of device data that has passed double verification is unified, and the event data collected by the corresponding device is associated through the primary key ID, and the responsible area and authority scope are associated through the personnel primary key ID; double verification of the primary key ID and equipment code is performed, and the device data that has not passed the verification is eliminated. The integrity of the data that has passed the verification is checked to ensure that the event time, location, associated device ID, and associated personnel ID are not missing, and the logic between fields is self-consistent; the device status data that has passed the verification is standardized and converted, and the original status descriptions of different devices are uniformly mapped to fixed codes; the unstructured data collected by AI devices is structured.

5. A community intelligent security management method according to claim 4, characterized in that: The specific process of obtaining the final level of the event is as follows: Based on the potential impact of an event on personal safety, property safety, and public order, three basic levels are preset; the probability of multiple event associations is analyzed through a Bayesian network model; the basic level, associated risk probability, time period weight, seasonal weight, historical association weight, and regional sensitivity weight are input into the final event level calculation formula to solve the final event level, and the final event level is compared with the set threshold to determine the event type.

6. A community intelligent security management method according to claim 5, characterized in that: The handling method is determined based on the final level of the incident: If the final level of the event is greater than or equal to 4, it is considered a high-level event and an emergency response must be initiated within 5 minutes. The information will be pushed to the terminal of the person handling the event and an audible and visual alarm will be triggered. If the final level of the incident is greater than or equal to 2.5 and less than 4, it is judged as a medium-level incident, and personnel must be arranged to handle the incident on site within 15 minutes, and the handling results must be recorded; if the final level of the incident is less than 2.5, it is judged as a low-level incident, and rectification reminders must be pushed through the APP within 2 hours, and feedback must be tracked.

7. A community intelligent security management method according to claim 6, characterized in that: The specific content of the real-time risk heat map is as follows: A hybrid model based on LSTM and Transformer is constructed. The AI ​​camera extracts target behavior features, associates them with the primary key ID of the area where the incident occurred, and uses environmental sensors to collect environmental data. Multi-source fused data is obtained through weighted summation. A time decay factor is used to highlight the impact of recent cases. Historical data is determined and then input into the hybrid model, along with the multi-source fused data, historical data, and the final incident level, to obtain a risk prediction value. The comprehensive risk value is obtained by weighted summation based on the weekly frequency of events, the final level of events, the associated risk probability and the risk prediction value. Three levels of visual identification are divided in the heat map according to the comprehensive risk value.

8. A community intelligent security management method according to claim 7, characterized in that: The specific process of the processing personnel is as follows: The system locates the incident area based on the device primary key ID associated with the event, and screens out qualified personnel based on the primary key ID and responsibilities and permissions of the personnel bound to the area. When pushing the task, the device primary key ID and code associated with the event are attached. After the processing personnel arrive at the site, they verify the consistency of the equipment code and the task information through the terminal; after the task is completed, the system will associate the disposal record with the equipment double verification information and archive it.

9. A residential intelligent security management system, characterized in that: The system is used to execute a community intelligent security management method as described in any one of claims 1-8.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement the community intelligent security management method described in any one of claims 1 to 8.

Images