Intelligent management system and data processing method

Abstract

The application relates to the technical field of intelligent community management, in particular to an intelligent management system and a data processing method, a data acquisition layer, which comprises various data acquisition devices arranged in a community and used for acquiring real-time data in the community; an edge computing node, which is in communication connection with the data acquisition layer and used for locally preprocessing the collected data; a data center, which is in communication connection with the edge computing node, used for receiving the preprocessed data, and performing desensitization processing and weighted partition storage on the data; a digital twin platform, which is in communication connection with the data center, used for constructing a digital twin three-dimensional model of the community based on the data provided by the data center; and an application display layer, which is in communication connection with the digital twin platform, used for providing a visual interface and an interaction function, so that the complex comprehensive data of the community is properly managed through the ways of unifying the formats of the same type of data, desensitizing sensitive information and performing partition storage.

Description

Technical Field

[0001] This invention relates to the field of intelligent community management technology, and in particular to an intelligent management system and data processing method. Background Technology

[0002] With the continuous advancement of urbanization, the scale of residential communities is expanding, and the complexity of management is increasing accordingly. In traditional community management systems, the various business systems are isolated from each other, with each unit only responsible for data processing within its own unit. For example, monitoring systems, fire protection systems, access control systems, and vehicle management systems are usually provided by different vendors and operate independently, making data sharing impossible. Managers need to log in to multiple platforms and switch between them, resulting in low management efficiency. The overall status of equipment lacks visualization; for example, the distribution and operational status of fire extinguishers, fire hydrants, and surveillance cameras are difficult to grasp intuitively. Equipment inspections rely on manual records, which are prone to omissions and cannot be reported in a timely manner. Current equipment suffers from problems such as malfunction or expiration, and response to abnormal events is delayed. For example, it is difficult to trace and collect evidence of violations such as objects being thrown from high-rise buildings, littering, and electric vehicles being charged on upper floors. Traditional monitoring can only record and play back videos after the fact, and cannot proactively detect and warn. Fire-fighting equipment such as smoke detectors are mostly local sound and light alarms, which cannot promptly notify the relevant responsible persons of the alarm. Data processing poses security risks. Community management involves a large amount of residents' personal information, such as facial data, permanent resident information, and vehicle information. Traditional data storage methods lack effective privacy protection mechanisms and pose a risk of data leakage. Therefore, a comprehensive and secure community management and control system is needed. Summary of the Invention

[0003] The purpose of this invention is to provide an intelligent management system and data processing method to solve the problems existing in the prior art.

[0004] The above-mentioned technical objective of the present invention is achieved through the following technical solution: An intelligent management system, comprising: The data acquisition layer includes various data acquisition devices deployed within the community to acquire real-time data within the community; Edge computing nodes, which are communicatively connected to the data acquisition layer, are used to perform localized preprocessing on the acquired data. The preprocessing includes data cleaning, format standardization, and preliminary analysis. The data platform is communicatively connected to the edge computing nodes and is used to receive preprocessed data, perform data anonymization processing, and store the data in weighted partitions. The digital twin platform is connected to the data platform and is used to construct a digital twin three-dimensional model of the community based on the data provided by the data platform, and to map and visualize the status information and real-time data of various devices in the three-dimensional model. The application presentation layer communicates with the digital twin platform and provides a visual interface and interactive functions.

[0005] By adopting the above technical solutions, the data format is unified through edge computing nodes, enabling the data platform to obtain data with consistent format. This avoids the data platform missing or being unable to read data due to inconsistent formats. The accurate and comprehensive capture of all data by the data platform is a prerequisite for providing real-time data analysis and monitoring. By integrating multiple monitoring strategies within the digital twin platform, it is possible to achieve the purpose of comprehensive monitoring and analysis within the community.

[0006] In a further embodiment, the data acquisition device includes at least: The video surveillance equipment consists of an environmental monitoring camera, a face recognition camera, a high-altitude object throwing monitoring camera, and a temperature measurement monitoring device. The face recognition camera is installed at the entrance and exit of the community to capture the facial data of people entering and exiting. The high-altitude object throwing monitoring camera is installed on the exterior of the building to monitor the behavior of throwing objects from high altitudes. The fire monitoring equipment consists of a smoke detector, a temperature sensor, and a water pressure sensor. The smoke detector is installed in areas such as non-motorized vehicle sheds. Access control equipment consists of door magnets and access controllers; The vehicle management equipment consists of a barrier gate and parking space status cameras. The environmental monitoring equipment consists of smart streetlights and waste sorting monitoring equipment. The charging pile equipment consists of voltage and current detectors and leakage protection sensors.

[0007] By adopting the above technical solutions, these devices are the basic components for data acquisition. The location and quantity of these devices need to be flexibly set according to the layout plan within the community in order to obtain more comprehensive and complete data.

[0008] In a further embodiment, the data platform includes: The data desensitization module is used to desensitize sensitive information in data; The weighted partition storage module is used to assign different weight values ​​to the data based on its sensitivity and importance, and then partition and store the data in different storage areas according to the weight values.

[0009] By adopting the above technical solution and desensitizing the data in the data platform, all classified data can be processed discreetly, eliminating the possibility of any non-highest-authority personnel obtaining classified information.

[0010] In a further embodiment, the application presentation layer includes: The monitoring and management module is used to display the distribution and operating status of monitoring devices in a 3D model, and supports clicking to view real-time images and video playback; The fire management module is used to display the distribution and status of fire protection equipment in a 3D model and automatically generate early warnings for equipment that has not been inspected on time. The personnel management module is used to display information such as the number of permanent residents in the community, real-time pedestrian flow, and personnel. The vehicle management module is used to display vehicle entry and exit records, parking space occupancy status, and charging pile usage status; The alarm management module is used to collect various abnormal alarm events and display and assign tasks in a unified manner.

[0011] The alarm push module, connected to the application presentation layer, is used to push alarm information to the terminal devices of designated administrators when an abnormal event occurs.

[0012] By adopting the above technical solution and displaying it in modules, the monitoring strategy within the monitored area can be more flexible, making it easier to highlight key areas that need to be monitored and to trigger early warnings in the event of emergencies.

[0013] This invention also discloses a data processing method for an intelligent management system, the data processing method comprising the following steps: Step S1: Multi-source data acquisition. Multi-source data is acquired by deploying various data acquisition devices in the community. The multi-source data includes video surveillance data, fire monitoring data, access control data, vehicle data, environmental monitoring data, and pedestrian flow data. Step S2: Data preprocessing. The collected multi-source data is preprocessed locally through edge computing nodes. The preprocessing includes data cleaning and format standardization. Step S3: Sensitive data identification and desensitization. Identify sensitive data in the multi-source data and desensitize the sensitive data. The sensitive data includes face image data, resident identity information data and vehicle information data. The desensitization process adopts at least one of the following methods: hash mapping replacement of sensitive fields, random distribution processing of numerical data near the mean, and generalization processing based on the K-anonymity algorithm. Step S4, Weight Calculation and Partitioned Storage: For the data after anonymization, calculate the weight value of each data item according to the data category, sensitivity, and usage frequency; and allocate the data to different storage partitions according to the weight values. Step S5: Digital twin mapping. Based on the data after desensitization and partitioned storage, a three-dimensional digital twin model of the community is constructed, and the status information and real-time data of various devices are mapped in the three-dimensional model. Step S6, Anomaly Analysis and Alarm: Perform real-time analysis on the multi-source data, identify abnormal events and generate alarm information; mark the alarm information in the digital twin 3D model, and push the alarm information to the terminal device of the designated administrator; Step S7: Closed-loop processing. Receive feedback from management personnel on the processing of the alarm information, update the alarm status, and form a closed-loop management system from anomaly detection to task allocation and processing completion.

[0014] By adopting the above technical solution, the de-identified data is stored in partitions, so that even if illegal means are used to obtain sensitive data, it will require breaking through multiple layers of defense systems, making it impossible to obtain all sensitive data at once, thus increasing data security.

[0015] In a further embodiment, the format of the video surveillance data is uniformly corrected to MP4 format.

[0016] In a further embodiment, the image data is processed locally using YOLOv5.

[0017] By adopting the above technical solution, MP4 format videos have better quality and smaller data packets, making them suitable for managing complex areas like residential communities. The image data consists mainly of facial images used for identity verification. After training the YOLOv5 system by extracting X core features locally, the comparison can be performed based on these core features, eliminating the need to upload all data to the terminal for comparison. This approach ensures that even if someone obtains facial image data illegally, it will be incomplete, further reducing the risk of data leakage.

[0018] In a further embodiment, a high-altitude object throwing warning triggering strategy is embedded in the digital twin mapping area.

[0019] By adopting the above technical solution, the amount of data that needs to be analyzed in the anomaly analysis and alarm steps can be reduced. Since objects thrown from high altitudes are sudden and unpredictable, placing them in step S5 can effectively reduce the subsequent amount of computation.

[0020] In summary, the present invention has the following beneficial effects: 1. By standardizing the format of similar data, desensitizing sensitive information, and using partitioned storage, a system was established to properly manage the complex comprehensive data of the community. Attached Figure Description

[0021] Figure 1 This is a topology diagram of the intelligent management system of the present invention; Figure 2 This is a schematic diagram of the deployment structure of the intelligent management system equipment of the present invention; Figure 3This is a flowchart of the data processing method of the present invention; Figure 4 This is a flowchart of the anomaly analysis and alarm triggering process in the data processing method of this invention; Figure 5 This is the high-altitude object throwing image recognition process in the data processing method of the present invention. Detailed Implementation

[0022] The present invention will be further described in detail below with reference to the accompanying drawings.

[0023] It should be noted that in the description of this invention, any descriptions of orientation, such as up, down, front, back, left, right, etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings. They are only for the purpose of facilitating the description of this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed or operated in a specific orientation, and should not be construed as a limitation of this invention.

[0024] Example 1: like Figure 1 and Figure 2 As shown, an intelligent management system includes: The data acquisition layer includes various data acquisition devices deployed within the community to acquire real-time data within the community; the data acquisition devices include at least video surveillance equipment, fire monitoring equipment, access control equipment, vehicle management equipment, and charging pile equipment; The video surveillance equipment consists of environmental monitoring cameras, facial recognition cameras, high-altitude object throwing monitoring cameras, and temperature monitoring equipment. The facial recognition cameras are installed at the entrances and exits of the residential area to capture facial data of people entering and exiting. The high-altitude object throwing monitoring cameras are installed on the exterior facades of buildings to monitor the act of throwing objects from heights. Assuming a building has four exterior facades, at least one high-altitude object throwing monitoring camera is installed on each facade. Each high-altitude object throwing monitoring camera can monitor a horizontal width of 25-50 meters on the facade. If a facade is too wide, additional cameras are needed. Multiple high-altitude littering surveillance cameras work together; the area within 2 meters around each building in a community is designated as the recognition zone for high-altitude littering surveillance cameras, and the area at the entrances and exits of the community is designated as the recognition zone for facial recognition cameras. The remaining areas of the community need to be monitored by environmental monitoring cameras. Multiple environmental monitoring cameras form a distributed community internal environmental monitoring network, and an environmental monitoring camera can be set at the entrance of each unit building to monitor the information of people entering and leaving the unit building, so as to have more monitoring data in the event of an accident.

[0025] Fire monitoring equipment consists of smoke detectors, temperature sensors, and water pressure sensors, with the smoke detectors installed in areas such as non-motorized vehicle sheds; access control equipment consists of door magnets and access controllers; vehicle management equipment consists of barriers and parking space status cameras. To manage parking spaces within the community and ensure sufficient parking for residents, temporary parking by external vehicles needs to be controlled. Parking space status cameras can determine the availability of parking spaces within the community, and by obtaining information on the daily fixed parking space usage, excess parking spaces can be flexibly disposed of; environmental monitoring equipment consists of smart streetlights and waste sorting monitoring equipment; charging pile equipment consists of voltage and current detectors and leakage protection sensors.

[0026] Edge computing nodes, communicating with the data acquisition layer, are used to perform localized preprocessing on the acquired data. This preprocessing includes data cleaning, format standardization, and preliminary analysis. Because various cameras with different functions are installed in this system, and these cameras have different requirements and models, many types of video files are generated. Therefore, it is necessary to standardize the format of the generated video files to facilitate their use by the data platform. The data within the community mainly consists of video data, image data, and audio data. Other data can be uniformly recorded, organized, and summarized using tables. For example, personnel information and vehicle information within the community can be uniformly recorded and summarized using tables. Image data mainly comes from vehicle gate data and identity verification data.

[0027] The data platform is communicatively connected to the edge computing nodes and is used to receive preprocessed data, perform data anonymization processing, and perform weighted partition storage. The data platform is equipped with a data anonymization module and a weighted partition storage module. The data anonymization module is used to anonymize sensitive information in the data. The weighted partition storage module is used to assign different weight values ​​according to the sensitivity and importance of the data, and to partition and store the data in different storage areas according to the weight values.

[0028] The digital twin platform communicates with the data platform and is used to construct a digital twin 3D model of the community based on the data provided by the data platform, and to map and visualize the status information and real-time data of various devices in the 3D model.

[0029] The application presentation layer, communicating with the digital twin platform, provides a visual interface and interactive functions. It includes a monitoring management module, a fire management module, a personnel management module, a vehicle management module, an alarm management module, and an alarm push module. The monitoring management module displays the distribution and operating status of monitoring equipment in a 3D model and supports clicking to view real-time footage and video playback. The fire management module displays the distribution and status of fire-fighting equipment in a 3D model and automatically generates warnings for equipment not inspected on time. The personnel management module displays information such as the number of permanent residents, real-time pedestrian flow, and personnel. The vehicle management module displays vehicle entry and exit records, parking space occupancy, and charging pile usage status. The alarm management module collects various abnormal alarm events, displays them uniformly, and assigns tasks. The alarm push module, connected to the application presentation layer, pushes alarm information to the terminal devices of designated management personnel when abnormal events occur.

[0030] Example 2: like Figures 3-5 As shown, a data processing method suitable for a community intelligent management system includes the following steps: Step S1: Multi-source data acquisition. Multi-source data is acquired by deploying various data acquisition devices in the community. The multi-source data includes video surveillance data, fire monitoring data, access control data, vehicle data, environmental monitoring data, and pedestrian flow data. Step S2: Data preprocessing. The collected multi-source data is preprocessed locally through edge computing nodes. The preprocessing includes data cleaning and format standardization. The format of the video surveillance data is uniformly corrected to MP4 format. The image data is processed locally using YOLOv5. It should be noted that the data storage time of the high-altitude object throwing monitoring camera is 6 hours. The 6-hour setting is mainly for buffering and pre-storage to avoid the lack of data source after someone reports a case if the high-altitude object throwing monitoring camera fails. The image boundary of the high-altitude object throwing monitoring camera is fixed. Its recognition strategy adopts a simple recognition method. First, the location where high-altitude objects may be thrown is marked with a box by manual means. When color spots appear in these areas and the color spots gradually increase in size at the edges according to the time sequence, an alarm is triggered. The image of the throwing path is fixed as a photo by frame extraction. While fixing the evidence, the alarm is sent. The final confirmation is made by manual identification of the video segment. The initial color spot size of each box is determined by intelligent algorithm. Step S3: Sensitive Data Identification and Desensitization. Sensitive data is identified from the multi-source data, and desensitized. The sensitive data includes facial image data, resident identity information data, and vehicle information data. The desensitization process employs at least one of the following methods: hash mapping replacement of sensitive fields, random distribution processing near the mean for numerical data, and generalization processing based on the K-anonymity algorithm. Image data primarily consists of facial images used for identity verification. After training the YOLOv5 system by extracting X core features locally, comparison can be performed using these core features. It is not necessary to upload all data to the terminal for comparison. This approach ensures that even if someone illegally obtains facial image data, it will be incomplete, further reducing the risk of data leakage. For example, by using a grid partitioning method to separate the size, proportion, and orientation of the face... A dataset is generated, and only these datasets are compared. If the datasets match the characteristics, the comparison is defined as successful. This setting prevents the collection and comparison of other data, thereby reducing the risk of data leakage. For example, taking the bridge of a person's nose as a benchmark, the length and maximum width of the bridge of the nose are two values, the size of the left and right eyes are two values, the tilt angle between the left eye and the length of the bridge of the nose and the tilt angle between the right eye and the length of the bridge of the nose are two values, and finally, the maximum width and thickness of the mouth are two values. Basic feature information can be obtained through these values. Based on this feature information, it is possible to compare whether the person belongs to an authorized person. The reason why a very complete data comparison method is not needed is that this application is mainly aimed at the internal management of the community, not a highly confidential or classified area. Therefore, an overly strict identity verification method is not required. This method can also effectively reduce the amount of data processing, thereby achieving a lower cost to complete the deployment of the community's smart system. Step S4: Weight Calculation and Partitioned Storage. For the anonymized data, calculate the weight value of each data item based on data category, sensitivity, and usage frequency; allocate the data to different storage partitions according to the weight values. Step S5: Digital twin mapping. Based on the data after desensitization and partitioned storage, a digital twin 3D model of the community is constructed. The status information and real-time data of various devices are mapped in the 3D model. A high-altitude object throwing warning triggering strategy is implanted in the digital twin mapping area. Step S6, Anomaly Analysis and Alarm: Perform real-time analysis on the multi-source data, identify abnormal events and generate alarm information; mark the alarm information in the digital twin 3D model, and push the alarm information to the terminal device of the designated administrator; Step S7: Closed-loop processing. Receive feedback from management personnel on the processing of the alarm information, update the alarm status, and form a closed-loop management system from anomaly detection to task allocation and processing completion.

[0031] Example 3: The system adopts a five-layer architecture, using edge computing nodes to perform localized preprocessing such as video transcoding, image inference, data cleaning, and format standardization near the data source. This unifies heterogeneous data from various manufacturers and models of equipment into a standardized format, effectively solving the core pain point of isolated subsystems such as monitoring, fire protection, access control, and vehicle management in traditional systems, where data cannot be shared. Taking a typical medium-to-large residential community as the deployment target, the system is set at a building area of ​​approximately 150,000 square meters, 12 high-rise residential buildings, a total of 864 households, and a resident population of approximately 2,600. Six 2-megapixel facial recognition cameras with a capture frame rate of 25 frames per second are deployed at the community entrances and exits. One high-altitude object throwing monitoring camera is installed on each of the four facades of each building, totaling 48 4-megapixel high-altitude object throwing monitoring cameras with 40x optical zoom, installed at a height of 15 meters above the ground with a tilt angle of 30 degrees. The monitoring range is 45°, with a single unit having an effective monitoring width of 30 meters and an overlap rate of no less than 15% between adjacent devices. 120 environmental monitoring cameras with 2-megapixel starlight-level low-light capabilities are deployed in public areas. At the same time, 432 smoke detectors, 8 temperature sensors, 4 water pressure sensors, 12 access control controllers, 2 gate systems, 180 parking space status cameras, 60 charging pile interfaces with 6 sets of voltage and current detectors, 60 smart streetlights, and 4 sets of garbage classification monitoring equipment are deployed to form a comprehensive, blind-spot-free real-time data acquisition network within the community. The edge computing nodes use Huawei Atlas500Pro intelligent edge stations, with 12 stations deployed in each building. Each station is equipped with an Ascend 310 AI processor, requiring only 22 TOPSINT8 computing power, 16GB of memory, and 512GB of SSD storage, and is responsible for data processing within the building and a 50-meter radius around it. During use, the RTSP real-time video streams from each camera are uniformly transcoded to H.264 encoded MP4 format, with a resolution normalized to 1920×1080 and a bitrate of 2Mbps. The streams are automatically segmented and timestamped every 5 minutes. A lightweight YOLOv5s model is run to perform real-time target detection on the high-altitude projectile video stream, limiting the detection category to falling objects, with a confidence threshold of 0.65 and an NMS IoU threshold of 0.4. After a target is detected, 5-second segments before and after the target are uploaded. Footage from lens damage, overexposure due to strong light, etc., is removed. Invalid frames caused by noise at night, with a Laplacian variance less than 100 or a mean luminance less than 30; structured data such as access control card swipes, vehicle gates, and smoke alarms are uniformly converted into JSON format and fields are mapped according to a standard dictionary to ensure that the data platform accurately and comprehensively captures all data. The data platform performs sensitive information desensitization processing on the received preprocessed data. Among them, facial image data is processed locally by YOLOv5 to extract the length and maximum width of the bridge of the nose, the size of the left and right eyes, the tilt angle of the left and right eyes and the bridge of the nose, and the maximum width of the mouth. Training and comparison with core feature vectors such as thickness eliminates the need to upload complete face images to the terminal, preventing the reconstruction of complete face information even if the data is illegally obtained. Simultaneously, weight values ​​are calculated based on data category, sensitivity, and usage frequency: sensitivity accounts for 60%, usage frequency for 30%, and data volume for 10%. Data is allocated to four encrypted partitions: L1 public data, L2 internal use, L3 sensitive de-identified features, and L4 highly sensitive original data. L3 and L4 partitions use AES-256 encryption for storage, with keys rotated every 7 days by KMS. Access control employs the RBAC model, classifying four roles: administrator, security supervisor, ordinary security guard, and operations engineer. All access operation records are logged for at least 180 days. TLS 1.3 encryption is used between edge nodes and the data platform, and SM4 encryption is used between the data platform and the digital twin platform. This constructs a comprehensive data security protection system from collection, transmission, storage to access, fundamentally eliminating the risk of leakage of residents' facial data, identity information, vehicle information, and other private data.

[0032] At the digital twin platform level, the system constructs a 3D digital twin model of the community based on anonymized and partitioned data. All device status information and real-time data are mapped and visualized within the model. The application presentation layer achieves a unified interactive interface through six modules: monitoring management, fire management, personnel management, vehicle management, alarm management, and alarm push. Administrators can complete all monitoring operations without switching between multiple platforms. Furthermore, the system pre-configures multi-scenario intelligent linkage control strategies, significantly improving the automated handling of abnormal events: In fire linkage, when a smoke detector or temperature sensor in a non-motorized vehicle shed triggers an alarm, edge nodes instantly upload alarm signals and on-site video clips. The digital twin platform highlights the alarm location in the 3D model and automatically displays surrounding monitoring footage. The alarm push module sends information including precise location, alarm type, and on-site screenshots to the property duty manager and fire safety officer via WeChat and SMS. Simultaneously, the system sends a power-off command to the charging pile equipment to cut off power to the faulty area to prevent the spread of electrical fires. In vehicle illegal parking management, parking space status cameras detect vehicles parked in fire lanes or unit entrances for extended periods. At 3 minutes, the edge node identifies the license plate and reports the illegal parking incident. The digital twin platform automatically generates an illegal parking mark, and the alarm management module generates a pending task and pushes it to the handheld terminal of the security guard. If it is a registered vehicle, a text message reminder is sent to the owner's mobile phone at the same time. If the vehicle does not leave within 10 minutes, security personnel are notified to handle the situation on-site. In terms of tracing the source of objects thrown from high-rise buildings, after the system detects an object throwing incident, it automatically captures the image sequence of the throwing path. At the same time, it retrieves the environmental monitoring video of the building unit entrance at the same timestamp and the access control card swipe / facial recognition records within 5 minutes before and after. It packages and generates an independent evidence package of the object throwing incident, which is encrypted and stored in the L4 dedicated evidence partition. It can only be accessed by the administrator with judicial evidence collection authority, which effectively solves the pain points of scattered evidence and difficulty in linking and tracing the source in traditional monitoring.To further verify the actual operational performance of the system, the applicant continuously ran it in the aforementioned deployment environment for 90 days and conducted performance tests. The results showed that the average daily online rate of all data acquisition devices reached 99.7%, the data reporting completeness rate was 98.9%, and the final consistency was 99.5%. The average end-to-end latency from camera acquisition to completion of preprocessing and uploading to the data platform was 1.8 seconds, meeting the design target of ≤3 seconds. After parameter optimization, the accuracy of high-altitude object throwing detection improved from the initial 78.3% to 92.6%, and the recall rate improved from 88.2% to 94.1%. A total of 17 manually confirmed events were detected within 90 days. Sixteen alarms were detected, with one missed alarm due to insufficient lighting at night and the small size of the object. There were 28 false alarms, mainly from birds and leaves. Further optimization can be achieved by adjusting the detection area mask and adding a lighting compensation algorithm. The success rate of alarm information reaching the responsible person's terminal within 30 seconds was 99.2%, with an average push time of 4.7 seconds. The initial loading time of the digital twin 3D model on a regular management terminal was 6.3 seconds, the viewpoint switching response was 0.5-1.2 seconds, and the device status refresh frequency was 2 seconds, meeting the smoothness requirements of daily monitoring. The average daily increase in storage was approximately 180GB, and a 48TB RAID5 storage array can meet approximately 8 months of online storage needs. The storage cycle can be further extended with an automatic cold data archiving strategy. The system also implements a complete closed-loop management mechanism: from anomaly detection, alarm generation, 3D model marking, terminal push, task allocation to responsible personnel, to processing feedback reception and alarm status update, a complete closed-loop process is formed to ensure that every abnormal event has a response, follow-up, and result. Through the above deployment scheme, linkage strategy, security reinforcement measures, and actual test data, it is fully demonstrated that this invention effectively solves many problems in traditional community management systems, such as the isolation of various business systems, the inability to communicate data, the lack of visualization of equipment status, the reliance on manual records for inspections leading to omissions, the lag in response to abnormal events making it difficult to trace and collect evidence, and the lack of privacy protection for data storage leading to leakage risks. The system has high engineering feasibility, operational stability, and security capabilities, and can be widely applied to the intelligent upgrading and transformation of medium and large-sized residential communities, industrial parks, and commercial complexes, with significant social benefits and promotional value.

[0033] In the description of this invention, unless otherwise explicitly defined, terms such as "setting," "installing," and "connecting" should be interpreted broadly, and those skilled in the art can reasonably determine the specific meaning of the above terms in this invention in conjunction with the specific content of the technical solution.

[0034] This specific embodiment is merely an explanation of the present invention and is not intended to limit the invention. After reading this specification, those skilled in the art can make modifications to this embodiment without contributing any inventive step, but such modifications are protected by patent law as long as they are within the scope of the claims of the present invention.

Claims

1. An intelligent management system, characterized in that, include: The data acquisition layer includes various data acquisition devices deployed within the community to acquire real-time data within the community; Edge computing nodes, which are communicatively connected to the data acquisition layer, are used to perform localized preprocessing on the acquired data. The preprocessing includes data cleaning, format standardization, and preliminary analysis. The data platform is communicatively connected to the edge computing nodes and is used to receive preprocessed data, perform data anonymization processing, and store the data in weighted partitions. The digital twin platform is connected to the data platform and is used to construct a digital twin three-dimensional model of the community based on the data provided by the data platform, and to map and visualize the status information and real-time data of various devices in the three-dimensional model. The application presentation layer communicates with the digital twin platform and provides a visual interface and interactive functions.

2. The intelligent management system according to claim 1, characterized in that, The data acquisition device includes at least: The video surveillance equipment consists of an environmental monitoring camera, a face recognition camera, a high-altitude object throwing monitoring camera, and a temperature measurement monitoring device. The face recognition camera is installed at the entrance and exit of the community to capture the facial data of people entering and exiting. The high-altitude object throwing monitoring camera is installed on the exterior of the building to monitor the behavior of throwing objects from high altitudes. The fire monitoring equipment consists of a smoke detector, a temperature sensor, and a water pressure sensor, with the smoke detector installed in the non-motorized vehicle shed area. Access control equipment consists of door magnets and access controllers; The vehicle management equipment consists of a barrier gate and parking space status cameras. The environmental monitoring equipment consists of smart streetlights and waste sorting monitoring equipment. The charging pile equipment consists of voltage and current detectors and leakage protection sensors.

3. The intelligent management system according to claim 1, characterized in that, The data platform includes: The data desensitization module is used to desensitize sensitive information in data; The weighted partition storage module is used to assign different weight values ​​to the data based on its sensitivity and importance, and then partition and store the data in different storage areas according to the weight values.

4. The intelligent management system according to claim 1, characterized in that, The application presentation layer includes: The monitoring and management module is used to display the distribution and operating status of monitoring devices in a 3D model, and supports clicking to view real-time images and video playback; The fire management module is used to display the distribution and status of fire protection equipment in a 3D model and automatically generate early warnings for equipment that has not been inspected on time. The personnel management module is used to display the number of permanent residents in the community, real-time pedestrian flow, and personnel information; The vehicle management module is used to display vehicle entry and exit records, parking space occupancy status, and charging pile usage status; The alarm management module is used to collect various abnormal alarm events and display and assign tasks in a unified manner; The alarm push module, connected to the application presentation layer, is used to push alarm information to the terminal devices of designated administrators when an abnormal event occurs.

5. A data processing method for an intelligent management system according to any one of claims 1-4, characterized in that, The data processing method includes the following steps: Step S1: Multi-source data acquisition. Multi-source data is acquired by deploying various data acquisition devices in the community. The multi-source data includes video surveillance data, fire monitoring data, access control data, vehicle data, environmental monitoring data, and pedestrian flow data. Step S2: Data preprocessing. The collected multi-source data is preprocessed locally through edge computing nodes. The preprocessing includes data cleaning and format standardization. Step S3: Sensitive data identification and desensitization. Sensitive data in the multi-source data is identified and desensitized. The sensitive data includes facial image data, resident identity information data, and vehicle information data. The desensitization process adopts at least one of the following methods: hash mapping replacement of sensitive fields, random distribution processing of numerical data near the mean, and generalization processing based on the K-anonymity algorithm. Step S4, Weight Calculation and Partitioned Storage: For the data after anonymization, calculate the weight value of each data item according to the data category, sensitivity, and usage frequency; and allocate the data to different storage partitions according to the weight values. Step S5: Digital twin mapping. Based on the data that has been de-identified and partitioned for storage, a three-dimensional digital twin model of the community is constructed, and the status information and real-time data of various devices are mapped in the three-dimensional model. Step S6, Anomaly Analysis and Alarm: Perform real-time analysis on the multi-source data, identify abnormal events and generate alarm information; mark the alarm information in the digital twin 3D model, and push the alarm information to the terminal device of the designated administrator; Step S7: Closed-loop processing. Receive feedback from management personnel on the processing of the alarm information, update the alarm status, and form a closed-loop management system from anomaly detection to task allocation and processing completion.

6. The data processing method according to claim 5, characterized in that: The format of the video surveillance data has been uniformly corrected to MP4 format.

7. The data processing method according to claim 5, characterized in that: The image data is processed locally using YOLOv5.

8. The data processing method according to claim 5, characterized in that: A high-altitude object throwing warning triggering strategy is embedded in the digital twin mapping area.

Images