Intelligent machine room multi-dimensional dynamic monitoring management method and system based on digital twinning

Abstract

The invention provides an intelligent machine room multi-dimensional dynamic monitoring management method and system based on digital twinning, and relates to the technical field of machine room monitoring, and the method comprises the steps: collecting a three-dimensional point cloud set and an RGB image set of a machine room space, and building a space mapping table; mapping the access control event data to a corresponding position in a digital twin space, generating an access control event space positioning set, identifying access control abnormal events, and constructing an access control abnormal event space positioning set; performing abnormal behavior recognition, and constructing an abnormal event graph structure; reconstructing and visualizing an abnormal track based on the abnormal event graph structure; in combination with the access control abnormal event space positioning set, the abnormal behavior category prediction probability distribution and the track risk level, executing a grading early warning mechanism, and generating an access control abnormal event log; according to the invention, intelligent positioning, track visualization and risk grading early warning of access control abnormal events in the intelligent machine room are realized, and the method is suitable for digital operation and maintenance management systems such as the intelligent machine room.

Description

Technical Field

[0001] The present invention relates to the field of computer room monitoring technology, and in particular to a multi-dimensional dynamic monitoring and management method and system for a smart computer room based on digital twins. Background Art

[0002] With the rapid development of digital twin technology and intelligent perception systems, smart computer rooms, as a next-generation information infrastructure, are becoming increasingly important for ensuring business continuity and efficient operations and maintenance. The key challenge in building smart computer rooms is how to achieve spatial localization, semantic understanding, and intelligent response to key anomalies, such as human behavior and access control events, based on dynamic perception data from real physical spaces.

[0003] The Chinese patent application with authorization announcement number CN118466258B provides a smart computer room BA automatic control management system based on big data analysis. The method includes: a real-time monitoring module, the real-time monitoring module is communicatively connected to a controller and a startup management module, the startup management module is communicatively connected to a startup control module, and the startup control module is communicatively connected to the controller; the real-time monitoring module is used to perform real-time monitoring and analysis of the environment in the smart computer room; the startup management module is used to manage and analyze the startup environment and energy consumption of the smart computer room; and the startup control module is used to perform environmental regulation and analysis when the smart computer room is in the startup state.

[0004] However, existing technologies still face numerous challenges in smart computer room security management. Most current access control systems in smart computer rooms are relatively limited in functionality, recording only basic log information such as card swipes and door openings. They lack the ability to connect with systems like video surveillance and personnel location tracking. This information silo makes it impossible to accurately reconstruct the events and personnel movements immediately when critical security incidents occur. In particular, when security incidents occur, such as unauthorized intrusion, unusual late-night card swiping, or repeated access control failures, the system lacks the ability to instantly trace behavioral trajectories and identify risks through 3D visualization, which can easily lead to missed alerts or misjudgments. Furthermore, due to a lack of effective integration between event alerts and operations and maintenance processes, the system often only issues alerts without further linking them to record the operations and maintenance response process and results. This results in a manual response and tracing system for critical incidents, which is inefficient and prone to errors, limiting the closed-loop security response capabilities of the computer room. If unauthorized entry is not promptly detected and addressed, unauthorized operators may enter core areas, potentially damaging servers, power systems, or communications equipment. This can lead to data leaks, equipment downtime, or even widespread business interruptions, compromising network security. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art, the present invention provides a multi-dimensional dynamic monitoring and management method and system for smart computer rooms based on digital twins. The method constructs a digital twin space model of the smart computer room, realizes the accurate spatial positioning of access control events in the digital twin space, breaks the limitations of the traditional planar event recording method, generates a device space mapping table, makes the spatial information expression more intuitive, and effectively solves the problem of multi-source data fusion and access control abnormal event identification in the complex environment of the smart computer room, assists operation and maintenance personnel to quickly lock key areas and key time periods, accurately carry out spatiotemporal tracing and security control of access control abnormal events, and ensures closed-loop management of the entire process of risk control.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] The multi-dimensional dynamic monitoring and management method for smart computer rooms based on digital twins includes:

[0008] Collect 3D point cloud collection of computer room space and RGB image collection , build a digital twin space model that represents the physical state of the computer room and supports monitoring and control, and establish a spatial mapping table between the digital twin space and the physical space ;

[0009] Based on spatial mapping table , map the personnel entry and exit event data generated by the access control system to the corresponding spatial position in the digital twin space, conduct continuous real-time status detection of personnel behavior to identify abnormal event behavior, and build a spatial positioning set of access control abnormal events ;

[0010] Based on the spatial positioning set of access control abnormal events , perform abnormal behavior identification and correlation analysis, and build abnormal event graph structure ;

[0011] Based on abnormal event graph structure , dynamically reconstruct abnormal trajectories and realize visual control decision support through the three-dimensional scene of the control interface;

[0012] Combined with the spatial positioning set of access control abnormal events , abnormal behavior category prediction probability distribution and trajectory risk level , determine the corresponding control strategy and issue linkage control instructions accordingly, build a hierarchical early warning response control mechanism for access control abnormal events, and automatically generate access control abnormal event logs to achieve closed-loop feedback control of the monitoring system; where k is the trajectory index and N' is the total number of trajectories.

[0013] Furthermore, the space mapping table The establishment methods include:

[0014] Based on 3D point cloud collection and RGB image collection Constructing multimodal fusion data ;

[0015] Based on multimodal fusion data set , combined with the building information model BIM, to establish a spatial index structure ;

[0016] Based on device collection , spatial model with semantic labels With spatial index structure Perform spatial mapping, introduce a time series index update mechanism, and establish a spatial mapping table ;

[0017] The device set Obtained through installation records, RFID tags, QR codes, or automatic camera recognition;

[0018] The temporal index update mechanism adopts KD-Tree to dynamically maintain the spatial index structure.

[0019] Furthermore, the spatial index structure The establishment methods include:

[0020] Based on multimodal fusion data set The voxel gridding algorithm is used to perform structured conversion on the 3D point cloud data to obtain a 3D point cloud voxel set. ;

[0021] Combine the structural component information provided by the Building Information Model (BIM) with the 3D point cloud voxel collection Perform spatial alignment and perform multimodal semantic segmentation to obtain a spatial model with semantic labels ;

[0022] Based on spatial model with semantic labels Building a spatial index structure .

[0023] Furthermore, the access control abnormal event spatial positioning set The construction methods include:

[0024] Get the original access control event collection , and based on the original access control event set Build a high-quality access control event collection ;

[0025] Based on high-quality access control event collection and spatial mapping table , get the access control event spatial positioning set ;

[0026] Based on the spatial positioning set of access control events , build a spatial positioning set of access control abnormal events .

[0027] Furthermore, the abnormal event graph structure The construction methods include:

[0028] Based on the spatial positioning set of access control abnormal events , obtain the characteristic representation of access control abnormal events and video behavior segment feature representation , construct multimodal features ;

[0029] Based on multimodal features , introduce the temporal convolutional network to construct the probability distribution of abnormal behavior category prediction ;

[0030] Predicting probability distribution based on abnormal behavior categories and historical operation and maintenance event data collection , build abnormal event graph structure .

[0031] Furthermore, the abnormal event graph structure The construction method also includes:

[0032] Defining the initial abnormal event graph , where the node set ; Indicates the nodes, that is, the corresponding historical or current abnormal events, is the total number of nodes; ≤ ;

[0033] Based on each event node , construct the initial feature matrix ;in, For one The set of real numbers with d rows and d columns; R represents a real number;

[0034] Based on historical operation and maintenance event data collection , define the initial adjacency matrix based on event correlation ;

[0035] The initial adjacency matrix Elements Representing an event Determine whether an edge exists;

[0036] Introducing a dynamic graph update mechanism to adapt to new events in real time and dynamically update the adjacency matrix ;

[0037] The graph convolutional network GCN is used for information propagation and embedding learning between events to obtain the first Layer-wise high-level semantic feature embedding;

[0038] pass After the convolution operation of the first layer, the Layer high-level semantic embedding vector , combined with the node set And the final dynamic adjacency matrix , building a structured abnormal event graph structure ;

[0039] Furthermore, the steps of dynamic trajectory reconstruction and three-dimensional scene visualization of the abnormal trajectory include:

[0040] Based on abnormal event graph structure Access control event spatial positioning collection , reconstruct the movement trajectory of abnormal personnel and output the trajectory point sequence ;

[0041] Based on trajectory point sequence , using the trajectory prediction algorithm based on the fusion of Kalman filter and long short-term memory network LSTM to obtain the predicted trajectory curve ;

[0042] Based on the predicted trajectory curve , Spatial positioning set of access control abnormal events and abnormal behavior category prediction probability distribution , perform three-dimensional trajectory mapping visualization and abnormal brightness display.

[0043] Furthermore, the steps of visualizing the three-dimensional trajectory mapping and displaying abnormal brightness include:

[0044] The predicted trajectory curve Each trajectory point Map to the digital twin 3D simulation scene coordinate system to construct the corresponding 3D space trajectory ;in, Indicates the The timestamp corresponding to each trajectory point record is Indicates the The timestamp corresponding to each track point record The corresponding spatial position based on the fusion of Kalman filter and long short-term memory network LSTM;

[0045] All By timestamp Connect them sequentially to form dynamic trajectory lines, and play them in a time-driven manner in the 3D visualization platform;

[0046] Based on the spatial positioning set of access control abnormal events and abnormal behavior category prediction probability distribution , for each predicted trajectory curve Each trajectory point Rendering color Make a judgment.

[0047] Determining the rendering color of each trajectory point of each predicted trajectory curve includes:

[0048] Set probability distribution thresholds, including upper threshold, middle threshold, and lower threshold;

[0049] , then the risk level High-risk warning. ; Predict the probability distribution of abnormal behavior category for the k-th trajectory point;

[0050] ,but This is a medium risk warning. ;

[0051] ,but Low risk reminder. ;

[0052] ,but Normal, .

[0053] Furthermore, when High risk warning, , the system immediately triggers a level 1 high-risk warning event; when Medium risk warning, , the system immediately triggers a secondary medium risk warning event; when Low risk warning or normal, , the system does not actively trigger an alarm.

[0054] Furthermore, the access control abnormal event log includes event number, event time, spatial location, abnormality category and risk level, and abnormality warning record.

[0055] A multi-dimensional dynamic monitoring and management system for a smart computer room based on digital twins is used to implement the multi-dimensional dynamic monitoring and management method for a smart computer room based on digital twins. The system includes:

[0056] Digital twin space modeling module: used to collect 3D point cloud collection of the computer room space With RGB image collection , build a digital twin space that represents the physical state of the computer room and supports monitoring and control, and establish a spatial mapping table between the digital twin space and the physical space ;

[0057] Access control abnormal event positioning module: based on spatial mapping table , map the personnel entry and exit event data generated by the access control system to the corresponding spatial position in the digital twin space, conduct continuous real-time status detection of personnel behavior to identify abnormal event behavior, and build a spatial positioning set of access control abnormal events ;

[0058] Abnormal behavior recognition and abnormal event graph construction module: based on the spatial positioning set of access control abnormal events , perform abnormal behavior identification and correlation analysis, and build abnormal event graph structure .

[0059] Abnormal trajectory reconstruction and 3D visualization control module: Based on the abnormal event graph structure, it dynamically reconstructs abnormal trajectories and implements visual control decision support through the 3D scene of the control interface.

[0060] Linkage warning control and log generation module: It is used to combine the spatial positioning set of access control abnormal events, the predicted probability distribution of abnormal behavior categories and the trajectory risk level to determine the hierarchical warning linkage response control mechanism and issue closed-loop linkage control instructions to the controlled devices in the target area accordingly. It also retrieves the execution status of the controlled devices in real time as status feedback and generates a closed-loop access control abnormal event log containing the device execution feedback and response status.

[0061] Compared with the prior art, the present invention has the following beneficial effects:

[0062] The present invention constructs a digital twin space model with spatial semantics by fusing three-dimensional point cloud and RGB image data, establishes a dynamically updateable spatial index structure, realizes accurate mapping and real-time updating of access control abnormal events in three-dimensional space, avoids the problems of ambiguous device positioning and unclear event response in traditional two-dimensional records, and improves the spatial visualization guidance capability and rapid response efficiency of abnormal events.

[0063] The present invention constructs a high-precision mapping mechanism between access control abnormal events and digital twin space, introduces a Bayesian filtering model and an abnormal behavior recognition model to perform semantic analysis and spatial reasoning on the original access control data, and realizes the identification and positioning of access control abnormal events such as multiple consecutive card swiping failures and passing through non-duty time periods in three-dimensional space, avoiding the problems of abstract event description and location ambiguity in traditional access control systems, and improving the spatial visibility and response efficiency of operation and maintenance personnel when dealing with sudden abnormal events.

[0064] The present invention constructs a multimodal alignment mechanism for access control abnormal events and video behavior information, integrates contextual semantics and video spatiotemporal features, and breaks through the spatiotemporal information breakpoints between multimodal data sources; introduces graph convolution and dynamic graph update strategies to establish spatiotemporal semantic associations between abnormal behaviors and historical events, forming a dynamically evolving abnormal event map, avoiding the problems of isolated behavior recognition and evolutionary relationship defects in traditional systems, and providing an accurate graph model foundation for subsequent event tracing, multi-source early warning linkage, and behavior reasoning.

[0065] This paper addresses the issue of missing trajectories in access control records by constructing a spatial mapping relationship between anomaly event graphs and access control events, and by integrating a Kalman filter with an LSTM network to dynamically repair and predict trajectory data. By visually representing predicted trajectories, access control anomaly locations, and predicted probabilities and risk levels for abnormal behavior categories, the paper dynamically reconstructs and displays risk levels for abnormal individuals in the three-dimensional space of a digital twin, providing operations and maintenance personnel with intuitive, continuous, and traceable analysis.

[0066] The present invention combines the access control abnormality positioning results, behavior category prediction probability and trajectory risk level to judge the risk level of each trajectory point one by one, and automatically triggers the corresponding level of early warning response; at the same time, it generates an abnormal event log in a structured manner to avoid the problems of early warning trigger delay, risk level ambiguity and non-traceability of event records in traditional systems, and realizes hierarchical alarm, full process recording and precise spatial positioning of abnormal events in smart computer rooms. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0068] Figure 1 This is a principle flow chart of the multi-dimensional dynamic monitoring and management method of a smart computer room based on digital twins of the present invention;

[0069] Figure 2is a flow chart of a method for constructing a spatial mapping table according to the present invention;

[0070] Figure 3 It is a flow chart of the method for constructing an abnormal event graph structure of the present invention;

[0071] Figure 4 It is a flow chart of the method for constructing an abnormal event graph structure of the present invention;

[0072] Figure 5 is a flow chart of a method for trajectory reconstruction and visualization of abnormal trajectories according to the present invention;

[0073] Figure 6 It is a flow chart of the method for generating access control abnormal event logs of the present invention;

[0074] Figure 7 This is an example diagram of the access control abnormal event log of the present invention;

[0075] Figure 8 It is a functional module diagram of the multi-dimensional dynamic monitoring and management system of the smart computer room based on digital twins of the present invention. DETAILED DESCRIPTION

[0076] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0077] Example 1

[0078] See also Figure 1 As shown, this embodiment provides a multi-dimensional dynamic monitoring and management method for a smart computer room based on digital twins, including:

[0079] Step S1000: Collect the 3D point cloud collection of the computer room space and RGB image collection , build a digital twin space model that represents the physical state of the computer room and supports monitoring and control, and establish a spatial mapping table between the digital twin space and the physical space .

[0080] Furthermore, if Figure 2 As shown, step S1000 includes:

[0081] Step S1100: Obtain a 3D point cloud set of the computer room space and RGB image collection , and based on a 3D point cloud collection and RGB image collection Constructing multimodal fusion data .

[0082] Specifically, a LiDAR sensor is arranged in the computer room to scan and collect high-density three-dimensional point cloud data in the space, record the precise three-dimensional coordinate information of each sampling point, and obtain the three-dimensional point cloud collection of the computer room space. Expressed as:

[0083] ;

[0084] in, Indicates the Point cloud sampling points, including three-dimensional coordinate information; Indicates the Point cloud sampling points The three-dimensional coordinate information of space, is the horizontal coordinate, is the vertical longitudinal coordinate, is the depth coordinate; Indicates the total number of sampling points in the three-dimensional point cloud set of the computer room space.

[0085] In order to improve the quality of point cloud data, a dynamic threshold function is set The original three-dimensional point cloud collection of the computer room space Perform preprocessing to filter out point cloud data that may contain anomalies. Specifically, it includes:

[0086] First, calculate each point cloud sampling point The neighborhood center mean point , the specific process formula is as follows:

[0087] ;

[0088] in, Indicates the Point cloud sampling points The neighborhood center mean point The three-dimensional coordinate information of space, yes The horizontal coordinate of is the vertical longitudinal coordinate, is the depth coordinate.

[0089] Then, based on the Euclidean distance and dynamic threshold function Determine point cloud sampling points Whether it is an abnormal point, the specific process formula is as follows:

[0090] ;

[0091] in, Represents the Euclidean distance, used to calculate the point cloud sampling points and the mean point of its neighborhood center The Euclidean distance between .

[0092] Finally, the retained valid point cloud sampling points constitute the filtered three-dimensional point cloud set of the computer room space , the specific expression formula is as follows:

[0093] ;

[0094] Represents the filtered three-dimensional point cloud set of the computer room space, which is used to retain the new set composed of valid point cloud sampling points; Represents a dynamic threshold function used to determine the point cloud sampling point Is it abnormal? Indicates the current timeframe; Represents the estimated value of environmental noise, which is used to dynamically adjust the threshold.

[0095] At the same time, multiple cameras with fixed viewing angles are arranged synchronously in the computer room space, and a unified time synchronization acquisition mechanism is adopted to ensure that RGB image frames are collected synchronously at each viewing angle to enrich the texture and color feature information in the three-dimensional geometric information. The collected RGB image set of the computer room space Expressed as:

[0096] ;

[0097] in, Indicates the Image frames collected by a camera; Indicates the total number of cameras.

[0098] To ensure the consistency of multi-source data, a time synchronization mechanism is used during collection to ensure that the data collected by the lidar and camera remain consistent in the time frame, providing a reliable foundation for subsequent multimodal fusion.

[0099] Finally, based on the intrinsic and extrinsic parameters of the lidar and camera obtained by sensor calibration, the projection mapping relationship between the 3D point cloud coordinates and the image pixel coordinates is established. Specifically, each point cloud sampling point after filtering is Project it to the corresponding RGB image plane and get the color value corresponding to its pixel coordinates , thereby generating a multimodal fusion data set of the computer room space , the specific expression formula is as follows:

[0100] ;

[0101] in, Represents the color value of the pixel in the image (ie, RGB three-channel values), is the red channel value, is the green channel value; is the blue channel value.

[0102] Step S1200: Based on the multimodal fusion data set , combined with the building information model BIM, to establish a spatial index structure .

[0103] Specifically, this step aims to combine the multimodal fusion data set in step S1100 It is converted into a structured three-dimensional voxel grid representation and integrated with the semantic information in the Building Information Modeling (BIM) to realize a three-dimensional digital model with spatial structure and semantic attributes, providing a basis for subsequent equipment mapping and spatial analysis.

[0104] First, based on the multimodal fusion data set The voxel grid filtering algorithm is used to perform structured conversion on the three-dimensional point cloud data to achieve the discretization expression of the continuous space. The specific process formula is as follows:

[0105] ;

[0106] in, represents a set of 3D point cloud voxels, The center coordinates of each voxel unit are used to discretize the point cloud data into a regular grid; represents the index of the voxel unit, is the total number of voxel units; Represents a voxelization algorithm, an operation for converting a point cloud into a voxel representation; Indicates the voxel size in meters (m), which is used to divide the side length of the spatial grid unit and control the spatial discretization accuracy.

[0107] Then, the structural component information provided in BIM (such as walls, cabinets, doors, etc.) is combined with the 3D point cloud voxel collection Perform spatial alignment and perform multimodal semantic segmentation. The specific process formula is as follows:

[0108] ;

[0109] in, Represents the output spatial model with semantic labels for each voxel unit Assign semantic label categories , Voxel unit The corresponding semantic label category, is a set of predetermined semantic categories (such as wall, ceiling, cabinet, pipe, etc.); Represents a semantic segmentation model that combines voxelized spatial data with BIM information and outputs semantic labeling results; Represents a building information model, which contains prior knowledge such as spatial structure, component type, and semantic labels, and provides additional semantic and structural information to assist or guide semantic segmentation.

[0110] Finally, based on the spatial model with semantic labels Building a spatial index structure , the specific process formula is as follows:

[0111] ;

[0112] in, Represented as a spatial index structure, it is used to efficiently query, locate, and match the structure of semantic voxel units in space; Represents the Build KD Tree function (Build K-Dimensional Tree), which is used to construct voxel center points into a tree index structure that supports fast search.

[0113] For example, taking a smart computer room as an example, a space model with semantic labels is constructed. , , ,but will be , , For the nodes, a 3D KD-Tree is constructed to facilitate fast search.

[0114] Step S1300: Based on the device set , spatial model with semantic labels With spatial index structure Perform spatial mapping, introduce a time series index update mechanism, and establish a spatial mapping table .

[0115] Specifically, this step aims to map the device entity set to the generated spatial model with semantic labels to achieve accurate spatial positioning and management of devices in the digital twin system.

[0116] Device Collection Obtained by installation records or RFID or QR code or camera automatic recognition, each Represents a physical device (such as a server, switch, UPS power supply, air conditioner, access control system, monitoring equipment and other intelligent terminal devices). Indicates the total number of physical devices. Mapping to a spatial model with semantic labels , establish the corresponding relationship between the device and the spatial position, and obtain the device space mapping table .

[0117] To support dynamic updates of device collections (such as new devices being added, moved, or removed), a time series index update mechanism is introduced, using KD-Tree to dynamically maintain the spatial index structure. The specific expression formula is as follows:

[0118] ;

[0119] in, Indicates the current time step The spatial index structure of Represents a dynamic update index function, which is used to efficiently insert or delete nodes in the existing index structure to maintain spatial search efficiency; Represents the spatial index structure of the previous time step; Indicates the set of device changes that are added, moved, or deleted at the current moment.

[0120] Step S2000: Based on the space mapping table , map the personnel entry and exit event data generated by the access control system to the corresponding spatial position in the digital twin space, conduct continuous real-time status detection of personnel behavior to identify access control abnormal event behavior, and build a spatial positioning set of access control abnormal event .

[0121] This step aims to collect data on personnel entry and exit events through the access control system, and combine it with spatial mapping to achieve spatial positioning of access control events and abnormal behavior detection, providing basic data support for subsequent personnel trajectory analysis and security management.

[0122] Furthermore, if Figure 3 As shown, step S2000 includes:

[0123] Step S2100: Get the original access control event set , and based on the original access control event set Build a high-quality access control event collection .

[0124] Specifically, this step aims to collect original access control event data from access control devices , and cleans the data to ensure data quality. Each event record contains a timestamp, personnel ID, access point ID, and card swiping status. Based on the collected data, duplicate removal and outlier filtering are performed to eliminate invalid data and obtain a high-quality access control event set for subsequent analysis. .

[0125] First, access control event data is collected in real time using access control collection devices (such as IC card swipe terminals, fingerprint readers, and facial recognition terminals) deployed in the computer room. The specific formula is as follows:

[0126] ;

[0127] in, Represents the original access control event set; Indicates a single access control event. Indicates the The time of each access control event is used to record the specific time when the event occurs. It is the unique ID of the personnel, indicating the employee number corresponding to the card swipe or fingerprint. It is the unique ID of the access control point, indicating the specific location of the access control device where the event occurred. Indicates the card swiping status (success or failure); Indicates the total number of raw access control events.

[0128] Then, based on the original access control event set , perform deduplication and outlier filtering, eliminate invalid data, and obtain a high-quality access control event set The specific process steps are as follows:

[0129] First, deduplication operation. For the same person and the same access control point, Multiple card swipes within a period are considered repeated events, and the earliest event is retained. satisfy: , , , , then the event is determined to be repeated. Keep the earliest event, that is: keep Finally, the set of events with duplicates removed is expressed as .

[0130] Second, abnormal data filtering. For the data after deduplication, perform outlier screening and remove invalid data. Abnormal elimination means invalid personnel IDs will be eliminated, among which Represents the set of legal personnel IDs; if , then the invalid access control point ID will be eliminated, where Represents the legal access control point ID set; if Abnormal elimination: timestamp abnormalities are eliminated, where Represents the range of reasonable events.

[0131] Finally, after the above two steps, a high-quality access control event set is obtained. , the specific expression formula is as follows:

[0132] ;

[0133] Step S2200: Based on the high-quality access control event set and spatial mapping table , get the access control event spatial positioning set .

[0134] Specifically, this step aims to locate each access control event to its physical coordinates by mapping the access control point ID to spatial coordinates. Time series modeling and Bayesian filtering are also used to infer the person's likely real-world location.

[0135] First, for the current Access control events , its corresponding spatial coordinates The specific expression formula is as follows:

[0136] ;

[0137] in, Represents a mapping function used to uniquely identify access control points Mapping to coordinate positions in space;

[0138] Next, to improve positioning accuracy, we further combine the time series characteristics of access control events with the spatial transfer law to construct a positioning model based on Bayesian filtering. The specific process formula is as follows:

[0139] ;

[0140] in, Indicates the The employee's spatial location corresponding to the access control event; Indicates that employees The spatial position at the time of the event is used to describe the employee's movement path; Indicates that from the 1st to the A time series collection of access control events; Indicates that from the 1st to the A time series collection of access control events, used to predict the current location; represents the posterior probability for the current inference target (the probability distribution of the employee's spatial state); Indicates proportionality; Represents the observation likelihood, used for events and location The degree of matching, such as whether the access point belongs to the location; Represents the state transition probability, which is used to describe the employee mobility model (such as floor corridor structure and channel accessibility); Represents the prior probability, which is used to infer the distribution of people in the previous location from historical data; Indicates all possible Perform summation;

[0141] Finally, we get the access control event spatial positioning set .

[0142] Step S2300: Based on the access control event spatial positioning set , build a spatial positioning set of access control abnormal events .

[0143] This step is based on the access control event spatial positioning set , comprehensively adopt rule base and learning model to detect access control anomalies, identify potential violations or abnormal access control behaviors, and build a spatial positioning set of access control anomaly events The specific basic rules are as follows:

[0144] First, if the card swipe fails multiple times, In continuous time periods Memory Card swipe identification records are considered as potential risk behaviors such as tailgating, fraudulent use, and identity theft, and are considered abnormal:

[0145] ;

[0146] in, Indicates the number of events, used to indicate the number of consecutive events; Indicates the threshold for the number of failed card swipes. If the number exceeds this, it is considered abnormal. Indicates the existence symbol, which is used to indicate that a subset of events meets the conditions; Indicates the To A collection of consecutive access control events; Indicates the The card swipe status of each access control event, such as "success" or "failure"; Indicates the The card swiping status of the access control event is failed; Indicates the The user ID of the first access control event is The user ID of the access control events is the same, indicating the same user; Indicates the length of the time window, which is used to limit the continuous time range of card swiping failure events; Indicates that the time interval between consecutive events does not exceed the set time window.

[0147] Second, enter during off-duty hours. Attempting to enter the access control area during unauthorized time periods will be considered as risky behavior such as nighttime crossing, illegal overtime, illegal intrusion, etc. Set the guard time period ,but:

[0148] ;

[0149] in, Indicates the legal duty time period, that is, the time range during which passage is allowed; Indicates the start time of the duty period; Indicates the end time of the duty time; Indicates the The timestamp of the access control event (i.e., the time when the card was swiped); It means "if and only if", indicating a logical equivalence relationship; Indicates "not belonging to".

[0150] Third, illegal area passage. If employees Permission area Can be obtained from the permission mapping table, if , then it is judged to be abnormal:

[0151] ;

[0152] in, Indicates employees The set of permission areas (such as the floor of the workstation, rooms allowed to enter, etc.).

[0153] Finally, through the screening of the above multiple exception rules, the spatial positioning set of access control exception events is obtained , the specific expression formula is as follows:

[0154] ;

[0155] Step S3000: Based on the spatial positioning set of access control abnormal events , perform abnormal behavior identification and correlation analysis, and build abnormal event graph structure .

[0156] This step aims to collect data on personnel entry and exit events through the access control system, and combine it with spatial mapping to achieve spatial positioning of access control events and abnormal behavior detection, providing basic data support for subsequent personnel trajectory analysis and security management.

[0157] Furthermore, if Figure 4 As shown, step S3000 includes:

[0158] Step S3100: Based on the spatial positioning set of access control abnormal events , obtain the characteristic representation of access control abnormal events and video behavior segment feature representation , construct multimodal features .

[0159] Specifically, this step aims to construct a multi-modal feature vector of multi-source fusion to support subsequent abnormal behavior identification.

[0160] First, based on the access control abnormal event spatial positioning set constructed in step S2300 , a bidirectional long short-term memory network (BiLSTM) is used to extract the contextual semantic features of access control abnormal events and obtain the feature representation of access control abnormal events .

[0161] Next, based on the access control abnormal event spatial positioning set constructed in step S2300 Locating abnormal segments in video surveillance streams , the specific process formula is as follows:

[0162] ;

[0163] in, Indicates that the video segment calls a function to extract a time segment containing an access control abnormality event from the complete video surveillance stream; Indicates the access control abnormal event time period, starting from the time when the access control abnormal event started End time of abnormal access control event ; Indicates the Access point ID.

[0164] For example, if The access control abnormality event time period is 08:01~08:06 on March 26, 2025, and the abnormal access control point ID is "West Gate of Building 1". Then, the abnormal clip from 08:01~08:06 on March 26, 2025 is intercepted from the video surveillance stream of the camera corresponding to the West Gate of Building 1. .

[0165] Based on abnormal segments in video surveillance streams , the Inflated 3D ConvNet model is used to model the actions in the video and obtain the feature representation of the video behavior fragment .

[0166] Finally, based on the feature representation of access control abnormal events and video behavior segment feature representation Multi-source data, such as building multi-modal features , the specific process formula is as follows:

[0167] ;

[0168] in, Indicates passing The unified feature vector of the access control modality obtained after mapping; Represents the linear mapping matrix of the access control mode, Convert to an embedding representation of uniform dimension; Indicates passing The unified feature vector of the video modality obtained after mapping; Represents the linear mapping matrix of the video modality, Convert to an embedding representation of uniform dimension.

[0169] ;

[0170] in, Indicates the The attention weight of each modality satisfies ; Indicates the Unified feature representation of multiple modalities; Indicates the 'Unified feature representation of multiple modalities; Represents a trainable modality weight query vector, used to measure the The importance of each modality feature determines the attention score; express The transpose of Represents an exponential function, which is used to map the scores to non-negative weights, helping to highlight significant modal features.

[0171] Based on attention weight , and obtain multimodal features , the specific process formula is as follows:

[0172] ;

[0173] Step S3200: Based on multimodal features , introduce the temporal convolutional network to construct the probability distribution of abnormal behavior category prediction .

[0174] Specifically, this step aims to build the multimodal features based on step S3100. , using a deep neural network classifier to determine the specific type of abnormal behavior, including but not limited to accidental intrusion, tailgating, malicious card swiping, etc. By introducing a temporal convolutional network (TCN), it can accurately identify the types of abnormal behavior that may occur in access control scenarios.

[0175] First, based on multimodal features , introduce the temporal convolutional network TCN to extract temporal features , the specific process formula is as follows:

[0176] ;

[0177] in, Represents the temporal convolutional network (TCN) function, which is used to capture temporal information, such as the occurrence pattern, duration, and dependencies of abnormal behaviors.

[0178] Then, based on the time series features , construct the predicted probability distribution of abnormal behavior categories , the specific process formula is as follows:

[0179] ;

[0180] in, Indicates the total number of abnormal behavior categories; Represents the activation function, which is used to convert the linear output into a probability distribution so that the sum of the predicted probabilities of all categories is 1, which facilitates classification decisions; Represents the classification weight matrix, which is used to transform the time series features Mapping to classification space; Represents the bias term, which is the bias corresponding to each category and is used to help adjust the classification boundary.

[0181] For example, if the system identifies three types of abnormal behavior ,in Indicates mistaken entry, Indicates trailing, Indicates malicious card swiping. The predicted probability distribution of abnormal behavior categories output by the model is: , it means that the model believes that the current behavior has the highest probability of "tailgating" (up to 75%), and the system can respond accordingly.

[0182] Step S3300: predicting probability distribution based on abnormal behavior categories and historical operation and maintenance event data collection , build abnormal event graph structure .

[0183] Specifically, this step aims to predict the probability distribution of abnormal behavior categories based on the constructed step S3200. Combined with the historical operation and maintenance event data obtained in step S3100, the abnormal event graph structure is constructed , to model the spatiotemporal relationships and semantic behavioral associations between abnormal events. By introducing a graph convolutional network (GCN) and a dynamic graph update mechanism, the system can deeply mine the potential correlations between abnormal events and adaptively learn from them, providing structured graph support for subsequent trajectory reconstruction and abnormal behavior tracing.

[0184] The historical operation and maintenance event data set The characteristic representation of the access control abnormal event obtained in step S3100 and video behavior segment feature representation .in Indicates the Historical operation and maintenance event data of abnormal events, which may include information such as access control event features and abnormal segment features extracted from surveillance videos; Indicates that the set has Historical abnormal data records.

[0185] First, define the initial abnormal event graph ,in Represents a node set, each node Corresponding to a historical or current abnormal event; Represents a set of edges, each edge It's an event and The potential connections between events in time, space or semantics. , construct the initial feature matrix , the specific expression formula is as follows:

[0186] ;

[0187] in, Represents the initial feature matrix, including the feature set of all nodes (i.e., abnormal events) in the initial abnormal event graph at the initial moment, with rows representing different events and columns representing features of each dimension; Indicates a OK, The set of real numbers in the column; Indicates the feature dimension, the feature dimension of each event node (e.g., 5-dimensional access control feature + 10-dimensional video feature, a total of 15 dimensions); Indicates the The feature vector of event nodes has the dimension ; Represents vector transpose.

[0188] ;

[0189] in, Indicates the Access control behavior characteristics of each event node (such as illegal card swiping time, card number category); Indicates the Video surveillance behavior characteristics of each event node (such as abnormal trajectory distribution, regional residence time); Represents feature concatenation operation; .

[0190] Subsequently, based on the historical operation and maintenance event data collection Define the initial adjacency matrix based on event correlation , the adjacency matrix Elements Representing an event The specific expression formula for determining whether an edge exists is as follows:

[0191] ;

[0192] The The basis for judging between them may include: 1. Time proximity (e.g. two events occurred within 5 minutes), ; 2. Spatial proximity (such as occurring on the same floor or area); 3. Similarity in the type of abnormal behavior (such as stalking). .

[0193] in, Represents the initial adjacency matrix The , If the item is 1, it means it is associated, otherwise it is 0, there is no association; and Indicates the abnormal event time, which is the , No. The time when an event occurred (which can be a timestamp, minutes, etc.), used to determine time proximity; Indicates the time threshold, used for determining time proximity; and is the predicted probability distribution of abnormal behavior categories constructed in step S3200, which is used to indicate which type of abnormality the event is judged to be (e.g., tailgating, forced entry, etc.); Indicates the consistency of behavior categories, which determines whether the behavior category predictions of two events are consistent. If they are consistent, it means that they may belong to the same type of abnormal behavior.

[0194] Next, we introduce the Dynamic Graph Update mechanism to adapt to new events in real time and dynamically update the adjacency matrix. The specific process formula is as follows:

[0195] ;

[0196] in, Indicates the The adjacency matrix after the first propagation; represents a nonlinear activation function; Indicates that all nodes in the graph are The feature matrix at the iteration; A trainable weight matrix representing graph attention; Represents the feature matrix The transpose operation.

[0197] On this basis, the graph convolutional network GCN is used for information propagation and embedding learning between events. The specific process formula of layer propagation is as follows:

[0198] ;

[0199] in, Indicates the The high-order semantic feature embedding of the layer, after being updated by the graph convolution of this layer, has the shape of , is the number of nodes in the graph, It is The characteristic dimension of the layer; Represents the normalized preprocessed self-loop adjacency matrix, with the shape ,Right now , is the original adjacency matrix, is the identity matrix; express The degree matrix is ​​a diagonal matrix with the shape ; express The inverse square root of the matrix is ​​used to normalize the adjacency matrix to prevent features from being too large or too small during propagation; Indicates the The weight matrix of the layer has the shape .

[0200] Finally, through After the convolution operation of the first layer, the Layer high-level semantic embedding vector , combined with node-based And the final dynamic adjacency matrix , building a structured abnormal event graph structure The specific expression formula is as follows:

[0201] ;

[0202] Step S4000: Based on the abnormal event graph structure , dynamically reconstruct abnormal trajectories, and realize visual control decision support through the three-dimensional scene of the control interface.

[0203] This step aims to collect data on personnel entry and exit events through the access control system, and combine it with spatial mapping to achieve spatial positioning of access control events and abnormal behavior detection, providing basic data support for subsequent personnel trajectory analysis and security management.

[0204] Furthermore, if Figure 5 As shown, step S4000 includes:

[0205] Step S4100: Based on the abnormal event graph structure Access control event spatial positioning collection , reconstruct the movement trajectory of abnormal personnel and output the trajectory point sequence .

[0206] Specifically, this step aims to construct the abnormal event graph structure based on step S3300. The access control event spatial positioning set obtained in step S2200 , reconstruct the movement trajectory of abnormal personnel in time sequence, and finally output a continuous sequence of trajectory points , which provides basic data support for subsequent abnormal behavior chain restoration analysis.

[0207] Aiming at the situation where the access control event location points are sparse or missing, a trajectory interpolation algorithm based on adaptive time interpolation weighting is adopted to consider the distance difference between the location values ​​at different time points and realize the smooth completion of the missing segments of the trajectory.

[0208] First, for the current time point to be estimated , according to the historical trajectory points , use the weighted average interpolation function to calculate the corresponding spatial position , the specific process formula is as follows:

[0209] ;

[0210] in, Indicates a time point The spatial position is the interpolation at any time The spatial position of the personnel on (i.e., the interpolation point); Indicates the interpolation target time point, which is the time point at which the spatial position needs to be estimated; The sum is used to perform weighted calculations on all historical points (numbered ); Represents the interpolation weight, which is the Historical trajectory record points to target time point The degree of influence of interpolation; Represents the spatial coordinates of a known time point, which is the The coordinate values ​​of each historical track record point.

[0211] ;

[0212] in, Indicates a known time point, which is the The timestamp corresponding to each historical trajectory record point; Represents the interpolation weight index, which is used to control the sensitivity of time distance to weight. The larger the value, the smaller the weight of distant points. The summation symbol is used to sum all trajectory points and normalize each , ensuring the total weight is 1; Indicates the total number of historical trajectory points.

[0213] Furthermore, in order to adapt to the changes in behavioral characteristics of different abnormal people, an adaptive interpolation weight index is proposed Dynamic adjustment mechanism. Adaptive interpolation weight index The specific process formula is as follows:

[0214] ;

[0215] in, Indicates the initial interpolation weight index benchmark value; The coefficient that controls the adjustment amplitude; The standard deviation of the trajectory's historical velocity (a volatility indicator) is used to quantify the trajectory's volatility (i.e., whether the trajectory is stable or fluctuates dramatically).

[0216] Volatility indicator based on historical velocity of trajectory Automatic adjustment is performed. If it is very large, it means that the speed of the person changes dramatically, the behavior jumps, and the movement is irregular, so a smaller interpolation weight index is set. , to improve interpolation responsiveness; if If it is very small, it means the trajectory is stable and linear, so a larger interpolation weight index is set. , increase local weight. Volatility index The specific process formula is as follows:

[0217] ;

[0218] ;

[0219] in, Indicates shared The speed calculation interval of each trajectory segment (i.e., the interval between two trajectory points is a trajectory segment); Represents an index variable; From the first track to the Sum operation of segment trajectories; express The average velocity of the trajectory segment; Represents trajectory points and The Euclidean distance between Represents the time interval between two adjacent trajectory points; represents the average velocity of all trajectory segments;

[0220] Finally, a continuous sequence of trajectory points is output , used to represent the continuous trajectory points interpolated on the complete time axis (including missing segments). Indicates the Each trajectory point records the corresponding timestamp; Indicates the The timestamp corresponding to each track point record The corresponding interpolation estimated spatial position, that is, in time The spatial coordinates of the personnel are calculated by the interpolation algorithm.

[0221] Step S4200: Based on the trajectory point sequence , a trajectory prediction algorithm based on the fusion of Kalman filter and long short-term memory network LSTM is used to obtain the predicted trajectory curve .

[0222] Specifically, this step is based on the trajectory point sequence output in step S4100 , uses a trajectory prediction algorithm based on the fusion of Kalman filter and long short-term memory network (LSTM), which outputs a continuous and high-confidence predicted trajectory curve by smoothing the trajectory and predicting missing points. .

[0223] In the specific process, firstly based on the trajectory point sequence Perform Kalman filter trajectory smoothing. The Kalman filter is an optimal recursive filtering algorithm that can be used for dynamic system state estimation. In this application, it is used to estimate the true spatial position of the target individual at each time point and filter out noise and mutation points in the trajectory. The specific process formula is as follows:

[0224] ;

[0225] in, Indicates the The timestamp corresponding to each track point record Corresponding to the spatial position processed by Kalman filter; Indicates the spatial position corresponding to the last moment after Kalman filtering; ' represents the state transition matrix, which is used to describe the transition from arrive The state evolution of Represents the Kalman Gain matrix, which is used to weigh the credibility between the measured value and the predicted value; Indicates the current time point Observed values ​​of Represents the observation matrix, which is used to map the state to the observable space.

[0226] Next, for situations where there are continuous missing segments or complex motion patterns, a long short-term memory network (LSTM) is introduced as an auxiliary prediction module to enhance the Kalman filter state and learn the dynamic patterns of the time series in the historical trajectory. The specific process formula is as follows:

[0227]

[0228] ;

[0229] in, Indicates the timestamp corresponding to the k-th trajectory point record The corresponding spatial position based on the fusion of Kalman filter and long short-term memory network LSTM; Indicates the weight of the LSTM prediction item in the current state update; Indicates the The timestamp corresponding to each track point record The corresponding LSTM predicted spatial position; Represents the LSTM model function obtained by training based on historical time series trajectory points.

[0230] Finally, the smooth and continuous predicted trajectory curve predicted by the output fusion model is .

[0231] Step S4300: Based on the predicted trajectory curve , Spatial positioning set of access control abnormal events and abnormal behavior category prediction probability distribution , perform three-dimensional trajectory mapping visualization and abnormal brightness display.

[0232] Specifically, this step is based on the predicted trajectory curve output in step S4200 , the access control abnormal event spatial positioning set constructed in step S2300 And the abnormal behavior category prediction probability distribution of step S3200 , the trajectory data is mapped into the digital twin 3D simulation environment, and the color coding mechanism is combined to visually highlight the abnormal vision, so as to realize the spatial visualization of the target individual's movement path and the spatiotemporal tracing of risk events.

[0233] First, the predicted trajectory curve Each trajectory point Map it to the digital twin 3D simulation scene coordinate system and construct the corresponding 3D space trajectory. The specific process formula is as follows:

[0234] ;

[0235] in, Indicates the The spatial position coordinates of each trajectory point after mapping in the digital twin 3D environment; Represents the mapping transformation function that maps each trajectory point to the 3D visualization scene.

[0236] Then, all By timestamp Connect them sequentially to form dynamic trajectory lines, and play them in a time-driven manner in a 3D visualization platform (such as one based on Cesium or Unity) to reproduce the movement process of the target individual, realizing dynamic playback and spatial restoration of dynamic behaviors in real scenes in a virtual environment.

[0237] Furthermore, based on the spatial positioning set of access control abnormal events and the predicted probability distribution of abnormal behavior categories , for each predicted trajectory curve Each trajectory point Rendering color Make detailed judgments to achieve a visual expression of confidence risk. The specific color rendering rules and trajectory risk levels are as follows:

[0238] Set probability distribution thresholds, including upper threshold, middle threshold, and lower threshold;

[0239] , then the risk level High-risk warning. ;

[0240] ,but This is a medium risk warning. ;

[0241] ,but Low risk reminder. ;

[0242] ,but Normal, .

[0243] In this embodiment, the upper threshold is set to 0.85, the middle threshold is set to 0.6, and the lower threshold is set to 0.4.

[0244] in, Indicates the Risk level of each trajectory point; Indicates the The rendering color of each trajectory point; Indicates the The predicted probability distribution of abnormal behavior categories for each trajectory point.

[0245] Step S5000: Combine the access control abnormal event spatial positioning set , abnormal behavior category prediction probability distribution and trajectory risk level , determine the hierarchical early warning linkage response control mechanism and issue closed-loop linkage control instructions to the controlled devices in the target area based on it, collect the execution status of the controlled devices in real time as status feedback, and generate a closed-loop access control abnormal event log including device execution feedback and response status.

[0246] Specifically, this step is based on the access control abnormal event spatial positioning set constructed in step S2300 , the predicted probability distribution of abnormal behavior category output in step S3200 , trajectory risk level This method implements a complete closed-loop control process. Its core is to: first, determine a hierarchical early warning linkage response control mechanism based on risk analysis results; second, issue closed-loop linkage control instructions to controlled devices in the physical world based on this mechanism; then, retrieve the execution status of these devices in real time as status feedback; and finally, generate a closed-loop access control anomaly event log containing device execution feedback and response status. Through this series of operations, this method achieves a complete implementation from intelligent analysis to closed-loop control, effectively enhancing the proactive security defense capabilities of smart computer rooms.

[0247] Furthermore, if Figure 6 As shown, step S5000 includes:

[0248] Step S5100: Setting the trajectory risk level A hierarchical early warning linkage response control mechanism.

[0249] This step makes a real-time point-by-point judgment on all track points in the trajectory and Automatically trigger the preset, differentiated, hierarchical early warning linkage response control mechanism to ensure the timeliness and accuracy of the system response. The specific rules are as follows:

[0250] First, high-risk warning and mandatory intervention response. High risk warning, When a high-risk alert occurs, the system immediately triggers a Level 1 high-risk warning event. Not only does it trigger a Level 1 red alert notification, it also automatically generates a target device control parameter set and issues closed-loop linkage control instructions to the controlled devices in the target area, proactively intervening in the situation.

[0251] The closed-loop linkage control instructions include but are not limited to driving at least one of the following operations: 1. Forcibly locking the relevant access control actuator, changing the door lock to a closed or no-entry state, and preventing the target person from further entering; 2. Instructing the pan / tilt head in the video surveillance system to automatically focus on and continuously track abnormal target persons to ensure uninterrupted recording of key images; 3. Activating the security warning lights and sound and light alarms in the target area, switching to a high-intensity alarm mode, warning the parties involved and reminding surrounding people.

[0252] Second, medium risk warning and manual decision response. Medium risk warning, When an abnormal situation occurs, the system immediately triggers a secondary orange medium-risk warning, highlights the abnormal trajectory and related information on the management control interface, and pushes a notification. The manager can choose whether to respond further, such as issuing inspection instructions.

[0253] Third, low-risk attention and silent recording response. Low risk warning or normal, When the system does not actively trigger an alarm, it will still silently record the relevant trajectory points and their abnormal prediction probabilities as a data source for potential risk trend analysis.

[0254] Step S5200: Generate a closed-loop access control abnormal event log including device execution feedback and response status.

[0255] Specifically, on the basis of completing the hierarchical early warning, this step further realizes the institutional recording and archiving management of abnormal events by automatically generating a closed-loop access control abnormal event log. Its key feature is that the log content contains physical-level device execution feedback and system-level response status, thereby building a complete and irrefutable traceability evidence chain.

[0256] The access control abnormal event log adopts a structured data format with a unified format and a globally unique identifier, including seven key information elements: First, the event number: as the unique identifier of each access control abnormal event log, it facilitates efficient indexing and precise positioning of the system. The number is automatically generated by "timestamp + spatial location code + behavior category code + serial number" to ensure the global uniqueness and traceability of the event record. For example, the number "EVT20250317-A13-B05" indicates that the event occurred on March 17, 2025, in area A13 (such as a certain access control floor area of ​​the smart computer room), and is the fifth Class B abnormal event identified in the area that day (where "B" represents a certain abnormal category, such as tailgating), with good uniqueness and traceability. Second, the event time: records the specific time when the abnormal trajectory point is determined to be an abnormal behavior by the system, using the international standard ISO8601 format as an important time benchmark for behavior pattern analysis and response timeliness evaluation. Third, the spatial position: contains the relative coordinate information of the trajectory point in the building space. The system can accurately identify the area where the abnormal event occurred in the smart computer room or visual map by using the ID number of the access control point to support fine-grained regional risk analysis and precise scheduling response. The fourth is the abnormal category and risk level: including the abnormal behavior categories identified by the system (such as tailing, forced entry, long stay, malicious card swiping, etc.), and record the corresponding abnormal behavior category prediction probability distribution. and trajectory risk level (high / medium / low), providing information management personnel with a reference for determining the nature of anomalies and prioritizing responses. Fifth, abnormal warning level records: This records in detail whether the system triggers the warning mechanism, and if so, the warning level. Sixth, device execution feedback: This records the execution results retrieved from each controlled device, such as the door lock status "confirmed locked," the camera status "tracking enabled," and the alarm status "sounded." Seventh, response closed-loop status: This automatically determines and records the handling status of this incident based on integrated instructions and feedback information, such as "successfully closed," "execution failed," or "awaiting manual intervention."

[0257] For example, Figure 7The figure shows an example of an access control exception event log in a visual monitoring scenario for a smart computer room. This log displays five key event information items in a visual dashboard, including the event number, event time, spatial location, exception category and risk level, exception warning record, device execution feedback, and response closed-loop status. For example, event number EVT20250317-A13-B01, recorded at 08:22:57 on March 17, 2025, occurred in area A13, access control number D-17, and three-dimensional coordinates (12.3, 8.7, 0.0). The system determined the exception category to be "tailgating entry," with a confidence level of 0.92 and a high risk level, triggering a high-level warning. Furthermore, the "Device Execution Feedback" field in the log further records the actual results of the system's coordinated control, specifically: "Access control D-17: Status confirmed locked; Camera: Focus tracking activated; Audible and visual alarm: Activated sounding." Its "response closed-loop status" field automatically determines and records it as "successfully closed" based on the above-mentioned successful instruction execution and status recall, thus forming a complete traceability evidence chain covering the entire process of perception, decision-making, action and feedback.

[0258] Specifically, this step has the following technical advantages: First, it supports the transformation of computer room security management from static audit to dynamic perception: in the past, computer room security management relied more on regular manual review of access control records and video materials, which resulted in a large number of blind spots and delays. The present invention gives the system real-time dynamic perception and intelligent archiving capabilities, and promotes the transformation of the operation and maintenance system from "post-audit" to "in-process perception + immediate response", which is in line with the future trend of data centers evolving towards intelligent and autonomous security management. Second, it enhances the ability to prevent and control traffic risks in important protection areas: in high-security areas, the traditional mechanism based on "door identification + log recording" cannot meet the system tracking requirements for intrusion paths, behavior patterns and subsequent evidence chains. The present invention establishes a complete semantic chain of "time + space + behavior + response" for abnormal behavior through the automatic archiving of high-confidence access control abnormal event logs and spatial annotation capabilities, so that managers have a clear, traceable and controllable decision-making basis in emergency response. The third is the structured expression and cross-platform linkage tracking of abnormal access control behaviors: through the unified format of the access control abnormal event log standard, multi-angle and quantifiable recording of access control behaviors from data dimension, spatial dimension to time dimension is achieved, enabling the video surveillance system and the dispatch response platform to accurately index, quickly replay, and intelligently compare, supporting the full process analysis and closed-loop control of abnormal events.

[0259] Example 2

[0260] Based on Example 1, this embodiment provides a multi-dimensional dynamic monitoring and management system for smart computer rooms based on digital twins, such as Figure 8 As shown, including:

[0261] Digital twin space modeling module: used to collect 3D point cloud collection of the computer room space With RGB image collection , build a digital twin space that represents the physical state of the computer room and supports monitoring and control, and establish a spatial mapping table between the digital twin space and the physical space ;

[0262] Access control abnormal event positioning module: based on spatial mapping table , map the personnel entry and exit event data generated by the access control system to the corresponding spatial position in the digital twin space, conduct continuous real-time status detection of personnel behavior to identify abnormal event behavior, and build a spatial positioning set of access control abnormal events ;

[0263] Abnormal behavior recognition and abnormal event graph construction module: based on the spatial positioning set of access control abnormal events , perform abnormal behavior identification and correlation analysis, and build abnormal event graph structure .

[0264] Abnormal trajectory reconstruction and 3D visualization control module: Based on the abnormal event graph structure, it dynamically reconstructs abnormal trajectories and implements visual control decision support through the 3D scene of the control interface.

[0265] Linkage warning control and log generation module: It is used to combine the spatial positioning set of access control abnormal events, the predicted probability distribution of abnormal behavior categories and the trajectory risk level to determine the hierarchical warning linkage response control mechanism and issue closed-loop linkage control instructions to the controlled devices in the target area accordingly. It also retrieves the execution status of the controlled devices in real time as status feedback and generates a closed-loop access control abnormal event log containing the device execution feedback and response status.

[0266] The parts of the above technical solutions provided in the embodiments of the present application that are consistent with the implementation principles of the corresponding technical solutions in the prior art are not described in detail to avoid excessive redundancy.

[0267] The above-described specific embodiments further illustrate the objectives, technical solutions, and beneficial effects of the present invention. It should be understood that the above description is merely a specific embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A multi-dimensional dynamic monitoring and management method for smart computer rooms based on digital twins, characterized by: include: Collect a 3D point cloud set and RGB image set of the computer room space, build a digital twin space that represents the physical state of the computer room and supports monitoring and control, and establish a spatial mapping table between the digital twin space and the physical space; Based on the spatial mapping table, the personnel entry and exit event data generated by the access control system is mapped to the corresponding spatial location in the digital twin space. Continuous real-time status detection of personnel behavior is performed to identify abnormal event behavior and construct a spatial location set of access control abnormal events. Based on the spatial location set of access control abnormal events, perform abnormal behavior identification and correlation analysis, and build an abnormal event graph structure; Based on the abnormal event graph structure, the abnormal trajectory is dynamically reconstructed, and the three-dimensional scene of the control interface is used to realize visual control decision support; Combined with the spatial positioning set of access control abnormal events, the predicted probability distribution of abnormal behavior categories and the trajectory risk level, a hierarchical early warning linkage response control mechanism is determined and a closed-loop linkage control instruction is issued to the controlled devices in the target area based on this mechanism. The execution status of the controlled devices is retrieved in real time as status feedback, and a closed-loop access control abnormal event log containing the device execution feedback and response status is generated.

2. The multi-dimensional dynamic monitoring and management method for a smart computer room based on digital twins according to claim 1 is characterized in that: The method for establishing the space mapping table includes: Construct multimodal fusion data based on 3D point cloud sets and RGB image sets; Based on the multimodal fusion data set and combined with the building information model, a spatial index structure is established; Based on the device collection, the spatial model with semantic labels and the spatial index structure, spatial mapping is performed, a temporal index update mechanism is introduced, and a spatial mapping table is established; The temporal index update mechanism adopts KD-Tree to dynamically maintain the spatial index structure.

3. The multi-dimensional dynamic monitoring and management method for a smart computer room based on digital twins according to claim 2 is characterized in that: The method for establishing the spatial index structure includes: Based on the multimodal fusion data set, a voxel gridding algorithm is used to perform structured conversion on the three-dimensional point cloud data to obtain a three-dimensional point cloud voxel set; Combine the structural component information provided by the building information model with the 3D point cloud voxel set for spatial alignment, and perform a multimodal semantic segmentation process to obtain a spatial model with semantic labels; Construct a spatial index structure based on a spatial model with semantic labels.

4. The multi-dimensional dynamic monitoring and management method for a smart computer room based on digital twins according to claim 1 is characterized in that: The method for constructing the access control abnormal event spatial positioning set includes: Obtain the original access control event set and construct a high-quality access control event set based on the original access control event set; Based on the high-quality access control event set and the spatial mapping table, the access control event spatial positioning set is obtained; Based on the access control event spatial positioning set, a spatial positioning set of access control abnormal events is constructed.

5. The multi-dimensional dynamic monitoring and management method for a smart computer room based on digital twins according to claim 1 is characterized in that: The method for constructing the abnormal event graph structure includes: Based on the spatial positioning set of access control abnormal events, the feature representation of access control abnormal events and the feature representation of video behavior segments are obtained to construct multimodal features; Based on multimodal features, a temporal convolutional network is introduced to construct the probability distribution of abnormal behavior category prediction; Based on the predicted probability distribution of abnormal behavior categories and the historical operation and maintenance event data set, an abnormal event graph structure is constructed.

6. The multi-dimensional dynamic monitoring and management method for a smart computer room based on digital twins according to claim 5 is characterized in that: The method for constructing the abnormal event graph structure further includes: Define the initial abnormal event graph and construct the initial feature matrix based on each event node; Based on the historical operation and maintenance event data set, the initial adjacency matrix is ​​defined according to the event correlation; The elements of the initial adjacency matrix represent the determination of whether there is an edge between events; Introducing a dynamic graph update mechanism to adapt to new events in real time and dynamically update the adjacency matrix; Using graph convolutional networks for information propagation and embedding learning between events, we get Layer-wise high-level semantic feature embedding; pass After the convolution operation of the first layer, the The high-level semantic embedding vector is combined with the node set and the final dynamic adjacency matrix to construct a structured abnormal event graph structure.

7. The multi-dimensional dynamic monitoring and management method for a smart computer room based on digital twins according to claim 1 is characterized in that: The steps of dynamic reconstruction of abnormal trajectories and three-dimensional scene visualization include: Map each trajectory point of the predicted trajectory curve to the digital twin 3D simulation scene coordinate system to construct the corresponding 3D space trajectory line; Connect all 3D space trajectory points in the order of timestamps to form a dynamic trajectory line, and play it in a time-driven manner on the 3D visualization platform; Based on the spatial location set of access control abnormal events and the predicted probability distribution of abnormal behavior categories, the rendering color of each trajectory point of each predicted trajectory curve is determined; Determining the rendering color of each trajectory point of each predicted trajectory curve includes: Set probability distribution thresholds, including upper threshold, middle threshold, and lower threshold; If the trajectory point belongs to the access control abnormal event spatial positioning set and the abnormal behavior category prediction probability distribution is greater than the upper threshold, the risk level is high risk warning and the rendering color is red; If the trajectory point belongs to the access control abnormal event spatial positioning set and the abnormal behavior category predicted probability distribution is greater than the middle threshold and less than or equal to the upper threshold, the risk level is medium risk prompt and the rendering color is orange; If the trajectory point belongs to the access control abnormal event spatial positioning set and the abnormal behavior category predicted probability distribution is greater than or equal to the lower threshold and less than or equal to the middle threshold, the risk level is low risk and the rendering color is yellow; If the trajectory point does not belong to the access control abnormal event spatial positioning set, the risk level is normal and the rendering color is green.

8. The multi-dimensional dynamic monitoring and management method for a smart computer room based on digital twins according to claim 7 is characterized in that: When the trajectory risk level is high risk warning and the rendering color is red, the system immediately triggers a level one high risk warning event; when the trajectory risk level is medium risk warning and the rendering color is orange, the system immediately triggers a level two medium risk warning event; When the trajectory risk level is low risk warning or normal, and the rendering color is yellow or green, the system does not actively trigger an alarm.

9. A multi-dimensional dynamic monitoring and management system for a smart computer room based on digital twins, which is used to implement the multi-dimensional dynamic monitoring and management method for a smart computer room based on digital twins according to any one of claims 1 to 8, characterized in that: The system comprises: Digital twin space modeling module: This module collects 3D point cloud sets and RGB image sets of the computer room space, constructs a digital twin space that represents the physical state of the computer room and supports monitoring and control, and establishes a spatial mapping table between the digital twin space and the physical space. Access control abnormal event location module: Based on the spatial mapping table, the personnel entry and exit event data generated by the access control system is mapped to the corresponding spatial location in the digital twin space. Continuous real-time status detection of personnel behavior is performed to identify abnormal event behavior and construct a spatial location set of access control abnormal events. Abnormal behavior identification and abnormal event graph construction module: Based on the spatial positioning set of access control abnormal events, it performs abnormal behavior identification and correlation analysis and constructs the abnormal event graph structure; Abnormal trajectory reconstruction and 3D visualization control module: Based on the abnormal event graph structure, it dynamically reconstructs abnormal trajectories and provides visual control decision support through the 3D scene of the control interface; Linkage warning control and log generation module: It is used to combine the spatial positioning set of access control abnormal events, the predicted probability distribution of abnormal behavior categories and the trajectory risk level to determine the hierarchical warning linkage response control mechanism and issue closed-loop linkage control instructions to the controlled devices in the target area accordingly. It also retrieves the execution status of the controlled devices in real time as status feedback and generates a closed-loop access control abnormal event log containing the device execution feedback and response status.

Images