Intelligent building three-dimensional security situation awareness method and system based on artificial intelligence

Abstract

The invention provides an intelligent building three-dimensional security situation perception method and system based on artificial intelligence, and the method comprises the steps: obtaining real-time three-dimensional security collection data of intelligent building three-dimensional security collection equipment, and fusing the real-time three-dimensional security collection data with preset building security reference situation data to generate security situation seed data; inputting the seed data into an artificial intelligence derivation module to generate a multi-path link set; inputting the multi-path link set and real-time three-dimensional security collection data into a three-dimensional security situation awareness artificial intelligence model to generate an initial three-dimensional security situation; a final three-dimensional security and protection situation is generated through a built-in cross-link verification unit of the model; generating a security control instruction based on the final situation, and collecting execution feedback data to be reversely input into the artificial intelligence fusion module to calibrate seed data generation parameters. According to the invention, the three-dimensional security situation of the intelligent building can be sensed comprehensively and accurately, and the adaptability and accuracy of the security system are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of artificial intelligence, in particular to a three-dimensional security situation awareness method and system for intelligent buildings based on artificial intelligence. BACKGROUND

[0002] In the field of intelligent building security, traditional security situation awareness methods mainly rely on two-dimensional plane data and simple rule matching mechanisms. Two-dimensional plane data can only present the layout and status of building security facilities in the plane, and cannot fully reflect the complex three-dimensional spatial structure inside the building and the dynamic correlation between different security scenarios. For example, for security monitoring of different floors in a multi-story building, two-dimensional data cannot accurately present the linkage relationship of security facilities between floors and the change of security situation when personnel move between different floors. At the same time, when processing security data, traditional methods usually use fixed rules and preset thresholds for judgment, lacking flexibility in adapting to real-time data and complex scenarios. When the security scenario changes or new abnormal situations occur, traditional methods are difficult to quickly and accurately adjust security strategies, resulting in serious impact on the accuracy and timeliness of security situation awareness. In addition, after generating security control instructions, traditional methods lack feedback and calibration mechanisms for instruction execution effects, and cannot optimize the security situation awareness model according to actual execution, further limiting the performance improvement of the security system. SUMMARY

[0003] In view of the above-mentioned problems, in combination with the first aspect of the present application, the embodiments of the present application provide a three-dimensional security situation awareness method for intelligent buildings based on artificial intelligence, which comprises: Obtaining real-time three-dimensional security collection data output by an intelligent building three-dimensional security collection device, accessing preset building security reference situation data, performing fusion processing, and generating security situation seed data through an artificial intelligence fusion module; Inputting the security situation seed data into an artificial intelligence derivation module, performing link derivation processing, generating multiple sets of security correlation links respectively adapted to different security scenarios, and integrating to form a multi-path link set; Synchronously inputting the multi-path link set and the real-time three-dimensional security collection data into a three-dimensional security situation awareness artificial intelligence model, performing situation generation processing through a multi-link linkage unit of the model, and generating an initial three-dimensional security situation; Inputting the initial three-dimensional security situation into a cross-link verification unit built-in the model, performing verification processing, eliminating link conflict data, and generating a final three-dimensional security situation; Based on the final three-dimensional security situation, generating security control instructions, issuing the instructions to a building control terminal, collecting security state feedback data after execution of the control instructions, and inputting the security state feedback data into the artificial intelligence fusion module in reverse, and performing calibration processing of the security situation seed data generation parameters.

[0004] In still another aspect, the embodiments of the present application also provide an intelligent building three-dimensional security situation awareness system based on artificial intelligence, comprising a processor, a machine-readable storage medium, the machine-readable storage medium and the processor are connected, the machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to realize the above-mentioned method.

[0005] Based on the above aspects, the embodiments of the present application can fully utilize the spatial information of three-dimensional data, comprehensively and accurately reflect the three-dimensional security situation inside the building, and overcome the limitations of traditional two-dimensional data in spatial expression by acquiring real-time three-dimensional security collection data output by the intelligent building three-dimensional security collection device, and combining and processing the preset building security reference situation data to generate security situation seed data. The security situation seed data is input into the artificial intelligence derivation module to generate multiple sets of security-related links adapted to different security scenes, and the multiple-path link set is formed by integration, so that the security system can flexibly adjust the strategy according to different security scenes, and the adaptability and flexibility of the security situation awareness are improved. The multi-path link set and the real-time three-dimensional security collection data are synchronously input into the three-dimensional security situation awareness artificial intelligence model, the initial three-dimensional security situation is generated through the multi-link linkage unit, the link conflict data is removed through the cross-link verification unit, and the final three-dimensional security situation is generated, which further improves the accuracy and reliability of the security situation awareness. Based on the final three-dimensional security situation, security control instructions are generated, and security state feedback data after execution of the control instructions is collected, which is reversely input into the artificial intelligence fusion module for calibration processing, forming a closed-loop security situation awareness and optimization system. The system can continuously adjust the parameters of the security situation awareness model according to the actual execution effect, improve the performance and accuracy of the model, realize real-time, accurate and dynamic awareness of the security situation of the intelligent building, and effective control, and provide a strong guarantee for the safe operation of the intelligent building. BRIEF DESCRIPTION OF DRAWINGS

[0006] Figure 1 is the execution flow diagram of the intelligent building three-dimensional security situation awareness method based on artificial intelligence provided by the embodiments of the present application.

[0007] Figure 2 is the schematic diagram of the exemplary hardware and software components of the intelligent building three-dimensional security situation awareness system based on artificial intelligence provided by the embodiments of the present application. DETAILED DESCRIPTION

[0008] The present application will be specifically described below in conjunction with the drawings of the specification, Figure 1It is a flowchart of an intelligent building three-dimensional security situation awareness method based on artificial intelligence provided by an embodiment of the present application, and the intelligent building three-dimensional security situation awareness method based on artificial intelligence will be described in detail below.

[0009] Step S110: Real-time three-dimensional security collection data output by the intelligent building three-dimensional security collection device is acquired, the preset building security reference situation data is accessed, fusion processing is performed, and security situation seed data is generated by an artificial intelligence fusion module.

[0010] In this embodiment, the intelligent building three-dimensional security collection device includes three-dimensional cameras deployed at entrances and exits of each floor of the building, three-dimensional sensors in elevators, infrared three-dimensional detection devices of fire passages, etc., and the real-time three-dimensional security collection data output by these devices covers three-dimensional position information of personnel in the building, movement trajectory information of personnel, operation state information of devices, etc. The preset building security reference situation data is pre-stored in the building security data center and contains personnel density reference range in the normal operation state of the building, device normal operation parameter reference range, safety state reference description of each area, etc. The artificial intelligence fusion module is a software module deployed in the building security data processing server, and the artificial intelligence fusion module has the function of fusion processing of data from different sources. When performing fusion processing, firstly, the real-time three-dimensional security collection data and the preset building security reference situation data are uniformly converted in data format to ensure that the two kinds of data are compatible in format, and then the two kinds of data in uniform format are input into the artificial intelligence fusion module. The artificial intelligence fusion module matches and integrates the features in the two kinds of data through internal feature extraction and correlation algorithms, and finally generates security situation seed data, which contains the fusion feature information of the real-time collection data and the reference data.

[0011] Step S111: Extract security scene features in real-time three-dimensional security collection data to generate scene feature data.

[0012] In this embodiment, the security scene features in the real-time three-dimensional security collection data include scene type information embodied in the data, such as entrance and exit scene, elevator scene, fire passage scene, etc., and also include personnel quantity information in the scene, behavior feature information of personnel, working state information of devices, etc. When extracting these features, the real-time three-dimensional security collection data is analyzed segment by segment through a data feature extraction algorithm to identify information that can reflect the characteristics of the security scene, and the information is encoded according to a preset feature coding rule to generate scene feature data. The scene feature data is stored in the form of a feature vector, and each vector element corresponds to a security scene feature.

[0013] Step S112: Split the reference features in the preset building security reference situation data to generate a reference feature set.

[0014] In this embodiment, the preset building security reference situation data includes reference features such as personnel density reference features, equipment operation parameter reference features, and regional security state reference features. When splitting these reference features, the reference situation data is classified and processed according to the types of the features, reference features belonging to the same type are grouped together, reference features of different types are processed respectively, and finally a reference feature set is generated, each element in the reference feature set corresponds to a type of reference feature.

[0015] Step S113: input the scene feature data and the reference feature set into a feature alignment unit of the artificial intelligence fusion module, perform feature alignment processing, and generate a feature alignment result.

[0016] In this embodiment, the feature alignment unit of the artificial intelligence fusion module is a functional sub-module inside the artificial intelligence fusion module, which has the ability to align features of different sources. After inputting the scene feature data and the reference feature set into the unit, the unit first identifies the feature types of the features in the scene feature data and the reference feature set, determines which features belong to the same type, then adjusts the numerical ranges of the features of the same type to make the features of the same type consistent in numerical range, and finally matches the positions of the adjusted features to generate a feature alignment result, which reflects the correspondence between the features in the scene feature data and the reference feature set.

[0017] Step S114: based on the feature alignment result, associate the matching features in the scene feature data and the reference feature set to generate a feature association pair.

[0018] In this embodiment, based on the correspondence between the features in the scene feature data and the reference feature set reflected in the feature alignment result, each feature in the scene feature data is associated with the corresponding matching feature in the reference feature set. The association method is to establish a mapping relationship between the features, each mapping relationship corresponds to a feature association pair, and the feature association pair contains one feature in the scene feature data and the corresponding matching feature in the reference feature set.

[0019] Step S115: input the feature association pair into a seed generation unit of the artificial intelligence fusion module, perform initial seed generation processing, and generate initial seed data.

[0020] In this embodiment, the seed generation unit of the artificial intelligence fusion module is a functional sub-module responsible for generating initial seed data inside the artificial intelligence fusion module. After the feature association pair is input into the unit, the unit performs fusion processing on the scene features and reference features in the feature association pair through an internal feature fusion algorithm. The importance of the two types of features is evaluated during the fusion process, different weights are assigned according to the evaluation results, and then the weighted features are integrated to generate initial seed data, which contains the fused feature information.

[0021] Step S116: associate the initial seed data with the collection scene information corresponding to the real-time three-dimensional security collection data to generate scene association data.

[0022] In this embodiment, the collection scene information of the real-time three-dimensional security collection data includes specific location information of data collection, collection time information, and number information of collection equipment. When associating the initial seed data with these collection scene information, a corresponding relationship between the initial seed data and the collection scene information is established through a data association algorithm, each feature in the initial seed data is bound with the corresponding collection scene information, and scene association data is generated, which contains the initial seed data and the corresponding collection scene information.

[0023] Step S117: based on the scene association data, perform feature weight adjustment processing of the initial seed data to generate security situation seed data.

[0024] In this embodiment, when performing feature weight adjustment processing based on the scene association data, first, analyze the collection scene information in the scene association data to determine the importance of the scene, such as the importance of the entrance scene being higher than that of the elevator interior scene. Then, according to the importance of the scene, determine the weight adjustment proportion of each feature in the initial seed data, adjust the feature weight in the initial seed data, and the adjusted feature weight can more accurately reflect the security situation under the scene. Finally, the initial seed data with adjusted weight is taken as the security situation seed data.

[0025] Step S118: store the security situation seed data to the seed database of the artificial intelligence fusion module, and synchronize to the input cache unit of the artificial intelligence derivation module.

[0026] In this embodiment, the seed database of the artificial intelligence fusion module is a database specially used for storing security situation seed data. The database is deployed in a building security data storage server and has data storage and query functions. After storing the security situation seed data in the database, the data is synchronized to the input cache unit of the artificial intelligence derivation module through a data synchronization interface. The input cache unit is a storage area inside the artificial intelligence derivation module for temporarily storing input data.

[0027] Step S1171: Parse the collected scene attributes in the scene association data and generate scene attribute data.

[0028] In this embodiment, the scene attributes collected in the scene association data include scene type attributes, such as entrance / exit scenes, elevator scenes, etc.; scene spatial attributes, such as the three-dimensional spatial coordinate range of the scene; and scene functional attributes, such as the passage function and the stay function of the scene. When parsing these attributes, the scene association data is analyzed using an attribute parsing algorithm to extract the scene attribute information. This information is then encoded according to a preset attribute encoding rule to generate scene attribute data. This data is stored in the form of attribute vectors, with each vector element corresponding to a scene attribute.

[0029] Step S1172: Call the scene weight rule library built into the artificial intelligence fusion module, match the feature weight adjustment rules corresponding to the scene attribute data, and generate weight rule matching results.

[0030] In this embodiment, the scene weight rule library built into the AI ​​fusion module stores feature weight adjustment rules corresponding to different scene attributes. For example, the feature weight adjustment rule for the entrance / exit scene is to increase the weight of the personnel location feature, and the feature weight adjustment rule for the elevator scene is to increase the weight of the equipment operating status feature. When calling this scene weight rule library, scene attribute data is input into the rule library. The rule library uses its internal rule matching algorithm to match the scene attribute data with the rules in the rule library to determine the feature weight adjustment rule corresponding to the scene attribute data, generating a weight rule matching result. This weight rule matching result contains the matched feature weight adjustment rule.

[0031] Step S1173: Based on the weight rule matching results, determine the weight adjustment direction of each feature in the initial seed data and generate weight direction data.

[0032] In this embodiment, the feature weight adjustment rules in the weight rule matching results clarify the weight adjustment direction of each feature in the initial seed data, such as increasing or decreasing the weight of a certain feature. Based on these rules, each feature in the initial seed data is analyzed to determine its corresponding weight adjustment direction. These adjustment directions are then arranged according to the feature order to generate weight direction data. This weight direction data is stored in the form of a direction vector, with each vector element corresponding to the weight adjustment direction of a feature.

[0033] Step S1174: Associate the scene importance descriptions in the associated weight direction data and the scene association data to generate weight adjustment association data.

[0034] In this embodiment, the scene importance description in the scene association data is a textual description of the importance of the collected scene, such as "the entrance / exit scene is an important security area of ​​the building, with a high importance level." When associating the weight direction data with the scene importance description, a data association algorithm is used to establish a correspondence between each adjustment direction in the weight direction data and the scene importance description. The adjustment direction is then bound to the corresponding scene importance description to generate weight adjustment association data, which includes the weight direction data and the corresponding scene importance description.

[0035] Step S1175: Based on the weight adjustment correlation data, determine the weight adjustment range of each feature and generate feature weight adjustment data.

[0036] In this embodiment, when determining the feature weight adjustment range based on the weight adjustment correlation data, the importance level in the scene importance description is first analyzed, such as high, medium, and low. Then, the corresponding weight adjustment range is determined according to the importance level. For example, when the importance level is high, the weight adjustment range is a larger interval; when the importance level is medium, the weight adjustment range is a medium interval; and when the importance level is low, the weight adjustment range is a smaller interval. Finally, the adjustment direction in the weight direction data is combined to determine the specific weight adjustment range of each feature within the corresponding range, generating feature weight adjustment data. This feature weight adjustment data is stored in the form of an amplitude vector, with each vector element corresponding to the weight adjustment range of a feature.

[0037] Step S1176: Adjust the data according to the feature weights, adjust the feature weights of the initial seed data, and generate the adjusted seed data.

[0038] In this embodiment, each feature in the initial seed data has a corresponding initial weight value. The adjustment range in the feature weight adjustment data is adjusted according to the feature weight adjustment range. The adjustment method is to increase or decrease the corresponding adjustment range value based on the initial weight value. After the adjustment is completed, the adjusted seed data is generated, which contains the feature information after the weight adjustment.

[0039] Step S1177: Associate the adjusted seed data with the corresponding scene association data to generate seed scene binding data.

[0040] In this embodiment, when associating the adjusted seed data with the corresponding scene-related data, a mapping relationship between each feature in the adjusted seed data and the corresponding information in the scene-related data is established through a data binding algorithm. The adjusted seed data and the scene-related data are then bound together to generate seed scene-bound data, which includes the adjusted seed data and the corresponding scene-related data.

[0041] Step S1178: Input the seed scene binding data into the seed verification unit of the artificial intelligence fusion module, perform verification processing, and generate seed verification results.

[0042] In this embodiment, the seed verification unit of the artificial intelligence fusion module is a functional submodule within the module responsible for verifying seed data. After the seed scene binding data is input into this unit, the unit uses its internal verification algorithm to verify whether the feature weights in the adjusted seed data meet the scene requirements and whether the feature information is complete, generating a seed verification result. This result includes information on whether the verification passed or failed, as well as the specific reason for failure.

[0043] Step S1179: Based on the seed verification result, fine-tune the feature details of the adjusted seed data to generate security situation seed data.

[0044] In this embodiment, if the seed verification result is successful, the adjusted seed data is directly used as the security situation seed data; if the seed verification result is unsuccessful, the feature details of the adjusted seed data are fine-tuned according to the specific reasons in the verification result, such as adjusting the feature weight values, supplementing missing feature information, etc. After the fine-tuning is completed, the seed data is input into the seed verification unit again for verification until the verification is successful, and the security situation seed data is generated.

[0045] Step S11710: Associate and store the security situation seed data with the seed verification result.

[0046] In this embodiment, when the security situation seed data and the seed verification result are associated and stored, the association between the two is established through a data storage algorithm, and the associated data is stored in a designated storage area of ​​the building security data storage server.

[0047] Step S120: Input the security situation seed data into the artificial intelligence derivation module, perform link derivation processing, generate multiple sets of security association links that are adapted to different security scenarios, and integrate them to form a multi-path link set.

[0048] In this embodiment, the AI-generated module is a software module deployed in the building security data processing server. This AI-generated module has the function of generating security-related links based on input data. After the security situation seed data is input into the AI-generated module, the module analyzes the features in the seed data through its internal link generation algorithm to identify different security scenarios, such as densely populated scenarios and equipment malfunction scenarios. For each security scenario, a corresponding security-related link is generated. This security-related link contains information on key nodes in the scenario and the relationships between nodes. When performing link generation processing, the security situation seed data is first used to identify different security scenarios. Then, a corresponding security-related link is generated for each scenario. Finally, multiple sets of security-related links are integrated to form a multi-path link set. This multi-path link set contains all security-related links adapted to different security scenarios.

[0049] Step S121: Analyze the core features in the security situation seed data and generate seed core features.

[0050] In this embodiment, the core features in the security situation seed data include features that can reflect key information about the security situation, such as the core features of personnel's three-dimensional location, the core features of equipment's operating status, and the core features of the area's security status. When parsing these core features, a core feature extraction algorithm is used to analyze the security situation seed data, identify the core feature information, and encode this information according to a preset core feature encoding rule to generate seed core features. These seed core features are stored in the form of feature vectors, with each vector element corresponding to a core feature.

[0051] Step S122: Call the security scenario rule library built into the artificial intelligence derivative module to extract the link generation rules corresponding to various types of security scenarios.

[0052] In this embodiment, the security scenario rule base built into the AI ​​derivative module stores link generation rules corresponding to various types of security scenarios. For example, the link generation rule for densely populated scenarios generates associated links around the location nodes of personnel, while the link generation rule for equipment malfunction scenarios generates associated links around the operating status nodes of equipment. When this rule base is invoked, the content in the rule base is extracted using a rule extraction algorithm to obtain the link generation rules corresponding to various types of security scenarios, generating a rule extraction result that includes the extracted link generation rules.

[0053] Step S123: Match the seed core features with the link generation rules corresponding to various types of security scenarios to generate rule matching results.

[0054] In this embodiment, when matching seed core features with link generation rules, a rule matching algorithm is used to match each feature in the seed core features with the corresponding conditions in the link generation rules to determine the security scenario type corresponding to the seed core features and the link generation rule corresponding to the scenario type, and generate a rule matching result. The rule matching result includes the matched security scenario type and the corresponding link generation rule.

[0055] Step S124: Based on the rule matching results, determine the link derivation direction corresponding to each type of security scenario and generate link direction data.

[0056] In this embodiment, the link generation rules in the rule matching results clearly define the link derivation direction corresponding to each type of security scenario. For example, the link derivation direction for a densely populated scenario is from the location node of the personnel to the surrounding area nodes, and the link derivation direction for a device malfunction scenario is from the device's operating status node to the device's control node. Based on these rules, each type of security scenario is analyzed to determine its corresponding link derivation direction. These directions are arranged in order of scenario type to generate link direction data. This link direction data is stored in the form of a direction vector, with each vector element corresponding to a link derivation direction for a specific scenario type.

[0057] Step S125: Based on the link direction data, derive the initial security association link and generate the initial link for a single security scenario.

[0058] In this embodiment, the node information in the seed core features is analyzed according to the derived direction in the link direction data to determine the starting node and ending node of the initial link. Then, based on the association between the starting node and the ending node, an initial security association link is generated. This initial security association link contains the connection order information and association information between nodes. Each single security scenario corresponds to one initial link, that is, the initial link of a single security scenario.

[0059] Step S126: Associate the initial link of a single security scenario with the corresponding security scenario features to generate scenario link association data.

[0060] In this embodiment, the corresponding security scene feature refers to the feature information corresponding to a single security scene. For example, the feature information corresponding to a densely populated scene includes personnel density features and personnel movement features. When associating the initial link with the corresponding security scene feature, a mapping relationship between each node in the initial link and the corresponding security scene feature is established through a data association algorithm. The initial link and the corresponding security scene feature are bound together to generate scene link association data, which contains the initial link and the corresponding security scene feature.

[0061] Step S127: Based on the scene link association data, optimize the link node connection relationship of the initial link of a single security scene to generate the optimized link of the single security scene.

[0062] In this embodiment, when optimizing the link node connection relationship of the initial link based on scene link association data, the security scene characteristics in the scene link association data are first analyzed to identify the unreasonable aspects of the node connection relationship in the initial link, such as the node connection order not conforming to the requirements of the scene characteristics, or the weak association between nodes. Then, the node connection relationship of the initial link is adjusted according to the analysis results. The adjustment methods include changing the connection order of nodes, adding or deleting the association between nodes, etc. After the adjustment is completed, an optimized link for a single security scene is generated, which contains the optimized node connection relationship information.

[0063] Step S128: Integrate all optimized links for a single security scenario, add scenario identifiers and link priority markers, and generate a multi-path link set.

[0064] In this embodiment, when integrating all optimized links for a single security scenario, all links are first classified according to scenario type. Then, a corresponding scenario identifier is added to each link. The scenario identifier is used to distinguish different security scenarios, such as "personnel-dense scenario link identifier" and "device malfunction scenario link identifier". At the same time, a link priority marker is added to each link. The priority marker is determined according to the importance of the security scenario corresponding to the link. For example, the priority of the personnel-dense scenario link is higher than the priority of the device malfunction scenario link. Finally, all links with added identifiers and markers are integrated to generate a multi-path link set. This multi-path link set contains all optimized security-related links and their corresponding identifiers and markers.

[0065] Step S1271: Extract security scene detail features from scene link association data to generate scene detail data.

[0066] In this embodiment, the security scene detail features in the scene link association data include specific behavioral details of personnel in the scene, such as details of personnel walking posture and duration of stay; specific operational details of equipment in the scene, such as details of equipment temperature changes and voltage fluctuations; and specific environmental details of areas in the scene, such as details of area light intensity and area air quality. When extracting these detail features, a detail feature extraction algorithm is used to analyze the scene link association data, identify the detail feature information, and encode this information according to a preset detail feature encoding rule to generate scene detail data. This scene detail data is stored in the form of detail feature vectors, with each vector element corresponding to a specific detail feature.

[0067] Step S1272: Analyze the distribution of link nodes in the initial link of a single security scenario and generate node distribution data.

[0068] In this embodiment, the node distribution of the initial link in a single security scenario includes the location distribution information of nodes in the link, the number distribution information of nodes, and the type distribution information of nodes. When analyzing the node distribution, the node information in the initial link is analyzed through a node distribution analysis algorithm to determine the distribution of nodes. The analysis results are encoded according to a preset distribution coding rule to generate node distribution data. This node distribution data is stored in the form of a distribution vector, with each vector element corresponding to a distribution situation.

[0069] Step S1273: Associate scene detail data with node distribution data, mark node connection points that adapt to scene details, and generate node adaptation tags.

[0070] In this embodiment, when associating scene detail data and node distribution data, a mapping relationship is established between each detail feature in the scene detail data and the corresponding node in the node distribution data through a data association algorithm. Based on the mapping relationship, node connection points that adapt to the scene details are determined. Adapted node connection points refer to node connection positions that can accurately reflect scene detail features. These connection points are marked by adding a corresponding marker symbol to each adapted connection point to generate a node adaptation marker. The node adaptation marker contains the location information and marker symbol of the adapted connection point.

[0071] Step S1274: Based on the node adaptation tag, adjust the node connection order of the initial link in a single security scenario to generate the link after the order adjustment.

[0072] In this embodiment, when adjusting the node connection order of the initial link based on the node adaptation mark, the order of the adaptation connection points in the node adaptation mark is first determined, and then the node connection order in the initial link is adjusted according to the order. The adjustment method is to rearrange the nodes in the initial link according to the order of the adaptation connection points. After the adjustment is completed, the link after the order adjustment is generated, which contains the node information after the connection order is adjusted.

[0073] Step S1275: Supplement the missing scenario adaptation nodes in the link after the order adjustment, and generate the supplemented link.

[0074] In this embodiment, the missing scene adaptation nodes in the sequence-adjusted link refer to nodes that can reflect scene details but are not included in the link. When supplementing these nodes, the scene detail data is first analyzed to determine the type and number of scene adaptation nodes that need to be supplemented. Then, based on the determined node type and number, the corresponding nodes are added to the sequence-adjusted link. The addition method is to insert the missing nodes at appropriate positions in the link. After the supplementation is completed, a node-supplemented link is generated, which contains the node information after the supplementation.

[0075] Step S1276: Delete redundant and invalid nodes in the link after node supplementation, and generate the link after node optimization.

[0076] In this embodiment, redundant and invalid nodes in the supplemented link refer to nodes that do not contribute to reflecting the characteristics of the security scenario. When deleting these nodes, each node in the supplemented link is first analyzed to determine whether it is a redundant and invalid node. The criterion for judgment is whether the node can reflect the detailed characteristics of the scenario. If the node cannot reflect the detailed characteristics of the scenario, it is determined to be a redundant and invalid node. Then, the nodes determined to be redundant and invalid nodes are deleted from the link. After the deletion is completed, a node-optimized link is generated, which contains the node information after the deletion of redundant and invalid nodes.

[0077] Step S1277: After the associated node is optimized, the link is associated with the scene link data, the compatibility verification between the link and the scene is performed, and the compatibility verification result is generated.

[0078] In this embodiment, when associating optimized links with scene link data, a data association algorithm is used to establish a mapping relationship between each node in the optimized link and the corresponding information in the scene link association data, thus binding the optimized links with the scene link association data. During adaptation verification, an adaptation verification algorithm is used to verify whether the bound links accurately reflect scene characteristics. The verification includes whether the nodes in the link can fully reflect the detailed features of the scene, whether the connection relationships between nodes meet the scene requirements, etc., generating an adaptation verification result. This result includes information on whether the verification passed or failed, and the specific reason for failure.

[0079] Step S1278: Based on the adaptation verification results, fine-tune the connection parameters of the link nodes to generate an optimized link for a single security scenario.

[0080] In this embodiment, if the adaptation verification result is successful, the optimized link is directly used as the link for the optimized single security scenario; if the adaptation verification result is unsuccessful, the node connection parameters in the optimized link are fine-tuned according to the specific reasons in the verification result, such as adjusting the connection strength parameters between nodes, adjusting the position parameters of nodes, etc. After the fine-tuning is completed, the adaptation verification is performed again until the verification is successful, and the optimized link for the single security scenario is generated.

[0081] Step S1279: Bind the optimized link of a single security scenario with the adaptation verification result and store it in the link optimization database of the artificial intelligence derivative module.

[0082] In this embodiment, when the optimized link of a single security scenario is bound and stored with the adaptation verification result, the association between the two is established through a data storage algorithm, and the associated data is stored in the link optimization database of the artificial intelligence derivative module. This database is deployed in the building security data storage server and has data storage and query functions.

[0083] Step S130: Synchronously input the multi-path link set and real-time 3D security data collection data into the 3D security situation awareness artificial intelligence model, and perform situation generation processing through the multi-link linkage unit of the model to generate the initial 3D security situation.

[0084] In this embodiment, the 3D security situation awareness AI model is an AI model deployed in a building security data processing server. This AI model has the function of processing multi-path link sets and real-time 3D security acquisition data to generate a 3D security situation. The multi-link linkage unit is a functional unit within this AI model, capable of performing linkage processing on multi-path links. After synchronously inputting the multi-path link set and real-time 3D security acquisition data into the AI ​​model, the AI ​​model first performs data preprocessing to ensure that the data meets the model's processing requirements in terms of format and content. Then, the preprocessed data is input into the multi-link linkage unit. This multi-link linkage unit uses its internal linkage algorithm to perform linkage analysis on the link information in the multi-path link set and the information in the real-time 3D security acquisition data, identify the situation characteristics, and generate an initial 3D security situation. This initial 3D security situation includes personnel situation information, equipment situation information, and area situation information within the building.

[0085] Step S131: Input the multi-path link set into the link preprocessing unit of the 3D security situation awareness artificial intelligence model, perform splitting processing, and extract an independent security-related link and its corresponding scene identifier.

[0086] In this embodiment, the link preprocessing unit of the 3D security situational awareness AI model is a functional unit within the model responsible for preprocessing link data. After the multi-path link set is input into this unit, it uses an internal splitting algorithm to split the set, separating each security-related link and extracting the scene identifier corresponding to each link. After splitting, an independent security-related link and its corresponding scene identifier are extracted. This security-related link contains node information and node connection relationship information, and the scene identifier is used to distinguish different security scenarios.

[0087] Step S132: Match the scene identifier of an independent security association link with the collection scene of real-time 3D security data to generate scene matching results.

[0088] In this embodiment, the acquisition scenario of real-time 3D security data refers to the specific security scenario in which the data is acquired, such as a densely populated scenario or a scenario with equipment malfunction. When matching the scenario identifier with the acquisition scenario, a scenario matching algorithm is used to analyze the scenario identifier of an independent security association link with the acquisition scenario of the real-time 3D security data to determine whether the two match and generate a scenario matching result. This scenario matching result includes information on whether the match was successful or failed, as well as the scenario correspondence information when the match was successful.

[0089] Step S133: Based on the scene matching results, filter the security-related links that are compatible with the current collection scene and generate a set of compatible links.

[0090] In this embodiment, when filtering security-related links that are compatible with the current collection scenario based on the scenario matching results, if the scenario matching result is successful, the corresponding security-related link is filtered out. If the scenario matching result is unsuccessful, the security-related link is not filtered out. After the filtering is completed, all successfully matched security-related links are integrated to generate an adapted link set. This adapted link set contains all security-related links that are compatible with the current collection scenario.

[0091] Step S134: Input the set of adapted links and the real-time 3D security data collected in the corresponding scenario into the multi-link linkage unit of the artificial intelligence model, perform linkage processing, and generate the linkage results of the link data.

[0092] In this embodiment, after the set of adapted links and the real-time 3D security data acquisition data of the corresponding scenario are input into the multi-link linkage unit, the multi-link linkage unit performs linkage analysis on the link information in the set of adapted links and the information in the real-time 3D security data acquisition data through its internal linkage algorithm. The analysis includes the correlation between the node information in the link and the node information in the real-time data, the correlation between the situation information in the link and the situation information in the real-time data, etc. After the analysis is completed, a link data linkage result is generated, which contains the information after linkage analysis.

[0093] Step S135: Extract situation features from the link data linkage results through the multi-link linkage unit and generate linkage situation features.

[0094] In this embodiment, the situational features in the link data linkage results include features that reflect the building security situation, such as personnel movement characteristics, equipment operation characteristics, and area security characteristics. When extracting these situational features, the multi-link linkage unit analyzes the link data linkage results using an internal feature extraction algorithm, identifies the situational feature information, and encodes this information according to preset situational feature encoding rules to generate linkage situational features. These linkage situational features are stored in the form of feature vectors, with each vector element corresponding to a situational feature.

[0095] Step S136: Associate the linkage situation characteristics with the link node information in the adapted link set to generate node situation association data.

[0096] In this embodiment, the link node information in the adapted link set includes node type information, node location information, node function information, etc. When associating linkage situation features with link node information, a mapping relationship is established between each feature in the linkage situation features and the corresponding node in the link node information through a data association algorithm. The linkage situation features are bound to the corresponding node information to generate node situation association data, which contains the linkage situation features and the corresponding link node information.

[0097] Step S137: Integrate all node situational data to generate an initial 3D security situational data containing scene situational details.

[0098] In this embodiment, when integrating all node situational association data, all node situational association data are first classified according to the node's location information, and then the classified node situational association data are integrated. The integrated content includes the node's situational feature information, node's location information, node's functional information, etc. After integration, an initial three-dimensional security situation containing scene situational details is generated. This situation includes personnel situational details, equipment situational details, and area situational details in various areas of the building.

[0099] Step S1351: Split the link features and collected data features in the link data linkage result to generate a combined set containing both link features and collected data features.

[0100] In this embodiment, the link features in the link data linkage result refer to feature information from a multi-path link set, such as node features and node connection relationship features in the link. The collected data features refer to feature information from real-time 3D security data collection, such as personnel location features and equipment operating status features. When splitting these features, a feature splitting algorithm is used to analyze the link data linkage result, separating the link features from the collected data features. After separation, a combined set containing both link features and collected data features is generated. This combined set includes both the separated link features and collected data features.

[0101] Step S1352: Associate the link features and the collected data features in the combined set containing link features and collected data features to generate feature linkage pairs.

[0102] In this embodiment, when associating link features and collected data features, the link features and collected data features in the combined set are analyzed by a feature association algorithm to determine the association relationship between the two. The basis for association is whether the security scene information reflected by the features is consistent. If the security scene information reflected by the features is consistent, the two are associated to generate a feature linkage pair. This feature linkage pair includes the associated link features and collected data features.

[0103] Step S1353: Extract the core association information from the feature linkage pairs and generate linkage core features.

[0104] In this embodiment, the core association information in a feature linkage pair refers to information that reflects the key association relationship between the feature linkage pair, such as the association information between node location information in the link feature and personnel location information in the data collection feature, or the association information between node operating status information in the link feature and equipment operating status information in the data collection feature. When extracting this core association information, the feature linkage pair is analyzed using a core association information extraction algorithm to identify the core association information. This information is then encoded according to a preset core association information encoding rule to generate linkage core features. These linkage core features are stored in the form of feature vectors, with each vector element corresponding to a type of core association information.

[0105] Step S1354: Track the changes of the core linkage features in the continuous link data linkage results and generate feature change trajectories.

[0106] In this embodiment, the continuous link data linkage result refers to the linkage results of multiple links generated within a certain period of time. When tracking the changes of the linkage core features in the continuous link data linkage results, the linkage core features in the continuous link data linkage results are analyzed by a feature tracking algorithm. The changes of the features at different time points are recorded. The recorded content includes the changes in the value of the features, the changes in the type of the features, etc. After the recording is completed, a feature change trajectory is generated. This feature change trajectory contains the change information of the linkage core features at continuous time points.

[0107] Step S1355: Associate the characteristic change trajectory with the corresponding link node information to generate trajectory node association data.

[0108] In this embodiment, the corresponding link node information refers to the node information in the link corresponding to the feature change trajectory, such as the node's location information and functional information. When associating the feature change trajectory with the corresponding link node information, a mapping relationship between each change point in the feature change trajectory and the corresponding link node information is established through a data association algorithm. The feature change trajectory and the corresponding link node information are then bound together to generate trajectory node association data, which contains the feature change trajectory and the corresponding link node information.

[0109] Step S1356: Based on the trajectory node association data, extract key feature points in the feature change trajectory and generate key feature data.

[0110] In this embodiment, key feature points in the feature change trajectory refer to points that can reflect key information about feature changes, such as points where feature values ​​change abruptly or feature types change. When extracting these key feature points, a key feature point extraction algorithm is used to analyze the trajectory node association data, identify the key feature points, and encode the information of these points according to a preset key feature point encoding rule to generate key feature data. This key feature data is stored in the form of key feature point vectors, with each vector element corresponding to one key feature point.

[0111] Step S1357: Integrate key feature data with core linkage features to generate initial linkage situation features.

[0112] In this embodiment, when integrating key feature data and linkage core features, a data integration algorithm is used to integrate the key feature point information in the key feature data with the core association information in the linkage core features. The integration method is to add the key feature point information to the linkage core features. After integration, an initial linkage situation feature is generated, which includes key feature point information and core association information.

[0113] Step S1358: Input the initial linkage situation features into the feature enhancement subunit of the multi-link linkage unit, perform feature enhancement processing, and generate enhanced linkage situation features.

[0114] In this embodiment, the feature enhancement subunit of the multi-link linkage unit is a functional subunit within the multi-link linkage unit responsible for enhancing features. After the initial linkage situation features are input into the feature enhancement subunit, the subunit uses an internal enhancement algorithm to enhance the key feature point information and core correlation information in the initial linkage situation features. The enhancement method is to increase the weight of the key feature point information and core correlation information in the features. After enhancement, enhanced linkage situation features are generated, which contain the enhanced key feature point information and core correlation information.

[0115] Step S1359: The enhanced linkage situational characteristics are used as the final linkage situational characteristics and transmitted to the situational integration unit of the artificial intelligence model.

[0116] In this embodiment, the situation integration unit of the artificial intelligence model is a functional unit within the model responsible for integrating situation features. After the enhanced linked situation features are transmitted to the situation integration unit, the unit integrates these features with other situation features to generate a complete three-dimensional security situation.

[0117] Step S140: Input the initial 3D security situation into the cross-link verification unit built into the model, perform verification processing, remove link conflict data, and generate the final 3D security situation.

[0118] In this embodiment, the cross-link verification unit built into the model is a functional unit within the 3D security situation awareness artificial intelligence model, capable of verifying the initial 3D security situation. After the initial 3D security situation is input into this unit, the unit first performs an integrity check on the data in the initial 3D security situation to ensure that no key information is missing, and then performs a consistency check on the data to ensure that there is no conflicting information. If data conflicts are found during the verification process, the conflicting data is removed. After the verification is completed, the final 3D security situation is generated. This final 3D security situation includes verified information such as the situation of personnel in the building, equipment, and area.

[0119] Step S141: Extract situational data of all link nodes in the initial three-dimensional security situation and generate a node situational set.

[0120] In this embodiment, the link node situation data in the initial three-dimensional security situation includes node location situation data, node operational status situation data, and node association relationship situation data. When extracting this data, a node situation data extraction algorithm is used to analyze the initial three-dimensional security situation, identify all link node situation data, and encode this data according to a preset node situation data encoding rule to generate a node situation set. This node situation set contains the situation data of all link nodes.

[0121] Step S142: Associat the situation data of the same node with different links in the associated node situation set, and generate node situation comparison data.

[0122] In this embodiment, situational data corresponding to the same node across different links refers to situational data from different links but corresponding to the same node. When associating these data, a data association algorithm is used to analyze the data in the node situational data set to determine the situational data corresponding to the same node across different links. These data are then associated to generate node situational comparison data, which includes situational data corresponding to the same node across different links and comparison information between the data.

[0123] Step S143: Analyze the data differences in the node situation comparison data, mark the conflicting nodes and corresponding links with differences, and generate conflict marking data.

[0124] In this embodiment, when analyzing the data differences in the node situation comparison data, the data difference analysis algorithm is used to analyze the situation data of different links corresponding to the same node in the node situation comparison data to determine whether there are differences between the data. If there are differences, the conflicting nodes and corresponding links with differences are marked. The marking method is to add conflict marker symbols to the conflicting nodes and corresponding links to generate conflict marker data. The conflict marker data includes the information of the conflicting nodes, the information of the corresponding links, and the conflict marker symbols.

[0125] Step S144: Call the model's built-in conflict verification rule library, match the verification rules corresponding to the conflict-marked data, and generate conflict verification results.

[0126] In this embodiment, the model's built-in conflict verification rule library stores verification rules corresponding to different conflict types, such as verification rules for data value difference conflicts and data type difference conflicts. When this verification rule library is invoked, a rule matching algorithm analyzes the conflict-marked data against the rules in the rule library to determine the verification rule corresponding to the conflict-marked data and generate a conflict verification result. This conflict verification result includes the matched verification rule and the corresponding verification method information.

[0127] Step S145: Based on the conflict verification results, retain the valid situation data of the conflicting nodes, remove invalid conflict data, and generate a set of node situations after verification.

[0128] In this embodiment, when retaining valid situational data based on the conflict verification results, the situational data of the conflicting nodes are first analyzed according to the verification rules and verification methods in the conflict verification results to determine which data is valid situational data and which data is invalid conflict data. Then, the valid situational data is retained and the invalid conflict data is removed. After the processing is completed, a set of verified node situational data is generated. This set of verified node situational data contains the verified node situational data.

[0129] Step S146: Associate the verified node status set with the corresponding link information to generate verified link status data.

[0130] In this embodiment, the corresponding link information refers to the link information to which a node in the verified node situation set belongs. When associating the verified node situation set with the corresponding link information, a mapping relationship between the situation data of each node in the verified node situation set and the corresponding link information is established through a data association algorithm. The verified node situation set and the corresponding link information are then bound together to generate verified link situation data, which includes the verified node situation data and the corresponding link information.

[0131] Step S147: Integrate all verified link situation data, supplement the situation completeness description, and generate the final three-dimensional security situation.

[0132] In this embodiment, when integrating all verified link status data, a data integration algorithm is used to integrate all verified link status data. The integrated content includes node status data, link information, etc. After integration, a status integrity description is added, including a description of the time range of the status and a description of the spatial range of the status. After the addition is completed, a final three-dimensional security status is generated, which includes the integrated link status data and integrity description information.

[0133] Step S1451: Parse the valid data judgment criteria in the conflict check results and generate data validity criteria.

[0134] In this embodiment, the valid data judgment criteria in the conflict verification results refer to the standards used to determine whether the situational data of conflict nodes is valid, such as standards that the data value is within a preset range, standards that the data type meets preset requirements, etc. When parsing these standards, the conflict verification results are analyzed through a standard parsing algorithm to identify the valid data judgment criteria. These standards are then encoded according to preset standard encoding rules to generate data validity standards, which contain specific requirements for determining the validity of the data.

[0135] Step S1452: Compare conflicting data with the data validity criteria in the node situation comparison data, mark valid data that meets the criteria, and generate valid data tags.

[0136] In this embodiment, when comparing conflict data with the data validity standard, the conflict data in the node situation comparison data is analyzed by a data comparison algorithm to determine whether the conflict data meets the data validity standard. If it does, the data is marked as valid data. The marking method is to add a valid marker symbol to the valid data to generate a valid data marker. The valid data marker contains the information of the valid data and the valid marker symbol.

[0137] Step S1453: Extract the situation data corresponding to the valid data markers and generate valid situation data for the nodes.

[0138] In this embodiment, when extracting the situation data corresponding to the valid data markers, the valid data markers are analyzed by a data extraction algorithm to identify the corresponding situation data. These data are then encoded according to a preset situation data encoding rule to generate node valid situation data. This node valid situation data contains the situation data corresponding to the valid data markers.

[0139] Step S1454: Bind the valid situational data of the node with the corresponding conflicting node identifier to generate the valid data node binding result.

[0140] In this embodiment, the corresponding conflict node identifier refers to the identifier information of the conflict node corresponding to the node's valid situation data. When binding the node's valid situation data with the corresponding conflict node identifier, a mapping relationship between the node's valid situation data and the corresponding conflict node identifier is established through a data binding algorithm. The node's valid situation data and the corresponding conflict node identifier are bound together to generate a valid data node binding result. This valid data node binding result includes the node's valid situation data and the corresponding conflict node identifier.

[0141] Step S1455: Mark invalid conflicting data in the node situation set that does not meet the data validity criteria, and generate invalid data markers.

[0142] In this embodiment, when marking invalid conflict data that does not meet the data validity standard, the conflict data in the node situation set is analyzed by a data marking algorithm to determine which data does not meet the data validity standard. If the data does not meet the standard, it is marked as invalid conflict data. The marking method is to add invalid marker symbols to the invalid conflict data to generate invalid data markers. The invalid data markers contain the information of invalid conflict data and invalid marker symbols.

[0143] Step S1456: Remove invalid conflict data corresponding to invalid data markers from the node situation set, and retain the valid node situation data in the valid data node binding results.

[0144] In this embodiment, when removing invalid conflict data from the node situation set, the invalid conflict data corresponding to the invalid data marker in the node situation set is analyzed by the data removal algorithm, and these data are removed from the node situation set. At the same time, the valid node situation data in the valid data node binding result is retained. After processing, the node situation set with the valid data retained is generated.

[0145] Step S1457: Integrate all valid state data from all nodes, add data validity markers, and generate a set of verified node states.

[0146] In this embodiment, when integrating all node valid situation data, a data integration algorithm is used to integrate all node valid situation data. The integrated content includes the information of the node valid situation data. After integration, a data validity mark is added. The method of adding the mark is to add a validity mark symbol to the integrated node valid situation data, and a verified node situation set is generated. This verified node situation set contains the integrated node valid situation data and the data validity mark.

[0147] Step S1458: Associate and store the verified node situation set with the conflict verification results.

[0148] In this embodiment, when associating the node status set after verification with the conflict verification result, the association between the two is established through a data storage algorithm, and the associated data is stored in a designated storage area of ​​the building security data storage server.

[0149] Step S150: Generate security control instructions based on the final three-dimensional security situation, send them to the building management terminal, collect security status feedback data after the execution of the control instructions, input the security status feedback data back to the artificial intelligence fusion module, and perform calibration processing of the security situation seed data generation parameters.

[0150] In this embodiment, when generating security control instructions based on the final 3D security situation, the situation information in the final 3D security situation is first analyzed to determine the security control measures to be taken, such as measures to adjust the direction of personnel flow and measures to adjust equipment operating parameters. Then, corresponding security control instructions are generated based on the determined measures. These instructions include the specific content and execution requirements of the control measures. When the instructions are sent to the building control terminal, they are transmitted to the control terminals deployed in various areas of the building, such as access control terminals at entrances and exits and elevator operation control terminals, through the building security communication network. When collecting security status feedback data after the execution of the control instructions, the security status after the execution of the control instructions is collected through feedback data collection devices deployed in various areas of the building, such as cameras and sensors. The collected content includes personnel flow status information and equipment operating status information. When the security status feedback data is input to the AI ​​fusion module, the collected feedback data is sent to the AI ​​fusion module through the data transmission interface. The AI ​​fusion module uses an internal calibration algorithm to calibrate the security status seed data generation parameters. The calibration method is to adjust the value of the generation parameters according to the feedback data, and update the generation parameters after calibration is completed.

[0151] For example, step S151: Extract the control execution effect information from the security status feedback data and generate effect feedback data.

[0152] In this embodiment, the control implementation effect information in the security status feedback data includes information on the impact of control measures on personnel flow, the impact of control measures on equipment operation, and the impact of control measures on the area's security status. When extracting this information, the security status feedback data is analyzed using an effect information extraction algorithm to identify the control implementation effect information. This information is then encoded according to a preset effect information encoding rule to generate effect feedback data, which contains the control implementation effect information.

[0153] Step S152: Correlate the feedback data of the associated effect with the corresponding final three-dimensional security situation to generate situation effect correlation data.

[0154] In this embodiment, the corresponding final three-dimensional security situation refers to the final three-dimensional security situation corresponding to the effect feedback data. When associating the effect feedback data with the corresponding final three-dimensional security situation, a mapping relationship between the effect feedback data and the corresponding final three-dimensional security situation is established through a data association algorithm. The effect feedback data and the corresponding final three-dimensional security situation are bound together to generate situation effect association data, which includes the effect feedback data and the corresponding final three-dimensional security situation.

[0155] Step S153: Trace the security situation seed data corresponding to the situation effect correlation data and generate seed data to be calibrated.

[0156] In this embodiment, when tracing the security situation seed data corresponding to the situation effect correlation data, the situation effect correlation data is analyzed by the data tracing algorithm to determine the corresponding security situation seed data. These data are then encoded according to the preset seed data encoding rules to generate seed data to be calibrated. The seed data to be calibrated contains the security situation seed data corresponding to the situation effect correlation data.

[0157] Step S154: Analyze the correlation between the effect feedback data and the seed data to be calibrated, mark the feature parameters in the seed data that need to be adjusted, and generate calibration mark data.

[0158] In this embodiment, when analyzing the correlation between the effect feedback data and the seed data to be calibrated, the effect feedback data and the seed data to be calibrated are analyzed by the correlation analysis algorithm to determine which feature parameters in the seed data need to be adjusted. The basis for adjustment is the control execution effect reflected by the effect feedback data. If the control execution effect does not meet expectations, the corresponding feature parameter is marked as the parameter to be adjusted. The marking method is to add calibration mark symbols to the parameter to be adjusted to generate calibration mark data. The calibration mark data contains the feature parameter information to be adjusted and the calibration mark symbols.

[0159] Step S155: Call the calibration rule library built into the artificial intelligence fusion module, match the parameter adjustment rules corresponding to the calibration mark data, and generate calibration rule results.

[0160] In this embodiment, the calibration rule library built into the artificial intelligence fusion module stores parameter adjustment rules corresponding to different calibration mark data, such as adjustment rules corresponding to personnel position feature parameters that need to be adjusted, and adjustment rules corresponding to equipment operating status feature parameters that need to be adjusted. When this adjustment rule library is called, the calibration mark data and the rules in the rule library are analyzed by a rule matching algorithm to determine the parameter adjustment rules corresponding to the calibration mark data and generate calibration rule results. These calibration rule results contain the matched parameter adjustment rules.

[0161] Step S156: Based on the calibration rule results, determine the adjustment direction and magnitude of the parameters for generating the seed data to be calibrated, and generate parameter calibration data.

[0162] In this embodiment, when determining the adjustment direction and magnitude based on the calibration rule results, the parameter adjustment rules in the calibration rule results are first analyzed to determine the adjustment direction of the parameters generated by the seed data to be calibrated, such as increasing or decreasing the value of the parameters. Then, the adjustment magnitude is determined according to the rules. The magnitude of the adjustment magnitude is determined based on the degree of difference between the control execution effect and the expected effect. After the determination is completed, parameter calibration data is generated, which contains information on the adjustment direction and adjustment magnitude.

[0163] Step S157: According to the parameter calibration data, adjust the generation parameters of the security situation seed data in the artificial intelligence fusion module to generate the calibrated parameters.

[0164] In this embodiment, when adjusting the generated parameters according to the parameter calibration data, the current generated parameter value in the artificial intelligence fusion module is first obtained, and then the current generated parameter value is adjusted according to the adjustment direction and adjustment range in the parameter calibration data. The adjustment method is to increase or decrease the corresponding adjustment range value based on the current value. After the adjustment is completed, the calibrated generated parameters are generated, which contain the adjusted value information.

[0165] Step S158: Store the calibrated parameters to the parameter configuration unit of the artificial intelligence fusion module and update the seed generation rules.

[0166] In this embodiment, the parameter configuration unit of the artificial intelligence fusion module is a functional unit within the module responsible for storing the generated parameters. When storing the calibrated generated parameters to this functional unit, the calibrated generated parameters are stored in this functional unit through a data storage algorithm, while the seed generation rules are updated. The update method is to adjust the parameter settings in the seed generation rules according to the calibrated generated parameters.

[0167] Step S159: Based on the parameters generated after calibration, generate new security situation seed data, perform adaptability verification of the new seed data, and complete the calibration process.

[0168] In this embodiment, when generating new security situation seed data based on the calibrated generation parameters, the seed data generation algorithm uses the calibrated generation parameters to fuse real-time 3D security acquisition data with preset building security baseline situation data to generate new security situation seed data. During adaptability verification, the new security situation seed data is verified using an adaptability verification algorithm. The verification includes whether the seed data meets the requirements of the current security scenario and whether the seed data accurately reflects the security situation information. If the verification passes, the calibration process is complete; if the verification fails, the above calibration steps are repeated until the verification passes.

[0169] Figure 2 The illustration shows exemplary hardware and software components of an AI-based intelligent building 3D security situational awareness system 100 that can implement the ideas of this application, according to some embodiments of this application. For example, a processor 120 can be used in the AI-based intelligent building 3D security situational awareness system 100 and to perform the functions described in this application.

[0170] The AI-based intelligent building 3D security situational awareness system 100 can be a general-purpose server or a special-purpose server; both can be used to implement the AI-based intelligent building 3D security situational awareness method of this application. Although only one server is shown in this application, for convenience, the functions described in this application can be implemented in a distributed manner on multiple similar platforms to balance the load.

[0171] For example, an AI-based intelligent building 3D security situational awareness system 100 may include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and various forms of storage media 140, such as a disk, ROM, or RAM, or any combination thereof. Exemplarily, the AI-based intelligent building 3D security situational awareness system 100 may also include program instructions stored in ROM, RAM, or other types of non-transitory storage media, or any combination thereof. The methods of this application can be implemented according to these program instructions. The AI-based intelligent building 3D security situational awareness system 100 also includes an I / O interface 150 between the computer and other input / output devices.

[0172] For ease of explanation, only one processor is described in the AI-based intelligent building 3D security situational awareness system 100. However, it should be noted that the AI-based intelligent building 3D security situational awareness system 100 of this application may also include multiple processors. Therefore, the steps executed by one processor described in this application may also be executed jointly by multiple processors or individually. For example, if the processor of the AI-based intelligent building 3D security situational awareness system 100 executes steps A and B, it should be understood that steps A and B may also be executed jointly by two different processors or individually by one processor. For example, the first processor executes step A, the second processor executes step B, or the first processor and the second processor jointly execute steps A and B.

[0173] Furthermore, this embodiment of the invention also provides a readable storage medium, wherein computer-executable instructions are preset in the readable storage medium, and when the processor executes the computer-executable instructions, the above-mentioned intelligent building three-dimensional security situational awareness method based on artificial intelligence is realized.

[0174] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. A method for intelligent building 3D security situational awareness based on artificial intelligence, characterized in that, The method includes: The system acquires real-time 3D security data from intelligent building 3D security data acquisition equipment, integrates preset building security baseline status data, performs fusion processing, and generates security status seed data through an artificial intelligence fusion module. The security situation seed data is input into the artificial intelligence derivative module, and the link derivative processing is performed to generate multiple sets of security association links that are adapted to different security scenarios, and then integrated to form a multi-path link set. The multi-path link set and real-time 3D security data are synchronously input into the 3D security situation awareness artificial intelligence model. The situation generation process is performed through the multi-link linkage unit of the model to generate the initial 3D security situation. Input the initial 3D security situation into the model's built-in cross-link verification unit, perform verification processing, remove link conflict data, and generate the final 3D security situation; Based on the final 3D security situation, security control instructions are generated and sent to the building management terminal. Security status feedback data after the execution of the control instructions is collected and input back to the artificial intelligence fusion module to perform calibration processing of the security situation seed data generation parameters.

2. The intelligent building 3D security situational awareness method based on artificial intelligence according to claim 1, characterized in that, The process involves acquiring real-time 3D security data output from the intelligent building 3D security data acquisition device, integrating it with preset building security baseline situational data, performing fusion processing, and generating security situational seed data through an artificial intelligence fusion module, including: Extract security scene features from real-time 3D security data acquisition to generate scene feature data; Extract the baseline features from the preset building security baseline situation data to generate a baseline feature set; The scene feature data and the benchmark feature set are input into the feature alignment unit of the artificial intelligence fusion module to perform feature alignment processing and generate feature alignment results; Based on the feature alignment results, the scene feature data is associated with matching features in the benchmark feature set to generate feature association pairs; The feature association is input to the seed generation unit of the artificial intelligence fusion module, and the initial seed generation process is performed to generate initial seed data; By associating the initial seed data with the corresponding real-time 3D security data acquisition scene information, scene association data is generated. Based on scene-related data, feature weight adjustment processing of the initial seed data is performed to generate security situation seed data; The security situation seed data is stored in the seed database of the artificial intelligence fusion module and synchronized to the input cache unit of the artificial intelligence derivative module.

3. The intelligent building 3D security situational awareness method based on artificial intelligence according to claim 1, characterized in that, The process involves inputting security situation seed data into the artificial intelligence derivation module, performing link derivation processing, generating multiple sets of security association links adapted to different security scenarios, and integrating them to form a multi-path link set, including: Analyze the core features in the security situation seed data and generate seed core features; The system calls upon the built-in security scenario rule library of the artificial intelligence derivative module to extract the link generation rules corresponding to various types of security scenarios. Match seed core features with link generation rules corresponding to various types of security scenarios, and generate rule matching results; Based on the rule matching results, determine the link derivation direction corresponding to each type of security scenario and generate link direction data; Based on the link direction data, the initial security association link is derived, and the initial link for a single security scenario is generated; Associate the initial link of a single security scenario with the corresponding security scenario features to generate scenario link association data; Based on scenario link association data, optimize the link node connection relationship of the initial link of a single security scenario to generate the optimized link of the single security scenario. Integrate all optimized links for a single security scenario, add scenario identifiers and link priority markers, and generate a multi-path link set; The multi-path link set is transmitted to the link storage unit of the artificial intelligence derivative module and synchronized to the three-dimensional security situational awareness artificial intelligence model.

4. The intelligent building 3D security situational awareness method based on artificial intelligence according to claim 1, characterized in that, The synchronous input of the multi-path link set and real-time 3D security data acquisition data is fed into the 3D security situation awareness artificial intelligence model. The model's multi-link linkage unit performs situation generation processing to generate an initial 3D security situation, including: The multi-path link set is input into the link preprocessing unit of the 3D security situational awareness artificial intelligence model, and splitting is performed to extract an independent security-related link and its corresponding scene identifier. Match the scene identifier of an independent security-related link with the scene of real-time 3D security data collection to generate scene matching results; Based on the scene matching results, security-related links that are compatible with the current collection scene are selected, and a set of compatible links is generated. The set of adapted links and the real-time 3D security data collected in the corresponding scenario are input into the multi-link linkage unit of the artificial intelligence model to perform linkage processing and generate the linkage results of the link data. By using multi-link linkage units, situational features are extracted from the linkage results of link data to generate linkage situational features. By associating the situational characteristics with the link node information in the adapted link set, node situational correlation data is generated. Integrate all node situational data to generate an initial 3D security situational awareness that includes scene situational details.

5. The intelligent building 3D security situational awareness method based on artificial intelligence according to claim 1, characterized in that, The initial 3D security posture is input to the model's built-in cross-link verification unit, which performs verification processing, removes link conflict data, and generates the final 3D security posture, including: Extract situational data of all link nodes in the initial three-dimensional security situation to generate a node situation set; The situation data of the same node corresponding to different links in the associated node situation set are used to generate node situation comparison data; Analyze the data differences in the node situation comparison data, mark the conflicting nodes and their corresponding links, and generate conflict marking data; The model's built-in conflict verification rule library is called to match the verification rules corresponding to the conflict-marked data and generate conflict verification results. Based on the conflict verification results, the valid situation data of the conflicting nodes are retained, the invalid conflict data is removed, and a set of node situations after verification is generated. Associate the verified node status set with the corresponding link information to generate verified link status data; Integrate all verified link status data, supplement the status completeness description, and generate the final three-dimensional security status; The final 3D security situation is transmitted to the model's output unit and synchronized to the security control command generation module.

6. The intelligent building 3D security situational awareness method based on artificial intelligence according to claim 2, characterized in that, The process of adjusting the feature weights of the initial seed data based on scene-related data to generate security situation seed data includes: Parse the collected scene attributes in the scene association data to generate scene attribute data; The system calls upon the built-in scene weight rule library of the AI ​​fusion module to match the feature weight adjustment rules corresponding to the scene attribute data and generate weight rule matching results. Based on the weight rule matching results, the weight adjustment direction of each feature in the initial seed data is determined, and weight direction data is generated. The importance descriptions of scenarios in the associated weight direction data and scenario association data are used to generate weight adjustment association data; Based on the weight adjustment correlation data, determine the weight adjustment range of each feature and generate feature weight adjustment data; Adjust the data according to the feature weights, adjust the feature weights of the initial seed data, and generate the adjusted seed data; After the seed data is associated with the corresponding scene data, seed scene binding data is generated. The seed scene binding data is input into the seed verification unit of the artificial intelligence fusion module, and verification processing is performed to generate seed verification results; Based on the seed verification results, the feature details of the adjusted seed data are fine-tuned to generate security situation seed data. The security situation seed data is stored in association with the seed verification results.

7. The intelligent building 3D security situational awareness method based on artificial intelligence according to claim 3, characterized in that, The process of optimizing the link node connection relationships of the initial link in a single security scenario based on scenario link association data, and generating optimized links for the single security scenario, includes: Extract security scene details from scene link correlation data to generate scene detail data; Analyze the distribution of link nodes in the initial link of a single security scenario and generate node distribution data; Associate scene detail data with node distribution data, mark node connection points that adapt to scene details, and generate node adaptation tags; Based on node adaptation tags, the node connection order of the initial link in a single security scenario is adjusted to generate the link after the order adjustment. Add missing scenario adaptation nodes to the link after the order adjustment, and generate nodes to supplement the link; After deleting redundant and invalid nodes in the link to supplement nodes, a link with optimized nodes is generated. After optimizing the associated nodes, the link and scene are associated with the data, and the compatibility verification between the link and the scene is performed to generate the compatibility verification results. Based on the adaptation verification results, the connection parameters of the link nodes are fine-tuned to generate an optimized link for a single security scenario. The optimized links for a single security scenario are bound to the adaptation verification results and stored in the link optimization database of the AI-derived module.

8. The intelligent building 3D security situational awareness method based on artificial intelligence according to claim 4, characterized in that, The step of extracting situational features from the link data linkage results through the multi-link linkage unit and generating linkage situational features includes: The link features and collected data features in the link data linkage results are separated to generate a combined set containing both link features and collected data features. Link features and collected data features are associated in a combined set containing link features and collected data features to generate feature linkage pairs; Extract the core association information from the feature linkage pairs and generate linkage core features; Track the changes in core linkage features in the results of continuous link data linkage, and generate feature change trajectories; Associate feature change trajectories with corresponding link node information to generate trajectory node association data; Based on trajectory node association data, key feature points are extracted from the feature change trajectory to generate key feature data; Integrate key feature data with core linkage features to generate initial linkage situation features; The initial linkage situation features are input into the feature enhancement subunit of the multi-link linkage unit, and feature enhancement processing is performed to generate enhanced linkage situation features. The enhanced linked situational characteristics are used as the final linked situational characteristics and transmitted to the situational integration unit of the artificial intelligence model.

9. The intelligent building 3D security situational awareness method based on artificial intelligence according to claim 5, characterized in that, Based on the conflict verification results, the valid situational data of conflicting nodes is retained, invalid conflicting data is removed, and a set of verified node situations is generated, including: Analyze the valid data judgment criteria in the conflict check results and generate data validity criteria; Compare conflicting data with the data validity criteria in the node situation comparison data, mark valid data that meets the criteria, and generate valid data tags; Extract the situational data corresponding to the valid data tags and generate valid situational data for the nodes; Bind valid situational data of nodes with corresponding conflicting node identifiers to generate valid data node binding results; Invalid conflict data in the marker node situation set that does not meet the data validity criteria are marked with invalid data tags. Remove invalid conflict data corresponding to invalid data markers from the node situation set, and retain the valid node situation data in the valid data node binding results; Integrate all valid state data from all nodes, add data validity markers, and generate a verified set of node states. The set of verified node states is stored in association with the conflict verification results.

10. An intelligent building 3D security situational awareness system based on artificial intelligence, characterized in that, The AI-based intelligent building 3D security situation awareness system includes a processor and a memory, the memory and the processor being connected. The memory is used to store programs, instructions or code, and the processor is used to execute the programs, instructions or code in the memory to implement the AI-based intelligent building 3D security situation awareness method according to any one of claims 1-9.

Images