Intelligent property safety management platform based on data analysis

Abstract

The invention discloses an intelligent property safety management platform based on data analysis, and relates to the technical field of data management, and the platform comprises the steps: collecting property monitoring equipment, and constructing a monitoring data network; obtaining a property sample database, and determining a security event level and data relevance; constructing a training data set to obtain a security analysis decision model; establishing mapping association between the monitoring data and the security events, and segmenting a plurality of monitoring data blocks; obtaining a security identification event; and generating safety early warning information based on the matching early warning mode. The technical problems of incomplete property safety management coverage, low management efficiency and hysteresis of safety early warning caused by difficulty in coping with complex and changeable property management requirements and lack of deep integration and analysis capability on massive real-time monitoring data in existing property safety management are solved, intelligent monitoring management of property safety is realized, and the property safety management efficiency is improved. The technical effect of improving real-time performance and accuracy of property safety early warning is achieved.

Description

Technical Field

[0001] The present application relates to the technical field related to data management, and specifically to a smart property security management platform based on data analysis. Background Art

[0002] In the context of accelerated urbanization today, property management, as an important part of urban governance, is directly related to the quality of life of residents and social stability due to its security and intelligence level. With the rapid development of technologies such as the Internet of Things, big data, and artificial intelligence, the concept of smart property has emerged, aiming to improve the efficiency and effectiveness of property management through technological innovation. However, under the traditional property management model, security monitoring often relies on a simple combination of manual inspections and video surveillance, and is unable to deeply integrate and analyze massive real-time monitoring data. This method is not only inefficient, but also under the traditional method, the identification and early warning of security incidents are based on simple rule matching, and there are problems such as delayed response and incomplete coverage. Faced with increasingly complex and changeable security threats such as fire, theft, and illegal intrusion, traditional means can no longer meet the needs of modern property management.

[0003] Therefore, at the current stage, the technologies related to property safety management have difficulties in coping with the complex and changeable property management needs and lack the ability to deeply integrate and analyze massive real-time monitoring data, which in turn leads to technical problems such as incomplete property safety management coverage, low management efficiency and delayed safety warnings. Summary of the invention

[0004] This application solves the technical problems of incomplete property safety management coverage, low management efficiency and delayed safety warnings by providing a smart property safety management platform based on data analysis. It realizes intelligent monitoring and management of property safety and achieves the technical effect of improving the immediacy and accuracy of property safety warnings.

[0005] The present application provides a smart property security management platform based on data analysis, the platform includes: a monitoring data network construction module, the monitoring data network construction module is used to collect property monitoring equipment and build a monitoring data network, the monitoring data network includes a data source, a layout location, a data coverage area, and cross-data; a property sample database acquisition module, the property sample database acquisition module is used to acquire a property sample database, perform data analysis based on the property sample database, and determine the security event level and data relevance; a security analysis decision model acquisition module, the security analysis decision model acquisition module is used to construct a training data set based on the security event level and data relevance, and obtain a security analysis A decision model; a mapping association establishment module, the mapping association establishment module is used to perform correlation analysis on the monitoring data network according to the security event level and data correlation, establish a mapping association between monitoring data and security events, and divide multiple monitoring data blocks; a security identification event acquisition module, the security identification event acquisition module is used to connect multiple monitoring data blocks, import the real-time monitoring data of each block into the security analysis decision model for security event identification, and obtain security identification events; a security warning information generation module, the security warning information generation module is used to match the security identification event with the adaptive warning module, and generate security warning information based on the matching warning method.

[0006] In a possible implementation, the monitoring data network construction module also performs the following processing: collecting a target distribution map of the property management area; configuring security events according to the attributes of the property management area to determine the security management type; parsing the monitoring data source according to the security management category, and searching for a full-range coverage solution for the target distribution map according to the monitoring sensitivity and monitoring range of the monitoring data source to locate the monitoring equipment distribution information; establishing a mapping relationship between the monitoring data source and the monitoring equipment distribution information, and setting the property monitoring equipment.

[0007] In a possible implementation, the monitoring data network construction module also performs the following processing: identifying the security risk probability of the target distribution map, and dividing the risk probability area of ​​the security management type; setting the control search weight of the risk probability area, wherein the control search weight corresponds to the risk probability; fitting the control coverage evaluation function according to the monitoring sensitivity and monitoring range of the monitoring data source corresponding to the security management type, combined with the control search weight of the risk probability area; based on the control coverage evaluation function, constructing an optimization space with the goal of maximizing the search coverage range and minimizing the number of deployed monitoring sources, and optimizing the layout plan of each monitoring data source; and obtaining the monitoring equipment distribution information according to the optimization result.

[0008] In a possible implementation, the monitoring data network construction module also performs the following processing: locate the location distribution of the property monitoring equipment according to the monitoring equipment distribution information, and establish a monitoring data map with the property monitoring equipment as nodes and the location distribution as edges; type label the nodes according to the monitoring data source, and identify the monitoring coverage coordinates and coverage scale according to the monitoring equipment distribution information and the monitoring sensitivity and monitoring range of the monitoring data source; locate the nodes and edges corresponding to the identified monitoring coverage coordinates and coverage scale, set coverage parameter labels, fit the coverage parameter labels and the monitoring data source type labels of the nodes into the monitoring data map, and construct the monitoring data network.

[0009] In a possible implementation, the property sample database acquisition module also performs the following processing: connecting to the local property learning sample library and the federated learning collaborative sample library to obtain the property sample database, including property safety events, property monitoring data, and identification and prevention results; classifying the property safety events according to their event categories and security risk levels, and setting up multi-classification clusters; clustering the property sample database based on the feature vectors of the multi-classification clusters to construct a multi-category sample set; fitting the identification influence relationship between the identification and prevention results and the property monitoring data based on the property monitoring data and the identification and prevention results in the multi-category sample set, and establishing a mapping association between the identification influence relationship and the property safety event to determine the security event level of the property safety event and its corresponding data correlation.

[0010] In a possible implementation, the security analysis decision model acquisition module also performs the following processing: according to the data correlation, the property monitoring data is screened, the data relationship between the security event level and the property monitoring data is marked, and the training data set is constructed; based on the identification and prevention results, the training data set is divided into a positive training set and a negative training set, the positive training set is a sample with correct identification and prevention results, and the negative training set is a sample with incorrect or unidentified identification and prevention results; the preset machine learning model is trained and converged through the positive training set and the negative training set to obtain the security analysis decision model.

[0011] In a possible implementation, the mapping association establishment module also performs the following processing: configuring the monitoring linkage scope according to the security event level and the security event category; configuring the event sensitivity of the monitoring data according to the data correlation; using the security event level to configure the sensitive threshold to filter the event sensitivity of the monitoring data and determine the matching monitoring data source; segmenting the monitoring data network according to the monitoring linkage scope and the matching monitoring data source to obtain multiple monitoring data blocks, wherein the monitoring data block includes the monitoring data source, monitoring data and monitoring location in the segmented area, wherein each monitoring data block corresponds to a security event.

[0012] In a possible implementation, the monitoring data network construction module further performs the following processing: the control coverage evaluation function expression is: Among them, w i is the search weight for the control of the ith risk probability area, R i is the monitoring range of the i-th monitoring data source, k i is the sensitivity constant of the i-th monitoring data source, ρ i is the threshold of the i-th monitoring data source, x i is the actual monitoring data of the ith monitoring data source, and e is the base of the natural logarithm.

[0013] It is planned to collect property monitoring equipment and build a monitoring data network through the data analysis-based smart property security management platform proposed in this application. The monitoring data network includes data sources, layout locations, data coverage areas, and cross-data; obtain a property sample database, perform data analysis based on the property sample database, and determine the security event level and data correlation; construct a training data set based on the security event level and data correlation to obtain a security analysis decision model; perform correlation analysis on the monitoring data network based on the security event level and data correlation, establish a mapping association between monitoring data and security events, and divide multiple monitoring data blocks; connect multiple monitoring data blocks, and import the real-time monitoring data of each block into the security analysis decision model for security event identification to obtain security identification events; match the security identification events with the adaptive warning module, and generate security warning information based on the matching warning method. It solves the technical problems of existing property safety management, such as difficulty in coping with complex and changeable property management needs, lack of deep integration and analysis capabilities for massive real-time monitoring data, which in turn leads to incomplete property safety management coverage, low management efficiency and delayed safety warnings. It realizes intelligent monitoring and management of property safety, and achieves the technical effect of improving the immediacy and accuracy of property safety warnings. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solution of the embodiment of the present disclosure, the accompanying drawings of the embodiment of the present disclosure are briefly introduced below. A flow chart is used in the present application to illustrate the operations performed by the platform according to the embodiment of the present application. It should be understood that the previous or following operations are not necessarily performed precisely in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or more operations can be removed from these processes.

[0015] Figure 1 A schematic diagram of the structure of a smart property safety management platform based on data analysis provided in an embodiment of the present application;

[0016] Figure 2 A schematic diagram of the execution flow of the monitoring data network construction module in the smart property security management platform based on data analysis provided in the embodiment of the present application.

[0017] Explanation of the reference numerals: monitoring data network construction module 10, property sample database acquisition module 20, security analysis decision model acquisition module 30, mapping association establishment module 40, security identification event acquisition module 50, security warning information generation module 60. DETAILED DESCRIPTION

[0018] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0019] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0020] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict, and the terms "first\second" involved are merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "including" and "having" and any variations are intended to cover non-exclusive inclusions, for example, a process, platform, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or components that are not clearly listed or inherent to these processes, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by technicians in the technical field of this application. The terms used herein are for the purpose of describing the embodiments of the present application only.

[0021] The present application embodiment provides a smart property security management platform based on data analysis, such as Figure 1 As shown, the platform includes:

[0022] The monitoring data network construction module 10 is used to collect property monitoring equipment and build a monitoring data network. The monitoring data network includes data source, layout location, data coverage area, and cross data. The monitoring data is collected through various monitoring devices in the property area, and then the collected data of various monitoring devices are integrated to build a monitoring data network. The monitoring data network includes data source, layout location, data coverage area, and cross data. Specifically, the data source refers to the source of the monitoring data, that is, the monitoring equipment itself. In the property monitoring system, the data source mainly includes various types of monitoring cameras, infrared sensors, access control systems, vehicle identification systems, etc. These monitoring equipment generate a large amount of monitoring data by capturing video images, identifying people and vehicles entering and leaving, and monitoring environmental changes. The layout location refers to the installation location of the monitoring equipment in the property area. A reasonable layout location can ensure the effectiveness and comprehensiveness of the monitoring data, including public areas (such as halls, doors, corridors, elevators, etc.), vehicle parking areas, community facilities (such as swimming pools, gyms, children's play areas, etc.), community entrances and exits and internal passages, fire passages, stairways, etc.; the data coverage area refers to the physical space that the monitoring equipment can capture and record The scope depends on the performance of the monitoring equipment (such as the resolution, viewing angle, focal length of the camera, etc.) and the height and angle of the deployment location. In the property monitoring system, the data coverage area needs to ensure that there are no dead angles or blind spots, so as to be able to comprehensively and accurately record various situations in the property area; cross data refers to the intersection or association between data from different monitoring devices, which may include time association, space association and person association. Time association refers to the data captured by different monitoring devices in the same time period. The timestamp can be used for correlation analysis to understand the sequence and causal relationship of events; spatial association refers to the positional relationship between different monitoring devices in the physical space. The location information in the data (such as longitude and latitude, floor number, room number, etc.) can be used for correlation analysis to understand the specific location and spatial distribution characteristics of the event; person association refers to the use of technical means such as face recognition and license plate recognition to correlate and analyze the information of people or vehicles captured in different monitoring devices to track their movement trajectories and activity patterns.

[0023] The property sample database acquisition module 20 is used to acquire a property sample database, perform data analysis based on the property sample database, and determine the security event level and data relevance. The property sample database contains the original data collected by various monitoring devices in the property area, as well as the data after preliminary processing, which may include but is not limited to video images, environmental parameters, personnel flow information, vehicle entry and exit records, etc., as well as historical security event records, maintenance records and other related information. The data in the property sample database usually comes from the data collected in real time by the property monitoring equipment and the data recorded by the property internal management system. Performing data analysis based on the property sample database means to conduct in-depth mining and analysis of various monitoring data, event records, environmental parameters and other information stored in the database, extract valuable information, and determine the security event level and data relevance, so as to better understand and respond to the security situation in the property area. Among them, the security event level refers to the classification of security events into different levels based on factors such as the nature, severity, and scope of impact of the event. Specifically, potential security events are identified from the property sample database, such as abnormal behavior captured by surveillance cameras, environmental changes detected by environmental monitoring equipment, or owners and tenants. The identified events are classified according to their nature, such as fire, theft, intrusion, equipment failure, etc., and the events are evaluated according to the severity, urgency, scope of impact and other factors of the events, such as being divided into different security event levels such as minor, general, serious, and urgent. Data correlation refers to the connection or correlation between different data sources or data items, which is of great significance for discovering potential safety hazards and predicting future event trends. In addition to time correlation and space correlation, it may also include event correlation, that is, analyzing the causal relationship or mutual influence between different events. For example, by analyzing the correlation between fire events and environmental monitoring data, abnormal changes in parameters such as ambient temperature and smoke concentration before the fire can be found, thereby early warning and prevention of fire. Equipment correlation, that is, analyzing the correlation between data of different monitoring devices or systems. For example, by correlating and analyzing video surveillance with data from access control systems, vehicle identification systems, etc., all-round tracking and monitoring of targets such as personnel and vehicles can be achieved.

[0024] The security analysis decision model acquisition module 30 is used to construct a training data set according to the security event level and data relevance to obtain a security analysis decision model. By using the security event level and data correlation information in the historical data, a model that can automatically identify and predict security events is trained, thereby providing property companies with more intelligent and efficient security management decision support. Specifically, according to the security event level, the events are labeled to clarify the severity of each event. At the same time, according to the data correlation, the relevant data are labeled to reveal the connection between the data. The labeled data is preprocessed, including data cleaning (removing noise, outliers, etc.), data transformation (such as normalization, standardization, etc.) and data dimensionality reduction (such as feature selection, principal component analysis, etc.), etc., to improve data quality and reduce the complexity and overfitting risk of model training. Then, a training data set is constructed with the preprocessed data and divided into a training set and a test set. The training set is used to train the security analysis decision model, and the test set is used to evaluate the performance of the model. A prediction model is constructed based on a machine learning model (such as a decision tree, a neural network, a support vector machine, etc.), and a determined training set is used for training. The model parameters are iteratively optimized to enable the model to accurately identify and predict security events, and then the test set is used for evaluation. Finally, a security analysis decision model is obtained, which is used to more intelligently and accurately identify and predict security events in property management.

[0025] A mapping association establishment module 40 is used to perform a correlation analysis on the monitoring data network according to the security event level and data correlation, establish a mapping association between monitoring data and security events, and divide multiple monitoring data blocks. The monitoring data network is analyzed for correlation according to the security event level and data correlation, that is, by mining the inherent relationship in the monitoring data, the monitoring data and security events are closely linked, so as to achieve rapid response and effective prevention of security events. Specifically, the correlation between the monitoring data and security events in the monitoring data network is evaluated based on multiple dimensions, such as time, location, activity mode, object characteristics, etc. If a certain monitoring frequency records a suspicious activity that occurs at a specific time and place, and the activity has similarities with past security events, then the data has a high correlation. Then, data analysis is performed based on pattern recognition to find out the correlation between the monitoring data and the security events. The system identifies which monitoring data in the monitoring data network are associated with security events of specific types and levels based on the correlation analysis results. Then, based on the correlation analysis results, it establishes a mapping relationship between monitoring data and security events, that is, it associates a security event level and type with each monitoring data point (such as video clips, sensor readings, etc.). For example, a monitoring video is marked as a high-level security event, which helps to find potential signs of security events in the monitoring data and take preventive measures in advance. Finally, based on factors such as geographic location, monitoring objects, and data types, the monitoring data network is divided into multiple data blocks to manage the monitoring data network more finely and improve the efficiency and accuracy of data analysis.

[0026] The security identification event acquisition module 50 is used to connect multiple monitoring data blocks, import the real-time monitoring data of each block into the security analysis decision model for security event identification, and obtain security identification events. Connect multiple monitoring data blocks obtained by segmentation, and then import the monitoring data collected in real time in the monitoring data blocks (such as video images, environmental parameters, personnel flow information, vehicle in and out records, etc.) as input data into the constructed security analysis decision model for security event identification analysis. The security analysis decision model outputs the security identification event results, that is, obtains the security identification event, including the type of security event, time of occurrence, location, security level and scope of impact, etc.

[0027] The security warning information generation module 60 is used to match the security identification event with the adaptive warning module and generate security warning information based on the matching warning method. The adaptive warning module is a module that can generate warning information and dynamically adjust the warning strategy based on historical data, current environment, and user preferences, etc. It contains a variety of warning methods and rules for matching security identification events of different types and levels, and then generating corresponding warning information. Specifically, when a security identification event occurs, the system will match the relevant information of the event (such as event type, level, etc.) with the warning methods and rules in the adaptive warning module based on the urgency of the event, the importance of the location of the event, the security status of the current environment, etc. After the match is successful, the adaptive warning module generates corresponding security warning information according to the matching warning method, including a brief description of the security event, the time of occurrence, the location, the impact level, and the recommended countermeasures. Matching the security identification event with the adaptive warning module and generating security warning information based on the matching warning method ensures that when a potential or actual security threat is detected, the warning information can be quickly and accurately generated and released to improve the efficiency of the entire property safety management.

[0028] According to the embodiment of the present invention, the smart property safety management platform based on data analysis is used to solve the technical problems of the existing property safety management, such as the difficulty in coping with complex and changeable property management needs, the lack of deep integration and analysis capabilities of massive real-time monitoring data, and the incomplete coverage of property safety management, low management efficiency, and delayed safety warnings, so as to realize the intelligent monitoring and management of property safety and achieve the technical effect of improving the immediacy and accuracy of property safety warnings. The smart property safety management platform based on data analysis includes: a monitoring data network construction module 10, a property sample database acquisition module 20, a security analysis decision model acquisition module 30, a mapping association establishment module 40, a security identification event acquisition module 50, and a security warning information generation module 60.

[0029] The specific configuration of the monitoring data network construction module 10 will be described in detail below. The monitoring data network construction module 10 may further include: collecting a target distribution map of the property management area. The map data of the property management area is obtained through GIS (Geographic Information System) technology, and the accurate locations of buildings, facilities and other targets are marked on the map in combination with on-site surveys to obtain a target distribution map of the property management area, which is a graphical representation of the location, shape, size and other information of all important targets (such as buildings, facilities, roads, green belts, etc.) in the property management area.

[0030] The monitoring data network construction module 10 also includes configuring security events according to the attributes of the property management area and determining the type of security management. The attributes of the property management area include area size, building type, population density, surrounding environment, etc. Security event configuration refers to predicting the types of security events that may occur (such as fire, theft, intrusion, etc.) based on the attributes of the property management area, and formulating corresponding emergency plans and disposal processes, and configuring corresponding security resources according to the needs of the emergency plans, such as fire-fighting equipment, monitoring equipment, security personnel, etc. After the security event configuration, the type of security management is determined according to the actual situation of the property management area, such as comprehensive security management, key area security management (such as fire passages, elevator rooms, distribution rooms, etc.), and special security management (such as fire, theft, etc.).

[0031] The monitoring data network construction module 10 also includes parsing the monitoring data source according to the security management category, and searching for a full-range coverage plan for the target distribution map according to the monitoring sensitivity and monitoring range of the monitoring data source, and locating the monitoring equipment distribution information. Monitoring data sources refer to various information sources used for security monitoring, including but not limited to video surveillance systems, intrusion detection systems, environmental monitoring systems, personnel positioning systems, etc. Parsing monitoring data sources according to security management categories specifically refers to determining the specific type of monitoring data source, such as video streams, sensor data, log data, etc., understanding the format and encoding method of monitoring data, and evaluating the accuracy and reliability of monitoring data; searching and determining the best monitoring equipment distribution plan according to the monitoring sensitivity and monitoring range of the monitoring data source, wherein monitoring sensitivity refers to the sensitivity of the monitoring system to security incidents. A highly sensitive monitoring system can respond and alarm faster, but it may also lead to an increase in the false alarm rate. When determining the monitoring sensitivity, it is necessary to weigh the actual situation and security needs of the property management area. For example, For example, in areas with dense crowds or important facilities, a higher monitoring sensitivity may need to be set; the monitoring scope refers to the physical area or logical scope covered by the monitoring system. In the property management area, the monitoring scope should cover all key areas and potential risk points to ensure that there are no blind spots in monitoring; specifically, clarify the security needs and monitoring objectives of the property management area, conduct on-site surveys of the property management area, understand the terrain, building layout, personnel flow, etc., design a distribution plan for monitoring equipment based on demand analysis and on-site survey results, and consider factors such as the number, type, location, and angle of the monitoring equipment to achieve full coverage and efficient monitoring. After determining the distribution plan for the monitoring equipment, locate the specific distribution information of the monitoring equipment, for example, assign a unique number to each monitoring device and determine the specific installation location of the monitoring equipment.

[0032] The monitoring data network construction module 10 also includes establishing a mapping relationship between the monitoring data source and the monitoring equipment distribution information, and setting the property monitoring equipment. The mapping relationship refers to establishing a clear correspondence between the monitoring data source (such as video stream, sensor data, etc.) and the specific monitoring equipment (such as camera, sensor, etc.), ensuring that the monitoring data can be accurately received, processed and analyzed by the corresponding monitoring equipment. Through the mapping relationship, the system can quickly identify and process data from different monitoring equipment, reduce the possibility of data confusion and false alarms, and improve the efficiency and accuracy of the monitoring system. Then, according to the security requirements and monitoring plan of the property management area, the monitoring equipment (such as camera, sensor, etc.) is installed in a suitable location, and corresponding configuration and debugging are performed to ensure that it can work normally and meet the monitoring requirements, including clarifying the security requirements and monitoring targets of the property management area, selecting appropriate monitoring equipment according to the results of the demand analysis, and installing and setting the monitoring equipment.

[0033] The specific configuration of the monitoring data network construction module 10 will be described in detail below. The monitoring data network construction module 10 may further include: identifying the safety risk probability of the target distribution map and segmenting the risk probability area of ​​the safety management type. Safety risk probability identification refers to performing safety risk assessment on each area on the target distribution map of the property management area to determine the possibility of a safety incident (i.e., risk probability) in each area, using a mathematical model combined with specific safety incident data, environmental factors, human factors, etc., to quantitatively assess the safety risk, derive a specific risk probability value, and determine various factors that may cause safety incidents, such as equipment failure, human operation errors, natural disasters, etc. After completing the safety risk probability identification, according to different safety management types (such as comprehensive safety management, key area safety management, special safety management, etc.), each area on the target distribution map is segmented according to the level of risk probability to form different risk probability areas, that is, according to the size of the risk probability, the area is divided into different levels such as high risk area, medium risk area, low risk area, etc. After the segmentation is completed, the target distribution map will clearly display the risk probability areas of different safety management types.

[0034] The monitoring data network construction module 10 also includes setting the control search weight of the risk probability area, and the control search weight corresponds to the risk probability. In security risk management, for the different risk probability areas that have been identified and segmented, they are given corresponding monitoring and search weight values, that is, the control search weight is set according to different risk probabilities. For example, assuming that the risk probability P is between 0 and 1, the control search weight W can be set by a nonlinear function (such as a logarithmic function, etc.), and the expression is as follows:

[0035] W = a × log (1 + P);

[0036] Where a is a constant used to adjust the weight. Gradual setting, set multiple weight levels according to different risk probabilities. For example: 0≤P<0.3 low risk, weight is 1; 0.3≤P<0.7 medium risk, weight is 2; 0.7≤P≤1.0 high risk, weight is 3; The control and search weight reflects the importance and priority of each area in the overall security management, so that monitoring and search resources can be more effectively configured and used when resources are limited. Areas with high risk probability are more likely to have security incidents, and higher control and search weights should be assigned. Conversely, areas with low risk probability correspond to lower control and search weights.

[0037] The monitoring data network construction module 10 also includes fitting a control coverage evaluation function according to the monitoring sensitivity and monitoring range of the monitoring data source corresponding to the security management type and the control search weight of the risk probability area. The control coverage evaluation function is used to evaluate the control coverage effect of the monitoring system. It comprehensively considers multiple factors such as the monitoring sensitivity of the monitoring data source, the monitoring range, and the control search weight of the risk probability area. Specifically, the input variables of the evaluation function are determined, including the monitoring sensitivity of the monitoring data source, the monitoring range, and the control search weight of the risk probability area. According to the characteristics of the input variables and actual needs, that is, to evaluate whether the control of a certain area has achieved the expected effect, a corresponding multivariate logistic regression model is established based on the logistic function, including data preparation, model construction, parameter estimation and model evaluation. These data are fitted to the multivariate logistic regression model to obtain the control coverage evaluation function, which accurately reflects the control coverage effect of the monitoring system. Among them, the dependent variable is a categorical variable, which is used to represent the category or result to be predicted. Multivariate logistic regression is applicable to multinomial categorical variables, that is, the dependent variable has three or more categories (usually called classes). The dependent variable does not need to obey any specific distribution because logistic regression uses maximum likelihood estimation, which is applicable to categorical variables; the independent variable can be categorical Variables or continuous variables, such as the control coverage effect (good, medium, poor), can be converted into numerical form by one-hot encoding or dummy variable encoding. The independent variables should be as independent as possible to avoid multicollinearity, otherwise it may affect the stability and accuracy of the model. For continuous variables, if the numerical ranges vary greatly, it is recommended to standardize or normalize them to improve the training effect of the model. Specifically, collect monitoring data from actual monitoring data sources, assessment results of risk probability areas and other data, and preprocess the data, including cleaning, denoising and standardization. For example, normalize the sensitivity and monitoring range to ensure that variables of different dimensions can be effectively compared in the model. Through verification and adjustment of actual data, the evaluation function is continuously adjusted and optimized through the actual effect of the monitoring system. For example, the risk assessment results in some areas may change. The model can update the control search weights in real time to adapt to new risks, thereby continuously optimizing the parameters of the evaluation function to ensure that it can accurately evaluate the control coverage effect of the monitoring system.

[0038] The monitoring data network construction module 10 also includes, based on the control coverage evaluation function, with the goal of maximizing the search coverage and deploying the least monitoring sources, constructing an optimization space, and optimizing the deployment plan of each monitoring data source. With the optimization goal of maximizing the search coverage and deploying the least monitoring sources, considering the monitoring sensitivity and monitoring range of the monitoring data source and the control search weight of the risk probability area and other constraints, constructing the optimization space, forming a multi-objective optimization, generating multiple possible monitoring data source deployment plans according to the actual situation and monitoring needs of the target area, using the control coverage evaluation function to evaluate each alternative plan, calculating the search coverage and the number of deployed monitoring sources of each plan, adjusting and optimizing the plan according to the evaluation results, so as to improve the search coverage and reduce the number of deployed monitoring sources, and finally determining the optimal monitoring data source deployment plan.

[0039] The monitoring data network construction module 10 also includes obtaining the monitoring equipment distribution information according to the optimization result. According to the optimization result, that is, the optimal monitoring data source layout plan, the monitoring equipment distribution information is obtained, including the type, quantity, location and corresponding monitoring parameters of the monitoring equipment. According to the monitoring equipment distribution information, the monitoring equipment can be actually deployed in the property management area, and configured and debugged, so as to realize real-time monitoring and early warning of security risks in the property management area and improve the level of security management.

[0040] The specific configuration of the monitoring data network building module 10 will be described in detail below. Figure 2 As shown, the monitoring data network construction module 10 may further include: according to the monitoring equipment distribution information, locate the location distribution of the property monitoring equipment, take the property monitoring equipment as a node, and take the location distribution as an edge to establish a monitoring data map. According to the monitoring equipment distribution information, locate the location distribution of the property monitoring equipment, regard each monitoring equipment as a node, and abstract the location distribution relationship as an edge, that is, the node represents a specific monitoring equipment entity, and the edge can represent the direct physical distance between the monitoring equipment, the overlap of the line of sight coverage, or a connection based on a certain logical relationship, and then use the knowledge of graph theory and data structure to combine the nodes and edges to form a complete monitoring data map. Each node in the map is connected to other nodes through an edge to form a complex network structure.

[0041] The monitoring data network construction module 10 also includes labeling the nodes according to the monitoring data source, and identifying the monitoring coverage coordinates and coverage scale according to the monitoring device distribution information and the monitoring sensitivity and monitoring range of the monitoring data source. According to the characteristics and functions of the monitoring data source, multiple type labels are defined to accurately reflect the type, function, and technical parameters of the monitoring device. According to the data source characteristics of each monitoring device, the corresponding type label is assigned to the node in the map. For example, if a node represents a high-definition camera, it is assigned a "high-definition camera" type label. Then, the distribution information of the monitoring device (such as location coordinates, installation height, direction, etc.) is combined with the monitoring sensitivity and monitoring range information. Through spatial analysis and calculation, the specific area range that each monitoring device can cover is determined, that is, the monitoring coverage coordinates and coverage scale are identified. For example, through geometric calculation, it is determined that the monitoring range of a camera is a circular area centered on the device, and its coverage coordinates include the boundary points of the circle. The coverage area boundary of each monitoring device is marked on the map to form the monitoring coverage coordinates. According to the monitoring coverage coordinates, the coverage scale of each monitoring device is calculated and determined, that is, the actual size and shape of the coverage area is determined, such as the radius, rectangular, and rectangular of the circular area. The length and width of the fan-shaped area, or the angle and radius of the fan-shaped area, are used to comprehensively evaluate the coverage coordinates and coverage scales of all monitoring devices, including evaluating whether there are monitoring blind spots (completeness of coverage range) of the monitoring devices, whether there are overlapping areas, and calculation of monitoring device coverage density. It is evaluated whether the coverage of the monitoring devices can completely cover the designated area to ensure that there are no monitoring blind spots (i.e., uncovered areas). For example, the coverage of all monitoring devices is calculated and superimposed using geometric methods, and whether there are uncovered areas by analyzing the coverage radius, angle, and relative position of the devices; whether there are too many overlaps in the coverage of the monitoring devices, that is, the overlapping area of ​​the coverage areas between adjacent monitoring devices is calculated to ensure that the overlapping area is within a reasonable range; the coverage density of the monitoring devices in the entire area is evaluated to ensure that the coverage density of the key areas is high, while the coverage density of the general areas is moderate. For example, the area is divided into sub-areas (such as grids), and the number of monitoring devices in each sub-area is calculated to determine whether the preset coverage density requirements are met.

[0042] The monitoring data network construction module 10 also includes positioning the nodes and edges corresponding to the identified monitoring coverage coordinates and coverage scale, setting coverage parameter labels, fitting the coverage parameter labels and the monitoring data source type labels of the nodes into the monitoring data map, and constructing the monitoring data network. According to the identified monitoring coverage coordinates and coverage scales (i.e., the area that each monitoring device can cover), the corresponding nodes and edges are positioned, i.e., the nodes and their positional relationships contained in the coverage coordinate range, and according to the coverage parameter information of each monitoring device, corresponding coverage parameter labels are assigned to it, and the coverage parameter labels are labels used to describe the coverage characteristics of the monitoring device, including but not limited to coverage coordinates, coverage area, maximum monitoring distance, minimum monitoring target size, etc., and then the set coverage parameter labels and the monitoring data source type labels of the nodes (such as high-definition cameras, temperature sensors, etc.) are integrated into the monitoring data map. After completing node positioning, edge definition and label fitting, the monitoring data map forms a preliminary monitoring data network, which is connected together by nodes (monitoring devices) and edges (positional relationships) to form a comprehensive description of the monitoring environment. The monitoring data network can perform real-time monitoring, historical data query, anomaly detection, event warning, etc.

[0043] The specific configuration of the property sample database acquisition module 20 will be described in detail below. The property sample database acquisition module 20 may further include: connecting to a local property learning sample library and a federated learning collaborative sample library to acquire the property sample database, including property safety events, property monitoring data, and identification and prevention results. The local property learning sample library refers to a database containing information such as historical property safety events, property monitoring data, and identification and prevention results based on these data. The federated learning collaborative sample library refers to a database formed by collaboratively integrating sample data from multiple participants (such as different property companies, related institutions, etc.) through federated learning technology. Specifically, property safety events, property monitoring data, identification and prevention results, etc. are obtained from the local property learning sample library and the federated learning collaborative sample library to form a property sample database, which not only contains historical data within the company, but also integrates data resources from other participants. Among them, property safety events record various security incidents that have occurred in the past, such as theft, fire, natural disasters, etc., as well as the specific circumstances and handling results of these incidents; property monitoring data is data collected through the property monitoring system, including video recordings, image recognition results, sensor data, etc., reflecting the real-time situation in the property area; identification and prevention results refer to the results obtained after identification and analysis based on property monitoring data and other information through technical means such as artificial intelligence and machine learning, which are used to guide property safety management and prevention and control work.

[0044] The property sample database acquisition module 20 classifies the property security events according to their event categories and security risk levels, and sets up multi-classification clusters. Multi-classification clusters are set up according to the event categories and security risk levels of the property security events, that is, property security events with similar characteristics or attributes are classified into different clusters, and each cluster represents a set of events with common characteristics. For example, the theft and robbery cluster includes all property security events involving theft and robbery, which usually have a lower security risk level (unless they involve large amounts of property or violent acts); the fire and natural disaster cluster includes all property security events related to fire and natural disasters, which usually have a higher security risk level and may cause serious casualties and property losses, etc.

[0045] The property sample database acquisition module 20 clusters the property sample database based on the feature vector of the multi-classification cluster to construct a multi-class sample set. Features related to the multi-classification cluster are extracted from the property sample database to construct a feature vector for each data point, including the type of property safety event, time of occurrence, location, safety risk level, key indicators in the monitoring data (such as human flow, abnormal behavior detection results, etc.), and identification and prevention results. Specifically, for the type of safety event, one-hot encoding can be used to represent different event types as a multidimensional vector, with the corresponding type position being 1 and other positions being 0. For example, there are 5 types of events, and fire may be encoded as 1, 0, 0, 0, 0; the time of event occurrence can be represented by 0 to 6 from Monday to Sunday, and the hour part is directly used as a numerical feature; the location category (different buildings or areas) is also encoded using one-hot encoding; the safety risk level is directly represented by the level value (such as low, medium, and high correspond to 1, 2, and 3 respectively); the human flow per minute is used as a numerical features, and the abnormal behavior detection results are binarized and represented by 0 and 1; the results of identification and prevention (such as whether the prevention and control is successful) are converted into numerical form, for example, whether the prevention and control is successful can be represented by binary (0 / 1); the extracted feature vector is used as input, and K-means clustering is used to perform cluster analysis on the data in the property sample database, and the data points are divided into different groups or categories according to the similarity or distance between the feature vectors. Specifically, K points are randomly selected in the data set as the initial clustering centers, and for each data point in the data set, its distance to each clustering center is calculated (such as Euclidean distance), and each data point is assigned to the cluster to which the nearest clustering center belongs, and the mean (or median, etc.) of all data points in each cluster is recalculated, and the mean is used as the new clustering center. Multiple iterations are performed to obtain multiple cluster clusters, that is, a multi-category sample set.

[0046] The property sample database acquisition module 20 fits the identification influence relationship between the identification and prevention results and the property monitoring data according to the property monitoring data in the multi-category sample set and the identification and prevention results, and establishes a mapping association between the identification influence relationship and the property safety event, so as to determine the safety event level of the property safety event and its corresponding data correlation. Extract property monitoring data and corresponding identification and prevention results from multi-category sample sets, including video surveillance recordings, sensor data (such as temperature, humidity, smoke concentration, etc.), access control records, and conclusions drawn after processing monitoring data through an intelligent analysis system, such as abnormal behavior detection, fire warning, intrusion alarm, etc. Match and compare the identification and prevention results with the property monitoring data to verify the accuracy and effectiveness of the identification results, and identify the correlation between the prevention and control results and the property monitoring data. For example, a certain abnormal behavior may be related to abnormal personnel flow within a certain period of time, or a certain equipment failure may be related to changes in ambient temperature. Match and map the identified impact relationship with the property security event, which helps to quickly locate potential security events when abnormal situations are discovered, and take corresponding prevention and control measures. Finally, after determining the security event level, further analyze which property monitoring data the event is associated with, and the degree of these associations, that is, to determine the impact relationship between the security event level of the property security event and its corresponding data.

[0047] The specific configuration of the security analysis decision model acquisition module 30 will be described in detail below. The security analysis decision model acquisition module 30 may further include: according to the data correlation, screening the property monitoring data, marking the data relationship between the security event level and the property monitoring data, and constructing the training data set. According to the determined data correlation, the property monitoring data related to the specific security event level is screened out from the original property monitoring data. These data should be monitoring records that can directly or indirectly reflect the security event level. The screened data are labeled to clarify which security event level each data is associated with. The labeled data is organized into a training data set, which contains multiple samples, each of which contains a set of property monitoring data and corresponding security event level labels.

[0048] The safety analysis decision model acquisition module 30 also includes, based on the identification and control results, dividing the training data set into a positive training set and a negative training set, wherein the positive training set is a sample with correct identification and control results, and the negative training set is a sample with incorrect or unidentified identification and control results. According to the identification and prevention results, the training data set is divided into a positive training set and a negative training set. Specifically, the identification and prevention results of each sample are evaluated, including comparing the prevention and control results with some form of "real" or "standard" results. The positive training set contains samples whose identification and prevention results match the "real" or "standard" results. The identification and prevention results of these samples are correct. For example, if an algorithm correctly identifies abnormal behavior in a surveillance video and classifies it as a "high-risk event", and manual review also confirms this, then this sample belongs to the positive training set; the negative training set contains samples whose identification and prevention results do not match the "real" or "standard" results, or samples that are not correctly identified for some reason (such as data quality issues, limitations of the recognition algorithm, etc.), including samples with incorrect recognition results (i.e., non-high-risk events are misclassified as high-risk events, or high-risk events are misclassified as non-high-risk events) and unrecognized (i.e., the algorithm fails to detect actual abnormal behavior).

[0049] The security analysis decision model acquisition module 30 also includes training and converging the preset machine learning model through the positive training set and the negative training set to obtain the security analysis decision model. The decision model is constructed based on a convolutional neural network or a support vector machine, and the positive training set and the negative training set are provided as input data to the machine learning model, which are usually divided into features (i.e., various indicators or attributes in the property monitoring data) and labels (i.e., the corresponding security event level). The decision model processes these data and finds the best mapping relationship between features and labels so that new data can be accurately classified. During the training process, the model will continuously and iteratively adjust its internal parameters (such as weights and biases) to minimize the prediction error on the training set. As the number of iterations increases, the prediction error of the model on the training set will gradually decrease. When the error is reduced to a certain extent, or the model performance is no longer significantly improved on the verification set, it is considered that the model has converged, and a security analysis decision model is obtained, which is used to receive new property monitoring data as input and output the corresponding security event level prediction result.

[0050] The specific configuration of the mapping association establishment module 40 will be described in detail below. The mapping association establishment module 40 may further include: configuring the monitoring linkage range according to the security event level and security event category. After clarifying the security event level and category, the monitoring linkage range is configured according to this information. Specifically, according to the type of event, appropriate monitoring equipment is selected. For example, for terrorist attacks, it may be necessary to focus on deploying high-definition cameras, face recognition systems, behavioral analysis systems, etc., and delineate key monitoring areas according to the areas or places where the events may occur. For example, dense monitoring points are set around large-scale events or important facilities to ensure coverage without blind spots, and then determine the regional scope of linkage.

[0051] The mapping association establishment module 40 also includes configuring the event sensitivity of the monitoring data according to the data relevance. First, the monitoring data with the strongest relevance to a specific security event is identified, and different event sensitivity levels are set for these data based on the data relevance and importance. The sensitivity levels can be divided according to actual conditions, such as high, medium, and low sensitivity levels.

[0052] The mapping association establishment module 40 also includes using the security event level to configure the sensitivity threshold to screen the event sensitivity of the monitoring data and determine the matching monitoring data source. According to the determined security event level and the sensitivity threshold configured for the level, the monitoring data is evaluated for sensitivity in real time or regularly, so as to screen out the monitoring data source that matches the current security event level, improve the efficiency and accuracy of the monitoring system, and ensure that the relevant monitoring data can be quickly located and processed when an event occurs, wherein the sensitivity threshold is set according to the security event level and is used to evaluate the sensitivity of the monitoring data. Different levels of security events correspond to different sensitivity thresholds. Specifically, the monitoring data is collected and analyzed in real time, and the sensitivity of the data is evaluated according to the preset sensitivity threshold. When the data sensitivity reaches or exceeds the threshold, it is considered that the data matches the current security event level, and the monitoring data source that matches the current security event level is screened out, which may include different types of monitoring equipment such as cameras, sensors, and alarms.

[0053] The mapping association establishment module 40 also includes segmenting the monitoring data network according to the monitoring linkage range and the matching monitoring data source to obtain multiple monitoring data blocks, wherein the monitoring data block includes the monitoring data source, monitoring data and monitoring location in the segmented area, wherein each monitoring data block corresponds to a security event. According to the determined monitoring linkage range and the matching monitoring data source, the overall monitoring data network is divided into multiple relatively independent monitoring data blocks, each monitoring data block contains a specific monitoring data source, monitoring data and monitoring location information in the area, and these blocks usually correspond to a specific security event, wherein in each monitoring data block, all monitoring data sources, monitoring data and monitoring location information in the area are integrated, that is, all monitoring devices in the area related to the current security event, the real-time or historical monitoring data generated by these devices, including video recordings, pictures, sensor readings, etc. and the geographical location information of the monitoring devices, help to quickly locate the specific location where the event occurred, making it more efficient and accurate to deal with security events.

[0054] The specific configuration of the monitoring data network construction module 10 will be described in detail below. The monitoring data network construction module 10 may further include: the control coverage evaluation function expression is: Among them, w i is the search weight for the control of the ith risk probability area, R i is the monitoring range of the i-th monitoring data source, k i is the sensitivity constant of the i-th monitoring data source, ρ i is the threshold of the i-th monitoring data source, x i is the actual monitoring data of the ith monitoring data source, and e is the base of the natural logarithm. is a logistic function that converts the actual monitoring data x i Converted to a value between 0 and 1 to represent the monitoring sensitivity.

[0055] Although the present application makes various references to certain components in the platform according to the embodiments of the present application, any number of different components may be used and run on the user terminal and / or server, and the various units and components included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be implemented; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0056] The above specific implementations do not constitute a limitation on the protection scope of this application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of this application should be included in the protection scope of this application.

Claims

1. A smart property security management platform based on data analysis, characterized by: The smart property safety management platform based on data analysis includes: A monitoring data network construction module, which is used to collect property monitoring equipment and build a monitoring data network, which includes data sources, layout locations, data coverage areas, and cross-data; A property sample database acquisition module, which is used to acquire a property sample database, perform data analysis based on the property sample database, and determine the security event level and data relevance; A security analysis decision model acquisition module, wherein the security analysis decision model acquisition module is used to construct a training data set according to the security event level and data relevance to obtain a security analysis decision model; A mapping association establishment module, the mapping association establishment module is used to perform correlation analysis on the monitoring data network according to the security event level and data relevance, establish a mapping association between monitoring data and security events, and divide multiple monitoring data blocks; A security identification event acquisition module, which is used to connect multiple monitoring data blocks, import the real-time monitoring data of each block into the security analysis decision model to identify security events and obtain security identification events; A safety warning information generation module is used to match the safety identification event with the adaptive warning module and generate safety warning information based on the matching warning mode.

2. The smart property safety management platform based on data analysis as claimed in claim 1, characterized in that: The monitoring data network construction module executes the following steps: Collect target distribution map of property management area; Perform security event configuration according to the attributes of the property management area and determine the security management type; Parse the monitoring data source according to the security management category, and search the target distribution map for a full-range coverage solution according to the monitoring sensitivity and monitoring range of the monitoring data source to locate the monitoring equipment distribution information; A mapping relationship between the monitoring data source and the monitoring equipment distribution information is established, and the property monitoring equipment is set.

3. The smart property safety management platform based on data analysis as claimed in claim 2, characterized in that: The monitoring data network construction module executes the following steps: Performing safety risk probability identification on the target distribution map, and segmenting the risk probability areas of the safety management type; Setting a control search weight for the risk probability area, wherein the control search weight corresponds to the risk probability; According to the monitoring sensitivity and monitoring range of the monitoring data source corresponding to the security management type, respectively, combined with the control search weight of the risk probability area, a control coverage evaluation function is fitted; Based on the control coverage evaluation function, with the goal of searching for the largest coverage and the least number of monitoring sources, an optimization space is constructed to optimize the layout of each monitoring data source; According to the optimization result, the monitoring equipment distribution information is obtained.

4. The smart property safety management platform based on data analysis as claimed in claim 3, characterized in that: The monitoring data network construction module executes the following steps: According to the monitoring equipment distribution information, locate the location distribution of the property monitoring equipment, and establish a monitoring data map with the property monitoring equipment as a node and the location distribution as an edge; Type-label the nodes according to the monitoring data source, and identify monitoring coverage coordinates and coverage scale according to the monitoring device distribution information and the monitoring sensitivity and monitoring range of the monitoring data source; The nodes and edges corresponding to the identified monitoring coverage coordinates and coverage scale are located, coverage parameter labels are set, the coverage parameter labels and the monitoring data source type labels of the nodes are fitted into the monitoring data graph, and the monitoring data network is constructed.

5. The smart property safety management platform based on data analysis as claimed in claim 1, characterized in that: The steps performed by the property sample database acquisition module include: Connect to the local property learning sample library and the federated learning collaborative sample library to obtain the property sample database, including property safety events, property monitoring data, and identification and prevention results; Classify the property safety incidents according to their event categories and safety risk levels, and set up multi-classification clusters; Clustering the property sample database based on the feature vectors of the multi-classification clusters to construct a multi-class sample set; According to the property monitoring data in the multi-category sample set and the identification and prevention results, the identification influence relationship between the identification and prevention results and the property monitoring data is fitted, and a mapping association between the identification influence relationship and the property safety incident is established to determine the safety incident level of the property safety incident and its corresponding data correlation.

6. The smart property safety management platform based on data analysis as claimed in claim 5, characterized in that: The steps performed by the safety analysis decision model acquisition module include: According to the data correlation, the property monitoring data is screened, the data relationship between the security event level and the property monitoring data is marked, and the training data set is constructed; Based on the identified prevention and control results, the training data set is divided into a positive training set and a negative training set, wherein the positive training set is a sample with a correct identification and prevention result, and the negative training set is a sample with an incorrect or unidentified identification and prevention result; The preset machine learning model is trained and converged through the positive example training set and the negative example training set to obtain the security analysis decision model.

7. The smart property safety management platform based on data analysis as claimed in claim 5, characterized in that: The mapping association establishing module executes the following steps: Configure the monitoring linkage scope according to the security event level and security event category; According to the data correlation, configure the event sensitivity of the monitoring data; Using the security event level to configure a sensitive threshold, the event sensitivity of the monitoring data is screened to determine a matching monitoring data source; The monitoring data network is segmented according to the monitoring linkage range and the matching monitoring data source to obtain multiple monitoring data blocks, wherein the monitoring data blocks include the monitoring data source, monitoring data and monitoring location in the segmented area, wherein each monitoring data block corresponds to a security event.

8. The smart property safety management platform based on data analysis as claimed in claim 3, characterized in that: The control coverage evaluation function expression is: Among them, w i is the search weight for the control of the ith risk probability area, R i is the monitoring range of the i-th monitoring data source, k i is the sensitivity constant of the i-th monitoring data source, ρ i is the threshold of the i-th monitoring data source, x i is the actual monitoring data of the ith monitoring data source, and e is the base of the natural logarithm.

Images