An intelligent control method and system for community digital information

Abstract

The present invention relates to the technical field of data processing, and in particular to an intelligent control method and system for community digital information. The method includes the following steps: obtaining historical community device data; performing an optimized design of the data interface of the device based on the historical community device data to generate a device identification data interface; using the device identification data interface to receive data for the data collection task of the optimized community device to obtain effective community device data; extracting unlabeled personnel behavior data from the effective community device data to generate unlabeled personnel behavior data; performing behavior anomaly analysis on the unlabeled personnel behavior data to generate unlabeled personnel behavior anomaly analysis data; and transmitting the unlabeled personnel behavior anomaly analysis data to an automated warning engine to execute an automated warning control task. The present invention realizes more intelligent control of community digital information.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular, to an intelligent control method and system for community digital information. Background Art

[0002] With the rapid development of social informatization and the complexity of community management in today's society, the popularization of digital technology and communities are faced with a huge and diverse information flow, covering all aspects such as residents, equipment, safety, and environment. Effectively managing this information is crucial for improving the operation efficiency of communities and the quality of life of residents. By using intelligent control of community digital information to enhance the refinement level of community management, comprehensively understand the operation status of communities, achieve reasonable allocation of resources and timely early warning of risks, so that community management is more scientific and efficient, promote the sense of participation and satisfaction of community residents, apply intelligent information control, enable residents to more conveniently obtain community services and participate in community activities, realize information sharing and interaction, improve residents' sense of identity and satisfaction with the community, and form a more closely-knit community group. And by real-time monitoring and analyzing the safety information inside and outside the community, potential safety risks can be quickly responded to, and the emergency response ability of the community can be improved, thus ensuring the life and property safety of residents. However, the traditional intelligent control method of community digital information fails to timely monitor whether there are abnormal operating conditions of community equipment, resulting in incomplete collection of community digital information by community equipment, and for unlabeled foreign personnel, it is necessary to manually judge whether there are abnormal behaviors, consuming manpower and time, and making the early warning feedback of the community not timely enough. Summary of the Invention

[0003] Based on this, the present invention provides an intelligent control method and system for community digital information to solve at least one of the above technical problems.

[0004] To achieve the above object, an intelligent control method for community digital information includes the following steps:

[0005] Step S1: Obtain historical community equipment data; design a data interface for the equipment based on the historical community equipment data to generate an equipment data interface;

[0006] Step S2: Design an equipment identity identifier for the equipment data interface according to the historical community equipment data to generate equipment identity identifier data; optimize the data interface identity identifier of the equipment data interface based on the equipment identity identifier data to generate an equipment identifier data interface;

[0007] Step S3: When the community device performs data collection tasks, collect device signals from the community device to generate device signals; classify the community devices with signal anomalies based on the device signals to obtain effective community devices and ineffective community devices respectively; feedback the ineffective community devices to the terminal to execute the repair control task for the ineffective community devices.

[0008] Step S4: Use the device identification data interface to receive the data for the data collection task of the optimized community device to obtain effective community device data; collect the real-time monitoring data of the community device for the effective community device data to generate multimodal monitoring data; extract the unlabeled face image data from the multimodal monitoring data to generate unlabeled face image data; extract the unlabeled person behavior data from the multimodal monitoring data based on the unlabeled face image data to generate unlabeled person behavior data.

[0009] Step S5: Obtain historical person behavior anomaly training samples; establish a relationship model for unlabeled person behavior anomaly analysis using the convolutional neural network algorithm and the historical person behavior anomaly training samples to generate a behavior anomaly analysis model; transmit the unlabeled person behavior data to the behavior anomaly analysis model for unlabeled person behavior anomaly analysis to generate unlabeled person behavior anomaly analysis data.

[0010] Step S6: Establish an automated warning engine based on the unlabeled person behavior anomaly analysis data; transmit the unlabeled person behavior anomaly analysis data to the automated warning engine to execute the automated warning control task.

[0011] The present invention realizes a comprehensive understanding of the historical operation conditions of community devices and the historical data collected by community devices by obtaining historical community device data. Designing the device data interface based on the historical community device data to generate the device data interface helps to standardize and uniformly manage the device data. It can uniformly manage the data collected by devices of different brands, improve the scalability and interoperability of the system, provide a basis for real-time community data collection, and also provide an accurate reference for device management and maintenance. Designing the device identity identification of the device data interface according to the historical community device data generates the device identity identification data with a unique identifier, which helps to improve the system's recognition and management accuracy of community devices. Optimizing the identity identification of the device data interface based on the device identity identification data to generate the device identification data interface further enhances the readability and relevance of the device data, helps to more effectively identify and respond to the data interface requests of community devices and distinguish from which community device the collected data is obtained, improves the overall intelligence and operation efficiency of the system, strengthens the data interaction ability, and integrates community devices of multiple brands into the system for processing. Executing the data collection task for community devices realizes the timely collection of device signals, and then generates detailed device signal data. By analyzing these signals, it is possible to effectively identify the invalid devices with abnormal signals in the community devices, and at the same time obtain the normal and valid community devices. Feedback the invalid community devices to the terminal to execute the repair control task, realizing the automatic detection and repair of device anomalies, improving the stability and reliability of community devices, realizing the real-time monitoring and management of the status of community devices, effectively reducing the impact of potential failures on the stability of the entire system, and improving the reliability and maintenance efficiency of the community digital information control system. Using the device identification data interface to receive the data of the optimized community device for the data collection task, accurately generate the valid community device data, enabling real-time acquisition of multi-modal data of the community, and then performing real-time collection of multi-modal monitoring data, including various data sources such as video and audio of community monitoring. By processing the multi-modal monitoring data, further extract the unlabeled facial image data, thus having the ability to identify unlabeled personnel, providing a basis for subsequent analysis of abnormal behaviors of unlabeled personnel, and being able to deeply mine the behavior information of unlabeled personnel in multi-modal data. Through this in-depth data processing process, not only realizes the real-time monitoring of community devices, but also provides sufficient data support for advanced functions such as intelligent security and abnormal behavior detection, further improving the intelligence level and security of the community digital information control system.By obtaining historical training samples of abnormal human behaviors, it lays a foundation for the analysis of unlabeled abnormal human behaviors. By applying the convolutional neural network algorithm and using these training samples, an efficient relationship model is established to form an abnormal behavior analysis model. The establishment of the model can more accurately identify the abnormal behaviors of unlabeled persons and achieve in-depth analysis of community activities. The model includes data on the analysis of abnormal behaviors of unlabeled persons, which includes both abnormal behavior data and normal behavior data, making the system more comprehensive in identifying abnormalities. Through training the model with the training samples during training, the model can learn the characteristics of different behaviors, thereby improving the accuracy and sensitivity to abnormal behaviors and enabling it to more precisely distinguish the abnormal behaviors of unlabeled persons. Based on the data on the analysis of abnormal behaviors of unlabeled persons, an automated early warning engine is successfully established. By transmitting the detailed data on the analysis of abnormal behaviors of unlabeled persons to the early warning engine, real-time monitoring and identification of abnormal behaviors are achieved. The engine combines the abnormal behavior analysis model and can quickly respond to potential risks and execute automated early warning control tasks. Since it is based on a deep learning model, the engine can accurately identify abnormal behaviors from a large amount of data, improving the ability of automated early warning of potential threats. By transmitting analysis data in real time, the engine can promptly feedback abnormal situations and then take prompt and effective measures, including issuing alarms and linking other security devices, thereby enhancing the overall security of the community and achieving real-time monitoring and early warning of the abnormal behaviors of unlabeled persons, providing a strong security guarantee for community management and enabling community management to more intelligently respond to potential security risks. Therefore, the community digital information intelligent control method of the present invention can timely monitor whether there are abnormal operating conditions of community devices, and then timely remind managers to repair through the terminal, making the community devices comprehensively collect community digital information, and for unlabeled foreign personnel, intelligently judge whether there are abnormal behaviors, reducing the need for manpower to judge the behaviors of foreign personnel, and thus automatically giving early warning feedback for the community according to the judged abnormal behaviors.

[0012] This specification provides a community digital information intelligent control system for implementing the community digital information intelligent control method as described above. The community digital information intelligent control system includes:

[0013] A device data interface design module, which is used to obtain historical community device data; based on the historical community device data, perform device data interface design to generate a device data interface;

[0014] A device data interface optimization module, which is used to perform device identity identification design on the device data interface according to the historical community device data to generate device identity identification data; based on the device identity identification data, optimize the data interface identity of the device data interface to generate a device identification data interface;

[0015] Community device anomaly analysis module, which is used to collect device signals of community devices when the community devices perform data collection tasks, and generate device signals; classify the community devices with signal anomalies according to the device signals to obtain effective community devices and ineffective community devices respectively; feedback the ineffective community devices to the terminal to execute the repair control task of the ineffective community devices;

[0016] Unidentified person behavior data collection module, which uses the device identification data interface to receive the data for the data collection task of optimizing community devices to obtain effective community device data; collect the real-time monitoring data of community devices for the effective community device data to generate multi-modal monitoring data; extract the unidentified facial image data from the multi-modal monitoring data to generate the unidentified facial image data; extract the unidentified person behavior data from the multi-modal monitoring data according to the unidentified facial image data to generate the unidentified person behavior data;

[0017] Unidentified person behavior data analysis module, which obtains historical person behavior anomaly training samples; uses the convolutional neural network algorithm and the historical person behavior anomaly training samples to establish a relationship model for the analysis of unidentified person behavior anomalies, and generates a behavior anomaly analysis model, where the unidentified person behavior anomaly analysis data includes behavior anomaly data or behavior normal data;

[0018] Automated warning control module, which establishes an automated warning engine based on the unidentified person behavior anomaly analysis data; transmits the unidentified person behavior anomaly analysis data to the automated warning engine to execute the automated warning control task.

[0019] The beneficial effects of this application are as follows. By obtaining historical community device data and analyzing the format of device data, the system can better understand and process the structures of different device data. Based on the device format data and data transmission protocols, device data interface design is carried out, realizing the standardized and normalized transmission of device data. The ability to make full use of historical data is improved. Through format analysis and interface design, it can more intelligently adapt to different device types and data formats, thus improving the maintainability and integration of the system, helping to more efficiently collect, monitor, and analyze data from community devices of different brands, and providing a solid foundation for the intelligent control of community digital information. Through the device data source information, the device identity identification design of the device data interface is carried out, generating device identity identification data with source identification, more accurately identifying and marking different devices, enhancing the accuracy and integrity of device identity information, and optimizing the data interface identity identification, enabling the system to more flexibly adapt to device identity identification information at the data interface level, improving the recognition and response capabilities for different devices. By deeply understanding the data sources of devices and establishing accurate identity identifications, the data interface becomes more intelligent, which improves the recognition and adaptability to devices and strengthens the processing capabilities for diverse device data. Device signal acquisition is carried out to timely obtain device status information. Using Fourier transform technology to convert the device signal into a spectrogram and using the device signal spectrogram anomaly detection algorithm for anomaly detection helps the system to more comprehensively understand the spectral characteristics of the device signal, be able to identify abnormal situations of the device signal in real time, classify community devices into effective or ineffective community devices, realizing the automatic classification of device status. Feeding back the ineffective community devices to the terminal to execute the repair control task for ineffective community devices can take timely actions to handle ineffective devices, improving the overall reliability of the system. By real-time monitoring and automatically classifying and processing device signals, the interference of ineffective devices to the normal operation of the system is reduced, improving the efficiency and response speed of community device management, helping the system to better maintain community devices and ensuring the smooth operation of the digital information control system. Using the device identification data interface to receive data for the data collection task of effective community devices provides a basis for subsequent correlation analysis.Community device correlation analysis and establishment of a multi-modal community association data matrix based on optimized community devices can deeply understand the relationships between different devices, realize the integration and association of multi-modal data of different community devices, extract facial image data of monitoring data from the multi-modal community association data matrix, generate monitoring facial image data, and extract unlabeled facial image data from the monitoring facial image data according to the obtained owner identification facial image data, so as to analyze which foreign personnel are unlabeled personnel not in the community, and analyze the behaviors of unlabeled personnel, providing further guarantee for the management of the community. Through comprehensive data collection, association and analysis, intelligent monitoring and management of community devices and personnel behaviors are realized, which helps to improve community security and management efficiency, and provides more comprehensive and intelligent support for the intelligent control of community digital information. Use the convolutional neural network algorithm to establish the mapping relationship of unlabeled personnel behavior anomaly analysis, generate the initial behavior anomaly analysis model, and use historical personnel behavior anomaly training samples to train the model, so that the model can learn and understand the normal patterns of different personnel behaviors, and generate a more accurate and adaptable behavior anomaly analysis model. Transmit the unlabeled personnel behavior data to the behavior anomaly analysis model for unlabeled personnel behavior anomaly analysis, generate unlabeled personnel behavior anomaly analysis data, which can identify and analyze the abnormal behaviors of unlabeled personnel in real time, and then perform subsequent automated warning control according to the analysis data. By establishing a deep learning model, real-time intelligent analysis and monitoring of the behaviors of unlabeled personnel are carried out, which helps to improve the accuracy and real-time performance of the system for abnormal behaviors in the community, and thus enhances the perception and processing ability of abnormal situations. This provides strong support for the safety management of the community. An automated warning engine is established based on the unlabeled personnel behavior anomaly analysis data, which can automatically execute warning control. The automated warning engine discriminates whether to trigger warning control according to the unlabeled personnel behavior anomaly analysis data. When the data received by the engine corresponds to abnormal behavior, the device closest to the monitored abnormal behavior is warned to improve the positioning ability of the warning. On the other hand, when the data received by the engine corresponds to normal behavior, the unlabeled personnel behavior data corresponding to the normal behavior data is transmitted to the terminal for feedback on the non-abnormal behavior of the personnel, so as to eliminate the real-time monitoring of unlabeled personnel, enabling the administrator and monitoring equipment to monitor other unlabeled personnel. Through the automated warning engine, real-time monitoring and automated response to abnormal behaviors are realized, improving the automated warning ability and response speed. By feeding back the normal behavior data to the terminal, timely confirmation of the non-abnormal behavior of the personnel is realized, reducing the false alarm rate, making the community digital information intelligent control system more intelligent, accurate and reliable. Brief Description of the Drawings

[0020] Figure 1 It is a schematic flow chart of the steps of a method for intelligent control of community digital information according to the present invention;

[0021] Figure 2 is Figure 1 a schematic diagram of the detailed implementation steps of step S3 in

[0022] Figure 3 is Figure 1 a schematic diagram of the detailed implementation steps of step S4 in

[0023] Figure 4 is Figure 1 a schematic diagram of the detailed implementation steps of step S6 in

[0024] The realization, functional characteristics and advantages of the object of the present invention will be further described with reference to the embodiments and the accompanying drawings. Specific embodiments

[0025] The technical method of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative work belong to the scope of protection of the present invention.

[0026] In addition, the accompanying drawings are only schematic diagrams of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings represent the same or similar parts, and thus repeated descriptions thereof will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor methods and / or microcontroller methods.

[0027] It should be understood that although the terms "first", "second", etc. may be used here to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, the first unit can be called the second unit, and similarly the second unit can be called the first unit. The term "and / or" used here includes any and all combinations of one or more of the listed related items.

[0028] To achieve the above object, please refer to Figures 1 to 4 , the present invention provides a method for intelligent control of community digital information, including the following steps:

[0029] Step S1: Obtain historical community device data; design a data interface for the device based on the historical community device data to generate a device data interface;

[0030] Step S2: Design the device identity identification of the device data interface according to the historical community device data to generate device identity identification data; optimize the data interface identity of the device data interface based on the device identity identification data to generate a device identification data interface;

[0031] Step S3: When the community device executes the data collection task, collect the device signal of the community device to generate a device signal; divide the community devices with signal anomalies according to the device signal to obtain effective community devices and invalid community devices respectively; feedback the invalid community devices to the terminal to execute the repair control task of the invalid community devices;

[0032] Step S4: Use the device identification data interface to receive the data of the data collection task of the optimized community device to obtain effective community device data; collect the real-time monitoring data of the community device for the effective community device data to generate multi-modal monitoring data; extract the unlabeled facial image data from the multi-modal monitoring data to generate unlabeled facial image data; extract the unlabeled personnel behavior data from the multi-modal monitoring data according to the unlabeled facial image data to generate unlabeled personnel behavior data;

[0033] Step S5: Obtain historical personnel behavior anomaly training samples; use the convolutional neural network algorithm and historical personnel behavior anomaly training samples to establish a relationship model for unlabeled personnel behavior anomaly analysis to generate a behavior anomaly analysis model; transmit the unlabeled personnel behavior data to the behavior anomaly analysis model for unlabeled personnel behavior anomaly analysis to generate unlabeled personnel behavior anomaly analysis data;

[0034] Step S6: Establish an automated warning engine based on the unlabeled personnel behavior anomaly analysis data; transmit the unlabeled personnel behavior anomaly analysis data to the automated warning engine to execute the automated warning control task.

[0035] The present invention achieves a comprehensive understanding of the historical operation of community equipment and the historical data collected by community equipment by acquiring historical community equipment data. Designing the equipment data interface based on historical community equipment data and generating the equipment data interface are helpful to achieve the standardization and unified management of equipment data, and can manage the data collected by equipment of different brands in a unified manner, thereby improving the scalability and interoperability of the system, providing a basis for real-time community data collection, and also providing an accurate reference basis for equipment management and maintenance. Designing the equipment identity of the equipment data interface based on historical community equipment data and generating equipment identity data with a unique identifier are helpful to improve the system's recognition and management accuracy of community equipment. Based on the equipment identity data, the equipment data interface is optimized for identity identification and a equipment identification data interface is generated, thereby further enhancing the readability and relevance of the equipment data, helping to more effectively identify and respond to the data interface request of the community equipment and distinguish which community equipment the collected data is obtained from, thereby improving the overall system's intelligence and operating efficiency, strengthening the data interaction capability, and integrating multi-brand community equipment into the system for processing. By performing data collection tasks on community equipment, timely collection of equipment signals is achieved, and detailed equipment signal data is generated. By analyzing these signals, invalid equipment with abnormal signals in community equipment can be effectively divided, and normal and valid community equipment can be obtained at the same time. The invalid community equipment is fed back to the terminal to perform repair and control tasks, and automatic detection and repair of equipment anomalies are achieved, which improves the stability and reliability of community equipment, realizes real-time monitoring and management of community equipment status, effectively reduces the impact of potential faults on the stability of the entire system, and improves the reliability and maintenance efficiency of the community digital information control system. The device identification data interface is used to optimize the data reception of community equipment for data collection tasks, and accurate and effective community equipment data is successfully generated, so that the multimodal data of the community can be obtained in real time, and then the real-time collection of multimodal monitoring data can be carried out, including video, audio and other data sources of community monitoring. By processing the multimodal monitoring data, the extraction of unidentified facial image data is further realized, thereby having the ability to identify unidentified persons, providing a basis for subsequent abnormal behavior analysis of unidentified persons, and being able to deeply mine the behavior information of unidentified persons in multimodal data. Through this in-depth data processing process, not only the real-time monitoring of community equipment is realized, but also sufficient data support is provided for advanced functions in intelligent security, abnormal behavior detection, etc., further improving the intelligence level and security of the community digital information control system.By obtaining historical training samples of abnormal personnel behaviors, it lays a foundation for the analysis of unlabeled personnel behavior anomalies. By applying the convolutional neural network algorithm and using these training samples, an efficient relationship model is established to form a behavior anomaly analysis model. The establishment of the model can more accurately identify the abnormal behaviors of unlabeled personnel and achieve in-depth analysis of community activities. The model includes data on the analysis of unlabeled personnel behavior anomalies, which includes both abnormal behavior data and normal behavior data, making the system more comprehensive in identifying anomalies. Through training the model with the training samples during training, the model can learn the characteristics of different behaviors, thereby improving the accuracy and sensitivity to abnormal behaviors and enabling it to more precisely distinguish the abnormal behaviors of unlabeled personnel. Based on the analysis data of unlabeled personnel behavior anomalies, an automated warning engine is successfully established. By transmitting detailed analysis data of unlabeled personnel behavior anomalies to the warning engine, real-time monitoring and identification of abnormal behaviors are achieved. The engine combines the behavior anomaly analysis model and can quickly respond to potential risks and execute automated warning control tasks. Since it is based on a deep learning model, the engine can accurately identify abnormal behaviors from a large amount of data, improving the ability to automatically warn of potential threats. Through real-time transmission of analysis data, the engine can timely feedback abnormal situations and then take prompt and effective measures, including issuing alarms and linking other security devices, thereby enhancing the overall security of the community and achieving real-time monitoring and warning of unlabeled personnel behavior anomalies, providing a strong security guarantee for community management and enabling community management to more intelligently respond to potential security risks. Therefore, the community digital information intelligent control method of the present invention can timely monitor whether there are abnormal operating conditions of community devices, and then timely remind the management personnel to repair through the terminal, making the community devices comprehensively collect community digital information, and for unlabeled foreign personnel, intelligently judge whether there are abnormal behaviors, reducing the manpower to judge the behaviors of foreign personnel, and then automatically giving early warning feedback to the community according to the judged abnormal behaviors.

[0036] In an embodiment of the present invention, referring to Figure 1 As described, it is a schematic flow chart of the steps of a community digital information intelligent control method of the present invention. In this embodiment, the community digital information intelligent control method includes the following steps:

[0037] Step S1: Obtain historical community device data; based on the historical community device data, design the data interface of the device to generate a device data interface;

[0038] In the embodiments of the present invention, historical community device data is obtained from the community database. The historical community device data includes data related to community devices such as historical acquisition data of community devices. Taking the surveillance cameras in a certain community as an example, historical surveillance data has been collected, including information in aspects such as videos, images, and device operating status. Device data interface design is carried out to ensure that the historical community device data can be efficiently utilized by the system. A flexible and extensible RESTful API design is selected to meet the access requirements of different types of devices. Based on the historical community device data, data exchange formats, protocols, and access endpoints are defined, etc., to establish a unified data interface for community devices. This includes the reception of camera video streams, the acquisition of sensor data, and the interaction of other device information, generating a device data interface.

[0039] Step S2: Device identity identification design of the device data interface is carried out according to the historical community device data to generate device identity identification data; based on the device identity identification data, data interface identity identification optimization of the device data interface is carried out to generate a device identification data interface;

[0040] In the embodiments of the present invention, the historical community device data is analyzed to identify the key identity information of the device, such as device model, manufacturer, unique identification code, etc. Taking the surveillance camera as an example, we may focus on information such as the location, model, and installation time of the camera. These information will form the basis of the device identity identification. Design a device identity identification data structure that includes these key identity information, including a JSON object that contains fields such as device model field, manufacturer field, unique identification code field, etc., so that the device identity information can be clearly represented and transmitted. To optimize the identity identification of the data interface, the device identity identification data is embedded into the device data interface to form a device identification data interface. The specification of RESTful API is adopted, and the device identity identification information is transmitted as part of the HTTP header or request body, making it easier to track and manage the data of specific devices and improving the maintainability of the data.

[0041] Step S3: When the community device executes a data acquisition task, device signals are collected from the community device to generate device signals; according to the device signals, community devices with signal anomalies are divided to obtain effective community devices and ineffective community devices respectively; the ineffective community devices are fed back to the terminal to execute the repair control task for the ineffective community devices;

[0042] In the embodiments of the present invention, when community devices perform data collection tasks, device signal collection is implemented for each device. Taking a warning device as an example, data reception and transmission signals may be collected, or data traffic signals may be collected in a surveillance camera. The data collected in this way constitutes device signals. The community devices are classified according to the device signals for signal anomalies. By setting appropriate thresholds and historical signal conditions, the Fourier transform technology can be used to convert the signals into spectrograms, which is more conducive to signal point analysis and calculation. Signals contrary to normal operations can be detected, and the devices can be divided into effective community devices and ineffective community devices. For example, when the received signal of a warning device is significantly higher than normal, it can be determined as an ineffective device. A feedback mechanism is implemented for ineffective community devices, and this information is fed back to the terminal. In the terminal system, a task queue can be established to identify and transmit ineffective community devices to the module that executes the repair control tasks for ineffective community devices. This module may include an automatic repair program, notifying relevant personnel, or triggering maintenance personnel to come for repair. To improve accuracy and timeliness, device signals can be collected in real time, and device signals can be checked regularly to ensure that the classification and repair of abnormal devices can be completed in the shortest time.

[0043] Step S4: Use the device identification data interface to receive data for the data collection task of optimizing community devices to obtain effective community device data; perform real-time collection of monitoring data of community devices on the effective community device data to generate multimodal monitoring data; extract unlabeled facial image data from the multimodal monitoring data to generate unlabeled facial image data; extract unlabeled personnel behavior data from the multimodal monitoring data according to the unlabeled facial image data to generate unlabeled personnel behavior data;

[0044] In the embodiments of the present invention, a standardized protocol for communicating with community devices is established through the device identification data interface, ensuring stable data transmission. In the data reception phase, this device identification data interface is used to connect with community devices to achieve the task of optimizing data collection from community devices. For example, accessing a surveillance camera to obtain video streams and other relevant data. During the process of real-time collection of monitoring data from valid community device data, real-time image processing technologies, such as the open-source image processing library (OpenCV), are utilized to extract key information, such as the location and movement direction of people, to ensure the timeliness of the data. The steps for generating multi-modal monitoring data include integrating different types of data, such as video, temperature, humidity, etc., to form a comprehensive monitoring data set. This helps to comprehensively understand the status of community devices and provides more information for subsequent analysis of abnormal behaviors. Through image processing technologies, such as face recognition models, unlabeled facial image data is extracted from the multi-modal monitoring data. This includes steps such as face detection and feature extraction to obtain unlabeled facial images in the monitoring data. Based on the unlabeled facial image data, unlabeled person behavior data is extracted. By analyzing information such as the actions and locations of people in the monitoring data, detailed data about the behaviors of unlabeled people is obtained, such as movement trajectories and residence times.

[0045] Step S5: Obtain historical training samples of abnormal person behaviors; establish a relationship model for analyzing abnormal behaviors of unlabeled persons using the convolutional neural network algorithm and the historical training samples of abnormal person behaviors to generate an abnormal behavior analysis model; transmit the unlabeled person behavior data to the abnormal behavior analysis model for analyzing the abnormal behaviors of unlabeled persons to generate unlabeled person abnormal behavior analysis data;

[0046] In the embodiments of the present invention, in terms of obtaining historical training samples of abnormal human behaviors, surveillance data over a past period within a community was collected, including video clips of human behaviors. These video clips were labeled by an administrator with tags for abnormal behaviors, such as which ones were abnormal behaviors, including people entering prohibited areas, abnormal movement patterns, or staying for an extended period, etc. These samples constituted our training dataset. The convolutional neural network (CNN) was selected as the algorithm for analyzing abnormal behaviors. The training samples were preprocessed, including steps such as video frame extraction and image enhancement, to ensure the quality of the input data. Based on the architecture of the convolutional neural network, an architecture for the abnormal behavior analysis model was designed, including convolutional layers, pooling layers, and fully connected layers, to capture the spatial features and temporal information in the images. An open-source deep learning framework (such as TensorFlow or PyTorch) was used to implement the training of the abnormal behavior analysis model. The abnormal behavior analysis model was repeatedly trained with the training dataset, and the model parameters were adjusted to capture the normal and abnormal patterns of human behaviors to the greatest extent, thereby obtaining the abnormal behavior analysis model. This model can analyze the behaviors of unlabeled people in real time through surveillance videos and detect whether there are abnormalities by setting thresholds. When the abnormal probability output by the model exceeds the threshold, an alarm is triggered or corresponding control tasks are taken. The unlabeled human behavior data is transmitted to this model for analysis, including pedestrian behaviors and vehicle movement trajectories captured by surveillance cameras. The model will evaluate this data and identify possible abnormal behaviors, such as running and staying for an abnormally long time.

[0047] Step S6: Establish an automated early warning engine based on the unlabeled human behavior abnormal analysis data; transmit the unlabeled human behavior abnormal analysis data to the automated early warning engine to execute the automated early warning control task.

[0048] In the embodiments of the present invention, an automated warning engine is established based on the analysis data of the abnormal behavior of unidentified personnel, and a real-time stream processing framework, such as Apache Flink or Spark Streaming, is adopted for the established automated warning engine to ensure a quick response to abnormal behaviors, including establishing automated warning rules, automated warning decisions, etc. based on the analysis data of the abnormal behavior of unidentified personnel, which are used to analyze and judge the abnormal behavior of unidentified personnel. The automated warning rules include those established based on the analysis data of the abnormal behavior of unidentified personnel, including triggering an alarm when abnormal speed, abnormal stay time, abnormal aggregation, etc. are detected, and these rules will be integrated into an automated rule. If the automated warning rule is triggered to execute the warning task of the automated warning decision, and if the automated warning rule is not triggered, the security feedback task of the automated warning decision is executed. The analysis data of the abnormal behavior of unidentified personnel is transmitted to the automated warning engine in real time, and after the engine receives the data, an automated warning decision is realized based on the automated warning rules. The automated warning decision includes a warning control task and a security feedback task. The warning control task includes triggering a real-time alarm notification or strengthening the real-time monitoring of the abnormal area, etc. The security feedback task includes that the behavior of the unidentified personnel monitored is normal to cancel the monitoring. The engine can also generate a detailed report, recording the occurrence time, location and relevant information of the abnormal event for subsequent analysis and system optimization.

[0049] Preferably, step S1 includes the following steps:

[0050] Step S11: Obtain historical community device data;

[0051] Step S12: Analyze the format of the device data according to the historical community device data to generate device format data;

[0052] Step S13: Design the data interface of the device based on the device format data and the data transmission protocol to generate a device data interface.

[0053] By obtaining the historical community device data, the present invention realizes a comprehensive understanding of the historical operation conditions of community devices and the historical data collected through community devices. The format of the device data is analyzed according to the historical community device data. By analyzing the data format of the data collected by community devices, and then designing based on the device format data during the data interface design, different brands of community devices can be uniformly processed in the system. By designing the device data interface based on the device format data and the data transmission protocol, the standardization and normalization of device data are realized, enabling the data to be better understood and processed within the system, ensuring the efficient transmission of device data and the good docking of different community devices, and improving the stability and scalability of the overall system.

[0054] In the embodiments of the present invention, historical community device data of various community devices is obtained from the community database, taking monitoring devices, warning devices, etc. as examples. These data may include collected data such as the video stream of a camera, the signal transmission traffic of the video, the warning signal transmission traffic, etc., to ensure that there is sufficient historical information for subsequent analysis and interface design. A detailed format analysis is performed on the historical community device data. For example, if the camera data is a video stream, parameters such as the video coding format, resolution, and frame rate will be analyzed; for the warning audio data of the warning device, the vibration magnitude, sampling frequency, etc. of the audio data will be analyzed, which helps to determine the specific characteristics of each type of device data, so as to obtain standardized device format data. For example, the camera information represented in JSON format includes not only the video format data of the camera, but also key information such as location, model, manufacturer, etc. This device format data becomes the basis for subsequent interface design. The device format data is used for the design of the data interface. According to the generality of the data interface, a suitable data transmission protocol such as HTTP or MQTT is selected, and the corresponding API endpoints are defined. For example, for camera data, an API endpoint that can obtain the real-time video stream is designed, so that the data of different types of devices in the community can be received and processed by the system in a unified manner, generating a device data interface.

[0055] Preferably, step S2 includes the following steps:

[0056] Step S21: Extract the device data source information from the historical community device data to generate device data source information;

[0057] Step S22: Design the device identity identifier of the device data interface according to the device data source information to generate device identity identifier data;

[0058] Step S23: Optimize the data interface identity identifier of the device data interface based on the device identity identifier data to generate a device identifier data interface.

[0059] The present invention extracts the device data source information from the historical community device data to comprehensively understand from which community device the device data is obtained, providing a more detailed reference for subsequent identity identifier design. By designing the device identity identifier of the device data interface through the device data source information, device identity identifier data with a source identifier is generated. By associating the device data with its source information, not only can the device be identified, but also the source of the data can be traced more accurately, improving the accuracy and traceability of the system for data traceability. Optimize the data interface identity identifier of the device data interface based on the device identity identifier data. Through the optimization of the data interface identity identifier, the device identity identifier information is more flexibly adapted at the data interface level, improving the ability to identify and respond to different devices, and the data traceability ability of the subsequent collected real-time data.

[0060] In the embodiments of the present invention, the data source information of the devices is extracted from the historical community device data, including determining the data sources of each device, such as cameras, sensors, access control systems, etc. In the embodiments, a device data source information table containing device IDs and corresponding data source types is obtained for subsequent identity identification design. Based on the device data source information, device identity identification design is carried out for the device data interface. Taking the device ID and the data source type as an example, a device identity identification data structure containing these two pieces of information is created, and this structure can be a unique identifier to ensure the uniqueness of the device identity in the system. The identity identification of the device data interface is optimized using the device identity identification data, including adding device identity identification information in the API request header or request body. For example, in the RESTful API design, the device ID and the data source type are included in each request to accurately identify and process the data of specific devices.

[0061] Preferably, step S3 includes the following steps:

[0062] Step S31: When the community device performs a data collection task, device signals are collected from the community device to generate device signals;

[0063] Step S32: Using Fourier transform technology, the spectrum diagram conversion of the device signals is performed to generate a device signal spectrum diagram;

[0064] Step S33: Using the device signal spectrum diagram anomaly detection algorithm, spectrum diagram anomaly detection calculations are performed on the device signal spectrum diagram to generate spectrum diagram anomaly detection data;

[0065] Step S34: According to the spectrum diagram anomaly detection data, the community devices are divided. When the spectrum diagram anomaly detection data is less than the preset anomaly spectrum diagram detection threshold, the community device corresponding to the spectrum diagram anomaly detection data is marked as a valid community device. When the spectrum diagram anomaly detection data is not less than the preset anomaly spectrum diagram detection threshold, the community device corresponding to the spectrum diagram anomaly detection data is marked as an invalid community device;

[0066] Step S35: The invalid community devices are fed back to the terminal to perform the repair control task of the invalid community devices.

[0067] The present invention collects device signals of community devices, enabling real-time acquisition of the operating signals of the devices and providing a data basis for subsequent analysis of community devices. The Fourier transform technology is used to convert the device signals into spectrograms of the device signals, and the spectrograms reflect the distribution of the device signals in the frequency domain, which helps to gain a deeper understanding of the signal characteristics and more clearly reflect signal anomalies, improving the analysis accuracy of the device signals. The device signal spectrogram anomaly detection algorithm is used to perform spectrogram anomaly detection calculations on the device signal spectrograms. By calculating the anomaly situation of the spectrograms through this device signal spectrogram anomaly detection algorithm, if the frequency of the spectrogram deviates significantly from the normal situation, the spectrogram anomaly detection data result output by the algorithm is relatively high. This algorithm realizes the anomaly detection of device signals, further improving the sensitivity to changes in device states. Dividing the community devices according to the spectrogram anomaly detection data helps to accurately identify abnormal devices and improves the intelligent discrimination ability of the device operating states. Feeding back the invalid community devices to the terminal to execute the repair control task of the invalid community devices realizes the maintenance of the device states by automatically feeding back the information of the invalid devices to the terminal, reducing the time for manual monitoring of device anomalies, improving the maintainability and stability of the entire system, and improving the health management and maintenance efficiency of community devices. It realizes the comprehensive monitoring, anomaly detection, and automatic repair of the community device states, improving the real-time performance, intelligence, and automation level of the community digital information control system.

[0068] As an example of the present invention, refer to Figure 2 shown in Figure 1 which is a schematic diagram of the detailed implementation steps of step S3 in

[0069] Step S31: When the community device executes the data collection task, collect the device signals of the community device to generate device signals;

[0070] In the embodiment of the present invention, when the community device executes the data collection task, we use corresponding sensors, monitoring cameras and other devices to collect data. Taking the monitoring camera as an example, we collect the video transmission signals of the camera, which form the basis of the device signals.

[0071] Step S32: Use the Fourier transform technology to perform spectrogram conversion of the device signals to generate device signal spectrograms;

[0072] In the embodiment of the present invention, the Fourier transform technology is applied to perform spectrogram conversion on the device signals, converting the device signals from the time domain to the frequency domain to obtain the spectrograms of the device signals. For the spectrogram of the video transmission signal of the monitoring camera, it helps us analyze the frequency domain information of the signal transmission, such as the peak value, mean value and other conditions of the transmission signal.

[0073] Step S33: Use the abnormal detection algorithm for the device signal spectrogram to perform spectrogram abnormal detection calculation on the device signal spectrogram, and generate spectrogram abnormal detection data;

[0074] In the embodiment of the present invention, the abnormal detection algorithm for the device signal spectrogram is used to calculate the spectrogram. This abnormal detection algorithm for the device signal spectrogram calculates the deviation of the frequency data in the spectrogram from the frequency data of the spectrogram under normal conditions, and considers the frequency error of the spectrogram caused by external signal interference, so as to judge whether there is an abnormality in the device signal spectrogram and generate spectrogram abnormal detection data. Conventional abnormal detection algorithms for spectrograms can also perform spectrogram abnormal detection calculations, but the accuracy of data accuracy is relatively low.

[0075] Step S34: Divide the community devices according to the spectrogram abnormal detection data. When the spectrogram abnormal detection data is less than the preset abnormal spectrogram detection threshold, mark the community device corresponding to the spectrogram abnormal detection data as a valid community device; when the spectrogram abnormal detection data is not less than the preset abnormal spectrogram detection threshold, mark the community device corresponding to the spectrogram abnormal detection data as an invalid community device;

[0076] In the embodiment of the present invention, the community devices are divided based on the spectrogram abnormal detection data. When the spectrogram abnormal detection data is less than the preset abnormal spectrogram detection threshold, mark the community device corresponding to the smaller spectrogram abnormal detection data as a valid community device, that is, the abnormal condition of this community device is within the abnormal error range; when the spectrogram abnormal detection data is not less than the preset abnormal spectrogram detection threshold, mark the community device corresponding to the smaller spectrogram abnormal detection data as an invalid community device, that is, the abnormal condition of this community device is outside the abnormal error range.

[0077] Step S35: Feed back the invalid community devices to the terminal to execute the repair control task for the invalid community devices.

[0078] In the embodiment of the present invention, the information of the marked invalid community devices is fed back to the terminal, so that a task queue is set up at the terminal, and the invalid community devices are identified and transmitted to the module that executes the repair control task for the invalid community devices, including sending notifications to relevant personnel, triggering maintenance tasks, or performing other targeted repair operations to ensure the normal operation of the community devices.

[0079] Preferably, the abnormal detection algorithm for the device signal spectrogram in step S33 is as follows:

[0080]

[0081] Wherein, D(f) represents the spectrogram anomaly detection data, f represents the frequency parameter of the device signal spectrogram, σ represents the standard deviation of the conventional frequency parameter, μ represents the conventional frequency parameter, α represents the correction adjustment value of the external interference frequency parameter of the device signal spectrogram, γ represents the frequency attenuation rate of the device signal spectrogram, δ represents the offset of the attenuation frequency parameter of the device signal spectrogram, and η represents the attenuation frequency parameter of the device signal spectrogram.

[0082] The present invention utilizes a device signal spectrogram anomaly detection algorithm, which includes the frequency parameter f of the device signal spectrogram, the standard deviation σ of the conventional frequency parameter, the conventional frequency parameter μ, the correction adjustment value α of the external interference frequency parameter of the device signal spectrogram, the frequency attenuation rate γ of the device signal spectrogram, the offset δ of the attenuation frequency parameter of the device signal spectrogram, the attenuation frequency parameter η of the device signal spectrogram, and the interaction relationship between functions to form a functional relationship:

[0083] That is, This functional relationship considers the size of the anomaly value of the device signal spectrogram by taking into account the frequency parameter of the device signal spectrogram and the differences between the corresponding signal spectrograms of conventional devices in history, so as to determine whether the device has an abnormal condition. For example, if the transmission signal of the device has a large deviation from the signals of conventional devices in the same category, the result of the generated spectrogram anomaly detection data will be higher.

[0084] Describes the expected distribution of the spectrogram under normal conditions, and then preliminarily determines whether the frequency of the spectrogram is abnormal. αcos(2πf) corrects the possible external interference in the spectrogram by introducing a cosine term. α is a parameter that adjusts the correction amplitude, which helps the algorithm to correct the external interference factors in the spectrogram and improves the robustness to interference frequencies, making the algorithm more robust in the presence of noise or interference. Describes the frequency attenuation factor, where γ, δ, and η are used to adjust the specific form of the attenuation, which can sensitively capture the attenuation characteristics of the spectrogram and has a certain adaptability to the time variation of the spectrogram. The entire functional relationship realizes the anomaly detection of the overall spectrogram through integral operations on various spectrogram features within the entire frequency domain range, making the comprehensive response of the algorithm to different frequency components more comprehensive and enhancing the analysis ability of the algorithm for complex spectrograms. This functional relationship comprehensively considers multiple factors such as the normal distribution, external interference, and frequency attenuation of the spectrogram in anomaly detection, enabling it to more accurately detect abnormal situations in a complex environment and providing a comprehensive and flexible solution for the anomaly monitoring of device signals.

[0085] Preferably, step S4 includes the following steps:

[0086] Step S41: Receive data for the data collection task of valid community devices through the device identification data interface to obtain valid community device data;

[0087] Step S42: Conduct an analysis of the relevance of community devices based on optimized community devices to generate community device relevance data;

[0088] Step S43: Establish multi-modal community association matrix nodes based on the community device relevance data, and transmit the valid community device data to the multi-modal community association matrix nodes for data filling to generate a multi-modal community association data matrix;

[0089] Step S44: Conduct real-time collection of monitoring data of community devices on the multi-modal community association data matrix to generate multi-modal monitoring data;

[0090] Step S45: Extract facial image data of the monitoring data from the multi-modal monitoring data to generate monitoring facial image data;

[0091] Step S46: Obtain the facial image data of the owner identification in the community database;

[0092] Step S47: Extract unlabeled facial image data from the monitoring facial image data based on the facial image data of the owner identification to generate unlabeled facial image data;

[0093] Step S48: Extract unlabeled person behavior data from the multi-modal monitoring data based on the unlabeled facial image data to generate unlabeled person behavior data.

[0094] The present invention receives data for data collection tasks of valid community devices through a device identification data interface, realizes the acquisition of real-time information of valid community devices, can accurately and timely obtain the latest data of valid community devices, and uniformly processes the data collected by community devices of different brands, providing a reliable data basis for subsequent analysis. According to the optimized community devices, the association analysis of community devices is carried out to generate community device association data. Through the association analysis, the relationship between devices can be better understood, so as to optimize the integration of multimodal data, improve the understanding and utilization of the association of community devices, and in the subsequent steps, when it is monitored that an unlabeled person has abnormal monitoring behavior, the device with the greatest association can be found among the monitoring devices that collect monitoring data for accurate early warning feedback, enabling the management personnel to obtain the early warning location and relevant information, etc. According to the community device association data, a multimodal community association matrix node is established, and the valid community device data is transmitted to the multimodal community association matrix node for data filling, realizing the integration and association of multimodal data, enabling a more comprehensive understanding of the association between community devices and the corresponding data, and improving the management ability of the overall community situation. Real-time collection of the monitoring data of community devices for the multimodal community association data matrix can obtain the monitoring data of community devices more timely, realizing the real-time monitoring of the status of community devices. Extracting the facial image data of the monitoring data from the multimodal monitoring data provides a basis for subsequent facial image analysis. Obtaining the facial image data of the owner identification in the community database helps the system to perform identity recognition and association in the multimodal monitoring data. Extracting the unlabeled facial image data from the monitoring facial image data according to the facial image data of the owner identification effectively identifies the facial images of non-owners, improving the monitoring accuracy of non-owner personnel in the community. Extracting the behavior data of unlabeled personnel from the multimodal monitoring data according to the unlabeled facial image data can deeply understand the behavior patterns of unlabeled personnel, providing strong support for subsequent abnormal behavior analysis.

[0095] As an example of the present invention, refer to Figure 3 shown in Figure 1 is a schematic diagram of the detailed implementation steps of step S4 in

[0096] Step S41: Use the device identification data interface to receive data for the data collection task of valid community devices to obtain valid community device data;

[0097] In the embodiment of the present invention, data collection is performed on valid community devices through a device identification data interface. This includes calling API endpoints to receive relevant data collected by valid community devices, thereby obtaining real-time valid community device data, such as monitoring video data and other data.

[0098] Step S42: Perform community device correlation analysis based on the optimized community devices to generate community device correlation data;

[0099] In the embodiment of the present invention, correlation analysis is performed on the optimized community devices. By analyzing the data correlation between devices, it can be determined through the positions between devices and the same type of devices. For example, the correlation between similar monitoring devices is relatively large, and the correlation between monitoring devices and nearby warning devices is relatively large. This facilitates the warning operation through the warning devices with relatively large correlation when the monitoring device detects abnormal personnel in subsequent steps, so as to obtain community device correlation data for establishing a multi-modal correlation matrix in the subsequent steps.

[0100] Step S43: Establish multi-modal community correlation matrix nodes based on the community device correlation data, and transmit the valid community device data to the multi-modal community correlation matrix nodes for data filling to generate a multi-modal community correlation data matrix;

[0101] In the embodiment of the present invention, multi-modal community correlation matrix nodes are established according to the community device correlation data, and the valid community device data is transmitted to the matrix nodes for data filling to generate a multi-modal community correlation data matrix. This matrix includes the correlation between each device, providing a basis for the subsequent extraction of monitoring data.

[0102] Step S44: Perform real-time collection of monitoring data of community devices on the multi-modal community correlation data matrix to generate multi-modal monitoring data;

[0103] In the embodiment of the present invention, real-time monitoring data collection is performed based on the multi-modal community correlation data matrix, including obtaining real-time monitoring data from each device, such as the monitoring data of monitoring cameras, the monitoring data of door lock cameras, etc., to form multi-modal monitoring data.

[0104] Step S45: Extract facial image data of the monitoring data from the multi-modal monitoring data to generate monitoring facial image data;

[0105] In the embodiment of the present invention, image processing technology, such as a face detection algorithm, is used to extract the facial images contained in the monitoring data from the multi-modal monitoring data. This can include using deep learning models or open-source face detection libraries (such as OpenCV, Dlib, etc.). The extracted facial images will be used for subsequent analysis to generate monitoring facial image data.

[0106] Step S46: Obtain the facial image data of the owner identification in the community database;

[0107] In the embodiment of the present invention, the facial image data of the owner identification is obtained from the community database. This can be the face image data of the owner pre-entered into the system, which is used to compare with the facial images in the monitoring data to identify the owner's identity.

[0108] Step S47: Extract unlabeled facial image data from the monitored facial image data according to the owner's labeled facial image data to generate unlabeled facial image data;

[0109] In the embodiment of the present invention, the owner's labeled facial image data is used to compare and match the previously extracted monitored facial image data, so as to filter out the unlabeled facial image data belonging to others than the owner, which is realized by using the face matching algorithm or feature matching technology (such as FaceNet, ArcFace, etc.) of the open source resource library of python. The generated unlabeled facial image data will be used for the subsequent extraction of unlabeled personnel behavior data.

[0110] Step S48: Extract unlabeled personnel behavior data from the multi-modal monitoring data according to the unlabeled facial image data to generate unlabeled personnel behavior data.

[0111] In the embodiment of the present invention, based on the extracted unlabeled facial image data, a behavior analysis algorithm or model, such as a convolutional neural network, can be used to analyze and extract the behavior of unlabeled personnel, including the recognition of information such as the actions, postures, and movement paths of personnel, to generate unlabeled personnel behavior data.

[0112] Preferably, step S5 includes the following steps:

[0113] Step S51: Use the convolutional neural network algorithm to establish a mapping relationship for the analysis of unlabeled personnel behavior anomalies to generate an initial behavior anomaly analysis model;

[0114] Step S52: Obtain historical personnel behavior anomaly training samples;

[0115] Step S53: Use the historical personnel behavior anomaly training samples to train the initial behavior anomaly analysis model to generate a behavior anomaly analysis model;

[0116] Step S54: Transmit the unlabeled personnel behavior data to the behavior anomaly analysis model for the analysis of unlabeled personnel behavior anomalies to generate unlabeled personnel behavior anomaly analysis data, where the unlabeled personnel behavior anomaly analysis data includes behavior anomaly data or behavior normal data.

[0117] The present invention establishes a mapping relationship for analyzing the abnormal behavior of unlabeled persons by using a convolutional neural network algorithm, generates an initial abnormal behavior analysis model, and can automatically learn and capture the complex mapping relationship of the behavior of unlabeled persons, which helps to deeply understand the behavior of unlabeled persons. Obtaining historical training samples of abnormal behavior of personnel provides representative historical data for the system, thereby improving the accuracy of the model during model training. Using the historical training samples of abnormal behavior of personnel to train the initial abnormal behavior analysis model to generate an abnormal behavior analysis model. By continuously training, the model parameters are optimized to better adapt to the characteristics of historical data, improving the generalization ability and accuracy of the model. Transmitting the behavior data of unlabeled persons to the abnormal behavior analysis model for analyzing the abnormal behavior of unlabeled persons, generating abnormal behavior analysis data of unlabeled persons, which can monitor the behavior of unlabeled persons in real time, and identify potential risks and abnormal behaviors by analyzing the abnormal analysis data output by the model. Through the combination of deep learning modeling and historical data training, the accurate analysis and abnormal detection of the behavior of unlabeled persons are realized, which not only improves the perception ability of potential threats, but also can better adapt to and understand the behavior of unlabeled persons in different scenarios, providing more intelligent and accurate monitoring and analysis support for community security.

[0118] In an embodiment of the present invention, a convolutional neural network (CNN) algorithm is used to establish a mapping relationship for analyzing the abnormal behavior of unlabeled persons, and an initial abnormal behavior analysis model is generated. A neural network structure including multiple convolutional layers, pooling layers, and fully connected layers may be designed to extract features from the behavior data of unlabeled persons and perform abnormal analysis. In terms of obtaining historical training samples of abnormal behavior of personnel, the monitoring data in the community over a past period of time is collected, including video clips of personnel behavior. These video clips are labeled by an administrator by designing labels for abnormal behaviors, such as which are abnormal behaviors, including a person entering a prohibited area, an abnormal movement pattern, or staying continuously, etc. These samples constitute our training data set. Using the historical training samples of abnormal behavior of personnel to train the initial abnormal behavior analysis model, the model parameters are adjusted through the backpropagation algorithm and an optimizer so that it can accurately identify normal and abnormal behavior patterns, generating an abnormal behavior analysis model. Transmitting the behavior data of unlabeled persons collected in real time to the trained abnormal behavior analysis model for analysis, and this abnormal behavior analysis model can determine whether the input behavior data is normal or abnormal according to the learned mapping relationship. The generated abnormal behavior analysis data of unlabeled persons may include the identification of abnormal behavior data and normal behavior data, providing a basis for subsequent automatic early warning.

[0119] Preferably, the execution of the automated early warning control task includes performing an early warning control task on a target early warning device or transmitting the unlabeled personnel behavior data corresponding to the normal behavior data to the terminal for personnel non-abnormal behavior feedback. Step S6 includes the following steps:

[0120] Step S61: Establish an automated early warning engine based on the unlabeled personnel behavior anomaly analysis data;

[0121] Step S62: Transmit the unlabeled personnel behavior anomaly analysis data to the automated early warning engine for automated early warning control. When the behavior anomaly data corresponding to the unlabeled personnel behavior anomaly analysis data received by the automated engine, perform Step S63. Or, when the automated engine receives the normal behavior data corresponding to the unlabeled personnel behavior anomaly analysis data, transmit the unlabeled personnel behavior data corresponding to the normal behavior data to the terminal for personnel non-abnormal behavior feedback;

[0122] Step S63: Mark the optimized community device as a target early warning device according to the unlabeled personnel data corresponding to the behavior anomaly data to obtain the target early warning device, and perform an early warning control task on the target early warning device.

[0123] The present invention establishes an automated early warning engine based on the unlabeled personnel behavior anomaly analysis data. By deep learning and model establishment of the anomaly analysis data, it can automatically execute early warning operations by identifying unusual personnel behavior patterns, laying a foundation for subsequent automated early warning control. Transmitting the unlabeled personnel behavior anomaly analysis data to the automated early warning engine for automated early warning control enables the automated early warning engine to intelligently discriminate abnormal behaviors and take different treatment measures in different situations. Marking the optimized community device as a target early warning device according to the unlabeled personnel data corresponding to the behavior anomaly data to obtain the target early warning device and performing an early warning control task on the target early warning device can quickly locate the involved device according to the abnormal behavior data. Through target early warning device marking and control tasks, rapid response and handling of abnormal situations are achieved. By establishing an automated early warning engine, accurate discrimination of abnormal behaviors and automatic execution of corresponding control tasks can be realized, improving the system's perception and response speed to potential risks and providing more comprehensive and intelligent protection for community safety.

[0124] As an example of the present invention, refer to Figure 4 shown, for Figure 1 the detailed implementation step flow diagram of Step S6 in

[0125] Step S61: Establish an automated early warning engine based on the unlabeled personnel behavior anomaly analysis data;

[0126] In an embodiment of the present invention, an automated warning engine is established based on unlabeled personnel behavior anomaly analysis data, and a real-time stream processing framework such as Apache Flink or Spark Streaming is adopted for the established automated warning engine to ensure a quick response to abnormal behaviors, including establishing automated warning rules, automated warning decisions, etc. based on unlabeled personnel behavior anomaly analysis data, which are used to analyze and judge unlabeled personnel behavior anomalies. The automated warning rules include those established based on unlabeled personnel behavior anomaly analysis data, including triggering an alarm when abnormal speed, abnormal stay time, abnormal aggregation, etc. are detected, and these rules will be integrated into an automated rule. If the automated warning rule is triggered to execute the warning task of the automated warning decision, and if the automated warning rule is not triggered, the security feedback task of the automated warning decision is executed.

[0127] Step S62: Transmit the unlabeled personnel behavior anomaly analysis data to the automated warning engine for automated warning control. When the behavior anomaly data corresponding to the unlabeled personnel behavior anomaly analysis data received by the automated engine, execute Step S63. Or, when the behavior normal data corresponding to the unlabeled personnel behavior anomaly analysis data received by the automated engine, transmit the unlabeled personnel behavior data corresponding to the behavior normal data to the terminal for personnel non-abnormal behavior feedback;

[0128] In an embodiment of the present invention, the unlabeled personnel behavior anomaly analysis data is transmitted to the automated warning engine in real time. After the engine receives the data, an automated warning decision is realized based on the automated warning rules. The automated warning decision includes a warning control task and a security feedback task. The warning control task includes triggering a real-time alarm notification or strengthening the real-time monitoring of the abnormal area, etc. The security feedback task includes that the unlabeled personnel behavior is normal as detected to cancel the monitoring.

[0129] Step S63: Mark the optimized community device as a target warning device according to the unlabeled personnel data corresponding to the behavior anomaly data to obtain the target warning device, and execute the warning control task for the target warning device.

[0130] In an embodiment of the present invention, when the behavior anomaly data corresponding to the unlabeled personnel behavior anomaly analysis data received by the automated engine, the warning control task executed by the automated warning engine is to locate the target warning device associated with the corresponding monitoring device according to the monitored data, and the target warning device executes the corresponding warning control task, such as triggering an alarm, adjusting monitoring parameters, feedbacking the location information of the warning device to the relevant administrator, etc. to cope with the detected abnormal behavior.

[0131] Preferably, Step S63 includes the following steps:

[0132] Locate abnormal data nodes in the multi-modal community association data matrix according to the unlabeled personnel data corresponding to the behavior abnormal data, and generate multi-modal abnormal data nodes;

[0133] Mark the optimized community devices as target warning devices according to the multi-modal abnormal data nodes and the community device correlation data, and generate target warning devices;

[0134] Execute warning control tasks on the target warning devices.

[0135] According to the unlabeled personnel data corresponding to the behavior abnormal data, the present invention locates abnormal data nodes in the multi-modal community association data matrix, and can accurately locate the nodes where there are unlabeled personnel behavior abnormalities in the monitoring data of the multi-modal community association data matrix, realizing the accurate positioning of abnormal events. When marking the optimized community devices as target warning devices according to the multi-modal abnormal data nodes and the community device correlation data, by combining the multi-modal abnormal data nodes and the community device correlation data, the devices closely related to the abnormal events can be determined and marked as target warning devices, so as to realize the timely attention and handling of potential problems. Executing warning control tasks on the target warning devices can quickly respond to abnormal events by controlling the target warning devices and take necessary measures to reduce the influence range of potential risk warnings. Through the accurate positioning of abnormal events and the marking and control of target warning devices, the intelligent warning and response of community devices are realized, which not only improves the system's perception and processing ability of abnormal situations, but also reduces the pressure of manual intervention, realizing the automatic and intelligent management of the community digital information control system.

[0136] In the embodiment of the present invention, the monitoring nodes of the multi-modal community association data matrix are located according to the unlabeled personnel data corresponding to the behavior abnormal data to obtain multi-modal abnormal data nodes corresponding to the behavior abnormal data. The multi-modal data nodes include monitoring device information corresponding to the monitoring data, etc. According to the multi-modal abnormal data nodes and the device association relationship of the multi-modal community association data matrix, the warning device most relevant to the monitoring device is found, and the warning device is marked as the target warning device. The target warning device executes corresponding warning control tasks, such as triggering an alarm, adjusting monitoring parameters, and feedbacking the location information of the warning device to relevant administrators, etc., to respond to the detected abnormal behavior.

[0137] This specification provides a community digital information intelligent control system for executing the community digital information intelligent control method as described above. The community digital information intelligent control system includes:

[0138] The device data interface design module is used to obtain historical community device data; based on the historical community device data, the data interface of the device is designed to generate a device data interface;

[0139] The device data interface optimization module is used to design the device identity identification of the device data interface according to the historical community device data, and generate device identity identification data; optimize the data interface identity of the device data interface based on the device identity identification data, and generate a device identification data interface;

[0140] The community device anomaly analysis module is used to collect device signals of community devices when the community devices execute data collection tasks, and generate device signals; divide the community devices with signal anomalies according to the device signals to obtain effective community devices and invalid community devices respectively; feedback the invalid community devices to the terminal to execute the repair control task of the invalid community devices;

[0141] The unlabeled personnel behavior data collection module uses the device identification data interface to receive data for the data collection task of optimizing community devices, so as to obtain effective community device data; collect real-time monitoring data of community devices for the effective community device data, and generate multimodal monitoring data; extract unlabeled facial image data from the multimodal monitoring data, and generate unlabeled facial image data; extract unlabeled personnel behavior data from the multimodal monitoring data according to the unlabeled facial image data, and generate unlabeled personnel behavior data;

[0142] The unlabeled personnel behavior data analysis module obtains historical personnel behavior anomaly training samples; uses the convolutional neural network algorithm and the historical personnel behavior anomaly training samples to establish a relationship model for unlabeled personnel behavior anomaly analysis, and generates a behavior anomaly analysis model; transmits the unlabeled personnel behavior data to the behavior anomaly analysis model for unlabeled personnel behavior anomaly analysis, and generates unlabeled personnel behavior anomaly analysis data;

[0143] The automated warning control module establishes an automated warning engine based on the unlabeled personnel behavior anomaly analysis data; transmits the unlabeled personnel behavior anomaly analysis data to the automated warning engine to execute the automated warning control task.

[0144] The beneficial effects of this application are as follows: obtaining historical community device data and performing format analysis of device data enables the system to better understand and process the structures of different device data. Based on the device format data and data transmission protocols, device data interface design is carried out, realizing the standardized and normalized transmission of device data. The ability to make full use of historical data is improved. Through format analysis and interface design, it can more intelligently adapt to different device types and data formats, thus enhancing the maintainability and integration of the system, and contributing to more efficient data collection, monitoring, and analysis of community devices of different brands, providing a solid foundation for the intelligent control of community digital information. Through the device data source information, the device identity identification design of the device data interface is carried out, generating device identity identification data with source identification, more accurately identifying and labeling different devices, enhancing the accuracy and integrity of device identity information, and optimizing the data interface identity identification, making the system more flexible in adapting to device identity identification information at the data interface level, improving the ability to identify and respond to different devices. By deeply understanding the data sources of devices and establishing accurate identity identifications, the data interface becomes more intelligent, which improves the ability to identify and adapt to devices and strengthens the ability to process diverse device data. Device signal acquisition is carried out to timely obtain device status information. Using Fourier transform technology to convert the device signal into a spectrogram and using the device signal spectrogram anomaly detection algorithm for anomaly detection helps the system to more comprehensively understand the spectral characteristics of device signals, be able to identify abnormal situations of device signals in real time, classify community devices as effective or ineffective community devices, realizing the automatic classification of device states. Feeding back ineffective community devices to the terminal to execute the repair control task of ineffective community devices can take timely actions to handle ineffective devices, improving the overall reliability of the system. By real-time monitoring and automatically classifying and processing device signals, the interference of ineffective devices on the normal operation of the system is reduced, improving the efficiency and response speed of community device management, and helping the system to better maintain community devices, ensuring the smooth operation of the digital information control system. Using the device identification data interface to receive data for the data collection task of effective community devices provides a basis for subsequent correlation analysis.Community device correlation analysis and establishment of a multi-modal community correlation data matrix based on optimized community devices can deeply understand the relationships between different devices, realize the integration and correlation of multi-modal data of different community devices, extract facial image data of monitoring data from the multi-modal community correlation data matrix to generate monitoring facial image data, and extract unlabeled facial image data from the monitoring facial image data according to the obtained facial image data of the property owner's identification, so as to analyze which outsiders are unlabeled personnel not in the community, and analyze the behaviors of unlabeled personnel, providing further guarantee for the management of the community. Through comprehensive data collection, correlation and analysis, intelligent monitoring and management of community devices and personnel behaviors are realized, which helps to improve community security and management efficiency, and provides more comprehensive and intelligent support for the intelligent control of community digital information. Using the convolutional neural network algorithm to establish the mapping relationship of unlabeled personnel behavior anomaly analysis, generating the initial behavior anomaly analysis model, and using historical personnel behavior anomaly training samples to train the model, enabling the model to learn and understand the normal patterns of different personnel behaviors, generating a more accurate and adaptable behavior anomaly analysis model. Transmitting the unlabeled personnel behavior data to the behavior anomaly analysis model for unlabeled personnel behavior anomaly analysis, generating unlabeled personnel behavior anomaly analysis data, which can identify and analyze the abnormal behaviors of unlabeled personnel in real time, and then perform subsequent automated early warning control according to this analysis data. By establishing a deep learning model, real-time intelligent analysis and monitoring of the behaviors of unlabeled personnel are carried out, which helps to improve the accuracy and real-time performance of the system for abnormal behaviors in the community, thus enhancing the perception and processing capabilities of abnormal situations. This provides strong support for the safety management of the community. An automated early warning engine is established based on the unlabeled personnel behavior anomaly analysis data, which can automatically execute early warning control. The automated household early warning engine discriminates whether to trigger early warning control according to the unlabeled personnel behavior anomaly analysis data. When the data received by the engine corresponds to abnormal behavior, the device closest to the monitored abnormal behavior is warned to improve the positioning ability of the early warning. On the other hand, when the data received by the engine corresponds to normal behavior, the unlabeled personnel behavior data corresponding to the normal behavior data is transmitted to the terminal for personnel non-abnormal behavior feedback, so as to eliminate the real-time monitoring of unlabeled personnel, enabling the administrator and monitoring devices to monitor other unlabeled personnel. Through the automated early warning engine, real-time monitoring and automated response to abnormal behaviors are realized, improving the automated early warning ability and response speed. By feeding back the normal behavior data to the terminal, timely confirmation of personnel non-abnormal behaviors is realized, reducing the false alarm rate, making the community digital information intelligent control system more intelligent, accurate and reliable.

[0145] Therefore, in any case, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the application documents are intended to be encompassed within the present invention.

[0146] The above are only specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the broadest scope consistent with the principles and novel features invented herein.

Claims

1. A method for intelligent control of community digital information, characterized in that, It includes the following steps: Step S1: Obtain the historical community device data of various community devices; Based on the historical community device data, conduct device data interface design to generate a device data interface; Among them, step S1 includes: Step S11: Obtain the historical community device data of various community devices; Step S12: Analyze the format of the device data according to the historical community device data to generate device format data; Step S13: Based on the device format data and the data transmission protocol, conduct device data interface design to generate a device data interface; Step S2: According to the historical community device data, conduct device identity identification design for the device data interface to generate device identity identification data; optimize the data interface identity of the device data interface based on the device identity identification data to generate a device identification data interface; Among them, step S2 includes: Step S21: Extract the device data source information from the historical community device data to generate device data source information; Step S22: According to the device data source information, conduct device identity identification design for the device data interface to generate device identity identification data; Step S23: Optimize the data interface identity of the device data interface based on the device identity identification data to generate a device identification data interface; Step S3: When the community device performs a data collection task, collect the device signal of the community device to generate a device signal; divide the community devices with signal anomalies according to the device signal to obtain effective community devices and invalid community devices respectively; feedback the invalid community devices to the terminal to execute the repair control task of the invalid community devices; Step S31: When the community device performs a data collection task, collect the device signal of the community device to generate a device signal; Step S32: Use the Fourier transform technology to convert the spectrum of the device signal to generate a device signal spectrum diagram; Step S33: Use the device signal spectrum diagram anomaly detection algorithm to perform spectrum anomaly detection calculation on the device signal spectrum diagram to generate spectrum anomaly detection data; Step S34: Divide the community devices according to the spectrum anomaly detection data. When the spectrum anomaly detection data is less than the preset anomaly spectrum detection threshold, mark the community device corresponding to the spectrum anomaly detection data as an effective community device. When the spectrum anomaly detection data is not less than the preset anomaly spectrum detection threshold, mark the community device corresponding to the spectrum anomaly detection data as an invalid community device; Step S35: Feedback the invalid community devices to the terminal to execute the repair control task of the invalid community devices; Step S4: Use the device identification data interface to receive the data of the data collection task of the effective community device to obtain effective community device data; collect the real-time monitoring data of the community device for the effective community device data to generate multi-modal monitoring data; extract the unlabeled face image data from the multi-modal monitoring data to generate unlabeled face image data; extract the unlabeled person behavior data from the multi-modal monitoring data according to the unlabeled face image data to generate unlabeled person behavior data; Among them, step S4 includes: Step S41: Receive data for the data collection task of valid community devices through the device identification data interface to obtain valid community device data; Step S42: Conduct an analysis of the relevance of community devices based on the valid community devices to generate community device relevance data; Step S43: Establish multi-modal community association matrix nodes according to the community device relevance data, and transmit the valid community device data to the multi-modal community association matrix nodes for data filling to generate a multi-modal community association data matrix; Step S44: Conduct real-time collection of monitoring data of community devices on the multi-modal community association data matrix to generate multi-modal monitoring data; Step S45: Extract facial image data of the monitoring data from the multi-modal monitoring data to generate monitored facial image data; Step S46: Obtain the facial image data of the owner identification in the community database; Step S47: Extract unlabeled facial image data from the monitored facial image data according to the facial image data of the owner identification to generate unlabeled facial image data; Step S48: Extract unlabeled personnel behavior data from the multi-modal monitoring data according to the unlabeled facial image data to generate unlabeled personnel behavior data; Step S5: Obtain historical personnel behavior anomaly training samples; Use the convolutional neural network algorithm and the historical personnel behavior anomaly training samples to establish a relationship model for unlabeled personnel behavior anomaly analysis to generate a behavior anomaly analysis model; Transmit the unlabeled personnel behavior data to the behavior anomaly analysis model for unlabeled personnel behavior anomaly analysis to generate unlabeled personnel behavior anomaly analysis data; Step S6: Establish an automated warning engine based on the unlabeled personnel behavior anomaly analysis data; Transmit the unlabeled personnel behavior anomaly analysis data to the automated warning engine to execute the automated warning control task.

2. The method for intelligent control of community digital information according to claim 1, characterized in that, The device signal spectrogram anomaly detection algorithm in Step S33 is as follows: In the formula, D(f) represents the spectrogram anomaly detection data, f represents the frequency parameter of the device signal spectrogram, σ represents the standard deviation of the conventional frequency parameter, μ represents the conventional frequency parameter, α represents the correction adjustment value of the external interference frequency parameter of the device signal spectrogram, γ represents the frequency attenuation rate of the device signal spectrogram, δ represents the offset of the attenuation frequency parameter of the device signal spectrogram, and η represents the attenuation frequency parameter of the device signal spectrogram.

3. The method for intelligent control of community digital information according to claim 1, characterized in that, Step S5 includes the following steps: Step S51: Use the convolutional neural network algorithm to establish a mapping relationship for unlabeled personnel behavior anomaly analysis to generate an initial behavior anomaly analysis model; Step S52: Obtain historical personnel behavior anomaly training samples; Step S53: Use the historical personnel behavior anomaly training samples to train the initial behavior anomaly analysis model to generate a behavior anomaly analysis model; Step S54: Transmit the unlabeled personnel behavior data to the behavior anomaly analysis model for unlabeled personnel behavior anomaly analysis to generate unlabeled personnel behavior anomaly analysis data, where the unlabeled personnel behavior anomaly analysis data includes behavior anomaly data or behavior normal data.

5. The intelligent control method for community digital information according to claim 3, characterized in that, Among them, the execution of the automated early warning control task includes performing an early warning control task on the target early warning device or transmitting the unlabeled personnel behavior data corresponding to the normal behavior data to the terminal for personnel non-abnormal behavior feedback. Step S6 includes the following steps: Step S61: Establish an automated early warning engine based on the unlabeled personnel behavior abnormal analysis data; Step S62: Transmit the unlabeled personnel behavior abnormal analysis data to the automated early warning engine for automated early warning control. When the behavior abnormal data corresponding to the unlabeled personnel behavior abnormal analysis data received by the automated engine, execute Step S63. Or, when the behavior normal data corresponding to the unlabeled personnel behavior abnormal analysis data received by the automated engine, transmit the unlabeled personnel behavior data corresponding to the behavior normal data to the terminal for personnel non-abnormal behavior feedback; Step S63: Mark the optimized community device as the target early warning device according to the unlabeled personnel data corresponding to the behavior abnormal data to obtain the target early warning device, and perform an early warning control task on the target early warning device.

6. The intelligent control method for community digital information according to claim 4, characterized in that, Step S63 includes the following steps: Locate the abnormal data node of the multi-modal community association data matrix according to the unlabeled personnel data corresponding to the behavior abnormal data, and generate a multi-modal abnormal data node; Mark the optimized community device as the target early warning device according to the multi-modal abnormal data node and the community device association data, and generate the target early warning device; Perform an early warning control task on the target early warning device.

7. An intelligent control system for community digital information, characterized in that, For executing the community digital information intelligent control method as described in Claim 1, the community digital information intelligent control system includes: The device data interface design module is used to obtain historical community device data; perform device data interface design based on the historical community device data to generate a device data interface; The device data interface optimization module is used to perform device identity identification design of the device data interface according to the historical community device data to generate device identity identification data; perform data interface identity identification optimization on the device data interface based on the device identity identification data to generate a device identification data interface; The community device abnormal analysis module is used to collect device signals of the community device when the community device performs a data collection task to generate device signals; divide the community device with signal abnormalities according to the device signals to obtain an effective community device and an ineffective community device respectively; feedback the ineffective community device to the terminal to execute an ineffective community device repair control task; The unlabeled personnel behavior data collection module uses the device identification data interface to receive data for the data collection task of the optimized community device to obtain effective community device data; perform real-time collection of monitoring data of the community device on the effective community device data to generate multi-modal monitoring data; extract unlabeled facial image data from the multi-modal monitoring data to generate unlabeled facial image data; extract unlabeled personnel behavior data from the multi-modal monitoring data according to the unlabeled facial image data to generate unlabeled personnel behavior data; An unlabeled personnel behavior data analysis module obtains historical personnel behavior anomaly training samples; uses a convolutional neural network algorithm and the historical personnel behavior anomaly training samples to establish a relationship model for unlabeled personnel behavior anomaly analysis, generating a behavior anomaly analysis model; transmits the unlabeled personnel behavior data to the behavior anomaly analysis model for unlabeled personnel behavior anomaly analysis, generating unlabeled personnel behavior anomaly analysis data; An automated early warning control module establishes an automated early warning engine based on the unlabeled personnel behavior anomaly analysis data; transmits the unlabeled personnel behavior anomaly analysis data to the automated early warning engine to execute the automated early warning control task.

Images