Space-air-ground integrated intelligent security system and method

Abstract

The problems of low security and protection monitoring intelligent level, limited data communication and the like under the scenes of defense, protection, production, safety, rescue and the like in the near-ground space are solved. The invention provides a space-air-ground integrated intelligent security system and method. The system is composed of an intelligent sentry cluster, a wide-area converged communication network and an intelligent security information platform. Each intelligent sentry in the intelligent sentry cluster is an integrated thunder-vision fusion monitoring smart station, and comprises a fusion monitoring unit, an edge computing unit, a satellite communication unit, a power distribution and energy storage unit and a bearing platform. The edge computing unit provides computing support for intelligence of the front-end sentry, and the rear-end intelligent security information platform realizes intelligent recognition learning, training, verification and comprehensive analysis on the basis of the front end. Three-dimensional situation awareness, real-time high-speed communication, intelligent analysis and early warning and efficient intelligent decision making are comprehensively realized. Compared with the traditional security and protection means, the scheme of the invention has higher technical content, greater benefit and wider applicability in different application scenes.

Description

Technical Field

[0001] The present invention belongs to the field of security monitoring, and in particular relates to a monitoring method and device applied to smart security scenarios. Background Art

[0002] With the development of science and technology and the changes in security prevention and emergency situations, it has become an important development trend to integrate multi-source perception, multi-mode communication, intelligent identification and other technologies to achieve smart security systems that can respond accurately and prevent and deal with problems across the entire area.

[0003] Smart security is a diversified, cross-domain, three-dimensional, collaborative, and intelligent technology system that meets the needs of defense, protection, production, safety, and rescue in the local space. Compared with traditional security methods, smart security solutions in different application scenarios have higher technical content, greater benefits, and wider applicability.

[0004] The application of AI technology can transform human defense into "intelligent defense". Through AI empowerment, a smart security brain can be established to enhance the ability to analyze and link massive intelligence data, providing efficient support for command and decision-making. High-throughput satellites can achieve three-dimensional communications with full coverage of wide areas. Summary of the invention

[0005] The purpose of this invention is to propose an integrated air-space-ground-integrated smart security system and method, integrating AI recognition, video surveillance, radar detection, satellite communication and other technologies to construct an integrated air-space-ground-integrated smart security solution with digital and visual fusion, intelligent collaboration and precise response, so as to achieve a comprehensive improvement in the perception system of security prevention and control and the level of intelligent management and control.

[0006] The technical solution of the present invention is: an air-ground-integrated intelligent security system, comprising Front-end stereoscopic multi-source sensing unit: completes the comprehensive aggregation of personnel, locomotives and ships, infrastructure, and weather and environmental data; Detection drones: to realize auxiliary data collection in key areas; Wide-area converged communication network: Realize real-time transmission of collected data; Smart security information platform: realize front-end equipment access, basic data management, unified early warning monitoring, intelligent analysis, comprehensive situation display, operation and maintenance management, event management and model training; It is necessary to combine application scenarios and collect a large amount of data in actual applications for model training, optimization and iteration; Command and dispatch center: complete basic data management, unified early warning monitoring, comprehensive situation awareness, and AR real-time command and dispatch; Through the collaborative work of the above units, the smart security goals of multi-source data comprehensive analysis, target identification, intrusion detection, risk warning, hidden danger assessment, comprehensive command and dispatch, and emergency response and rescue can be achieved.

[0007] Furthermore, the front-end stereoscopic multi-source perception unit is composed of a cluster of smart sentinels. Each smart sentinel in the smart sentinel cluster is an integrated radar-vision fusion monitoring smart site, including a fusion monitoring unit, an edge computing unit, a satellite communication unit, a power distribution and energy storage unit, and a carrier platform. The fusion monitoring unit uses visible light and thermal imaging fusion monitoring to achieve all-weather and all-round perception; it is equipped with a high-definition video monitoring system to achieve comprehensive coverage of the monitoring front-end points; it uses ground-based low-altitude monitoring radar to detect, monitor, track and measure low-altitude active targets; The edge computing unit collects and integrates multi-source data from the monitoring unit; uses AI technology and combines effective linkage of multiple systems to achieve detailed capture, tracking and positioning of suspected targets in important areas, and target classification; uses pre-trained target recognition algorithm models to classify the threat levels of detected intruding vehicles, personnel, and animals; according to different threat levels, it automatically warns in real time, processes accordingly, and performs linkage control; it subdivides threat levels to achieve intelligent recognition and operation status monitoring of different targets and personnel; and completes data storage and data transmission; and configures and manages target recognition sensitivity, detection range, and target type. The satellite communication unit uses high-throughput satellites to achieve three-dimensional communication with full coverage of a wide area, and transmits intelligence information and command and dispatch information at high speed around the clock, so that the monitoring equipment has wide-area high-speed communication capabilities; The energy storage and distribution unit provides continuous energy security for the equipment; The load-bearing platform is a high-strength integrated modular steel structure frame platform with reserved space for installing major equipment.

[0008] An air-ground-integrated intelligent security method adopts the above-mentioned air-ground-integrated intelligent security system, wherein the edge computing unit provides computing support for the intelligence of the front-end sentinel, and the back-end intelligent security information platform realizes intelligent recognition learning, training, verification and comprehensive analysis based on the front-end; The edge computing unit performs intrusion detection based on the AI ​​configuration, accurately identifies and classifies intrusion events according to the pre-set AI configuration, and pushes the events to the AI ​​event cloud platform for subsequent analysis and processing; The AI ​​event cloud platform is responsible for receiving and storing alarm events. Once an alarm event is received, the AI ​​event cloud platform will push the event to the administrator in a timely manner in various forms for timely processing. At the same time, the AI ​​event cloud platform will also push the event to the AI ​​training cloud platform for subsequent data analysis and model training. The AI ​​training cloud platform is responsible for receiving events pushed by the AI ​​event cloud platform and storing them for backup. Subsequently, the targets in the video are labeled and classified by manually annotating and classifying the data, providing accurate labels and data for model training. Through continuous iteration and optimization, a more accurate model is trained. After the training is completed, the AI ​​training cloud platform updates the new model to the edge computing unit to improve the accuracy and efficiency of intrusion detection.

[0009] Furthermore, the AI ​​configuration includes: AI detection sensitivity configuration: adjust the sensitivity of AI detection according to actual needs. By adjusting the sensitivity, the system can control the recognition accuracy and false alarm rate of the target; Area configuration: Set a specific area to specify the scope of AI detection. Through area configuration, the system can be limited to detect only the specified area, improving the efficiency and accuracy of the system.

[0010] Type configuration: Configure the target type for AI detection. Through type configuration, different types of targets can be detected and identified according to actual needs.

[0011] Furthermore, high-throughput satellites provide high-speed broadband access and adopt a data transmission mode that is mainly based on periodic status warning transmission and supplemented by emergency transmission in emergency situations. Combining the application characteristics of terminal sites and data transmission requirements, and considering bandwidth limitations, the node transmission bandwidth requirements are divided into two categories: periodic monitoring data transmission bandwidth requirements under daily monitoring conditions; and emergency communication data transmission bandwidth requirements under emergency conditions.

[0012] Furthermore, node data as a whole can be divided into structured data and unstructured data; Structured data includes node registration data, periodic keep-alive data, irregular warning data, command interaction data, and periodic status collection data; Unstructured data includes image data, compressed video data, and centralized surveillance video data; Furthermore, in daily monitoring status, the data transmission required for basic functions such as node online, equipment deployment, and status monitoring is ensured, so that services can be implemented with less bandwidth and more nodes can be accommodated; In an emergency, bandwidth occupancy is mainly due to unstructured video data, and more bandwidth is needed for emergency real-time communication. If necessary, the transmission bit rate can be lowered to reduce bandwidth occupancy. At the same time, applicable satellite link video compression and transmission technology should be used to ensure the quality of information transmission with limited bandwidth, and this can be achieved by sharing bandwidth among multiple nodes.

[0013] Compared with the prior art, the present invention has the following characteristics and beneficial effects: 1. Establish an integrated air-ground-ground smart security system, use the air-ground-ground multi-source data collection capabilities to achieve all-round three-dimensional perception, and use the rapidly developing artificial intelligence technology to change manual passive monitoring into video active warning. Through AI empowerment, establish a smart brain to provide efficient support for command decision-making. Through high-throughput satellite real-time wide-area communication, solve the pain points of "four no" application scenarios, achieve "deep" surveillance, and control the cross-domain macro situation.

[0014] 2. Smart Sentinel is an integrated radar-vision fusion monitoring smart site that integrates AI intelligent recognition technology, satellite communication technology, and wind-solar complementary technology to achieve all-weather ground perception capabilities. It can be used in areas without network coverage or city power coverage.

[0015] 3. Use front-end and back-end intelligence to achieve deep integration of intelligence and application scenarios. The front-end sentinel uses AI technology and combines the effective linkage of multiple systems to achieve automatic early warning, detail capture, tracking and positioning, and target classification of suspected targets in important areas. Change manual passive monitoring to active video early warning. The intelligence of the platform is reflected in the establishment of various targeted professional models, artificial intelligence models, and professional databases to provide support for management decision-making, command and dispatch.

[0016] 4. By using AI technology and combining the effective linkage of multiple systems, automatic early warning, detail capture, tracking and positioning, and target classification of suspected targets in important areas can be achieved, effectively avoiding the various drawbacks of traditional manual monitoring systems and greatly improving the accuracy of system early warning and the efficiency of manual investigation.

[0017] 5. High-throughput satellites can achieve full coverage of wide-area three-dimensional communications, and intelligence information and command and dispatch information can be transmitted at high speed around the clock. Ground monitoring equipment, vehicles, and personnel can have mobile, high-bandwidth, and real-time communication capabilities, greatly improving the efficiency of intelligence transmission and no longer being affected by terrain and climate. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of the air-ground-integrated smart security system; Figure 2 This is a schematic diagram of the composition of the Smart Sentinel; Figure 3 It is a schematic diagram of the AI ​​application system architecture; Figure 4 It is the flow chart of the cloud platform system model training; Figure 5 It is a transmission planning diagram. DETAILED DESCRIPTION

[0019] The present invention will be further described below in conjunction with the accompanying drawings and embodiments, but the practical application of the present invention is not limited to the following embodiments.

[0020] The air-space-ground integrated intelligent security system of the present invention is composed of a front-end three-dimensional multi-source sensing unit composed of a cluster of intelligent sentinels, which integrates data such as personnel, vehicles and ships, infrastructure, and weather environment, and uses detection drones to realize auxiliary data collection in key areas.

[0021] A converged communication network is formed by integrating high-throughput satellite communications and multiple networks to achieve wide-area transmission of collected data.

[0022] Establish a smart security information platform, develop applications for specific businesses, and realize business applications based on an integrated air-space-ground-land perception system.

[0023] With the command and dispatch center as the support, basic data management, unified early warning monitoring, comprehensive situation awareness, AR real-life command and dispatch, etc. are completed.

[0024] Through the collaborative work of the above units, we can achieve intelligent security goals such as comprehensive analysis of multi-source data, target identification, intrusion detection, risk warning, hidden danger assessment, comprehensive command and dispatch, emergency response and rescue. Figure 1 .

[0025] The front-end unit of the system is the intelligent sentinel cluster. The intelligent sentinel is an integrated radar-vision fusion monitoring smart site. It integrates AI intelligent recognition technology, satellite communication technology, and wind-solar complementary technology to achieve all-weather ground perception capabilities. It can be used in areas without network coverage or city power coverage. It mainly includes fusion monitoring units, edge computing units, satellite communication units, power distribution and energy storage units, and carrier platforms. See the schematic diagram of the intelligent sentinel composition. Figure 2 .

[0026] 1) The fusion monitoring unit uses visible light and thermal imaging fusion monitoring to achieve all-weather and all-round perception. It is equipped with a high-definition video monitoring system, and can choose a variety of cameras such as panoramic cameras, gun-ball linkage, double-cabin cameras, and high-precision ball-type turntables to achieve comprehensive coverage of the monitoring front-end points. A ground-based low-altitude monitoring radar is used to discover, monitor, track, and measure low-altitude active targets.

[0027] 2) The edge computing unit uses AI technology and combines the effective linkage of multiple systems to achieve automatic early warning, detail capture, tracking and positioning, and target classification of suspected targets in important areas. According to the threat level, it can realize intrusion detection of different targets and units, and realize automatic identification and monitoring of different personnel and targets. According to different threat levels, the corresponding processing mode is realized, and linkage processing is realized. According to the intrusion identification of vehicles, personnel, and animals, targeted scene training is carried out to realize the identification of targets of different threat levels.

[0028] 3) Satellite communication units are the key to three-dimensional perception and the central nervous system of smart security. High-throughput satellites can be used to achieve three-dimensional communication with full coverage of a wide area, and intelligence information and command and dispatch information can be transmitted at high speed around the clock. Monitoring equipment, vehicles, and personnel can have mobile, high-bandwidth, and real-time communication capabilities, greatly improving the efficiency of intelligence transmission and no longer being affected by terrain and climate.

[0029] 4) The energy storage and distribution unit provides continuous energy security for the equipment. It consists of the following components: solar cell modules, wind power generation modules, wind-solar hybrid controllers, battery packs, inverters, distribution management modules and cabinets. The power supply system is designed according to outdoor application scenarios, and is designed and calculated for different working modes such as sleep mode, early warning mode, and full-load working mode.

[0030] 5) The bearing platform is a high-strength integrated modular steel structure platform with a frame structure. It can be selected in the form of poles, trusses, iron towers, etc. according to the actual situation of the point and the detection distance. The installation position of the main equipment is reserved; it can be installed with monitoring ball cameras, gun cameras, solar energy, perimeters, antennas, cabinets and other equipment, taking into account aesthetics, safety, and convenient installation. The bearing platform is designed in combination with a variety of application scenarios and different terrains, which is easy to transport, easy to install and construct, and adapts to different installation conditions such as mountains, deserts, and Gobi.

[0031] The present invention uses the comprehensive application of front-end and back-end intelligence to achieve deep integration with application scenarios. The front-end sentinel uses AI technology and combines the effective linkage of multiple systems to realize automatic early warning, detail capture, tracking and positioning, and target classification of suspected targets in important areas. Change manual passive monitoring to active video early warning. By analyzing and refining various behavioral patterns of targets such as personnel and vehicles, a core algorithm is formed. In the application, through comparison and comparison, it is identified what kind of object and behavior the collected video image belongs to, and the frame circumference and movement trajectory of the target are labeled, and real-time alarms are triggered to trigger linkage mechanisms such as capture, recording, and early warning.

[0032] The application of AI technology relies on edge computing units and smart security information platforms. The edge computing unit provides computing support for the intelligence of the front-end sentinels. The smart security information platform implements intelligent recognition learning, training, verification and comprehensive analysis based on the front-end. Figure 3 .

[0033] The smart security information platform realizes front-end equipment access, basic data management, unified early warning monitoring, intelligent analysis, comprehensive display, operation and maintenance management, event management, and model training. The smart security information platform needs to be combined with application scenarios, and in actual applications, through a large amount of data collection, model training, optimization, and iteration, to achieve a deep integration of intelligence and scenarios.

[0034] The edge computing unit performs intrusion detection based on the AI ​​configuration. Through the intrusion detection function of the edge computing unit, the intrusion events are accurately identified and classified according to the pre-set AI configuration. At the same time, the events are pushed to the AI ​​event cloud platform for subsequent analysis and processing.

[0035] The main AI configurations include: AI detection sensitivity configuration: Adjust the sensitivity of AI detection according to actual needs. By adjusting the sensitivity, you can control the system's recognition accuracy and false alarm rate of the target.

[0036] Area configuration: Set a specific area to specify the range of AI detection. Through area configuration, the system can be limited to detect only the specified area, improving the efficiency and accuracy of the system.

[0037] Type configuration: Configure the target type for AI detection, such as people, cars, etc. Through type configuration, different types of targets can be detected and identified according to actual needs.

[0038] The AI ​​event cloud platform is responsible for receiving and storing alarm events. Once an alarm event is received, the platform will push the event to the administrator in various forms for timely notification and processing. At the same time, the platform will also push the event to the AI ​​training cloud platform for subsequent data analysis and model training.

[0039] The AI ​​training cloud platform is responsible for receiving events pushed by the AI ​​event cloud platform and storing them for backup. Subsequently, by manually annotating and classifying data, the targets in the video are annotated and classified to provide accurate labels and data for model training to improve the accuracy and performance of the model. Through continuous iteration and optimization, a more accurate model is trained. After the training is completed, the AI ​​training cloud platform updates the new model to the edge computing unit to improve the accuracy and efficiency of intrusion detection. See the flowchart of the model training of the smart security information platform. Figure 4 .

[0040] High-throughput satellites can provide high-speed broadband access, but due to the large number of service nodes, bandwidth needs to be reasonably planned to ensure the real-time and quality of data services. The system takes into account the bandwidth and cost issues of satellite links for multi-point monitoring, and adopts a data transmission mode that is mainly based on periodic status warning transmission and supplemented by emergency transmission in emergency situations. Combined with the application characteristics of terminal sites and data transmission requirements, and considering bandwidth limitations, the node transmission bandwidth requirements are divided into two categories: periodic monitoring data transmission bandwidth requirements under daily monitoring conditions; emergency communication data transmission bandwidth requirements under emergency conditions. Under daily monitoring conditions, the main guarantee is the data transmission required for basic functions such as node online, equipment deployment, and status monitoring. Services can be provided with less bandwidth to accommodate more nodes. In an emergency, real-time information and video need to be provided for emergency disposal to ensure real-time and visualization.

[0041] Example: Node data as a whole can be divided into structured data and unstructured data.

[0042] Structured data includes node registration data, periodic keep-alive data, irregular warning data, command interaction data, periodic status collection data, etc., and is generally about 100KB in size.

[0043] Unstructured data includes image data, compressed video data, centralized monitoring video data, etc. One image is about 100KB-1MB byte, and one second of video is about 100KB-1MB byte.

[0044] In daily monitoring mode: the data size is calculated based on a transmission packet size of 500KB (100KB for structured data + 400KB for unstructured data), a transmission cycle of 1 minute, and sudden warning transmission is also considered. Data volume in 1 hour: 500KB *60*2=60000KB≈60MB.

[0045] In an emergency state: In this state, bandwidth is mainly occupied by unstructured video data. In this state, more bandwidth is required for emergency real-time communication. The transmission bit rate determines the bandwidth occupation. If necessary, the transmission bit rate can be reduced to reduce bandwidth occupation. At the same time, satellite link video compression and transmission technology is used to ensure the quality of information transmission with limited bandwidth. This is achieved by sharing bandwidth among multiple nodes. See the transmission planning diagram for details. Figure 5 .

Claims

1. An air-ground-integrated intelligent security system, characterized by: include Front-end stereoscopic multi-source sensing unit: completes the comprehensive aggregation of personnel, locomotives and ships, infrastructure, and weather and environmental data; Detection drones: to realize auxiliary data collection in key areas; Wide-area converged communication network: Realize real-time transmission of collected data; Smart security information platform: realize front-end equipment access, basic data management, unified early warning monitoring, intelligent analysis, comprehensive situation display, operation and maintenance management, event management and model training; It is necessary to combine application scenarios and collect a large amount of data in actual applications for model training, optimization and iteration; Command and dispatch center: complete basic data management, unified early warning monitoring, comprehensive situation awareness, and AR real-time command and dispatch; Through the collaborative work of the above units, the smart security goals of multi-source data comprehensive analysis, target identification, intrusion detection, risk warning, hidden danger assessment, comprehensive command and dispatch, and emergency response and rescue can be achieved.

2. The air-ground-integrated intelligent security system as claimed in claim 1, characterized in that: The front-end stereoscopic multi-source perception unit is composed of a cluster of intelligent sentinels. Each intelligent sentinel in the cluster is an integrated radar-vision fusion monitoring smart site, including a fusion monitoring unit, an edge computing unit, a satellite communication unit, a power distribution and energy storage unit, and a carrier platform. The fusion monitoring unit uses visible light and thermal imaging fusion monitoring to achieve all-weather and all-round perception; it is equipped with a high-definition video monitoring system to achieve comprehensive coverage of the monitoring front-end points; it uses ground-based low-altitude monitoring radar to detect, monitor, track and measure low-altitude active targets; The edge computing unit collects and integrates multi-source data from the monitoring unit; uses AI technology and combines the effective linkage of multiple systems to achieve detailed capture, tracking and positioning of suspected targets in important areas, and target classification; uses pre-trained target recognition algorithm models to classify the threat levels of detected intruding vehicles, personnel, and animals; according to different threat levels, it automatically warns in real time, processes accordingly, and conducts linkage control; and subdivides threat levels to achieve intelligent identification and operation status monitoring of different targets and personnel; And complete data storage and data transmission; at the same time, configure and manage target recognition sensitivity, detection range, and target type; The satellite communication unit uses high-throughput satellites to achieve three-dimensional communication with full coverage of a wide area, and transmits intelligence information and command and dispatch information at high speed around the clock, so that the monitoring equipment has wide-area high-speed communication capabilities; The energy storage and distribution unit provides continuous energy security for the equipment; The load-bearing platform is a high-strength integrated modular steel structure frame platform with reserved space for installing major equipment.

3. A smart security method using the air-ground integrated smart security system according to any one of claims 1-2, characterized in that the edge The computing unit provides computing support for the intelligence of the front-end sentinels, while the back-end smart security information platform implements intelligent recognition learning, training, verification and comprehensive analysis based on the front-end. The edge computing unit performs intrusion detection according to the AI ​​configuration, accurately identifies and classifies intrusion events according to the pre-set AI configuration, and pushes the events to the AI ​​event cloud platform for subsequent analysis and processing. The AI ​​event cloud platform is responsible for receiving and storing alarm events. Once an alarm event is received, the AI ​​event cloud platform will push the event to the administrator in a timely manner in various forms for timely processing. At the same time, the AI ​​event cloud platform will also push the event to the AI ​​training cloud platform for subsequent data analysis and model training. The AI ​​training cloud platform is responsible for receiving events pushed by the AI ​​event cloud platform and storing them for backup. Subsequently, the targets in the video are labeled and classified by manually annotating and classifying the data, providing accurate labels and data for model training. Through continuous iteration and optimization, a more accurate model is trained. After the training is completed, the AI ​​training cloud platform updates the new model to the edge computing unit to improve the accuracy and efficiency of intrusion detection.

4. The air-ground-integrated intelligent security method as claimed in claim 3 is characterized in that: AI configuration includes: AI detection sensitivity configuration: adjust the sensitivity of AI detection according to actual needs. By adjusting the sensitivity, the system can control the recognition accuracy and false alarm rate of the target; Area configuration: Set a specific area to specify the scope of AI detection. Through area configuration, the system can be limited to detect only the specified area, improving the efficiency and accuracy of the system; Type configuration: Configure the target type for AI detection. Through type configuration, different types of targets can be detected and identified according to actual needs.

5. The air-ground-integrated intelligent security method as claimed in claim 4 is characterized in that: High-throughput satellites provide high-speed broadband access and adopt a data transmission mode that is mainly based on periodic status warning transmission and supplemented by emergency transmission in emergency situations. Combining the application characteristics of terminal sites and data transmission requirements, and considering bandwidth limitations, the node transmission bandwidth requirements are divided into two categories: periodic monitoring data transmission bandwidth requirements under daily monitoring status; emergency communication data transmission bandwidth requirements under emergency status.

6. The air-ground-integrated intelligent security method as claimed in claim 5, characterized in that Node data as a whole can be divided into structured data and unstructured data; Structured data includes node registration data, periodic keep-alive data, irregular warning data, command interaction data, and periodic status collection data; Unstructured data includes image data, compressed video data, and centralized monitoring video data.

7. The air-ground-integrated intelligent security method as claimed in claim 6, characterized in that: In daily monitoring state, the data transmission required for node online, equipment deployment, and status monitoring basic functions is ensured, and services are provided with less bandwidth to accommodate more nodes; In an emergency, bandwidth occupancy is mainly due to unstructured video data, and more bandwidth is needed for emergency real-time communication. If necessary, the transmission bit rate can be lowered to reduce bandwidth occupancy. At the same time, applicable satellite link video compression and transmission technology should be used to ensure the quality of information transmission with limited bandwidth, and this can be achieved by sharing bandwidth among multiple nodes.

Images