Remote transmission and round patrol method for fire-fighting map detection data

Abstract

The application discloses a kind of remote transmission and round patrol method of fire-fighting map detection data, through multi-source data fusion identification disaster and generate dynamic digital task card, construct self-adapting redundant communication link and intelligent layered coding mechanism, realize the efficient transmission of key information under complex network environment;Round patrol strategy is upgraded to event-driven mode, combined with dynamic priority hierarchical scheduling transmission and display resources, while linkage historical case library intelligent push disposal experience, realize whole-process cooperation by unified event bus;Fusion geographic information generates three-dimensional enhanced situation map, and establishes comprehensive monitoring and evaluation and strategy closed-loop optimization mechanism, significantly improves disaster perception depth, communication reliability and command decision efficiency, provides intelligent, high-reliability map detection data support for fire emergency rescue.

Description

Technical Field

[0001] This invention relates to the field of fire protection image reconnaissance data technology, specifically a method for remote transmission and rotation of fire protection image reconnaissance data. Background Technology

[0002] With the increasing height and complexity of urban buildings, and the growing safety risks associated with special spaces such as chemical plants and underground facilities, disaster sites place extremely high demands on real-time, accurate, and reliable image reconnaissance and data transmission. Currently, image reconnaissance data transmission and patrol methods in the field of fire emergency rescue are evolving from traditional one-way video transmission towards preliminary intelligent and multi-source approaches. However, when dealing with complex, dynamic, and harsh disaster sites, existing technical solutions still face significant limitations and shortcomings in practical applications, mainly in the level of intelligence in information acquisition and the reliability of transmission assurance, which restricts the further improvement of the overall effectiveness of emergency rescue. Currently, mainstream fire image reconnaissance systems can integrate multiple video sources such as drones, robots, and individual soldiers, and transmit real-time images back to the command center through communication networks, but the analysis and judgment of massive video stream content still heavily relies on commanders manually monitoring screens. Existing systems typically employ a simple time-slice rotation method for displaying and switching multiple video feeds. This approach fails to intelligently identify and automatically switch between feeds based on the urgency and importance of actual disaster events (such as trapped individuals, fire spread, or structural anomalies). This leads to two core problems: firstly, critical events (such as brief cries for help from trapped individuals) may be missed due to gaps in the rotation, resulting in missed opportunities for optimal rescue; secondly, commanders must continuously manually sift through multiple feeds under high pressure, easily leading to visual fatigue and judgment errors. Although some systems integrate geographic information, this is usually just a simple two-dimensional map overlay, unable to achieve precise spatial fusion and three-dimensional visualization with the video stream, making it difficult to intuitively construct a global spatial situation. While the value of historical case databases has been recognized, current technologies lack the ability to quickly and intelligently compare and match current disaster characteristics with historical cases in real-time tasks and accurately push relevant experiences. Current technologies have made limited progress in achieving the qualitative leap from simply seeing to understanding, and there is still a significant gap to bridge before realizing automated and intelligent deep disaster perception and proactive decision support. Summary of the Invention

[0003] The purpose of this invention is to provide a method for remote transmission and rotation of fire protection image reconnaissance data, so as to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for remote transmission and rotation of fire reconnaissance image data, comprising the following steps:

[0005] S1. Automatically identify disaster situations through multi-source data analysis and generate dynamic digital task cards containing resource deployment suggestions and initial strategy parameters;

[0006] S2. Integrate multiple communication modules and intelligent transmission control to establish an adaptive redundant communication guarantee link;

[0007] S3. Intelligent layered encoding and transmission optimization of video streams, adaptive transmission under different network conditions;

[0008] S4. Upgrade the polling strategy to event and rule-driven, automatically identify abnormal events and trigger switching and alarms;

[0009] S5. Based on the events identified in step S4, and according to the dynamic comprehensive priority, allocate transmission and display resources in a hierarchical manner, and perform differentiated scheduling of the video stream.

[0010] S6. Compare the video identified in step S4 with the historical database, associate and push historical cases, and simultaneously input the current task data into the database.

[0011] S7. Through a unified event bus and arbitration mechanism, information exchange, event response and strategy coordination are carried out for all steps from S1 to S6.

[0012] S8. Based on the video transmitted in step S2 and the video information optimized in step S3, integrate geographic information to generate an enhanced situation map that visualizes the location of the video source in three-dimensional space.

[0013] S9. Conduct comprehensive monitoring and visual evaluation of the communication quality of step S2, the video information density of step S3, and the polling efficiency of steps S4 and S5.

[0014] S10. Based on the monitoring data from step S9 and the historical cases from step S6, the strategy is optimized in a closed loop through dynamic parameter tuning, simulation, and case matching.

[0015] Preferably, step S1 specifically includes the following steps:

[0016] S11. Automatically access and integrate multi-source heterogeneous data, including alarm information, disaster type, meteorological information, geographic information, emergency plan database, and reports from on-site forces;

[0017] S12. Based on the results of the fusion analysis, identify the specific types of disasters, including high-rise building fires, chemical leaks, and earthquake collapses, and determine the relevant risk factors and key areas;

[0018] S13. Based on the preliminary judgment results, automatically construct a dynamic digital task card for this task, wherein the dynamic digital task card contains the basic information of the disaster that has been judged;

[0019] S14. When constructing the task card, two core pieces of information are automatically derived. These two core pieces of information specifically include:

[0020] Recommend initial deployment sites for mobile reconnaissance equipment such as drones and robots;

[0021] By matching the task model with a pre-built strategy template library, initial strategy parameters are automatically generated. These automatically generated initial strategy parameters specifically include:

[0022] Recommended initial primary and backup communication link combination;

[0023] Generate an initial video source priority list and display template;

[0024] Preset bitrate adjustment sensitivity parameters suitable for the current disaster scenario.

[0025] Preferably, step S2 specifically includes the following steps:

[0026] S21. Based on the initial primary and backup communication link combination recommended in the task card of step S1, establish a redundant communication link from the field image acquisition terminal to the rear command center.

[0027] S22. The acquisition terminal integrates multiple communication modules, including: fourth-generation mobile communication technology, fifth-generation mobile communication technology, satellite communication, mesh network self-organizing network and fire protection network;

[0028] S23. Establish a transmission control unit to monitor the operating quality of each link in real time. The specific indicators of real-time monitoring include signal strength, delay, and packet loss rate.

[0029] S24. The transmission control unit dynamically selects one of the available links as the primary link based on a preset strategy to carry the main data transmission.

[0030] S25. Implement a data splitting and parallel transmission mechanism to split the data stream and transmit data packets simultaneously through multiple links;

[0031] S26. When a link is detected to be interrupted, the traffic carried will be switched to other available links to ensure uninterrupted communication.

[0032] Preferably, step S3 specifically includes the following steps:

[0033] S31. Before the video data transmitted in step S2 is sent out, the video stream is analyzed and processed, and different elements are extracted and separated based on information entropy.

[0034] S32. Decompose the video frame into main components, the main components specifically including: static background layer, dynamic target layer and key semantic features;

[0035] S33. Combining the preset bitrate adjustment sensitivity parameters in the task card of step S1, adopt differentiated encoding and transmission strategies for different main components, specifically including the following steps:

[0036] S331. Structurally encode key semantic features and assign them high priority during transmission;

[0037] S332. A balanced compression strategy is adopted for the dynamic target layer;

[0038] S333. For static background layers, efficient compression or periodic updates are used instead of continuous full transmission.

[0039] S34. In order to cope with the network fluctuations of the link in step S2, when the network is congested, actively discard low information entropy data or instead transmit the abstracted and refined semantic graph.

[0040] S35. In the extreme weak network environment of the link in step S2, the video stream is converted into a dynamic graphic newsletter with spatial coordinate information for transmission.

[0041] Preferably, step S4 specifically includes the following steps:

[0042] S41. Upgrade the polling strategy from the traditional time-slice rotation to an event- and rule-driven model;

[0043] S42. Continuously analyze the video stream transmitted in step S3 in real time to identify preset key event types, including flame recognition, smoke detection, people falling to the ground, and abnormal building structure.

[0044] S43. When the above-mentioned abnormal event is detected in any video, the video source is automatically marked as a high-priority event source.

[0045] S44. Perform two core actions simultaneously, specifically including:

[0046] Automatically switch the main display screen of the command center to the high-priority event video of this channel;

[0047] An audible and visual alarm was sent to the command center to alert the personnel in charge.

[0048] Preferably, step S5 specifically includes the following steps:

[0049] S51. Manage the video stream and dynamically assign a comprehensive priority to each video source;

[0050] S52. The calculation of the overall priority is determined by multiple factors, including the fixed weight set by the task card in step S1, the dynamic events identified in step S4, and the manual specification by the operator.

[0051] S53. Based on the calculated video priority and the current available total network bandwidth of the link in step S2, decide on the allocation of transmission and display resources;

[0052] S54. Execute the hierarchical resource allocation strategy:

[0053] S541. For the highest priority video stream, allocate the dedicated transmission channel of step S2 and display it continuously and exclusively in the command center.

[0054] S542. For video streams with medium priority, display them using split-screen sharing or timed rotation methods while ensuring basic watchability.

[0055] S543. For low-priority or non-critical video streams, place them in a background polling queue and manually retrieve and view them as needed, or only upload periodic snapshots.

[0056] Preferably, step S6 specifically includes the following steps:

[0057] S61. Compare and analyze all received video frames with the database storing historical disaster scene feature data;

[0058] S62. Extract the scene features, situational information, and key structural features identified in the current video stream in step S4;

[0059] S63. The extracted scene features, situational information, and key structural features are matched with cases in the historical database, and the similarity is calculated.

[0060] S64. When the similarity calculated by matching exceeds the preset threshold, automatically associate and push detailed information of the high-similarity historical case, including process records, key decision points, handling points and final result summary.

[0061] S65. After structuring the complete data stream of the current task, including the three-dimensional situation evolution process generated in step S8, the performance data recorded in S9, and all key events in steps S4 and S5, the data is synchronously stored in the historical disaster database to enrich the case library.

[0062] Preferably, step S7 specifically includes the following steps:

[0063] S71, responsible for information exchange, event response and strategy coordination between all steps from S1 to S6;

[0064] S72. Implement a unified event publishing and subscription mechanism, specifically including:

[0065] S721. The events identified in step S4, the link quality anomalies monitored in step S9, the status changes in steps S2 and S3, and the commander's operation instructions in steps S5 and S6 are all published as events to the unified event bus.

[0066] S722, Steps S1 to S6 and other steps: Subscribe to the types of events that need to be responded to on the bus according to their own responsibilities;

[0067] S73. Execution Strategy Conflict Arbitration and Coordinated Response:

[0068] S731. When multiple concurrent events in steps S4, S5, and S6 trigger strategies for different steps and cause conflicts, arbitration is initiated.

[0069] S732. Arbitration is conducted based on the core objectives, preset rules, and current task stage set in step S1 of the task card.

[0070] S733. Finally, a unified collaborative decision is output and distributed to the relevant steps for execution.

[0071] S74. Drives adaptive switching of task phases, identifies key phase transition nodes in task progress by analyzing event flow;

[0072] S75. When it is determined that a phase transition is required, the corresponding phase coordination strategy package is called from the preset strategy library or optimized and generated by step S10.

[0073] S76. Use a new strategy package to refresh the strategy parameters of multiple related steps, including steps S2, S3, and S5, to adapt to the needs of the new stage.

[0074] Preferably, step S8 specifically includes the following steps:

[0075] S81. Utilize the video stream and data transmitted back from step S2, which contain precise timestamps and spatial stamps;

[0076] S82. The real-time data in step S81 is fused with the on-site architectural drawings or geographic information system data loaded in the task card of step S1.

[0077] S83. By estimating the pose, the viewpoint of each video stream is bound to specific coordinates and directions in three-dimensional space.

[0078] S84. Generate an interactive video-enhanced 3D situation map, in which each video source is visualized on the 3D model in the form of a virtual camera.

[0079] S85. Commanders can directly access and view the real-time video stream of any virtual camera on this 3D situation map by clicking on it.

[0080] Step S9 specifically includes the following steps:

[0081] S91. For the communication links in step S2, display the real-time performance indicators of each link. The real-time performance indicators specifically include bandwidth utilization, latency, packet loss rate and historical switching records. Give the transmission reliability rating of each link in the form of a comprehensive score, and issue an alarm for the links that are about to deteriorate.

[0082] S92. For the video stream output from step S3, monitor the video quality and information density, display the real-time bitrate, and evaluate the key information density through an algorithm to form a quality change curve and density map.

[0083] S93. Based on the event recognition log generated in step S4 and the resource scheduling record generated in step S5, generate and display an intelligent patrol efficiency heat map to reflect the distribution of the command center's attention to each video source.

[0084] Preferably, step S10 specifically includes the following steps:

[0085] S101. Based on the monitoring data collected in step S9, optimize and analyze the overall system strategy.

[0086] S102. Perform dynamic optimization of strategy parameters based on feedback: Based on the performance data provided in step S9 and the intervention records of the commander in steps S5 and S6, automatically or assistedly optimize the strategy parameters of the preceding related steps in steps S2, S3 and S5.

[0087] S103. Based on the video-enhanced 3D situation map generated in step S8 and the full and synchronous task data recorded in step S9, an interactive task process simulation copy is generated after the task is completed.

[0088] S104. In this copy, reset the initial strategy of step S1 and deduce the changes in the transmission delay, interruption time and key event discovery time index of the key screen under the new strategy.

[0089] S105. In the simulation copy, simulate the adjustment of the preset deployment position of the video source in step S1, and simulate the impact of this adjustment on the event detection probability in step S3 and the three-dimensional situation coverage integrity in step S8.

[0090] S106. Implement intelligent matching and strategy recommendation for historical cases:

[0091] S1061. Call the historical database of step S6 and perform a deep feature comparison between the process simulation copy generated by the current task and the most similar historical case in the database.

[0092] S1062. Through deep feature comparison, the key differences and risk evolution trends between the current task and historical cases are automatically identified.

[0093] S1063. Based on the lessons learned from historical cases, generate targeted strategy optimization suggestions for reviewing current tasks and planning future tasks.

[0094] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure.

[0095] Compared with existing technologies, the beneficial effects of this invention are as follows: Through multi-source data fusion and intelligent analysis in step S1, disaster types, risk factors, and key areas can be accurately identified, and dynamic digital task cards containing scientific deployment suggestions and optimized initial strategies can be generated, laying a precise decision-making foundation for the entire rescue mission; The introduction of steps S4, S6, and S8 upgrades the traditional video patrol and manual judgment to a collaborative perception mode driven by artificial intelligence event recognition, intelligent push of historical experience, and intuitive presentation of the three-dimensional spatial situation; This ensures that commanders and fighters can discover the most critical events, such as trapped personnel or gas leaks, obtain high-value experience references in a timely manner, and control the situation from a global three-dimensional perspective, significantly improving the depth, breadth, and speed of disaster perception, making command and decision-making more scientific, proactive, and efficient;

[0096] Step S2 establishes an intelligent redundant link integrating multiple communication modules, possessing dynamic routing, data splitting, and millisecond-level fault switching capabilities based on real-time quality monitoring, ensuring the continuous availability of the communication link at the physical layer. Step S3 performs revolutionary intelligent processing of the video stream at the application layer, stratifying video elements according to information entropy into key semantic features, dynamic targets, and static backgrounds, and implementing differentiated encoding and transmission strategies, adaptively adjusting based on network conditions, such as sacrificing background to preserve features during congestion and transmitting rapid text and image reports in extremely weak networks. Step S5 further performs precise scheduling of transmission and display resources based on dynamic priority. These three aspects work together to form a triple guarantee mechanism of elastic network, intelligent encoding and transmission, and precise scheduling, ensuring that under any network conditions, the command center can receive the most critical disaster information, such as personnel locations and leak points, with the lowest latency and highest priority, maximizing information transmission efficiency under limited bandwidth resources. Attached Figure Description

[0097] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings in the following description are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0098] Figure 1 This is the overall flowchart of fire protection image reconnaissance data transmission and patrol in this invention;

[0099] Figure 2 This is a flowchart of the intelligent coding and polling strategy sub-process of the present invention;

[0100] Figure 3 This is a simplified flowchart of the data acquisition and polling process of this invention;

[0101] Figure 4 This is a timing diagram of the video information density of the present invention;

[0102] Figure 5 This is a time-series diagram of video source attention in this invention;

[0103] Figure 6 This is a comparison chart of transmission delays at different priorities according to the present invention. Detailed Implementation

[0104] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses consistent with some aspects of this disclosure as detailed in the appended claims.

[0105] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0106] Example 1:

[0107] See Figures 1 to 6As shown, an embodiment of the present invention provides a method for remote transmission and rotation of fire-fighting image reconnaissance data, comprising the following steps: S1, automatically identifying disaster situations through multi-source data analysis and generating dynamic digital task cards containing resource deployment suggestions and initial strategy parameters; S2, integrating multiple communication modules and intelligent transmission control to establish an adaptive redundant communication guarantee link; S3, performing intelligent layered encoding and transmission optimization on the video stream, enabling adaptive transmission under different network conditions; S4, upgrading the rotation strategy to event- and rule-driven, automatically identifying abnormal events and triggering switching and alarms; S5, combining the events identified in step S4, allocating transmission and display resources hierarchically according to dynamic comprehensive priority, and performing differentiated scheduling of the video stream; S6, through the identification in step S4... The video is compared with the historical database, and historical cases are pushed along. The current task data is also entered into the database simultaneously. S7. Through a unified event bus and arbitration mechanism, information exchange, event response and strategy coordination are carried out for all steps from S1 to S6. S8. Based on the video transmitted in step S2 and the optimized video information in step S3, geographic information is integrated to generate an enhanced situation map in which the video source is visualized and located in three-dimensional space. S9. The communication quality of step S2, the video information density of step S3 and the patrol efficiency of steps S4 and S5 are comprehensively monitored and visualized for evaluation. S10. Based on the monitoring data of step S9 and the historical cases of step S6, the strategy is optimized in a closed loop through dynamic parameter adjustment, simulation and case matching.

[0108] Example 2:

[0109] Step S1 specifically includes the following steps:

[0110] S11. Automatically access and integrate multi-source heterogeneous data, including alarm information, disaster type, meteorological information, geographic information, emergency plan database, and reports from on-site forces;

[0111] S12. Based on the results of the fusion analysis, identify the specific types of disasters, including high-rise building fires, chemical leaks, and earthquake collapses, and determine the relevant risk factors and key areas.

[0112] S13. Based on the preliminary judgment results, automatically construct a dynamic digital task card for this task. The dynamic digital task card contains the basic information of the disaster that has been judged.

[0113] S14. When constructing the task card, two core pieces of information are automatically derived. These two core pieces of information specifically include:

[0114] Recommend initial deployment sites for mobile reconnaissance equipment such as drones and robots;

[0115] By matching the task model with a pre-built strategy template library, initial strategy parameters are automatically generated. These automatically generated initial strategy parameters specifically include:

[0116] Recommended initial primary and backup communication link combination;

[0117] Generate an initial video source priority list and display template;

[0118] Preset bitrate adjustment sensitivity parameters suitable for the current disaster scenario;

[0119] The matching mechanism between the task model and the strategy template library involves task model construction, strategy library organization, and matching between the two. The task model is based on the disaster type, risk factors, and key areas identified by S12, and integrates data such as meteorological, on-site forces, GIS geographic information, and emergency response procedures accessed by S11 to form a structured digital task description. The strategy template library pre-stores a set of verified initial response strategies for typical disaster scenarios such as high-rise building fires, chemical leaks, and earthquake collapses, as well as different environmental conditions such as severe weather and signal blockage. Each template is associated with the feature dimensions of the task model and includes complete initial parameter configurations such as communication link combinations, video source priority lists, display templates, and encoding parameters. The matching process is achieved by calculating the feature similarity between the current task model and all templates in the template library. A rule-based engine or weight-based matching algorithm can be used to find the template with the highest similarity. This process focuses on the similarity of multi-dimensional feature combinations to adapt to complex disaster situations.

[0120] The initial strategy parameters are derived as follows: Based on the predefined link selection rules in the matching template and combined with the real-time or predicted network status provided by S2, a recommended initial primary and backup communication link combination is generated. For example, in high-rise building fires, a private network or Mesh self-organizing network with strong signal penetration is prioritized as the primary link, and satellite communication is used as the backup link. According to the attention weight of different reconnaissance targets in the template and combined with the initial positions determined by the resource deployment suggestions in S1, an initial priority is assigned to each video acquisition device, and a display template for the initial layout of the specified video stream is generated. Based on the encoding control parameters preset for different scenarios in the template, Set the bitrate adjustment sensitivity parameters applicable to the current disaster situation to guide the video encoding strategy of S3; the dynamic digital task card generated by S1 is not a static file; throughout the entire task execution process, the S7 coordination hub continuously gathers event identification results from S4, monitoring and evaluation data from S9, and status and performance feedback from other steps; when this information indicates that there are significant changes in the disaster situation, task stage, or resource status, S7 can trigger S1 to perform a partial re-evaluation of the task model or re-match the strategy template, thereby dynamically updating the resource deployment suggestions and strategy parameters in the task card to achieve adaptive optimization of the task strategy;

[0121] High-rise building fire Mesh self-organizing network / Fire protection private network Satellite communication / 5G Drone reconnaissance footage (highest), soldier's perspective (medium), fixed surveillance (lowest) High (emphasis on dynamic target clarity) Chemical leak Explosion-proof Mesh Self-organizing Network 4G / Satellite Communication Robot reconnaissance view (highest), gas monitoring view (medium), surrounding area monitoring (lowest). Medium (focusing on the fidelity of smoke / gas characteristics) earthquake collapse 5G / Satellite Communication Emergency Broadband Private Network Life detection view (highest), structural monitoring view (medium), rubble monitoring (lowest) Extremely high (focusing on crack / deformation feature identification)

[0122] .

[0123] Step S2 specifically includes the following steps:

[0124] S21. Based on the initial primary and backup communication link combination recommended in the task card of step S1, establish a redundant communication link from the field image acquisition terminal to the rear command center.

[0125] S22. The acquisition terminal integrates multiple communication modules, including: fourth-generation mobile communication technology, fifth-generation mobile communication technology, satellite communication, mesh network self-organizing network and fire protection network;

[0126] S23. Establish a transmission control unit to monitor the operational quality of each link in real time. Specific indicators for real-time monitoring include signal strength, latency, and packet loss rate.

[0127] S24. The transmission control unit dynamically selects one of the available links as the primary link based on a preset strategy to carry the main data transmission.

[0128] S25. Implement a data splitting and parallel transmission mechanism to split the data stream and transmit data packets simultaneously through multiple links;

[0129] S26. When a link is detected to be interrupted, the traffic carried will be switched to other available links to ensure uninterrupted communication.

[0130] Based on the initial strategy parameters generated in the task card of step S1 (especially the recommended initial primary and backup communication link combination), a variety of heterogeneous communication modules, including fourth-generation mobile communication technology (4G), fifth-generation mobile communication technology (5G), satellite communication, mesh network self-organizing network, and fire protection network, are integrated at the field image acquisition terminal. The transmission control unit realizes real-time monitoring of the operating quality of each physical link (including indicators such as signal strength, transmission delay, and packet loss rate), and dynamically selects the primary link for core data transmission according to preset strategies (such as a greedy algorithm based on link quality scoring, or rules based on task stage and service priority). A data diversion and parallel transmission mechanism is implemented, which can split the uplink data stream (especially critical data) and transmit it synchronously through multiple links. It also has the ability to quickly switch over in case of failure. When the current primary link is detected to be interrupted or its quality is severely degraded, it can automatically and seamlessly switch the traffic it carries to other available high-quality links within a set time threshold (such as milliseconds).

[0131] Step S3 specifically includes the following steps:

[0132] S31. Before the video data transmitted in step S2 is sent out, the video stream is analyzed and processed, and different elements are extracted and separated based on information entropy.

[0133] S32. Decompose the video frame into its main components, which specifically include: static background layer, dynamic target layer and key semantic features;

[0134] S33. Combining the preset bitrate adjustment sensitivity parameters in the task card of step S1, adopt differentiated encoding and transmission strategies for different main components, specifically including the following steps:

[0135] S331. Structurally encode key semantic features and assign them high priority during transmission;

[0136] S332. A balanced compression strategy is adopted for the dynamic target layer;

[0137] S333. For static background layers, efficient compression or periodic updates are used instead of continuous full transmission.

[0138] S34. In order to cope with the network fluctuations of the link in step S2, when the network is congested, actively discard low information entropy data or instead transmit the abstracted and refined semantic graph.

[0139] S35. In the extreme weak network environment of the link in step S2, the video stream is converted into a dynamic graphic and text message with spatial coordinate information for transmission.

[0140] Before the video data transmitted in step S2 is sent, the original video stream undergoes real-time online analysis and processing. This processing utilizes computer vision algorithms and machine learning models to analyze the video frame sequence, specifically including: separating a relatively stable, slowly changing static background layer using background modeling algorithms; identifying and segmenting moving objects in the frame by performing motion detection and target tracking between consecutive frames, forming a dynamic target layer containing targets such as people, vehicles, flames, and smoke; performing semantic segmentation and target recognition on the video frames to extract predefined key semantic information, including flames, smoke, fallen people, safety exit signs, and building cracks, and organizing this information and its position, size, and confidence attributes in the frame into structured key semantic feature data; after completing the element extraction and separation, a differentiated encoding and transmission strategy is executed: the key semantic feature data is structured using a lightweight data format. The data is marked as having the highest priority during transmission and prioritized in the transmission queue to ensure that critical event information is delivered to the command center with minimal delay. For video data in the dynamic target layer, a balanced compression strategy is adopted. By using a video coding standard with adjustable quantization parameters, relatively fine quantization parameters are applied to the moving areas to control the bit rate while ensuring that moving objects are clearly distinguishable. For video data in the static background layer, an efficient compression strategy is adopted. A high-quality, fully coded background image is transmitted once at the beginning of the video sequence. Thereafter, it is transmitted periodically or when the background changes exceed a set threshold. In the frames between the two updates, only the background residual information is transmitted or the received background image is reused by the receiver, and the background area is encoded with a higher compression rate to avoid continuous full transmission of the static background.

[0141] Step S4 specifically includes the following steps:

[0142] S41. Upgrade the polling strategy from the traditional time-slice rotation to an event- and rule-driven model;

[0143] S42. Continuously analyze the video stream transmitted in step S3 in real time to identify preset key event types, including flame recognition, smoke detection, people falling to the ground, and abnormal building structure.

[0144] S43. When the above-mentioned abnormal event is detected in any video, the video source is automatically marked as a high-priority event source.

[0145] S44. Perform two core actions simultaneously, specifically including:

[0146] Automatically switch the main display screen of the command center to the high-priority event video of this channel;

[0147] Send an audible and visual alarm to the command center to alert the personnel in charge.

[0148] The video polling strategy is upgraded from the traditional time-slice rotation mode to an event- and rule-driven mode. Events refer to abnormal situations identified through real-time processing of video streams by a computer vision analysis module, while rules refer to pre-defined handling logic for different event types and levels. The specific implementation process is as follows: A trained deep learning model is deployed in the system backend. This model continuously analyzes each received video stream at the frame level and compares it with a pre-defined abnormal feature library. When the model identifies an abnormal event in any video stream that matches pre-defined characteristics, such as flames, smoke, a person falling, or a building structure cracking, the event is triggered. The system immediately identifies the video source to which the abnormal event belongs as a high-priority event source based on a pre-configured rule engine. The criteria for this rule engine include, but are not limited to: the inherent priority of the event type, the spatial location of the event in the frame, the duration of the event, the rate of change of the event's area or intensity, and other related sensor alarm information. This marking process is automatic and real-time, requiring no manual intervention, thus ensuring that key situational information can be captured instantly and its processing and display priority in the entire system is improved.

[0149] Step S5 specifically includes the following steps:

[0150] S51. Manage the video stream and dynamically assign a comprehensive priority to each video source;

[0151] S52. The calculation of the overall priority is determined by multiple factors, including the fixed weight set by the task card in step S1, the dynamic events identified in step S4, and the manual specification by the operator.

[0152] S53. Based on the calculated video priority and the current available total network bandwidth of the link in step S2, decide on the allocation of transmission and display resources;

[0153] S54. Execute the hierarchical resource allocation strategy:

[0154] S541. For the highest priority video stream, allocate the dedicated transmission channel of step S2 and display it continuously and exclusively in the command center.

[0155] S542. For video streams with medium priority, display them using split-screen sharing or timed rotation methods while ensuring basic watchability.

[0156] S543. For low-priority or non-critical video streams, place them in the background polling queue and manually retrieve and view them as needed, or only upload periodic snapshots.

[0157] Based on the calculated dynamic comprehensive priority of each video source and combined with the total available network bandwidth of the communication link monitored in step S2, a decision is made regarding the allocation of transmission and display resources. This decision-making process uses a preset bandwidth allocation strategy, such as a proportional fairness algorithm or a priority-based weighted bandwidth allocation algorithm, to dynamically divide the total bandwidth into video streams of different priorities. Specifically, the hierarchical resource allocation strategy is implemented as follows: For video streams marked as having the highest priority, the system allocates a dedicated logical or physical transmission channel for them in the communication link in step S2, ensuring that they have reserved bandwidth to meet their bitrate requirements, and simultaneously provides continuous, uninterrupted, exclusive full-screen display on the main display screen of the command center or a designated main display area; for medium-priority video streams, while ensuring that their bitrate is not lower than the threshold for maintaining basic viewability, Under the premise of [specific conditions], the system uses split-screen sharing or timed rotation for display. In split-screen mode, multiple medium-priority video streams are displayed synchronously in a grid pattern in the non-core areas of the secondary or main screen. In timed rotation mode, the system automatically cycles through and displays each medium-priority video stream in a non-exclusive window of the main display area at preset time intervals (e.g., every 5 seconds). For low-priority or non-key monitoring video streams, no dedicated display resources are allocated; instead, they are placed in a background rotation queue. The video streams in this queue are not actively pushed to the main display interface of the command center, but operators can manually select and retrieve them as needed through the console. At the same time, to save bandwidth, the system can perform periodic snapshot capture only on these video streams, extracting a key image frame every 30 seconds for compression and uploading, rather than transmitting continuous real-time video streams.

[0158] Step S6 specifically includes the following steps:

[0159] S61. Compare and analyze all received video frames with the database storing historical disaster scene feature data;

[0160] S62. Extract the scene features, situational information, and key structural features identified in the current video stream in step S4;

[0161] S63. The extracted scene features, situational information, and key structural features are matched with cases in the historical database, and the similarity is calculated.

[0162] S64. When the similarity calculated by matching exceeds the preset threshold, automatically associate and push detailed information of the high-similarity historical case, including process records, key decision points, handling points and final result summary.

[0163] S65. After structuring the complete data stream of the current task, including the three-dimensional situation evolution process generated in step S8, the performance data recorded in S9, and all key events in steps S4 and S5, the data is synchronously stored in the historical disaster database to enrich the case library.

[0164] All received video frames are compared and analyzed in real time or periodically with a database storing historical disaster scene feature data. Multidimensional key information is extracted from the current video stream identified and labeled in step S4. This information includes: scene features, such as flame color distribution, smoke concentration gradient, overall building outline and material; situational information, such as fire spread direction and speed, personnel gathering and evacuation dynamics, and visible changes in key structures (such as load-bearing walls and ventilation ducts); and structural features, such as crack morphology and development, local deformation, or signs of collapse. Subsequently, the extracted current multidimensional feature vector is matched with the corresponding feature vector for each case in the historical database. The matching process uses a feature similarity calculation algorithm, based on Euclidean distance, cosine similarity, or more complex algorithms. The multimodal feature fusion matching model calculates a comprehensive similarity score between the current task and each historical case. When the similarity score of one or more historical cases exceeds the system's preset threshold, the system automatically performs an association operation: retrieving the complete digital archives of these highly similar cases from the historical database, and extracting and pushing key auxiliary decision-making information in a structured manner. The pushed detailed information includes: a complete timeline record of the historical case, key decision nodes marked during the handling process and their context, key points and tactical measures for on-site handling that have been summarized and verified, and a summary of the final handling result and effectiveness evaluation of the case. This process realizes the automated association and intelligent push of the current real-time situation and historical experience knowledge.

[0165] Step S7 specifically includes the following steps:

[0166] S71, responsible for information exchange, event response and strategy coordination between all steps from S1 to S6;

[0167] S72. Implement a unified event publishing and subscription mechanism, specifically including:

[0168] S721. The events identified in step S4, the link quality anomalies monitored in step S9, the status changes in steps S2 and S3, and the commander's operation instructions in steps S5 and S6 are all published as events to the unified event bus.

[0169] S722, Steps S1 to S6 and other steps: Subscribe to the types of events that need to be responded to on the bus according to their own responsibilities;

[0170] S73. Execution Strategy Conflict Arbitration and Coordinated Response:

[0171] S731. When multiple concurrent events in steps S4, S5, and S6 trigger strategies for different steps and cause conflicts, arbitration is initiated.

[0172] S732. Arbitration is conducted based on the core objectives, preset rules, and current task stage set in step S1 of the task card.

[0173] S733. Finally, a unified collaborative decision is output and distributed to the relevant steps for execution.

[0174] S74. Drives adaptive switching of task phases, identifies key phase transition nodes in task progress by analyzing event flow;

[0175] S75. When it is determined that a phase transition is required, the corresponding phase coordination strategy package is called from the preset strategy library or optimized and generated by step S10.

[0176] S76. Use a new strategy package to refresh the strategy parameters of multiple related steps, including steps S2, S3, and S5, to adapt to the needs of the new stage.

[0177] A unified event bus mechanism based on a publish / subscribe pattern is implemented. This mechanism standardizes all internal state changes, identification results, and operation instructions generated in all other steps into event messages of a specific format. Abnormal events identified in step S4 (such as flames and smoke), communication link quality degradation or interruption alarms detected in step S9, changes in self-operating status reported in steps S2 and S3 (such as primary link switching and coding strategy adjustment), and manual operation instructions executed by the commander on the interfaces of steps S5 (resource scheduling) and S6 (historical case retrieval) are all encapsulated as event objects with timestamps, event sources, event types, and load data, and published to the unified event bus. Simultaneously, each functional module in steps S1 to S6 actively subscribes to specific types of events it needs to monitor on the event bus according to its predefined responsibilities and response logic, thus forming a loosely coupled, event-driven interactive architecture.

[0178] When steps S4, S5, and S6 trigger their respective independent handling strategies due to concurrent events (S4 simultaneously identifies multiple fire points requiring video retrieval; S5 is executing high-priority video transmission assurance; S6 pushes historical case suggestions), and these strategies conflict in resource allocation (such as bandwidth, display area) or actions, the arbitration engine for this step is automatically triggered. The arbitration engine, based on the core handling objectives set in the dynamic digital task card generated in step S1, the preset global conflict resolution rules (such as life rescue taking precedence over property protection, and key location monitoring taking precedence over general area inspection), and the current task stage determined by event flow analysis (such as reconnaissance stage, assault stage, evacuation stage), performs weighted evaluation and priority ranking of the conflicting strategies. The arbitration engine outputs a unified, conflict-free collaborative decision-making instruction set and distributes it to relevant steps such as S2, S3, and S5 for execution. It decides which fire point video stream identified by S4 should be prioritized when bandwidth is insufficient, driving the adaptive switching mechanism of the task stage. This step continuously analyzes the event sequence flowing through the event bus; through pre-set stage transition rules or machine learning models; identifies key nodes that mark fundamental changes in the task process; when specific event combinations such as continuous open flame extinguishing, discovery of multiple injured persons on site, and large-scale abnormal noises in the building structure occur in the event stream; it can be determined that the task has moved from the fire fighting stage to the personnel search and rescue or structural hazard assessment stage; when it is determined that a stage transition is required; this step will call the pre-configured strategy template for the new stage from the pre-set strategy library; or request step S10 to optimize and generate a new stage collaborative strategy package based on the current situation and historical data; this step will coordinate and refresh the strategy parameters of relevant steps: update the recommended communication link combination for this stage for step S2; update the video encoding parameters suitable for the new stage (such as the search and rescue stage where personnel morphology clarity is required) for step S3; update the priority calculation weight and display rules of the video source for step S5; thereby enabling the behavior of the entire system to adaptively switch to a collaborative working mode that matches the new stage.

[0179] Step S8 specifically includes the following steps:

[0180] S81. Utilize the video stream and data transmitted back from step S2, which contain precise timestamps and spatial stamps;

[0181] S82. The real-time data in step S81 is fused with the on-site architectural drawings or geographic information system data loaded in the task card of step S1.

[0182] S83. By estimating the pose, the viewpoint of each video stream is bound to specific coordinates and directions in three-dimensional space.

[0183] S84. Generate an interactive video-enhanced 3D situation map, in which each video source is visualized on the 3D model in the form of a virtual camera.

[0184] S85. Commanders can directly access and view the real-time video stream of any virtual camera on this 3D situation map by clicking on it.

[0185] The video stream and its metadata, transmitted from step S2 and accompanied by precise time and space stamps (including GPS / BeiDou coordinates, altitude, orientation, and time), are fused with the 2D / 3D geospatial data from the on-site Building Information Model (BIM), digital orthophoto, or Geographic Information System (GIS) loaded in the task card of step S1. Through visual synchronous positioning and map building or image feature matching and pose estimation algorithms, the shooting viewpoint (i.e., the position of the camera's optical center and the orientation of its optical axis) of each video frame is calculated and bound to specific 3D coordinates and orientations in a unified world coordinate system or geographic coordinate system. Based on this precise spatial binding relationship... The system generates an interactive video-enhanced 3D situation map in real time. In this map, each video source is superimposed and visualized on the corresponding spatial location of the 3D building model or terrain scene in the form of a virtual camera icon (or view frustum) that matches its actual pose. On this 3D situation map, the commander can select any virtual camera icon by clicking the mouse or using touch. The system will then automatically retrieve and play the real-time video stream corresponding to that camera from the link established in step S2 and display it in a designated area of ​​the interface (such as picture-in-picture, floating window, or main display area), achieving a seamless connection from 3D spatial situation awareness to viewing the details of specific video sources.

[0186] Step S9 specifically includes the following steps:

[0187] S91. For the communication links in step S2, display the real-time performance indicators of each link. The real-time performance indicators specifically include bandwidth utilization, latency, packet loss rate and historical switching records. Give the transmission reliability rating of each link in the form of a comprehensive score, and issue an alarm for the links that are about to deteriorate.

[0188] S92. For the video stream output from step S3, monitor the video quality and information density, display the real-time bitrate, and evaluate the key information density through an algorithm to form a quality change curve and density map.

[0189] S93. Based on the event identification log generated in step S4 and the resource scheduling record generated in step S5, generate and display the intelligent patrol efficiency heat map to reflect the distribution of the command center’s attention to each video source.

[0190] For each communication link in step S2, real-time performance indicators such as bandwidth utilization, end-to-end latency, and packet loss rate are dynamically displayed on a unified monitoring panel, and historical master / slave switchover records are listed. Based on these indicators, a dynamically updated comprehensive transmission reliability score is calculated and displayed for each link using a preset scoring model (which weights latency and packet loss rate indicators). Simultaneously, the monitoring logic continuously analyzes the trends of each indicator. When a signal strength of a link continuously decreases or latency jitter increases, indicating performance degradation, a warning alert is automatically issued to the operator on the panel in the form of a highlighted color, a flashing icon, or a pop-up message box. For the video stream processed and transmitted in step S3, its real-time transmission bitrate is displayed in another monitoring view. Simultaneously, by running background analysis algorithms, the decoded video stream content is evaluated, and the quantity and rate of change of key semantic information (such as event targets identified in S4) contained per unit time are calculated to quantify information density. The temporal changes of bitrate and information density are plotted as a coaxial curve to form a video quality and information density map, which intuitively reflects the relationship between the content value and transmission efficiency of the video stream. Based on the event identification log continuously output in step S4 (recording when and which video triggered what event) and the resource scheduling record generated in step S5 (recording the priority and display method of each video stream), the system background performs correlation analysis to count the frequency and duration of each video source being marked as an event source, switched to the main screen display, or assigned high priority within a certain period of time. This data is mapped onto the video source list or a two-dimensional monitoring point plan and visualized in the form of a heat map. The color depth represents the level of attention, thereby generating and displaying an intelligent patrol efficiency heat map, which intuitively reveals the distribution of visual focus of the command center and the actual effect of the patrol strategy.

[0191] Specifically, step S10 includes the following steps:

[0192] S101. Based on the monitoring data collected in step S9, optimize and analyze the overall system strategy.

[0193] S102. Perform dynamic optimization of strategy parameters based on feedback: Based on the performance data provided in step S9 and the intervention records of the commander in steps S5 and S6, automatically or assistedly optimize the strategy parameters of the preceding related steps in steps S2, S3 and S5.

[0194] The system performs dynamic optimization of strategy parameters based on feedback. Based on the quantified performance data provided in step S9, the deviation between the actual transmission reliability of each link and the target value, the statistical distribution of end-to-end latency of different priority video streams, and the information density and bitrate relationship curve of the video stream in step S3, combined with the commander's manual operations and confirmation records in steps S5 (resource scheduling) and S6 (historical case review), the system automatically analyzes the matching degree between the current strategy parameters (such as the link switching threshold in S2, the hierarchical coding bitrate allocation ratio in S3, and the priority calculation weight in S5) and the expected performance target. Based on the analysis results, the system automatically adjusts these parameters. If the data in S9 shows that the latency of a high-priority video stream exceeds the standard, its bandwidth weight in S5 may be automatically increased, or S2 may be adjusted to select a lower-latency backup link. If the commander frequently manually increases the video priority of a certain type of event, the system can learn this pattern and automatically increase the weight coefficient of the corresponding event type in S5.

[0195] S103. Based on the video-enhanced 3D situation map generated in step S8 and the full and synchronous task data recorded in step S9, an interactive task process simulation copy is generated after the task is completed.

[0196] S104. In this copy, reset the initial strategy of step S1 and deduce the changes in the transmission delay, interruption time and key event discovery time index of the key screen under the new strategy.

[0197] S105. In the simulation copy, simulate the adjustment of the preset deployment position of the video source in step S1, and simulate the impact of this adjustment on the event detection probability in step S3 and the three-dimensional situation coverage integrity in step S8.

[0198] S106. Implement intelligent matching and strategy recommendation for historical cases:

[0199] S1061. Call the historical database of step S6 and perform a deep feature comparison between the process simulation copy generated by the current task and the most similar historical case in the database.

[0200] S1062. Through deep feature comparison, the key differences and risk evolution trends between the current task and historical cases are automatically identified.

[0201] S1063. Based on the lessons learned from historical cases, generate targeted strategy optimization suggestions for reviewing current tasks and planning future tasks;

[0202] Based on the video-enhanced 3D situation evolution record with precise spatiotemporal markers generated in step S8, and the full system operation data recorded in step S9 that is strictly synchronized with the situation changes, an interactive and controllable task process simulation copy is generated offline after the task is completed. Analysts can reset the initial strategy parameters of step S1. The system will re-simulate and simulate the task process based on the boundary conditions such as real-time network conditions and event sequences recorded in the copy, and quantify the changes in the end-to-end transmission delay of key monitoring screens, cumulative duration of communication interruption, and first discovery time of key events identified in step S4 under the new strategy assumptions. This can be used to compare and evaluate the potential effects of different initial strategies.

[0203] In the same simulation, the initial deployment sites preset for UAV and robot video sources in step S1 can be adjusted. The system will combine a three-dimensional geographic environment model to simulate the video coverage and occlusion under the new sites, and based on the characteristics of the event recognition model in step S3, it will deduce the impact of site adjustments on the overall detection probability of various preset events (such as flames and smoke). The system will also evaluate the impact of the new site layout on the model coverage integrity and key angle missingness of the three-dimensional situation map to be constructed in step S8, providing a quantitative basis for optimizing the reconnaissance deployment plan.

[0204] Example 3:

[0205] After system startup, it automatically accesses data from the 119 emergency call center, weather (northeast wind level 3), on-site personnel (3 fire trucks, 1 drone), the building's fire emergency plan, and Geographic Information System (GIS) data. Through fusion analysis, it quickly identifies the fire as a high-rise building fire, determining core risks including rapid spread of smoke and fire, people trapped, and glass curtain wall shattering. It also identifies the floor where the fire originated and the evacuation routes as critical areas. Based on this analysis, the system automatically generates a dynamic digital task card, recommending that a drone be deployed to hover over floors 15-18 on the east side of the building for reconnaissance. Initial strategy parameters include prioritizing the use of fire mesh. The self-organizing network serves as the primary communication link, with the 5G public network as a backup; drone reconnaissance footage is set as the highest video priority; and a bitrate adjustment sensitivity is preset for dynamic targets (such as people and flames) with high clarity requirements; the drones on site integrate Mesh and 5G communication modules; the transmission control unit, based on the task card recommendation, first establishes the Mesh self-organizing network link and monitors its quality in real time; when the drone approaches a building, causing the Mesh signal to be blocked and degraded, the system automatically and seamlessly switches the primary link to the 5G network according to a preset algorithm to ensure uninterrupted video transmission; before the video stream is sent out, the system performs real-time monitoring. Intelligent analysis separates the static building exterior background from the dynamic elements of dense smoke, flames at windows, and people waving from balconies, extracting key semantic features and their coordinates for flames, smoke, and people. These key features are then structured and assigned the highest transmission priority. Equal compression is applied to dynamic smoke and fire areas to maintain discernible shapes, while efficient compression is used for the static background, significantly saving bandwidth. The system continuously analyzes drone footage; when a human event is detected, the video stream is immediately marked as a high-priority event source, automatically switching the command center's main screen to it and triggering audible and visual alarms. Combined with this event, the drone video... The overall priority of the stream is dynamically calculated to be the highest, thus obtaining a dedicated transmission channel and continuous exclusive display on the command screen, while other videos are rotated on split screens or placed in the background; the system also compares the current fire characteristics with the historical case database, and when a similar high-rise building fire case is matched, it automatically pushes historical handling points such as paying attention to the risks of glass curtain walls and avoiding evacuation downwind; all data of this task is synchronously stored in the historical database; as the coordination hub, the system uses an event bus mechanism to arbitrate based on the core objectives of life rescue when resource scheduling requests conflict with bandwidth constraints, prioritizing the protection of personnel-related video streams.As the event stream showed that the open flames were suppressed and injured personnel needed to be transferred, the system determined that the mission had entered a new phase of personnel search and rescue and transfer. It automatically invoked a new strategy package, adjusted video encoding parameters to emphasize personnel clarity, and updated display rules. The system integrated UAV pose data with the building information model (BIM) of the building to generate a 3D enhanced situation map. Commanders could click on the virtual camera icon on the map to directly view real-time footage and intuitively grasp the spatial situation. The monitoring interface evaluated the performance of each link and video information density in real time and generated a heat map of attention. After the mission was completed, the system found through deductive analysis that triggering link switching earlier could reduce lag. Based on historical case comparisons, it generated optimization suggestions, such as pre-setting satellite communication as a more aggressive backup link activation condition for future similar high-rise fire missions.

[0206] Example 4:

[0207] The system integrates leak alarms, meteorological (calm wind) data, underground pipeline geographic information, and hazardous chemical disposal plans. Analysis identifies a chemical tunnel leak, with risks including toxic gas diffusion, flammability and explosiveness, and low visibility. The key areas are the leak point and connected ventilation ducts. Based on this, a dynamic digital task card is generated, recommending the deployment of an explosion-proof robot along the tunnel's centerline for reconnaissance. Initial strategy parameters include: primary link is an explosion-proof Mesh self-organizing network, backup is 4G; robot video is set to the highest priority; and bitrate sensitivity parameters are preset to enhance image quality and fidelity in low-light and smoky environments.

[0208] The on-site explosion-proof robot integrates Mesh and 4G modules. Inside the tunnel, as the robot delves deeper, the Mesh signal gradually weakens. The transmission control unit implements data diversion, transmitting critical sensor data and low-bitrate video via the Mesh link, while simultaneously transmitting high-definition video streams in parallel via the 4G link to optimize overall communication efficiency. Before transmission, the robot's video is intelligently separated into a static tunnel wall background, dynamic leak mist, and illuminated areas, extracting key semantic features such as gas leaks and pipeline valves. When the network is unstable inside the tunnel, the system prioritizes the transmission of key features and dynamic mist patterns; in extreme cases, it switches to transmitting graphic and textual reports with coordinates. The system continuously analyzes the robot's footage; once smoke (toxic gas) or pipeline anomalies are detected, the video is immediately marked as high priority and switched to the command center's main screen, triggering a leak alarm. Robot video, as the core reconnaissance source, is given the highest priority and exclusively has access to transmission and display resources. When the system compares with the historical case database and matches similar leak cases, it automatically pushes key points for handling, such as closing specific upstream valves, detecting downwind concentration, and requiring personnel to wear positive pressure breathing apparatus. The coordination hub coordinates operations through an event bus. When a leak event is identified and a weak Mesh link signal is detected, a switch to 4G as the primary link is triggered. When the robot approaches the core leak area and the video becomes difficult to identify due to a surge in smoke concentration, the hub arbitrates and instructs to adjust the robot's movement speed, triggering a reassessment of the task model and updating it to a refined reconnaissance strategy. Simultaneously, the video encoding parameters are adjusted to maximize the detail of the core area. The system also integrates the robot's pose with the underground pipeline network BIM model to generate a 3D enhanced situation map, clearly showing the spatial layout of the leak point and the relationship between valves. Monitoring and assessment showed that while the video information density was high in the core area of ​​the leak, the transmission quality fluctuated, and attention was entirely focused on the robot's video. After the mission, simulation revealed that adjusting the robot's initial deployment position to be closer to the suspected leak entrance could significantly improve the probability of incident detection. Based on in-depth comparison of historical cases, the system generated strategy optimization suggestions, such as adding a multi-point robot collaborative reconnaissance deployment option to the task card strategy template for similar future missions.

[0209] Video priority allocated bandwidth percentage Average transmission delay (ms) Packet loss rate (%) Key information delivery rate (%) Typical application scenarios High priority 60%~70% ≤150 ≤0.5 ≥99.5 Footage of key events such as trapped personnel and leak points Medium priority 20%~25% 300~500 ≤2.0 ≥95.0 Images showing the spread of fire and structural changes. low priority 5%~10% ≥1000 ≤5.0 ≥90.0 Background environment, surveillance footage of non-key areas

[0210] .

[0211] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

[0212] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the disclosure herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope of this disclosure is indicated by the following claims.

Claims

1. A method for remote transmission and rotation of fire scene reconnaissance data, characterized in that, Includes the following steps: S1. Automatically identify disaster situations through multi-source data analysis and generate dynamic digital task cards containing resource deployment suggestions and initial strategy parameters; S2. Integrate multiple communication modules and intelligent transmission control to establish an adaptive redundant communication guarantee link; S3. Intelligent layered encoding and transmission optimization of video streams, adaptive transmission under different network conditions; S4. Upgrade the polling strategy to event and rule-driven, automatically identify abnormal events and trigger switching and alarms; S5. Based on the events identified in step S4, and according to the dynamic comprehensive priority, allocate transmission and display resources in a hierarchical manner, and perform differentiated scheduling of the video stream. S6. Compare the video identified in step S4 with the historical database, associate and push historical cases, and simultaneously input the current task data into the database. S7. Through a unified event bus and arbitration mechanism, information exchange, event response and strategy coordination are carried out for all steps from S1 to S6. S8. Based on the video transmitted in step S2 and the video information optimized in step S3, integrate geographic information to generate an enhanced situation map that visualizes the location of the video source in three-dimensional space. S9. Conduct comprehensive monitoring and visual evaluation of the communication quality of step S2, the video information density of step S3, and the polling efficiency of steps S4 and S5. S10. Based on the monitoring data from step S9 and the historical cases from step S6, the strategy is optimized in a closed loop through dynamic parameter tuning, simulation, and case matching.

2. The method for remote transmission and rotation of fire reconnaissance image data according to claim 1, characterized in that, Step S1 specifically includes the following steps: S11. Automatically access and integrate multi-source heterogeneous data, including alarm information, disaster type, meteorological information, geographic information, emergency plan database, and reports from on-site forces; S12. Based on the results of the fusion analysis, identify the specific types of disasters, including high-rise building fires, chemical leaks, and earthquake collapses, and determine the relevant risk factors and key areas; S13. Based on the preliminary judgment results, automatically construct a dynamic digital task card for this task, wherein the dynamic digital task card contains the basic information of the disaster that has been judged; S14. When constructing the task card, two core pieces of information are automatically derived. These two core pieces of information specifically include: Recommend initial deployment sites for mobile reconnaissance equipment such as drones and robots; By matching the task model with a pre-built strategy template library, initial strategy parameters are automatically generated. These automatically generated initial strategy parameters specifically include: Recommended initial primary and backup communication link combination; Generate an initial video source priority list and display template; Preset bitrate adjustment sensitivity parameters suitable for the current disaster scenario.

3. The method for remote transmission and rotation of fire reconnaissance image data according to claim 2, characterized in that, Step S2 specifically includes the following steps: S21. Based on the initial primary and backup communication link combination recommended in the task card of step S1, establish a redundant communication link from the field image acquisition terminal to the rear command center. S22. The acquisition terminal integrates multiple communication modules, including: fourth-generation mobile communication technology, fifth-generation mobile communication technology, satellite communication, mesh network self-organizing network and fire protection network; S23. Establish a transmission control unit to monitor the operating quality of each link in real time. The specific indicators of real-time monitoring include signal strength, delay, and packet loss rate. S24. The transmission control unit dynamically selects one of the available links as the primary link based on a preset strategy to carry the main data transmission. S25. Implement a data splitting and parallel transmission mechanism to split the data stream and transmit data packets simultaneously through multiple links; S26. When a link is detected to be interrupted, the traffic carried will be switched to other available links to ensure uninterrupted communication.

4. The method for remote transmission and rotation of fire reconnaissance image data according to claim 3, characterized in that, Step S3 specifically includes the following steps: S31. Before the video data transmitted in step S2 is sent out, the video stream is analyzed and processed, and different elements are extracted and separated based on information entropy. S32. Decompose the video frame into main components, the main components specifically including: static background layer, dynamic target layer and key semantic features; S33. Combining the preset bitrate adjustment sensitivity parameters in the task card of step S1, adopt differentiated encoding and transmission strategies for different main components, specifically including the following steps: S331. Structurally encode key semantic features and assign them high priority during transmission; S332. A balanced compression strategy is adopted for the dynamic target layer; S333. For static background layers, efficient compression or periodic updates are used instead of continuous full transmission. S34. In order to cope with the network fluctuations of the link in step S2, when the network is congested, actively discard low information entropy data or instead transmit the abstracted and refined semantic graph. S35. In the extreme weak network environment of the link in step S2, the video stream is converted into a dynamic graphic newsletter with spatial coordinate information for transmission.

5. The method for remote transmission and rotation of fire reconnaissance image data according to claim 4, characterized in that, Step S4 specifically includes the following steps: S41. Upgrade the polling strategy from the traditional time-slice rotation to an event- and rule-driven model; S42. Continuously analyze the video stream transmitted in step S3 in real time to identify preset key event types, including flame recognition, smoke detection, people falling to the ground, and abnormal building structure. S43. When the above-mentioned abnormal event is detected in any video, the video source is automatically marked as a high-priority event source. S44. Perform two core actions simultaneously, specifically including: Automatically switch the main display screen of the command center to the high-priority event video of this channel; An audible and visual alarm was sent to the command center to alert the personnel in charge.

6. The method for remote transmission and rotation of fire reconnaissance image data according to claim 5, characterized in that, Step S5 specifically includes the following steps: S51. Manage the video stream and dynamically assign a comprehensive priority to each video source; S52. The calculation of the overall priority is determined by multiple factors, including the fixed weight set by the task card in step S1, the dynamic events identified in step S4, and the manual specification by the operator. S53. Based on the calculated video priority and the current available total network bandwidth of the link in step S2, decide on the allocation of transmission and display resources; S54. Execute the hierarchical resource allocation strategy: S541. For the highest priority video stream, allocate the dedicated transmission channel of step S2 and display it continuously and exclusively in the command center. S542. For video streams with medium priority, display them using split-screen sharing or timed rotation methods while ensuring basic watchability. S543. For low-priority or non-critical video streams, place them in a background polling queue and manually retrieve and view them as needed, or only upload periodic snapshots.

7. The method for remote transmission and rotation of fire reconnaissance image data according to claim 6, characterized in that, Step S6 specifically includes the following steps: S61. Compare and analyze all received video frames with the database storing historical disaster scene feature data; S62. Extract the scene features, situational information, and key structural features identified in the current video stream in step S4; S63. The extracted scene features, situational information, and key structural features are matched with cases in the historical database, and the similarity is calculated. S64. When the similarity calculated by matching exceeds the preset threshold, automatically associate and push detailed information of the high-similarity historical case, including process records, key decision points, handling points and final result summary. S65. After structuring the complete data stream of the current task, including the three-dimensional situation evolution process generated in step S8, the performance data recorded in S9, and all key events in steps S4 and S5, the data is synchronously stored in the historical disaster database to enrich the case library.

8. The method for remote transmission and rotation of fire reconnaissance image data according to claim 7, characterized in that, Step S7 specifically includes the following steps: S71, responsible for information exchange, event response and strategy coordination between all steps from S1 to S6; S72. Implement a unified event publishing and subscription mechanism, specifically including: S721. The events identified in step S4, the link quality anomalies monitored in step S9, the status changes in steps S2 and S3, and the commander's operation instructions in steps S5 and S6 are all published as events to the unified event bus. S722, Steps S1 to S6 and other steps: Subscribe to the types of events that need to be responded to on the bus according to their own responsibilities; S73. Execution Strategy Conflict Arbitration and Coordinated Response: S731. When multiple concurrent events in steps S4, S5, and S6 trigger strategies for different steps and cause conflicts, arbitration is initiated. S732. Arbitration is conducted based on the core objectives, preset rules, and current task stage set in step S1 of the task card. S733. Finally, a unified collaborative decision is output and distributed to the relevant steps for execution. S74. Drives adaptive switching of task phases, identifies key phase transition nodes in task progress by analyzing event flow; S75. When it is determined that a phase transition is required, the corresponding phase coordination strategy package is called from the preset strategy library or optimized and generated by step S10. S76. Use a new strategy package to refresh the strategy parameters of multiple related steps, including steps S2, S3, and S5, to adapt to the needs of the new stage.

9. A method for remote transmission and rotation of fire reconnaissance image data according to claim 8, characterized in that, Step S8 specifically includes the following steps: S81. Utilize the video stream and data transmitted back from step S2, which contain precise timestamps and spatial stamps; S82. The real-time data in step S81 is fused with the on-site architectural drawings or geographic information system data loaded in the task card of step S1. S83. By estimating the pose, the viewpoint of each video stream is bound to specific coordinates and directions in three-dimensional space. S84. Generate an interactive video-enhanced 3D situation map, in which each video source is visualized on the 3D model in the form of a virtual camera. S85. Commanders can directly access and view the real-time video stream of any virtual camera on this 3D situation map by clicking on it. Step S9 specifically includes the following steps: S91. For the communication links in step S2, display the real-time performance indicators of each link. The real-time performance indicators specifically include bandwidth utilization, latency, packet loss rate and historical switching records. Give the transmission reliability rating of each link in the form of a comprehensive score, and issue an alarm for the links that are about to deteriorate. S92. For the video stream output from step S3, monitor the video quality and information density, display the real-time bitrate, and evaluate the key information density through an algorithm to form a quality change curve and density map. S93. Based on the event recognition log generated in step S4 and the resource scheduling record generated in step S5, generate and display an intelligent patrol efficiency heat map to reflect the distribution of the command center's attention to each video source.

10. A method for remote transmission and rotation of fire reconnaissance image data according to claim 9, characterized in that, Step S10 specifically includes the following steps: S101. Based on the monitoring data collected in step S9, optimize and analyze the overall system strategy. S102. Perform dynamic optimization of strategy parameters based on feedback: Based on the performance data provided in step S9 and the intervention records of the commander in steps S5 and S6, automatically or assistedly optimize the strategy parameters of the preceding related steps in steps S2, S3 and S5. S103. Based on the video-enhanced 3D situation map generated in step S8 and the full and synchronous task data recorded in step S9, an interactive task process simulation copy is generated after the task is completed. S104. In this copy, reset the initial strategy of step S1 and deduce the changes in the transmission delay, interruption time and key event discovery time index of the key screen under the new strategy. S105. In the simulation copy, simulate the adjustment of the preset deployment position of the video source in step S1, and simulate the impact of this adjustment on the event detection probability in step S3 and the three-dimensional situation coverage integrity in step S8. S106. Implement intelligent matching and strategy recommendation for historical cases: S1061. Call the historical database of step S6 and perform a deep feature comparison between the process simulation copy generated by the current task and the most similar historical case in the database. S1062. Through deep feature comparison, the key differences and risk evolution trends between the current task and historical cases are automatically identified. S1063. Based on the lessons learned from historical cases, generate targeted strategy optimization suggestions for reviewing current tasks and planning future tasks.

Images