A smart emergency management method, device and medium

Abstract

The present application relates to the technical field of emergency management. Specifically relates to a kind of wisdom emergency management method, device and medium, its method includes: obtaining emergency alarm information, monitoring center is based on emergency dispatch time datum synchronous acquisition monitoring information and on-site audio and video information corresponding to emergency alarm information, and monitoring information and audio and video information are fused and calculated, generate event credibility weight, when event credibility weight exceeds preset credibility weight threshold, group scheduling instruction is issued to emergency communication terminal and emergency monitoring terminal, in the process of group scheduling instruction issuing, when detecting that communication network delay exceeds preset network delay threshold, edge computing node is directly executed local broadcast and early warning instruction based on emergency dispatch time datum and synchronization deviation information, and the execution state of local broadcast and early warning instruction is fed back to cloud management platform.The present application has the effect of improving the timeliness of emergency management.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of emergency management, and in particular to a smart emergency management method, device and medium. BACKGROUND

[0002] In the prior art, to solve the problem of emergency management, the conventional means include: one is to transmit information and dispatch command through traditional communication methods such as telephone and intercom, but this method has low information transmission efficiency, poor real-time performance, and is easily affected by communication signals and other factors; another is to use simple sensor monitoring equipment, which can only obtain limited monitoring data and cannot fully and accurately grasp the on-site situation, and the monitoring data and communication information cannot be effectively integrated; and the third is a network access method based on different protocols, the clocks of various terminals are difficult to unify, resulting in inconsistent data acquisition and dispatching time, affecting the collaboration of emergency management.

[0003] The defects of the prior art are that there is a lack of unified time reference, which leads to difficulty in synchronizing data of emergency communication terminals and emergency monitoring terminals, affecting the accuracy and timeliness of emergency dispatching; at the same time, it is difficult to accurately judge the credibility of the event because the monitoring information and on-site audio and video information cannot be effectively fused and calculated; when the communication network is delayed, there is a lack of effective local processing mechanism, resulting in untimely emergency response. SUMMARY

[0004] In order to improve the timeliness of emergency management, the present application provides a smart emergency management method, device and medium.

[0005] The above invention purpose of the present application is achieved by the following technical scheme:

[0006] A smart emergency management method, the smart emergency management method comprising:

[0007] Based on the SIP protocol, the emergency communication terminal and the emergency monitoring terminal are connected to the network, and in the connection process, the unified clock synchronization processing is performed, the emergency dispatching time reference is generated, and the synchronization deviation information of the emergency communication terminal and the emergency monitoring terminal is recorded;

[0008] The emergency alarm information is obtained, the monitoring center synchronously obtains the monitoring information and the on-site audio and video information corresponding to the emergency alarm information based on the emergency dispatching time reference, and performs fusion calculation on the monitoring information and the audio and video information to generate an event credibility weight; when the event credibility weight exceeds a preset credibility weight threshold, the monitoring center generates a grouping dispatching instruction based on the emergency dispatching time reference and the synchronization deviation information, determines the execution order of the emergency communication terminal and the emergency monitoring terminal in the grouping dispatching instruction according to the synchronization deviation information, and sends the grouping dispatching instruction to the emergency communication terminal and the emergency monitoring terminal;

[0009] In the process of issuing the grouping scheduling instruction, when it is detected that the communication network delay exceeds the preset network delay threshold, the edge computing node directly executes the local broadcast and early warning instruction based on the emergency scheduling time reference and the synchronization deviation information, and feeds back the execution state of the local broadcast and early warning instruction to the cloud management platform. By adopting the above technical scheme, unified clock synchronization processing can be performed based on the SIP protocol when the emergency communication terminal and the emergency monitoring terminal access the network, the emergency scheduling time reference is generated and the synchronization deviation information is recorded, thereby solving the problem that the terminal data is difficult to synchronize due to the lack of unified time reference in the prior art. The monitoring center can synchronously obtain the monitoring information and the on-site audio and video information corresponding to the emergency alarm information based on the emergency scheduling time reference, and perform fusion calculation to generate an event credibility weight, thereby realizing effective integration of multi-source data, avoiding the deficiency that the information is dispersed and it is difficult to accurately judge the event situation in the prior art, and generating a grouping scheduling instruction when the event credibility weight exceeds a preset threshold, and determining the terminal execution order based on the synchronization deviation information, thereby ensuring the consistency and collaboration of the scheduling time sequence. When the communication network delay exceeds the preset threshold, the edge computing node can directly execute the local broadcast and early warning instruction and feed back the execution state, thereby realizing fast response under the condition that the network is blocked, and solving the problem that the emergency response is not timely due to the lack of local processing mechanism in the prior art.

[0010] Preferably: obtaining emergency alarm information, the monitoring center synchronously obtains monitoring information and on-site audio and video information corresponding to the emergency alarm information based on the emergency scheduling time reference, and performs fusion calculation on the monitoring information and the audio and video information to generate an event credibility weight, comprising:

[0011] extracting a sensor trigger intensity parameter and a trigger duration parameter from the monitoring information, and performing normalization processing on the sensor trigger intensity parameter and the trigger duration parameter to obtain a first credibility score;

[0012] extracting a target event feature parameter from the on-site audio and video information, the target event feature parameter including an audio feature parameter and a video feature parameter, and performing feature matching on the audio feature parameter and the video feature parameter to obtain a second credibility score;

[0013] performing fusion calculation on the first credibility score and the second credibility score according to a preset weighting rule to obtain the event credibility weight.

[0014] By adopting the technical scheme, the sensor trigger intensity parameter and the trigger duration parameter can be extracted from the monitoring information and normalized to obtain a first credibility score, so as to ensure the comparability of the sensor data under different sampling conditions, the audio feature parameter and the video feature parameter can be extracted from the live audio and video information and matched to obtain a second credibility score, so as to realize multi-dimensional description of the live environment, and the first credibility score and the second credibility score are fused and calculated according to a preset weighting rule to obtain an event credibility weight, so that the sensor data and the audio and video information are effectively integrated, the problem that different source data cannot be cooperatively analyzed and event judgment is inaccurate in the prior art is solved, and the reliability and accuracy of event recognition are improved, thereby providing a unified and credible quantitative basis for subsequent emergency dispatching decision.

[0015] Preferably, the sensor trigger intensity parameter and the trigger duration parameter are extracted from the monitoring information, the sensor trigger intensity parameter and the trigger duration parameter are normalized to obtain a first credibility score, including:

[0016] The sensor trigger intensity parameter is mapped to a preset intensity interval to obtain an intensity normalization result;

[0017] The trigger duration parameter is mapped to a preset time interval to obtain a duration normalization result;

[0018] A dynamic weighting coefficient is calculated based on the intensity normalization result and the duration normalization result, in the dynamic weighting coefficient, the weight of the intensity normalization result increases with the increase of the duration normalization result;

[0019] The intensity normalization result and the duration normalization result are weighted and fused according to the dynamic weighting coefficient to obtain the first credibility score.

[0020] By adopting the technical scheme, the sensor trigger intensity parameter can be mapped to a preset intensity interval, and the trigger duration parameter can be mapped to a preset time interval, so that the original parameters of different dimensions have a unified comparison reference after normalization, a dynamic weighting coefficient can be calculated based on the intensity normalization result and the duration normalization result, and in the dynamic weighting coefficient, the weight of the intensity normalization result increases with the increase of the duration normalization result, so as to reflect the coupling relationship between the event occurrence intensity and the duration, the intensity normalization result and the duration normalization result are further weighted and fused according to the dynamic weighting coefficient to obtain the first credibility score, so as to combine the instantaneous trigger characteristics of the sensor with the time accumulation effect, solve the problem of single parameter judgment distortion or large deviation in the prior art, and improve the rationality and accuracy of the credibility calculation.

[0021] Preferably, target event feature parameters are extracted from the live audio and video information, the target event feature parameters including audio feature parameters and video feature parameters, the audio feature parameters and the video feature parameters are matched in feature, a second confidence score is obtained, including:

[0022] Audio energy fluctuation parameters and spectrum distribution parameters are extracted from the live audio, the audio energy fluctuation parameters and the spectrum distribution parameters are normalized to obtain audio feature vectors;

[0023] Target region motion trajectory parameters and pixel change intensity parameters are extracted from the live video, the target region motion trajectory parameters and the pixel change intensity parameters are normalized to obtain video feature vectors;

[0024] The audio feature vectors and the video feature vectors are calculated in correspondence in the same time segment to obtain a cross-modal correlation coefficient; in the calculation process of the cross-modal correlation coefficient, when the audio energy fluctuation parameters and the video pixel change intensity parameters change synchronously in the same time segment, the cross-modal correlation coefficient is modified in weight to obtain a modified cross-modal correlation coefficient; the modified cross-modal correlation coefficient is compared with a preset threshold to obtain a comparison result, the audio-video matching degree in the time segment is determined according to the comparison result, and a corresponding second confidence score is generated.

[0025] By adopting the above technical solution, the audio energy fluctuation parameters and the spectrum distribution parameters can be extracted from the live audio and normalized to generate audio feature vectors, and the target region motion trajectory parameters and the pixel change intensity parameters can be extracted from the live video and normalized to generate video feature vectors, so that the audio and video data are comparable in a unified feature space; the audio feature vectors and the video feature vectors are calculated in correspondence in the same time segment to obtain a cross-modal correlation coefficient, and when the audio energy fluctuation parameters and the video pixel change intensity parameters change synchronously in the time dimension, the cross-modal correlation coefficient is modified in weight to obtain a modified cross-modal correlation coefficient, thereby enhancing the stability of cross-modal feature matching; finally, the modified cross-modal correlation coefficient is compared with a preset threshold to generate a second confidence score, thereby solving the problem of low event recognition accuracy caused by the split processing of audio features and video features in the prior art, and improving the robustness and reliability of event confidence calculation.

[0026] Preferably, when the event confidence weight exceeds a preset confidence weight threshold, the monitoring center generates a grouping dispatch instruction based on the emergency dispatch time reference and the synchronization deviation information, determines the execution order of the emergency communication terminal and the emergency monitoring terminal in the grouping dispatch instruction according to the synchronization deviation information, and issues the grouping dispatch instruction to the emergency communication terminal and the emergency monitoring terminal, including:

[0027] The synchronization deviation of the emergency communication terminal and the emergency monitoring terminal is compared with a preset synchronization deviation threshold interval based on the synchronization deviation information, and a corresponding deviation level identifier is generated;

[0028] The emergency communication terminal and the emergency monitoring terminal are divided into multiple priority levels based on the deviation level identifier;

[0029] The emergency communication terminal is arranged in front of the emergency monitoring terminal to form an initial scheduling queue within the priority level according to the device type;

[0030] An execution priority queue is generated based on the priority level and the initial scheduling queue, and the start timing of each emergency communication terminal and emergency monitoring terminal is determined in combination with the emergency scheduling time reference;

[0031] The execution priority queue and the start timing are combined to generate a grouping scheduling instruction, and the grouping scheduling instruction is issued to the emergency communication terminal and the emergency monitoring terminal.

[0032] By adopting the above technical solution, the synchronization deviation of the emergency communication terminal and the emergency monitoring terminal can be compared with a preset synchronization deviation threshold interval based on the synchronization deviation information, and a deviation level identifier can be generated, thereby establishing a hierarchical standard for terminal clock synchronization differences. The emergency communication terminal and the emergency monitoring terminal can be divided into multiple priority levels based on the deviation level identifier, and the emergency communication terminal can be arranged in front of the emergency monitoring terminal to form an initial scheduling queue within the priority level according to the device type, thereby ensuring the hierarchy and order of the scheduling order. An execution priority queue can be generated based on the priority level and the initial scheduling queue, and the start timing of each terminal can be determined in combination with the emergency scheduling time reference, thereby realizing the time consistency of scheduling execution. The execution priority queue and the start timing are further combined to generate a grouping scheduling instruction and issue it to the terminal, thereby solving the problems of lack of unified time reference in scheduling, random terminal ordering, and inconsistent execution timing in the prior art, and significantly improving the accuracy and cooperativeness of emergency scheduling.

[0033] Preferably, the execution priority queue is generated based on the priority level and the initial scheduling queue, and the start timing of each emergency communication terminal and emergency monitoring terminal is determined in combination with the emergency scheduling time reference, including:

[0034] The corresponding time segment is allocated to the emergency communication terminal and the emergency monitoring terminal based on the priority level;

[0035] The start order of the emergency communication terminal and the emergency monitoring terminal is determined according to the ordering result of the initial scheduling queue within the time segment;

[0036] The time segment and the start order are combined to form a start timing table with a start time and a duration.

[0037] align the start-up timing table based on the emergency dispatch time reference to generate a start-up timing.

[0038] By adopting the above technical solution, the corresponding time segments can be allocated to the emergency communication terminal and the emergency monitoring terminal based on the priority level, so as to ensure that the dispatch time ranges of different level terminals do not conflict with each other, the start-up order can be determined according to the sorting result of the initial dispatch queue within the time segment, the start-up process of the emergency communication terminal and the emergency monitoring terminal has a clear execution sequence, the time segment and the start-up order can be combined to form a start-up timing table with a start time and a duration, so as to realize the visual scheduling of the terminal in the time dimension, and further, the start-up timing table can be globally aligned based on the emergency dispatch time reference to generate a start-up timing, so as to ensure that the start-up processes of all terminals under the global time reference remain consistent and coordinated, thereby solving the problems of inconsistent start-up timing and disorderly scheduling process in the prior art, and improving the timing accuracy and overall reliability of emergency dispatch.

[0039] Preferably, aligning the start-up timing table based on the emergency dispatch time reference to generate a start-up timing further comprises:

[0040] obtaining synchronization deviation information of the emergency communication terminal and each emergency monitoring terminal;

[0041] comparing the synchronization deviation information with the emergency dispatch time reference to obtain a time correction amount of each emergency communication terminal and each emergency monitoring terminal;

[0042] dynamically correcting the start time and the duration in the start-up timing table based on the time correction amount to form a corrected start-up timing.

[0043] By adopting the above technical solution, the synchronization deviation information of the emergency communication terminal and each emergency monitoring terminal can be obtained, and the synchronization deviation information is compared with the emergency dispatch time reference to obtain the time correction amount of each terminal, thereby providing a quantitative basis for subsequent timing correction, the start time and the duration in the start-up timing table can be dynamically corrected based on the time correction amount to form a corrected start-up timing, so that the running intervals of different terminals under the global time reference are adjusted and unified, and further, the consistency and controllability of dispatch execution can still be maintained through the corrected start-up timing in the case that there is a clock difference between the emergency communication terminal and the emergency monitoring terminal, thereby solving the problem of inaccurate scheduling caused by terminal clock error in the prior art, and improving the accuracy and global coordination of emergency instruction execution.

[0044] Preferably, during the process of issuing a group scheduling instruction, when it is detected that the communication network delay exceeds a preset network delay threshold, the edge computing node directly executes a local broadcast and early warning instruction based on an emergency scheduling time reference and synchronization deviation information, and feeds back the execution state of the local broadcast and early warning instruction to a cloud management platform, including:

[0045] Comparing the communication network delay with the preset network delay threshold generates a delay level identifier;

[0046] When the delay level identifier is in the first delay level, the edge computing node triggers a local broadcast instruction based on the emergency scheduling time reference, and records a corresponding execution timestamp after execution is completed;

[0047] When the delay level identifier is in the second delay level, the edge computing node triggers a local early warning instruction based on the synchronization deviation information, and dynamically adjusts the coverage range and execution duration of the instruction during execution;

[0048] The execution timestamp, the coverage range and the execution duration recorded during execution form execution state information, and the execution state information is fed back to the cloud management platform.

[0049] By adopting the above technical solution, the communication network delay can be compared with the preset network delay threshold to generate a delay level identifier, thereby realizing hierarchical monitoring of the network state. When the delay level identifier is in the first delay level, the edge computing node triggers a local broadcast instruction based on the emergency scheduling time reference and records an execution timestamp, thereby ensuring that broadcast information can be quickly issued under low delay conditions. When the delay level identifier is in the second delay level, the edge computing node triggers a local early warning instruction based on the synchronization deviation information, and dynamically adjusts the coverage range and execution duration during execution, thereby ensuring effective coverage and reasonable duration of the early warning information under medium delay conditions. Finally, the execution timestamp, the coverage range and the execution duration form execution state information, and the execution state information is fed back to the cloud management platform, thereby solving the problem of lack of localized emergency response and execution tracking in the prior art, and improving the reliability and traceability of emergency instructions in unstable network environments.

[0050] The second invention purpose of the application is realized by the following technical solution:

[0051] A smart emergency management device, the smart emergency management device comprising:

[0052] A SIP access and clock synchronization module is configured to access an emergency communication terminal and an emergency monitoring terminal to a network based on a SIP protocol, perform unified clock synchronization processing during the access process, generate an emergency scheduling time reference, and record synchronization deviation information of the emergency communication terminal and the emergency monitoring terminal;

[0053] An event credibility calculation module is configured to acquire emergency alarm information, and the monitoring center synchronously acquires monitoring information and on-site audio and video information corresponding to the emergency alarm information based on the emergency dispatch time reference, and performs fusion calculation on the monitoring information and the audio and video information to generate an event credibility weight;

[0054] A grouping scheduling instruction generation module is configured to, when the event credibility weight exceeds a preset credibility weight threshold, generate a grouping scheduling instruction by the monitoring center based on the emergency dispatch time reference and the synchronization deviation information, determine an execution order of the emergency communication terminal and the emergency monitoring terminal in the grouping scheduling instruction according to the synchronization deviation information, and issue the grouping scheduling instruction to the emergency communication terminal and the emergency monitoring terminal;

[0055] An edge execution and feedback module is configured to, when detecting that a communication network delay exceeds a preset network delay threshold during the issuance of the grouping scheduling instruction, directly execute a local broadcast and early warning instruction by the edge computing node based on the emergency dispatch time reference and the synchronization deviation information, and feed back an execution state of the local broadcast and early warning instruction to the cloud management platform.

[0056] By adopting the above technical solutions, unified clock synchronization processing can be performed based on the SIP protocol when the emergency communication terminal and the emergency monitoring terminal access the network, the emergency dispatch time reference is generated and the synchronization deviation information is recorded, thereby solving the problem that the terminal data is difficult to synchronize due to the lack of a unified time reference in the prior art, the monitoring center can synchronously acquire the monitoring information and the on-site audio and video information corresponding to the emergency alarm information based on the emergency dispatch time reference, and perform fusion calculation to generate the event credibility weight, thereby realizing effective integration of multi-source data, avoiding the problem that the information is dispersed and it is difficult to accurately determine the event situation in the prior art, the grouping scheduling instruction can be generated when the event credibility weight exceeds the preset threshold, and the terminal execution order can be determined based on the synchronization deviation information, thereby ensuring the consistency and collaboration of the dispatch timing, when the communication network delay exceeds the preset threshold, the edge computing node can directly execute the local broadcast and early warning instruction and feed back the execution state, thereby realizing rapid response under the condition that the network is blocked, and solving the problem that the emergency response is not timely due to the lack of a local processing mechanism in the prior art.

[0057] The third aspect of the present application is achieved by the following technical solutions:

[0058] A computer readable storage medium stores a computer program, and the computer program is executed by a processor to realize the steps of the above-mentioned intelligent emergency management method.

[0059] In summary, the present application has at least one of the following beneficial technical effects:

[0060] 1. The application can perform unified clock synchronization processing based on the SIP protocol when the emergency communication terminal and the emergency monitoring terminal access the network, generate an emergency dispatch time reference, and record synchronization deviation information, thereby solving the problem of lack of unified time reference in the prior art, which makes it difficult to synchronize terminal data; the monitoring center can synchronize the emergency alarm information corresponding to the monitoring information and the on-site audio and video information based on the emergency dispatch time reference, and perform fusion calculation to generate an event credibility weight, thereby realizing effective integration of multi-source data and avoiding the problem of dispersed information in the prior art, which makes it difficult to accurately determine the event situation; when the event credibility weight exceeds the preset threshold, the grouping dispatch instruction is generated, and the terminal execution order is determined based on the synchronization deviation information, thereby ensuring the consistency and collaboration of the dispatch timing; when the communication network delay exceeds the preset threshold, the edge computing node can directly execute the local broadcast and warning instruction and feedback the execution state, thereby realizing fast response under network congestion conditions and solving the problem of lack of local processing mechanism in the prior art, which leads to delayed emergency response. BRIEF DESCRIPTION OF DRAWINGS

[0061] Figure 1 is a flowchart of a smart emergency management method in an embodiment of the application.

[0062] Figure 2 is an implementation flowchart of step S20 in a smart emergency management method in an embodiment of the application;

[0063] Figure 3 is another implementation flowchart of step S201 in a smart emergency management method in an embodiment of the application;

[0064] Figure 4 is an implementation flowchart of step S202 in a smart emergency management method in an embodiment of the application;

[0065] Figure 5 is a principle block diagram of a smart emergency management device in an embodiment of the application, DETAILED DESCRIPTION

[0066] The application will be further described in detail below with reference to the accompanying drawings.

[0067] In an embodiment, as shown in Figure 1 , the application discloses a smart emergency management method, which specifically includes the following steps:

[0068] S10: based on the SIP protocol, the emergency communication terminal and the emergency monitoring terminal are connected to the network, and unified clock synchronization processing is performed during the access process, an emergency dispatch time reference is generated, and the synchronization deviation information of the emergency communication terminal and the emergency monitoring terminal is recorded.

[0069] Specifically, the emergency communication terminal and the emergency monitoring terminal are accessed to the network based on the SIP protocol, the SIP protocol refers to the Session Initiation Protocol, which is used to establish and manage multimedia session connections, the emergency communication terminal refers to a device used to perform voice broadcast, telephone call and intercom functions, the emergency monitoring terminal refers to a device used to collect environmental parameters, video information and running state data, unified clock synchronization processing is performed in the access process, the unified clock synchronization processing refers to collecting and comparing the local clock time of each access terminal by using the Network Time Protocol, and generating a correction value by a time deviation calculation method, and writing the correction value into the clock register of each access terminal, so that the local clock of all emergency communication terminals and emergency monitoring terminals is consistent with the network reference clock, after synchronization is completed, an emergency dispatch time reference is generated, the emergency dispatch time reference refers to a global unified time scale established by the monitoring center based on the reference clock, which is used as a unified time basis for event scheduling and task issuance, and records the synchronization deviation information of the emergency communication terminal and the emergency monitoring terminal, the synchronization deviation information refers to the time difference between the local clock time of each access terminal and the emergency dispatch time reference, which is saved by establishing a deviation information record table in the database to support subsequent dynamic correction of dispatch order and start timing.

[0070] S20: Obtain emergency alarm information, the monitoring center synchronously obtains monitoring information and on-site audio and video information corresponding to the emergency alarm information based on the emergency dispatch time reference, and performs fusion calculation on the monitoring information and the audio and video information to generate an event credibility weight.

[0071] Specifically, the emergency alarm information refers to the emergency event alarm data reported by the emergency communication terminal, the monitoring center synchronously obtains the monitoring information and the on-site audio and video information corresponding to the emergency alarm information based on the emergency dispatch time reference, the monitoring information refers to the environmental parameter data collected by the emergency monitoring terminal, including temperature parameter, smoke concentration parameter and harmful gas concentration parameter, the audio and video information refers to the audio data and video data collected by the on-site camera device, after the acquisition is completed, the monitoring information and the audio and video information are fused and calculated to generate an event credibility weight, the event credibility weight refers to a numerical index used to represent the authenticity degree of the emergency alarm information. In the fusion calculation process, first, standardization processing is performed on various environmental parameters in the monitoring information, the standardization processing refers to interval mapping of parameters with different dimensions according to a preset maximum value and minimum value range, so that each parameter is converted into a standardized numerical value in the [0, 1] interval, let the temperature parameter be x1, the smoke concentration parameter be x2, and the harmful gas concentration parameter be x3, after standardization processing, the results f(x1), f(x2), f(x3) are obtained respectively, and the monitoring information credibility score is recorded as:

[0072] wherein, a1, a2, a3 are monitoring information weight coefficients. Then audio and video information is extracted to obtain audio feature parameters and video feature parameters, the audio energy fluctuation parameter is denoted as y1, and the video pixel change intensity parameter is denoted as y2. The audio energy fluctuation parameter and the video pixel change intensity parameter are converted into values in the interval [0, 1] through standardization processing to obtain results g(y1) and g(y2) respectively. The audio and video information credibility score is denoted as:

[0073] wherein, h(g(y1), g(y2)) is a cross-modal correlation function, used to represent the synchronization of audio and video in the same time segment. Finally, the monitoring information credibility score and the audio and video information credibility score are nonlinearly fused, and the event credibility weight is denoted as:

[0074] wherein, γ1, γ2, γ3 are fusion weight coefficients.

[0075] S30: When the event credibility weight exceeds the preset credibility weight threshold, the monitoring center generates a grouping scheduling instruction based on the emergency scheduling time reference and the synchronization deviation information, determines the execution order of the emergency communication terminal and the emergency monitoring terminal in the grouping scheduling instruction according to the synchronization deviation information, and issues the grouping scheduling instruction to the emergency communication terminal and the emergency monitoring terminal.

[0076] Specifically, when the event credibility weight exceeds the preset credibility weight threshold, the monitoring center generates a grouping scheduling instruction based on the emergency scheduling time reference and the synchronization deviation information, determines the execution order of the emergency communication terminal and the emergency monitoring terminal in the grouping scheduling instruction according to the synchronization deviation information, and issues the grouping scheduling instruction to the emergency communication terminal and the emergency monitoring terminal, wherein the grouping scheduling instruction is a control instruction used to identify the execution order and execution timing of the emergency communication terminal and the emergency monitoring terminal. In the process of generating the grouping scheduling instruction, first, the synchronization deviation information corresponding to each emergency communication terminal and emergency monitoring terminal is denoted as Δt i , and the synchronization deviation information is the deviation between the terminal local clock and the emergency scheduling time reference. The monitoring center calculates the synchronization deviation level of each terminal based on the synchronization deviation information, and the synchronization deviation level is denoted as:

[0077] wherein, θ represents a preset synchronization deviation threshold interval. Then, the monitoring center classifies the emergency communication terminals and the emergency monitoring terminals based on the synchronization deviation level, and the smaller the priority level is, the closer the synchronization is to the emergency dispatch time reference, and the higher the priority is. After the priority level classification is completed, the monitoring center sorts the devices within the same priority level according to the device type, and the device type is divided into an emergency communication terminal type and an emergency monitoring terminal type, and the emergency communication terminal is arranged before the emergency monitoring terminal, to obtain an initial dispatch queue, which refers to a terminal set sorted based on the device type within the same priority level. Subsequently, the monitoring center generates an execution priority queue based on the priority level and the initial dispatch queue, and the execution priority queue refers to a terminal execution sequence arranged from high to low according to the priority level, and the order of the initial dispatch queue is maintained within the same level. The execution priority queue is denoted as:

[0078] Q = {T (1) , T (2) ,..., T (n)}, wherein the jth terminal in the execution priority queue is denoted as T(j), wherein j ∈ [1, n], and n represents the total number of terminals in the queue. Finally, the monitoring center assigns a start timing to the execution priority queue in combination with the emergency dispatch time reference, and the start timing is determined by the following formula:

[0079] wherein, Sj represents the start time of the jth terminal in the execution priority queue, T0 represents the start time of the emergency dispatch time reference, λ represents the adjustment coefficient of the synchronization deviation level to the start interval, and μ represents the correction coefficient of the synchronization deviation information to the start interval. Based on the above start timing, the monitoring center combines the execution priority queue and the start time corresponding relationship to generate a grouping dispatch instruction, and sends the grouping dispatch instruction to the emergency communication terminal and the emergency monitoring terminal.

[0080] S40: During the process of sending the grouping dispatch instruction, when it is detected that the communication network delay exceeds the preset network delay threshold, the edge computing node directly executes the local broadcast and warning instruction based on the emergency dispatch time reference and the synchronization deviation information, and feeds back the execution state of the local broadcast and warning instruction to the cloud management platform.

[0081] Specifically, during the process of issuing the group scheduling instruction, when the communication network delay exceeds the preset network delay threshold, the edge computing node directly executes the local broadcast and early warning instruction based on the emergency scheduling time reference and the synchronization deviation information, and feeds back the execution state of the local broadcast and early warning instruction to the cloud management platform, wherein the communication network delay refers to the time difference value generated in the process of transmitting the group scheduling instruction from the monitoring center to the edge computing node, and the preset network delay threshold refers to the maximum allowed transmission delay set according to the network quality requirement. In the execution process, the edge computing node first compares the communication network delay Δd and the preset network delay threshold δ d

[0082] Wherein, M represents the delay level, the value of 0 indicates that the delay is within the threshold range, the value greater than 0 indicates that the delay exceeds the threshold, and the larger the value, the more serious the delay. When the delay level identifier is equal to 1, the edge computing node triggers the local broadcast instruction based on the emergency scheduling time reference, the local broadcast instruction refers to the unified voice broadcast or warning signal executed by the broadcast terminal accessed by the edge computing node, and after the execution is completed, the edge computing node records the corresponding execution timestamp, the execution timestamp refers to the actual time point when the broadcast terminal starts to execute the local broadcast instruction. When the delay level identifier is greater than 1, the edge computing node triggers the local early warning instruction based on the synchronization deviation information, the local early warning instruction refers to the alarm action executed by the emergency monitoring terminal or alarm device accessed by the edge computing node, and in the execution process, the edge computing node dynamically adjusts the coverage range and execution duration of the early warning instruction based on the synchronization deviation information, wherein the coverage range refers to the geographical area or device range that the early warning signal can reach, and the execution duration refers to the time length of the early warning signal. After the execution of the local broadcast instruction or the local early warning instruction is completed, the edge computing node forms the execution state information by combining the execution timestamp, the coverage range and the execution duration, and feeds back the execution state information to the cloud management platform, and the execution state information refers to the complete running data for representing the execution process of the emergency instruction by the edge computing node under the network delay condition.

[0083] In an embodiment, as shown in FIG. 2, in step S20, the emergency alarm information is obtained, the monitoring center synchronously obtains the monitoring information and the on-site audio and video information corresponding to the emergency alarm information based on the emergency scheduling time reference, and performs fusion calculation on the monitoring information and the audio and video information to generate an event credibility weight, including: Figure 2

[0084] S201: Extract the sensor trigger intensity parameter and the trigger duration parameter from the monitoring information, normalize the sensor trigger intensity parameter and the trigger duration parameter, and obtain a first credibility score.

[0085] ​​Specifically, the sensor trigger intensity parameter and the trigger duration parameter are extracted from the monitoring information, wherein the monitoring information refers to real-time physical quantity data collected by sensors deployed in the emergency scene, the sensor trigger intensity parameter refers to the signal amplitude size output by the sensor under the trigger of a specific event, and the trigger duration parameter refers to the length of time for which the sensor maintains signal output under the trigger of the same event. In the normalization process, the sensor trigger intensity parameter Ps is first mapped to the preset intensity interval [0, 1] to obtain the intensity normalization result:

[0086] wherein Pmin represents the minimum trigger intensity perceivable by the sensor in the monitoring scene, Pmax represents the maximum trigger intensity perceivable by the sensor in the monitoring scene, and Ns represents the intensity normalization result. Subsequently, the trigger duration parameter Ts is mapped to the preset time interval [0, 1] to obtain the duration normalization result:

[0087] wherein Tmin represents the minimum trigger duration recorded by the sensor in the monitoring scene, Tmax represents the maximum trigger duration recorded by the sensor in the monitoring scene, and Nt represents the duration normalization result. After obtaining the intensity normalization result and the duration normalization result, the dynamic weighting coefficient a is calculated, wherein the dynamic weighting coefficient refers to a parameter for adjusting the weight of the intensity normalization result according to the trigger duration normalization result, and the calculation method is:

[0088] a = a0 + γ · N t wherein a0 represents the initial weight coefficient, and γ represents the incremental factor of the duration on the intensity weight. Finally, the intensity normalization result and the duration normalization result are weighted and fused based on the dynamic weighting coefficient to obtain the first confidence score C1, and the calculation method is:

[0089] C1 = a · N s + (1 - a) · N t wherein C1 represents the first confidence score, which is used to represent the quantitative contribution of the monitoring information in the event confidence calculation.

[0090] S202: Extract target event feature parameters from the on-site audio and video information, wherein the target event feature parameters include audio feature parameters and video feature parameters. Perform feature matching on the audio feature parameters and the video feature parameters to obtain a second confidence score.

[0091] Specifically, target event feature parameters are extracted from the on-site audio and video information. These feature parameters include audio and video feature parameters. The on-site audio and video information refers to the original audio signal and video image sequence acquired by the audio pickup and video capture devices in the emergency scenario at the same time reference. Audio feature parameters are a set of parameters that characterize the energy and frequency distribution characteristics of the audio signal, while video feature parameters are a set of parameters that characterize the pixel changes and motion trajectory characteristics in the video image. In the audio feature parameter extraction process, the audio signal is first divided into frames at a fixed sampling rate. For each frame, the audio energy fluctuation parameter Ea and the spectral distribution parameter Fa are calculated. The audio energy fluctuation parameter refers to the short-term energy change of the audio signal in the time dimension, and the spectral distribution parameter refers to the amplitude distribution of the audio signal in the frequency dimension. The audio energy fluctuation parameter and the spectral distribution parameter are mapped to the interval [0,1] respectively, resulting in the normalized audio energy result Ne and the audio spectrum result Nf. Based on this, an audio feature vector is constructed:

[0092] V a ={N e N f In the video feature parameter extraction process, the video image sequence is first subjected to inter-frame differencing to obtain the pixel change intensity parameter Pv, which refers to the average difference in grayscale values ​​between adjacent frames. Then, a target detection algorithm is used to extract the target region motion trajectory parameter Mv, which refers to the displacement trajectory of the same target position over time in consecutive video frames. The pixel change intensity parameter and the target region motion trajectory parameter are mapped to the interval [0,1] respectively, yielding normalized results Np and Nm. Based on these, a video feature vector is constructed:

[0093] V v ={N p N m In the feature matching process, audio feature vectors and video feature vectors within the same time segment are selected for cross-modal correlation calculation. The cross-modal correlation coefficient is denoted as:

[0094] Where Va(t) represents the audio feature vector within time segment t, v(t) represents the video feature vector within time segment t, the numerator represents the vector dot product, and the denominator represents the product of vector magnitudes. After obtaining the cross-modal correlation coefficient, the second confidence score C2 is calculated as follows:

[0095]

[0096] Where T represents the total number of time segments, β represents the balance factor between cross-modal correlation coefficient and local feature matching relationship, and min{N e (t), Np (t) represents the minimum normalized value of audio energy fluctuation and video pixel change at the same time, which is used to enhance the punishment effect of low consistency segments. The final second credibility score C2 is used to represent the quantitative contribution of live audio and video information in event credibility calculation.

[0097] S203: The first credibility score and the second credibility score are fused and calculated according to a preset weighting rule to obtain an event credibility weight.

[0098] Specifically, the first credibility score and the second credibility score are fused and calculated according to a preset weighting rule to obtain an event credibility weight, wherein the first credibility score refers to a credibility score obtained by normalizing and dynamically weighting a sensor trigger intensity parameter and a trigger duration parameter extracted based on monitoring information, the second credibility score refers to a credibility score obtained after feature matching of audio feature parameters and video feature parameters extracted based on live audio and video information, and the event credibility weight refers to a quantitative value used to represent comprehensive evaluation of event reliability of multi-source monitoring information and live audio and video information under a unified time reference. In the fusion calculation process, first, a preset weighting rule is set, wherein the preset weighting rule refers to a calculation criterion for assigning weight coefficients to credibility scores of different information sources, denoted as ω1 for the first credibility score weight and ω2 for the second credibility score weight, satisfying ω1 + ω2 = 1. Then, the first credibility score C1 and the second credibility score C2 are weighted and averaged according to the preset weighting rule to obtain a preliminary event credibility weight:

[0099] W e = ω1 · C1 + ω2 · C2, wherein We represents the event credibility weight. In order to enhance the dynamic adaptation capability of different credibility scores in emergency situations, an adjustment factor η is introduced, wherein the adjustment factor refers to a coefficient for dynamically correcting the weighted calculation result based on the difference between the first credibility score and the second credibility score, and the calculation method is:

[0100] η = 1 - |C1 - C2|, after obtaining the adjustment factor, the corrected event credibility weight calculation formula is: W = W e · η, wherein W represents the final event credibility weight, which can reflect the enhancement effect of the first credibility score and the second credibility score on the overall credibility when the consistency is high, and the weakening effect on the overall credibility when the difference is large. The finally generated event credibility weight W is used as an input parameter for the subsequent grouping scheduling instruction generation step.

[0101] In an embodiment, as Figure 3As shown, in step S201, that is, from the monitoring information, the sensor trigger intensity parameter and the trigger duration parameter are extracted, the sensor trigger intensity parameter and the trigger duration parameter are normalized to obtain a first credibility score, including:

[0102] S2011: mapping the sensor trigger intensity parameter to a preset intensity interval to obtain an intensity normalization result.

[0103] Specifically, the sensor trigger intensity parameter is mapped to a preset intensity interval to obtain an intensity normalization result, wherein the sensor trigger intensity parameter refers to the signal amplitude size output by the sensor in the monitoring information when an event is triggered, and the preset intensity interval refers to a numerical range established according to the minimum trigger intensity and the maximum trigger intensity that the sensor can perceive in the actual working environment. In the mapping process, first set the minimum value Pmin and the maximum value Pmax of the sensor trigger intensity, wherein Pmin represents the minimum trigger intensity that the sensor can perceive in the monitoring scene, and Pmax represents the maximum trigger intensity that the sensor can perceive in the monitoring scene. The actually collected sensor trigger intensity parameter is denoted as P_s, and P_s is mapped to the interval [0, 1] by a linear normalization method, and the calculation formula is:

[0104] Wherein, Ns represents the intensity normalization result, used to represent the relative size of the sensor trigger intensity parameter in the preset intensity interval. When the sensor trigger intensity parameter is equal to the minimum value, the intensity normalization result takes the value 0, when the sensor trigger intensity parameter is equal to the maximum value, the intensity normalization result takes the value 1, and when the sensor trigger intensity parameter is between the minimum value and the maximum value, the intensity normalization result continuously changes in the interval between 0 and 1.

[0105] S2012: mapping the trigger duration parameter to a preset time interval to obtain a duration normalization result.

[0106] Specifically, the trigger duration parameter is mapped to a preset time interval to obtain a duration normalization result, wherein the trigger duration parameter refers to the length of time that the sensor maintains signal output in the monitoring information during a one-time event trigger process, and the preset time interval refers to a numerical range established according to the minimum trigger duration and the maximum trigger duration that the sensor can detect in the actual working environment. In the mapping process, first set the minimum value Tmin and the maximum value Tmax of the trigger duration, wherein Tmin represents the minimum trigger duration recorded by the sensor in the monitoring scene, and Tmax represents the maximum trigger duration recorded by the sensor in the monitoring scene. The actually collected trigger duration parameter is denoted as Ts, and Ts is mapped to the interval [0, 1] by a linear normalization method, and the calculation formula is:

[0107] wherein Nt represents a duration normalization result, used to represent the relative position of the trigger duration parameter in the preset time interval. When the trigger duration parameter is equal to the minimum value, the duration normalization result takes the value of 0, when the trigger duration parameter is equal to the maximum value, the duration normalization result takes the value of 1, and when the trigger duration parameter is between the minimum value and the maximum value, the duration normalization result continuously changes in the interval between 0 and 1.

[0108] S2013: Calculate a dynamic weighting coefficient based on the intensity normalization result and the duration normalization result, wherein the weight of the intensity normalization result increases with the increase of the duration normalization result in the dynamic weighting coefficient.

[0109] Specifically, a dynamic weighting coefficient is calculated based on the intensity normalization result and the duration normalization result, wherein the dynamic weighting coefficient refers to a proportion factor used to adjust the relative importance of the intensity normalization result and the duration normalization result in the weighting calculation process. The design principle of the dynamic weighting coefficient is to make the weight of the intensity normalization result increase with the increase of the duration normalization result, so as to reflect that in the case of a longer event trigger duration, the sensor trigger intensity has a more significant impact on the credibility. In the calculation process, first set an initial weight coefficient a0, wherein a0 represents the initial weight of the intensity normalization result when the duration normalization result is zero. Then introduce an increment adjustment factor γ, wherein γ represents the promotion amplitude of the duration normalization result to the weight of the intensity normalization result. Denote the duration normalization result as Nt and the intensity normalization result as Ns, and the calculation formula of the dynamic weighting coefficient is:

[0110] a = a0 + γ·N t wherein a represents the dynamic weighting coefficient, and the value range is limited to [0, 1]. When the duration normalization result N_t increases, the dynamic weighting coefficient increases accordingly, thereby promoting the weight of the intensity normalization result in the subsequent fusion calculation. The final obtained dynamic weighting is used to guide the weighted fusion of the intensity normalization result and the duration normalization result, so as to generate the first credibility score.

[0111] S2014: Weighted fusion of the intensity normalization result and the duration normalization result according to the dynamic weighting coefficient to obtain the first credibility score.

[0112] Specifically, the intensity normalization result and the duration normalization result are weighted and fused using a dynamic weighting coefficient to obtain the first confidence score. Here, the intensity normalization result refers to the value obtained after normalizing the sensor trigger intensity parameter, and the duration normalization result refers to the value obtained after normalizing the trigger duration parameter. The dynamic weighting coefficient is a proportional factor used in the fusion process to balance the relative contributions of the intensity normalization result and the duration normalization result. The first confidence score is a quantitative indicator that characterizes the reliability of the event, calculated based on monitoring information. In the calculation process, the intensity normalization result is denoted as Ns, the duration normalization result as Nt, and the dynamic weighting coefficient as α. The formula for calculating the first confidence score is:

[0113] C1=α·N s +(1-α)·N t Where C1 represents the first credibility score, α represents the dynamic weighting coefficient, Ns represents the intensity normalization result, and Nt represents the duration normalization result. When the dynamic weighting coefficient is large, the first credibility score is mainly affected by the intensity normalization result; when the dynamic weighting coefficient is small, the first credibility score is mainly affected by the duration normalization result; when the dynamic weighting coefficient is in the middle value, the first credibility score is jointly determined by the intensity normalization result and the duration normalization result. The final first credibility score C1 will be used as one of the input parameters in the calculation of event credibility weight, and will participate in the generation of event credibility weight together with the second credibility score.

[0114] In one embodiment, such as Figure 4 As shown, in step S202, target event feature parameters are extracted from the on-site audio and video information. These target event feature parameters include audio feature parameters and video feature parameters. The audio feature parameters and video feature parameters are then matched to obtain a second credibility score, including:

[0115] S2021: Extract audio energy fluctuation parameters and spectral distribution parameters from the live audio, normalize the audio energy fluctuation parameters and spectral distribution parameters to obtain audio feature vectors.

[0116] Specifically, audio energy fluctuation parameters and spectral distribution parameters are extracted from the on-site audio. These parameters are then normalized to obtain an audio feature vector. On-site audio refers to the raw sound signal collected by a pickup device under a unified time reference in an emergency scenario. The audio energy fluctuation parameter refers to the amplitude change value of the audio signal obtained through short-time energy calculation within a preset time segment. The spectral distribution parameter refers to the amplitude distribution characteristics of the audio signal in the frequency domain obtained through Fast Fourier Transform within a preset time segment. During the normalization process, a minimum value Emin and a maximum value Emax of the audio energy fluctuation parameter are first set. The actual collected audio energy fluctuation parameter is denoted as Ea. Ea is mapped to the interval [0,1] using a linear normalization method to obtain the normalized result.

[0117] Where Ne represents the normalized result of the audio energy fluctuation parameter. Then, the minimum value Fmin and the maximum value Fmax of the spectral distribution parameter are set, and the actual collected spectral distribution parameter is denoted as Fa. Fa is mapped to the interval [0,1] using a linear normalization method to obtain the normalized result:

[0118] Where Nf represents the normalized result of the spectral distribution parameters. After obtaining the normalized results of the audio energy fluctuation parameters and the spectral distribution parameters, an audio feature vector is constructed:

[0119] V a ={N e N f}, where Va represents the audio feature vector, used to characterize the quantitative features of live audio during event analysis.

[0120] S2022: Extract the target area motion trajectory parameters and pixel change intensity parameters from the live video, normalize the target area motion trajectory parameters and pixel change intensity parameters to obtain the video feature vector.

[0121] Specifically, the motion trajectory parameters and pixel change intensity parameters of the target area are extracted from the on-site video. These parameters are then normalized to obtain a video feature vector. The on-site video refers to a sequence of continuous image frames acquired by video capture equipment under a unified time reference in an emergency scenario. The motion trajectory parameters of the target area refer to the displacement trajectory of the event-related target obtained through target detection and tracking methods in the continuous image frames. The pixel change intensity parameters refer to the average grayscale difference value obtained through inter-frame difference methods in the continuous image frames, used to characterize the magnitude of pixel change over time. During the normalization process, the minimum value Mmin and the maximum value Mmax of the target area motion trajectory parameters are first set. The actual acquired target area motion trajectory parameters are denoted as Mv. A linear normalization method is used to map Mv to the interval [0,1] to obtain the normalization result.

[0122] Where Nm represents the normalized result of the target region's motion trajectory parameters. Then, the minimum value Pmin and the maximum value Pmax of the pixel change intensity parameter are set, and the actual collected pixel change intensity parameter is denoted as Pv. Pv is mapped to the interval [0,1] using a linear normalization method to obtain the normalized result:

[0123] Where Np represents the normalized result of the pixel change intensity parameter. After obtaining the normalized results of the target region motion trajectory parameters and the pixel change intensity parameter, a video feature vector is constructed:

[0124] V v ={N m N p}, where Vv represents the video feature vector, used to characterize the quantitative features of the live video during the event analysis process.

[0125] S2023: The audio feature vector and the video feature vector are calculated in the same time segment to obtain the cross-modal correlation coefficient.

[0126] Specifically, the audio feature vector and video feature vector are calculated correspondingly within the same time segment to obtain the cross-modal correlation coefficient. The audio feature vector is a vector composed of the normalized results of audio energy fluctuation parameters and spectral distribution parameters; the video feature vector is a vector composed of the normalized results of target area motion trajectory parameters and pixel change intensity parameters. A time segment refers to the smallest time unit obtained by dividing the on-site audio and video according to a fixed length under a unified time reference. The cross-modal correlation coefficient is a measure of the correlation between the audio and video feature vectors within the same time segment. In the calculation process, the total number of time segments is set to T. Within the t-th time segment, the audio feature vector is denoted as:

[0127] V a (t)={N e (t),N f , where Ne(t) represents the normalized result of the audio energy fluctuation parameter in the t-th time segment, and Nf(t) represents the normalized result of the spectral distribution parameter in the t-th time segment.

[0128] The video feature vector is denoted as:

[0129] V v (t)={N m (t),N p Nm(t)} represents the normalized result of the target region motion trajectory parameters in the t-th time segment, and Np(t) represents the normalized result of the pixel change intensity parameters in the t-th time segment. After obtaining the audio feature vector and video feature vector, the cosine similarity method is used to calculate the cross-modal correlation coefficient of the t-th time segment:

[0130] Where the numerator represents the inner product of the audio feature vector and the video feature vector, and the denominator represents the product of the magnitudes of the audio feature vector and the video feature vector. Finally, the cross-modal correlation coefficient sequence {R} is calculated. av (1),R av (2),...,R av (T)} is used for subsequent calculation of audio and video matching degree and generation of second credibility score.

[0131] S2024: In the process of calculating the cross-modal correlation coefficient, when the audio energy fluctuation parameter and the video pixel change intensity parameter change synchronously within the same time segment, the cross-modal correlation coefficient is weighted and corrected to obtain the corrected cross-modal correlation coefficient.

[0132] Specifically, in the calculation of the cross-modal correlation coefficient, when the audio energy fluctuation parameter and the video pixel change intensity parameter change synchronously within the same time segment, a weight correction is performed on the cross-modal correlation coefficient to obtain the corrected cross-modal correlation coefficient. Here, the audio energy fluctuation parameter refers to the normalized result obtained from short-time energy calculation in the live audio signal, and the video pixel change intensity parameter refers to the normalized result obtained from inter-frame difference calculation in the live video sequence. Synchronous change refers to the simultaneous increase or decrease of the audio energy fluctuation parameter and the video pixel change intensity parameter within the same time segment. In the t-th time segment, let the normalized result of the audio energy fluctuation parameter be Ne(t), the normalized result of the video pixel change intensity parameter be Np(t), and the cross-modal correlation coefficient be Rav(t). When the changing trends of Ne(t) and Np(t) are consistent, a correction weighting factor δ(t) is introduced. The correction weighting factor is a dynamically adjusted parameter used to amplify the cross-modal correlation coefficient under synchronous change conditions, and its calculation formula is:

[0133] δ(t)=1+κ·min{N e (t),N p κ represents the correction intensity coefficient, with a larger value indicating a stronger correction effect. The corrected cross-modal correlation coefficient is denoted as κ(t)}. The calculation formula is: in, This represents the corrected cross-modal correlation coefficient, which retains the correlation measurement function of the original cross-modal correlation coefficient while enhancing the matching effect of audio and video when they change synchronously by adjusting the weighting factors. Finally, the corrected cross-modal correlation coefficient sequence... This serves as input data for subsequent comparisons and credibility calculations.

[0134] S2025: Compare the corrected cross-modal correlation coefficient with a preset threshold to obtain the comparison result. Based on the comparison result, determine the degree of audio-visual matching within the time segment, and then generate the corresponding second confidence score.

[0135] Specifically, the modified cross-modal correlation coefficient is compared with a preset threshold to obtain a comparison result, and the audio-video matching degree in the timing segment is determined according to the comparison result, and then a corresponding second confidence score is generated, wherein the modified cross-modal correlation coefficient refers to a correlation measurement value obtained after introducing a modified weight factor in the calculation process of the cross-modal correlation coefficient, the preset threshold refers to a judgment reference value set according to the audio-video synchronization characteristics in the emergency scene, the comparison result refers to the size relationship between the modified cross-modal correlation coefficient and the preset threshold, the audio-video matching degree refers to the consistency level of the audio feature vector and the video feature vector in the unified time segment, and the second confidence score refers to a quantitative confidence result generated based on the audio-video matching degree. In the tthtime segment, the modified cross-modal correlation coefficient is denoted as The preset threshold is denoted as θ. The comparison result is defined as: Wherein, Δ(t) represents the comparison result. In the case that the comparison result is greater than zero, it means that the modified cross-modal correlation coefficient exceeds the preset threshold, and the audio-video matching degree is higher. In the case that the comparison result is less than or equal to zero, it means that the modified cross-modal correlation coefficient does not reach the preset threshold, and the audio-video matching degree is lower. Based on the comparison result, the audio-video matching degree in the timing segment is defined as: M(t) = σ(Δ(t)), wherein σ(·) represents a normalized step function or a Sigmoid function, which is used to map the comparison result to the interval [0,1] to obtain the matching degree of the timing segment. Finally, the second confidence score is denoted as C2, and the calculation formula is: Wherein, T represents the total number of time segments, and C2 represents the second confidence score, which is used to represent the overall matching degree of the on-site audio-video information in multiple time segments.

[0136] In an embodiment, in step S30, i.e., when the event confidence weight exceeds the preset confidence weight threshold, the monitoring center generates a grouping dispatch instruction based on the emergency dispatch time reference and the synchronization deviation information, determines the execution order of the emergency communication terminal and the emergency monitoring terminal in the grouping dispatch instruction according to the synchronization deviation information, and issues the grouping dispatch instruction to the emergency communication terminal and the emergency monitoring terminal, including:

[0137] S301: Based on the synchronization deviation information, the synchronization deviation of the emergency communication terminal and the emergency monitoring terminal is compared with the preset synchronization deviation threshold interval, and a corresponding deviation level identifier is generated.

[0138] Specifically, the synchronization deviation of the emergency communication terminal and the emergency monitoring terminal is compared with a preset synchronization deviation threshold interval based on synchronization deviation information, and a corresponding deviation level identifier is generated, wherein the synchronization deviation information refers to the difference between the local clock and the emergency dispatch time reference during the access process of the emergency communication terminal and the emergency monitoring terminal, the preset synchronization deviation threshold interval refers to a plurality of hierarchical intervals set according to the system tolerance, which is used to divide the synchronization deviation levels in different ranges, and the deviation level identifier refers to a level symbol representing the synchronization accuracy of the terminal. In the implementation process, first, the synchronization deviation value corresponding to each emergency communication terminal and emergency monitoring terminal is obtained, and the synchronization deviation value is compared with the preset synchronization deviation threshold interval level by level. When the synchronization deviation value falls into the first level interval, a first level deviation level identifier is generated, indicating that the terminal has high synchronization accuracy. When the synchronization deviation value falls into the second level interval, a second level deviation level identifier is generated, indicating that the terminal has medium synchronization accuracy. When the synchronization deviation value falls into the third level interval, a third level deviation level identifier is generated, indicating that the terminal has low synchronization accuracy. In the comparison process, each emergency communication terminal and emergency monitoring terminal will generate a unique deviation level identifier, which is used to provide input basis for subsequent priority division and execution order determination.

[0139] S302: Dividing the emergency communication terminal and the emergency monitoring terminal into a plurality of priority levels based on the deviation level identifier.

[0140] Specifically, the emergency communication terminal and the emergency monitoring terminal are divided into a plurality of priority levels based on the deviation level identifier, wherein the priority level refers to the hierarchical level of grouping and sorting the emergency communication terminal and the emergency monitoring terminal according to the synchronization accuracy, and the deviation level identifier refers to the level symbol generated after comparing the synchronization deviation information with the preset synchronization deviation threshold interval, which is used to represent the synchronization accuracy range of the terminal. In the implementation process, first, the deviation level identifier corresponding to each emergency communication terminal and emergency monitoring terminal is collected. The emergency communication terminal and the emergency monitoring terminal with a first level deviation level identifier are divided into a priority level one, which is used to represent the terminal set with the highest synchronization accuracy. The emergency communication terminal and the emergency monitoring terminal with a second level deviation level identifier are divided into a priority level two, which is used to represent the terminal set with intermediate synchronization accuracy. The emergency communication terminal and the emergency monitoring terminal with a third level deviation level identifier are divided into a priority level three, which is used to represent the terminal set with the lowest synchronization accuracy. After the division is completed, each emergency communication terminal and emergency monitoring terminal is uniquely mapped to a priority level, which will be used as the basic input for generating the initial dispatch queue subsequently.

[0141] S303: Perform sorting within the priority level according to the device type, so that the emergency communication terminal is arranged before the emergency monitoring terminal to form an initial scheduling queue.

[0142] Specifically, the sorting within the priority level is performed according to the device type, where the device type refers to two types of devices, i.e., the emergency communication terminal and the emergency monitoring terminal. First, the identification information of all emergency communication terminals and emergency monitoring terminals is extracted in each priority level respectively, all emergency communication terminals are sequentially arranged according to the access order to obtain a communication terminal sorting result, and all emergency monitoring terminals are sequentially arranged according to the access order to obtain a monitoring terminal sorting result. Then, the communication terminal sorting result is placed before the monitoring terminal sorting result, and the order is kept unchanged within the priority level to form an initial scheduling queue. The initial scheduling queue refers to a queue structure arranged according to the order of device type within the priority level, which is used as an input basis for subsequent execution of the priority queue generation.

[0143] S304: Generate an execution priority queue based on the priority level and the initial scheduling queue, and determine the start timing of each emergency communication terminal and emergency monitoring terminal in combination with the emergency dispatch time reference.

[0144] Specifically, the execution priority queue is generated based on the priority level and the initial scheduling queue. First, the identification information of the emergency communication terminal and the emergency monitoring terminal is read in order according to the order of the initial scheduling queue within each priority level, and the read identification information is combined in order according to the priority level to form an execution priority queue. The execution priority queue refers to a sequential queue generated in combination with the priority level and the initial scheduling queue, which is used to describe the execution order of the emergency communication terminal and the emergency monitoring terminal in the overall dispatch. Then, the start timing of each emergency communication terminal and emergency monitoring terminal is determined in combination with the emergency dispatch time reference. Specifically, a start time and a duration are allocated to each emergency communication terminal and each emergency monitoring terminal in the execution priority queue, and the allocation result is mapped to a unified clock system specified by the emergency dispatch time reference, so that a start timing table corresponding to time is obtained in the execution priority queue. The start timing table refers to a time scheduling structure formed based on the execution priority queue and in combination with the emergency dispatch time reference, which is used to indicate the start relationship and the duration of the emergency communication terminal and the emergency monitoring terminal.

[0145] S305: Combine the execution priority queue and the start timing to generate a group scheduling instruction, and issue the group scheduling instruction to the emergency communication terminal and the emergency monitoring terminal.

[0146] Specifically, the execution priority queue is combined with the start timing, the order position of each emergency communication terminal and each emergency monitoring terminal in the execution priority queue is read first, and is associated with the corresponding start time and duration in the start timing table to form a mapping relationship. The mapping relationship refers to a one-to-one correspondence relationship between the order information in the execution priority queue and the time information in the start timing table, which is used to ensure that the execution order of the emergency communication terminal and the emergency monitoring terminal is synchronized and consistent with the time scheduling. Then, a grouping scheduling instruction is generated based on the mapping relationship, the grouping scheduling instruction includes execution priority information and start time information, wherein the execution priority information is determined by the execution priority queue, and the start time information is determined by the start timing table, and the two are combined to form a complete grouping scheduling structure. The generated grouping scheduling instruction is encapsulated as a scheduling data packet, which is a scheduling instruction unit formatted according to a communication protocol and is used for transmission in the network. Finally, the grouping scheduling instruction is issued to the emergency communication terminal and the emergency monitoring terminal, and the emergency scheduling time reference is used as a unified clock reference in the issuing process to ensure that the emergency communication terminal and the emergency monitoring terminal are started synchronously according to the execution priority and the start time in the grouping scheduling instruction, and the scheduling control process is completed.

[0147] In an embodiment, in step S304, the execution priority queue is generated based on the priority level and the initial scheduling queue, and the start timing of each emergency communication terminal and emergency monitoring terminal is determined in combination with the emergency scheduling time reference, including:

[0148] S3041: Assign corresponding time segments to the emergency communication terminals and the emergency monitoring terminals based on the priority level.

[0149] Specifically, the corresponding time segments are assigned to the emergency communication terminals and the emergency monitoring terminals based on the priority level, first, the identification information of all emergency communication terminals and emergency monitoring terminals is extracted within each priority level, and the time allocation interval is determined according to the order of the priority level. The time allocation interval refers to a continuous time period divided according to the emergency scheduling time reference, each time allocation interval corresponds to a priority level, which is used to ensure that the start order between different priority levels does not conflict. Within the time allocation interval, the continuous time segments are sequentially assigned to each emergency communication terminal and each emergency monitoring terminal according to the arrangement order of the emergency communication terminals and the emergency monitoring terminals in the priority level. The time segment refers to the smallest time unit divided from the emergency scheduling time reference, which has a definite start time and duration, and is used to control the start time and running window of a single terminal. Through the above method, each emergency communication terminal and each emergency monitoring terminal obtains the time segment corresponding to the priority level, thereby forming a time segment allocation result covering all terminals, laying a foundation for subsequent determination of the start order and generation of the start timing table.

[0150] S3042: Within the time slice, the starting order of the emergency communication terminal and the emergency monitoring terminal is determined according to the sorting result of the initial scheduling queue.

[0151] Specifically, within the time slice, the starting order of the emergency communication terminal and the emergency monitoring terminal is determined according to the sorting result of the initial scheduling queue. First, the arrangement information of the initial scheduling queue is read in each time slice, the initial scheduling queue refers to the order queue arranged according to the device type within the priority level, which contains the order identifier of all emergency communication terminals and emergency monitoring terminals. Then, according to the arrangement position in the initial scheduling queue, each emergency communication terminal and each emergency monitoring terminal is mapped to the specific execution order within the time slice in turn, the execution order refers to the relative starting order within a time slice, which is used to distinguish the starting order of different terminals in the same time slice. After the mapping is completed, the starting order table corresponding to each time slice is generated, the starting order table refers to the order structure recording the execution order of the emergency communication terminal and the emergency monitoring terminal within the time slice, which is consistent with the initial scheduling queue and is used to form the starting time sequence table subsequently.

[0152] S3043: The time slice is combined with the starting order to form the starting time sequence table with the starting time and the duration.

[0153] Specifically, the time slice is combined with the starting order to form the starting time sequence table with the starting time and the duration. First, the time slice corresponding to each emergency communication terminal and each emergency monitoring terminal is read in the time slice allocation result, the time slice has a starting time and a duration, denoted as Tstart(i) and Tdur(i), where i represents the i-th terminal. Then, the execution order of each emergency communication terminal and each emergency monitoring terminal is read in the starting order table, the execution order is denoted as R(i), where the smaller the value of R(i) is, the earlier the starting order is. On this basis, the time slice and the execution order are calculated correspondingly, the calculation method is: T exec (i) = T start (i) + (R(i) - 1) × Δt, where Texec(i) represents the specific starting time point of the i-th terminal, and Δt represents the minimum time interval allocated for adjacent terminals within the time slice. The running time interval of the terminal can be represented as: [T exec (i), T exec (i) + T dur(i), and finally, arranging all the specific starting time points Texec(i) of the emergency communication terminals and the emergency monitoring terminals and the running time intervals according to the execution order R(i) to form a starting time sequence table. The starting time sequence table refers to a scheduling structure recording the starting time and the duration of each emergency communication terminal and each emergency monitoring terminal on the global time axis, which is used for subsequent global alignment and generation of grouping scheduling instructions.

[0154] S3044: performing global alignment on the starting time sequence table based on the emergency scheduling time reference to generate a starting time sequence.

[0155] Specifically, the starting time sequence table is globally aligned based on the emergency scheduling time reference to generate a starting time sequence. First, the starting time and the duration of each emergency communication terminal and each emergency monitoring terminal in the starting time sequence table are read, the starting time is denoted as Texec(i), and the duration is denoted as Tdur(i), where i represents the i-th terminal. Then, the emergency scheduling time reference is obtained, which is a global reference time point generated by the unified clock synchronization process during the access of the emergency communication terminal and the emergency monitoring terminal to the network, denoted as Tbase. In the global alignment process, the starting time in the starting time sequence table is calculated by difference with the emergency scheduling time reference to obtain the time offset of each terminal: ΔT(i) = T exec (i) - T base According to the time offset ΔT(i), each starting time in the starting time sequence table is modified, and the modified starting time is: T' exec (i) = Tbase + ΔT(i), while keeping the corresponding duration Tdur(i) unchanged, to obtain the globally aligned running time interval:

[0156] [T' exec (i), T' exec (i) + T dur (i)], and finally, all the globally aligned and modified running time intervals are sorted according to the execution order to form the starting time sequence. The starting time sequence refers to a time sequence structure of the running time of the emergency communication terminal and the emergency monitoring terminal after being globally aligned based on the emergency scheduling time reference, which is used to ensure the consistency and executability of the grouping scheduling instructions on the global time axis.

[0157] In an embodiment, after step S3044, i.e., performing global alignment on the starting time sequence table based on the emergency scheduling time reference to generate a starting time sequence, it further includes:

[0158] S30441: obtaining synchronization deviation information of the emergency communication terminal and each emergency monitoring terminal.

[0159] Specifically, the synchronization deviation information of the emergency communication terminal and each emergency monitoring terminal is acquired. First, the synchronization deviation information of the emergency communication terminal is recorded in the emergency communication terminal access process, and the synchronization deviation information is the time difference between the local clock time of the emergency communication terminal and the emergency dispatch time reference. Meanwhile, the synchronization deviation information of each emergency monitoring terminal is recorded in the emergency monitoring terminal access process, and the synchronization deviation information is the time difference between the local clock time of each emergency monitoring terminal and the emergency dispatch time reference.

[0160] S30442: Comparing the synchronization deviation information with the emergency dispatch time reference to obtain the time correction amount of each emergency communication terminal and each emergency monitoring terminal.

[0161] Specifically, the synchronization deviation information is compared with the emergency dispatch time reference to obtain the time correction amount of each emergency communication terminal and each emergency monitoring terminal. First, the synchronization deviation information of each emergency communication terminal and each emergency monitoring terminal is acquired, and the synchronization deviation information is the time difference between the local clock time of the terminal and the emergency dispatch time reference, denoted as δ(i), where i represents the i th terminal. Then, the emergency dispatch time reference is acquired, and the emergency dispatch time reference is denoted as Tbase. This time reference is a global reference time point generated by unified clock synchronization processing in the process of accessing the network by the emergency communication terminal and the emergency monitoring terminal. In the comparison process, the synchronization deviation information is calculated by difference with the emergency dispatch time reference to obtain the time correction amount of each terminal: DeltaT corr (i) = T base - δ(i), where ΔT corr (i) represents the time correction amount of the i th terminal, reflecting the time compensation value that needs to be adjusted by the terminal under the global time reference. Finally, the time correction amounts corresponding to all emergency communication terminals and emergency monitoring terminals are stored in the correction result table, and the correction result table is a data set recording the time difference compensation amount between each terminal and the emergency dispatch time reference, which is used for subsequent dynamic correction of the start time and duration of the start timing table.

[0162] S30443: Dynamically correcting the start time and duration in the start timing table based on the time correction amount to form the corrected start timing.

[0163] Specifically, the start time and duration in the start timing table are dynamically corrected based on the time correction amount to form the corrected start timing. First, the start time and duration of each emergency communication terminal and each emergency monitoring terminal are read in the start timing table, and the start time is denoted as Texec(i) and the duration is denoted as Tdur(i), where i represents the i th terminal. Then, the time correction amount corresponding to the i th terminal is read in the correction result table, and the time correction amount is denoted as ΔT corr(i). In the dynamic correction process, the starting time is superimposed with the time correction amount to obtain the corrected starting time: T' exec (i) = Texec(i) + ΔT corr (i) is superimposed with the time correction amount to obtain the corrected duration: wherein, T' exec (i) represents the starting time of the i-th terminal after time correction, T' dur (i) represents the duration of the i-th terminal after time correction. Finally, the corrected starting time and the corrected duration are combined to generate the corrected operation interval corresponding to each emergency communication terminal and each emergency monitoring terminal: [T'exec(i), T'exec(i) + T' dur (i), and all the corrected operation intervals are arranged in the order of execution priority queue to obtain the corrected start-up timing. The corrected start-up timing refers to the scheduling result after dynamically adjusting the starting time and the duration in combination with the time correction amount under the emergency scheduling time reference, which is used to ensure the global consistency and execution accuracy of the grouping scheduling instruction.

[0164] In an embodiment, in step S40, i.e. during the issuance of the grouping scheduling instruction, when it is detected that the communication network delay exceeds the preset network delay threshold, the edge computing node directly executes the local broadcast and early warning instruction based on the emergency scheduling time reference and the synchronization deviation information, and feeds back the execution state of the local broadcast and early warning instruction to the cloud management platform, including: S401: comparing the communication network delay with the preset network delay threshold to generate a delay level identifier.

[0165] Specifically, the network delay refers to the round-trip time of the data packet in the transmission link monitored during the issuance of the grouping scheduling instruction. Then the preset network delay threshold is obtained, which is a delay limit value set to ensure that the emergency communication terminal and the emergency monitoring terminal can complete instruction execution within the specified time limit. In the comparison process, the communication network delay is compared with the preset network delay threshold one by one first, if the communication network delay is less than or equal to the preset network delay threshold, the delay level identifier is generated as the first delay level, if the communication network delay is greater than the preset network delay threshold and less than or equal to twice the preset network delay threshold, the delay level identifier is generated as the second delay level, if the communication network delay exceeds twice the preset network delay threshold, the delay level identifier is generated as the third delay level. Finally, the delay level identifier is recorded in the delay state table, which is a data set for saving the level corresponding to the current communication network delay level, providing a basis for the edge computing node to select the triggering mode of the local broadcast instruction or the local early warning instruction subsequently

[0166] S402: When the delay level identifier is in the first delay level, the edge computing node triggers a local broadcast instruction based on the emergency scheduling time reference, and records the corresponding execution timestamp after execution is completed.

[0167] Specifically, when the delay level identifier is in the first delay level, the edge computing node triggers a local broadcast instruction based on the emergency scheduling time reference. First, it is detected in the delay state table that the delay level identifier is in the first delay level, which refers to a delay state in which the communication network delay is less than or equal to a preset network delay threshold. Then, the emergency scheduling time reference is called, which refers to a global reference time point generated by unified clock synchronization processing in the process of accessing the network by the emergency communication terminal and the emergency monitoring terminal. After the trigger condition is met, the edge computing node generates a local broadcast instruction, which refers to a broadcast signal for issuing warning information to the emergency communication terminal and the emergency monitoring terminal at the same time, started according to the emergency scheduling time reference. The broadcast instruction is executed immediately in the edge computing node and the signal transmission is completed through the local area communication link. After the execution of the broadcast instruction is completed, the edge computing node extracts the time point of the execution completion and records it as the execution timestamp. The execution timestamp refers to the recorded data indicating the actual completion time of the local broadcast instruction, which is used for subsequent feedback to the cloud management platform as part of the scheduling execution state information.

[0168] S403: When the delay level identifier is in the second delay level, the edge computing node triggers a local warning instruction based on the synchronization deviation information, and dynamically adjusts the coverage range and execution duration of the instruction during execution.

[0169] Specifically, when the delay level identifier is in the second delay level, the edge computing node triggers the local early warning instruction based on the synchronization deviation information. First, the delay level identifier is detected in the delay state table as the second delay level, and the second delay level refers to a delay state in which the communication network delay is greater than the preset network delay threshold and less than or equal to twice the preset network delay threshold. Then, the synchronization deviation information is obtained, which refers to the difference between the local clock time of the emergency communication terminal and the emergency monitoring terminal and the emergency dispatch time reference. The edge computing node corrects the early warning trigger time according to the synchronization deviation information to ensure the relative consistency of the local early warning instruction under the global reference time. In the triggering process, the edge computing node generates the local early warning instruction, which refers to a local alarm signal independently issued by the edge computing node under the communication network delay state, used to cover the emergency communication terminal and the emergency monitoring terminal within a specific range. In the execution process of the local early warning instruction, the edge computing node dynamically adjusts the coverage range, which refers to the geographical area or terminal set that the early warning signal can act on. The dynamic adjustment mode is: when the absolute value of the synchronization deviation information increases, the coverage range is expanded to ensure that the affected area is included in the warning range, and when the absolute value of the synchronization deviation information decreases, the coverage range is reduced to avoid excessive spread. At the same time, the edge computing node adjusts the execution duration according to the synchronization deviation information, which refers to the duration during which the local early warning instruction remains valid. When the absolute value of the synchronization deviation information approaches the upper limit of the preset deviation threshold, the execution duration is extended to ensure reliable instruction transmission, and when the absolute value of the synchronization deviation information approaches the lower limit of the deviation threshold, the execution duration is shortened to improve execution efficiency.

[0170] S404: Form the execution state information by combining the execution timestamp, the coverage range recorded during the execution process, and the execution duration, and feed back the execution state information to the cloud management platform.

[0171] Specifically, the execution timestamp is formed into execution state information together with the coverage range and the execution duration recorded in the execution process, and the execution state information is fed back to the cloud management platform. First, the execution timestamp is obtained after the local broadcast instruction or the local early warning instruction execution is completed, and the execution timestamp refers to the recorded data identifying the instruction completion time. Then, the coverage range information is collected in the instruction execution process, and the coverage range information refers to the geographical area or the terminal set actually affected by the local broadcast instruction or the local early warning instruction in the execution process. Then, the execution duration information is collected, and the execution duration information refers to the duration from triggering to ending of the local broadcast instruction or the local early warning instruction. After the data collection is completed, the edge computing node integrates the execution timestamp, the coverage range information and the execution duration information to form the execution state information. The execution state information refers to the state description data composed of the execution timestamp, the coverage range information and the execution duration information, and is used to reflect the execution process and the execution effect of the local broadcast instruction or the local early warning instruction. Finally, the edge computing node sends the execution state information to the cloud management platform through the uplink communication link with the cloud management platform, and the cloud management platform records and stores the execution state information after receiving, which is used for subsequent tracking of the emergency dispatch process and optimization of the dispatch strategy.

[0172] It should be understood that the size of the serial number of each step in the above embodiment does not mean the order of execution, and the execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present application.

[0173] In an embodiment, a smart emergency management device is provided, which corresponds to the smart emergency management method in the above embodiment. As shown in the figure, the smart emergency management device comprises a SIP access and clock synchronization module, an event credibility calculation module, a grouping dispatch instruction generation module and an edge execution and feedback module. Figure 5

[0174] The specific limitations of the smart emergency management device can be referred to the limitations of the smart emergency management method in the above, which will not be repeated here. Each module in the above smart emergency management device can be realized by software, hardware and their combinations in whole or in part. The above modules can be embedded in or independent of the processor in the computer device in hardware form, or can be stored in the memory in the computer device in software form, so as to be called and executed by the processor to perform the operations corresponding to each module.​

Claims

1. A smart emergency management method, characterized in that, The wisdom emergency management method comprises: Based on the SIP protocol, the emergency communication terminal and the emergency monitoring terminal are accessed to the network, and in the access process, the unified clock synchronization processing is performed, the emergency dispatch time reference is generated, and the synchronization deviation information of the emergency communication terminal and the emergency monitoring terminal is recorded; Obtain emergency alarm information, and the monitoring center synchronously obtains monitoring information and on-site audio and video information corresponding to the emergency alarm information based on the emergency dispatch time reference, and performs fusion calculation on the monitoring information and the audio and video information to generate an event credibility weight; When the event credibility weight exceeds a preset credibility weight threshold, the monitoring center generates a grouping dispatch instruction based on the emergency dispatch time reference and the synchronization deviation information, determines the execution order of the emergency communication terminal and the emergency monitoring terminal in the grouping dispatch instruction according to the synchronization deviation information, and issues the grouping dispatch instruction to the emergency communication terminal and the emergency monitoring terminal; In the process of issuing the grouping dispatch instruction, when it is detected that the communication network delay exceeds a preset network delay threshold, the edge computing node directly executes a local broadcast and early warning instruction based on the emergency dispatch time reference and the synchronization deviation information, and feeds back the execution state of the local broadcast and early warning instruction to the cloud management platform. 2.The intelligent emergency management method of claim 1, wherein, Obtain emergency alarm information, and the monitoring center synchronously obtains monitoring information and on-site audio and video information corresponding to the emergency alarm information based on the emergency dispatch time reference, and performs fusion calculation on the monitoring information and the audio and video information to generate an event credibility weight, comprising: Extract a sensor trigger intensity parameter and a trigger duration parameter from the monitoring information, normalize the sensor trigger intensity parameter and the trigger duration parameter, and obtain a first credibility score; Extract a target event feature parameter from the on-site audio and video information, the target event feature parameter comprising an audio feature parameter and a video feature parameter, and perform feature matching on the audio feature parameter and the video feature parameter to obtain a second credibility score; Fuse the first credibility score and the second credibility score according to a preset weighting rule to obtain the event credibility weight.

3. The intelligent emergency management method according to claim 2, characterized in that, Extract a sensor trigger intensity parameter and a trigger duration parameter from the monitoring information, normalize the sensor trigger intensity parameter and the trigger duration parameter, and obtain a first credibility score, comprising: Map the sensor trigger intensity parameter to a preset intensity interval to obtain an intensity normalization result; Map the trigger duration parameter to a preset time interval to obtain a duration normalization result; Calculate a dynamic weighting coefficient based on the intensity normalization result and the duration normalization result, wherein the weight of the intensity normalization result increases with the increase of the duration normalization result in the dynamic weighting coefficient; Weight fuse the intensity normalization result and the duration normalization result according to the dynamic weighting coefficient to obtain the first credibility score.

4. The intelligent emergency management method according to claim 2, characterized in that, Extract a target event feature parameter from the on-site audio and video information, the target event feature parameter comprising an audio feature parameter and a video feature parameter, and perform feature matching on the audio feature parameter and the video feature parameter to obtain a second credibility score, comprising: extracting an audio energy fluctuation parameter and a spectral distribution parameter from the live audio, normalizing the audio energy fluctuation parameter and the spectral distribution parameter to obtain an audio feature vector; extracting a target region motion trajectory parameter and a pixel change intensity parameter from the live video, normalizing the target region motion trajectory parameter and the pixel change intensity parameter to obtain a video feature vector; performing corresponding calculation on the audio feature vector and the video feature vector in the same time segment to obtain a cross-modal correlation coefficient; in the process of calculating the cross-modal correlation coefficient, when the audio energy fluctuation parameter and the video pixel change intensity parameter change synchronously in the same time segment, performing weight correction on the cross-modal correlation coefficient to obtain a corrected cross-modal correlation coefficient; comparing the corrected cross-modal correlation coefficient with a preset threshold to obtain a comparison result, determining the audio-video matching degree in the time sequence segment according to the comparison result, and further generating a corresponding second confidence score.

5. The intelligent emergency management method according to claim 1, characterized in that, When the event confidence weight exceeds the preset confidence weight threshold, the monitoring center generates a grouping dispatch instruction based on the emergency dispatch time reference and the synchronization deviation information, determines the execution order of the emergency communication terminal and the emergency monitoring terminal in the grouping dispatch instruction according to the synchronization deviation information, and issues the grouping dispatch instruction to the emergency communication terminal and the emergency monitoring terminal, including: comparing the synchronization deviation of the emergency communication terminal and the emergency monitoring terminal with the preset synchronization deviation threshold interval based on the synchronization deviation information, and generating a corresponding deviation level identifier; dividing the emergency communication terminal and the emergency monitoring terminal into multiple priority levels based on the deviation level identifier; performing sorting within the priority level according to the device type, arranging the emergency communication terminal in front of the emergency monitoring terminal to form an initial dispatch queue; generating an execution priority queue based on the priority level and the initial dispatch queue, and determining the start timing of each emergency communication terminal and emergency monitoring terminal in combination with the emergency dispatch time reference; combining the execution priority queue and the start timing to generate the grouping dispatch instruction, and issuing the grouping dispatch instruction to the emergency communication terminal and the emergency monitoring terminal.

6. The intelligent emergency management method according to claim 5, characterized in that, Generating an execution priority queue based on the priority level and the initial dispatch queue, and determining the start timing of each emergency communication terminal and emergency monitoring terminal in combination with the emergency dispatch time reference, including: allocating corresponding time segments for the emergency communication terminal and the emergency monitoring terminal based on the priority level; determining the start order of the emergency communication terminal and the emergency monitoring terminal according to the sorting result of the initial dispatch queue within the time segment; combining the time segment and the start order to form a start timing table with a start time and a duration; globally aligning the start timing table based on the emergency dispatch time reference to generate the start timing.

7. The intelligent emergency management method according to claim 6, characterized in that, Generating the start timing based on the emergency dispatch time reference to align the start timing table, further including: obtaining the synchronization deviation information of the emergency communication terminal and each emergency monitoring terminal; comparing the synchronization deviation information with the emergency dispatch time reference to obtain the time correction amount of each emergency communication terminal and each emergency monitoring terminal; The starting time and the duration in the starting timing table are dynamically corrected based on the time correction amount, to form a corrected starting timing. 8.The intelligent emergency management method of claim 1, wherein, In the process of issuing the grouping scheduling instruction, when it is detected that the communication network delay exceeds the preset network delay threshold, the edge computing node directly executes the local broadcast and early warning instruction based on the emergency scheduling time reference and the synchronization deviation information, and feeds back the execution state of the local broadcast and early warning instruction to the cloud management platform, including: Comparing the communication network delay with the preset network delay threshold to generate a delay level identifier; When the delay level identifier is in the first delay level, the edge computing node triggers the local broadcast instruction based on the emergency scheduling time reference, and records the corresponding execution timestamp after execution is completed; When the delay level identifier is in the second delay level, the edge computing node triggers the local early warning instruction based on the synchronization deviation information, and dynamically adjusts the coverage range and execution duration of the instruction during execution; The execution timestamp, the coverage range and the execution duration recorded during execution form the execution state information, and the execution state information is fed back to the cloud management platform.

9. A smart emergency management device, characterized in that, The intelligent emergency management device comprises: A SIP access and clock synchronization module is configured to access the emergency communication terminal and the emergency monitoring terminal to the network based on the SIP protocol, perform unified clock synchronization processing in the access process, generate an emergency scheduling time reference, and record synchronization deviation information of the emergency communication terminal and the emergency monitoring terminal; An event credibility calculation module is configured to obtain emergency alarm information, and a monitoring center synchronously obtains monitoring information and on-site audio and video information corresponding to the emergency alarm information based on the emergency scheduling time reference, and performs fusion calculation on the monitoring information and the audio and video information to generate an event credibility weight; A grouping scheduling instruction generation module is configured to generate a grouping scheduling instruction based on the emergency scheduling time reference and the synchronization deviation information when the event credibility weight exceeds a preset credibility weight threshold, determine an execution order of the emergency communication terminal and the emergency monitoring terminal in the grouping scheduling instruction according to the synchronization deviation information, and issue the grouping scheduling instruction to the emergency communication terminal and the emergency monitoring terminal; An edge execution and feedback module is configured to directly execute a local broadcast and early warning instruction based on the emergency scheduling time reference and the synchronization deviation information when it is detected that the communication network delay exceeds the preset network delay threshold in the process of issuing the grouping scheduling instruction, and feed back an execution state of the local broadcast and early warning instruction to a cloud management platform.

10. A computer-readable storage medium storing a computer program, the computer program comprising instructions that, when executed by a computer, cause the computer to perform the method of any one of claims 1 to 9. The computer program is executed by the processor to realize the steps of the intelligent emergency management method according to any one of claims 1 to 8. The computer program is executed by the processor to realize the steps of the intelligent emergency management method according to any one of claims 1 to 8.

Images