AI-based emergency rescue scene intelligent supervision system and method

Abstract

The invention relates to the technical field of scene supervision, in particular to an AI-based emergency rescue scene intelligent supervision system and method, and the system comprises a scene data collection module, a data feature analysis module, an intelligent decision support module and a rescue action execution module. According to the method, the dynamic change features in the rescue scene are accurately recognized by performing detailed color depth analysis, edge detection, waveform analysis and tone recognition on the collected images and audios, the recognition of key moment changes is more accurate through advanced data processing, and the construction of key morphological feature data is more comprehensive; through three-dimensional space mapping and analysis of time sequence information, accurate assessment of an accident range and rapid determination of resource demands can be realized, configuration and response time of rescue resources are greatly optimized, comprehensive application of technical means is realized, the rescue efficiency is remarkably improved, the risk of life and property loss is reduced, and the method is suitable for popularization and application. And the safety and the accuracy of rescue operation are improved.

Description

Technical Field

[0001] The present invention relates to the field of scene supervision technology, and in particular to an AI-based emergency rescue scene intelligent supervision system and method. Background Art

[0002] The field of scene monitoring technology mainly involves the use of sensors, cameras, data analysis tools and artificial intelligence algorithms to monitor and manage specific environments or occasions in real time, enabling institutions or individuals to obtain key information in real time and respond quickly to dynamic changes in the environment. In applications such as emergency rescue, scene monitoring technology can be used to track and coordinate rescue personnel, optimize resource allocation, and enhance the ability to respond to emergencies. The core lies in integrating advanced computing and communication technologies to improve monitoring efficiency and decision-making quality, thereby saving lives and protecting property at critical moments.

[0003] Among them, the emergency rescue scene intelligent supervision system is an intelligent system designed for rescue operations in emergency situations. Its main purpose is to monitor the rescue scene in real time through integrated sensors to provide real-time data support and decision-making assistance. The system can analyze various data in the rescue scene, such as personnel location, the status of rescue equipment and environmental conditions, and then guide the rescue team to make a more accurate response. The purpose is to maximize rescue efficiency and safety through technical means, ensure rapid and effective response to various emergencies, thereby saving lives and reducing losses.

[0004] Existing technologies are limited in their adaptability and responsiveness in rapidly changing environments, and are often unable to accurately process complex or ambiguous data, resulting in rescue operations being unable to accurately locate the accident site and the affected area in a timely manner. Lack of advanced data processing capabilities and insufficient agility in dynamic environments prevents rapid responses to various changes at the accident site. Technical limitations may result in resource allocation that is not timely or accurate, increasing rescue time and risk. Lack of three-dimensional data analysis and mapping technology also often leads to unscientific rescue priority assessments and resource allocation, affecting the overall efficiency and effectiveness of rescue operations. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an AI-based intelligent supervision system and method for emergency rescue scenarios.

[0006] In order to achieve the above objectives, the present invention adopts the following technical solutions: An AI-based emergency rescue scene intelligent supervision system includes: The scene data acquisition module captures images and sounds through the rescue scene video and audio equipment, synchronously collects geographic and climate data from GPS and environmental sensors, synchronizes the data in time, and outputs the original scene data records; The data feature analysis module performs color depth and edge detection processing on the image based on the original data record of the scene, performs waveform analysis and tone recognition on the sound, synchronously compares the timestamp data of the image sequence and the sound sequence according to the image and sound processing results, identifies dynamic change features, and constructs key morphological feature data; The intelligent decision support module analyzes the spatial position and time series information in the key morphological feature data, performs three-dimensional spatial mapping of the accident range, evaluates the rescue priority in real time according to the mapping results, determines the resource requirements, optimizes the resource allocation and response time, and outputs real-time decision guidance information; The rescue action execution module adjusts the rescue resource allocation, specifies the optimal rescue route, activates rescue equipment and personnel, and formulates and pushes the rescue action plan according to the real-time decision guidance information.

[0007] As a further solution of the present invention, the steps of obtaining the scene original data record are: Install video and audio equipment at the rescue site to capture real-time images and sounds of the rescue site and generate audio-visual data sets; Collect GPS location data and environmental sensor data of the rescue site in real time, construct a comprehensive geographical and climatic data set, and perform preliminary data association with the audio-visual data set to obtain an associated data set; The associated data set is time synchronized using the formula: Calculate the average value of each data type at the same time point , get the original scene data record, where Representative Class data in time The observed value of is the total number of data categories.

[0008] As a further solution of the present invention, the image and sound processing steps are: Based on the original data record of the scene, extract image data and sound data, perform image preprocessing, synchronously obtain preliminary waveforms of sound data, and generate image preprocessing results and preliminary sound waveform data; Based on the image preprocessing result and the preliminary sound waveform data, color depth analysis is performed on the image preprocessing result, edge detection is used to extract image edges, the waveform of the sound data is synchronously analyzed, the fundamental frequency and pitch of the sound are identified through spectrum analysis, and image depth edge data and sound spectrum analysis results are generated; The image depth edge data and the sound spectrum analysis results are integrated, the features of the image and sound data are compared, the visual and auditory information of the scene is identified, and the visual and auditory data analysis results are generated.

[0009] As a further solution of the present invention, the steps for acquiring key morphological feature data are: Extracting timestamps based on the visual and auditory data analysis results, confirming that the visual data and the auditory data are aligned on the time axis, and generating image and sound sequence data with synchronized timestamps; Based on the image and sound sequence data of the synchronization timestamp, the image sequence and the sound sequence are compared, and the formula is used, Calculate the edge change of image data and the spectrum change value of sound data at each time point , generate dynamic change feature data of images and sounds, where represents the edge strength of the image sequence, represents the frequency response value of the sound sequence, is the minimum cumulative distance of the previous state; The dynamic change feature data of the image and sound are analyzed, the features are classified, the motion trajectory of objects in the rescue scene and the change pattern of the sound are analyzed, and key morphological feature data are constructed.

[0010] As a further solution of the present invention, the three-dimensional space mapping step is: Based on the key morphological feature data, the spatial coordinates and timestamps of each feature are extracted, coordinate mapping is performed and time calibration is completed, and the spatial position and time series information are integrated to generate a preliminary data set; Synchronizing the coordinates and time series of the preliminary data set, matching and integrating the time and space points of key morphological features, and obtaining a complete three-dimensional time-space data set of the accident; According to the complete accident three-dimensional time-space data set, the formula is used: Calculate the The spatial location of key morphological features , generate the three-dimensional spatial mapping result of the accident range, where For the The coordinates of the locations, is the base coordinate of the center point, is the timestamp, Indicates the number of positions.

[0011] As a further solution of the present invention, the steps for obtaining the real-time decision guidance information are: Based on the three-dimensional spatial mapping results of the accident range, the key affected areas of the accident area are identified, and the spatial distribution characteristics and damage degree of each area are analyzed item by item. The formula is used. Calculate the impact index for each region , compare the impact index of each region and get the regional priority ranking table, where, Indicates the damaged area of ​​the region, Indicates the density of people affected in the area, Indicates the start time of the event. Indicates the current time of the event; According to the regional priority ranking table, call the real-time resource inventory information, compare the emergency demand and inventory, and get the actual demand gap of each region. Use the formula: Compute resource allocation priority , generate a resource configuration plan, where represents the impact index of the region, Indicates the emergency resource demand of the region, Indicates the current available resource inventory; Based on the resource allocation scheme and combined with the response data of the rescue path, the formula is adopted: Calculate resource arrival time for each priority zone , output real-time decision-making guidance information, where represents the path length, Indicates the current rescue vehicle speed. represents the average speed of the area, Indicates the waiting time.

[0012] As a further solution of the present invention, the steps for obtaining the rescue action plan are: Based on the real-time decision guidance information, the urgency and resource availability of the rescue mission are analyzed, and a preliminary matching of rescue teams and equipment is performed to obtain a preliminary rescue mission matching plan; According to the preliminary rescue mission matching plan, rescue route planning is carried out using the formula: Calculate the shortest effective time from each rescue point to the resource assembly point , get the optimal rescue route plan, where Represents the priority of the rescue point. Represents the emergency demand of the rescue point, Represents the distance between the resource and the rescue point; The optimal rescue route plan is used to activate the required rescue equipment and personnel, formulate a rescue action plan, and push it to the execution unit in real time to generate a rescue action plan.

[0013] An AI-based intelligent supervision method for emergency rescue scenarios includes the following steps: S1: Install video and audio equipment to capture real-time images and sounds of the rescue scene, synchronously collect GPS location data and environmental sensor data of the rescue scene in real time, perform time synchronization processing, and obtain the original data record of the scene; S2: Based on the original data record of the scene, perform synchronous processing of image data and sound data, extract image edges by edge detection, identify the fundamental frequency and tone of the sound by spectrum analysis, identify the visual and auditory information of the scene, and generate visual and auditory data analysis results; S3: extracting timestamps based on the visual and auditory data analysis results, comparing image sequences and sound sequences, classifying features, analyzing the motion trajectories of objects in the rescue scene and the changing patterns of sounds, and constructing key morphological feature data; S4: Based on the key morphological feature data, coordinate mapping is performed and time calibration is completed, coordinate and time series are synchronized, the spatial position of the key morphological feature is calculated and mapped, and a three-dimensional spatial mapping result of the accident range is generated; S5: Based on the three-dimensional spatial mapping result of the accident range, identify the key affected areas in the accident area, perform regional priority sorting, compare the emergency demand and inventory, combine the response data of the rescue path, calculate the resource arrival time of each priority area, and output real-time decision guidance information; S6: Based on the real-time decision guidance information, analyze the urgency and resource availability of the rescue mission, plan the rescue route, activate the required rescue equipment and personnel, and push it to the execution unit in real time to generate a rescue action plan.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, by performing detailed color depth analysis, edge detection, waveform analysis and tone recognition on the collected images and audio, the dynamic change characteristics in the rescue scene can be accurately identified. Advanced data processing makes the identification of changes at critical moments more accurate and the construction of key morphological feature data more comprehensive. Through the analysis of three-dimensional spatial mapping and time series information, it is possible to accurately assess the scope of the accident and quickly determine resource requirements, greatly optimize the allocation of rescue resources and response time, and the comprehensive application of technical means can significantly improve rescue efficiency, reduce the risk of loss of life and property, and improve the safety and accuracy of rescue operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is a system flow chart of the present invention; Figure 2 A flowchart for obtaining scene original data records of the present invention; Figure 3The figure is a flow chart of image and sound processing of the present invention; Figure 4 It is a flowchart of obtaining key morphological feature data of the present invention; Figure 5 It is a three-dimensional space mapping flow chart of the present invention; Figure 6 A flowchart for obtaining real-time decision-making guidance information of the present invention; Figure 7 The figure is a flowchart for obtaining the rescue action plan of the present invention. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0017] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0018] See also Figure 1 , an AI-based intelligent supervision system for emergency rescue scenarios includes: The scene data acquisition module captures images and sounds through the rescue scene video and audio equipment, synchronously collects geographic and climate data from GPS and environmental sensors, synchronizes the data in time, and outputs the original scene data records; The data feature analysis module performs color depth and edge detection processing on images, waveform analysis and tone recognition on sounds based on the original scene data records. It also compares the timestamp data of image sequences and sound sequences synchronously based on the image and sound processing results, identifies dynamic change features, and constructs key morphological feature data. The intelligent decision support module analyzes the spatial position and time series information in the key morphological feature data, performs three-dimensional spatial mapping of the accident range, evaluates the rescue priority in real time based on the mapping results, determines resource requirements, optimizes resource allocation and response time, and outputs real-time decision guidance information; The rescue action execution module adjusts the allocation of rescue resources, specifies the optimal rescue route, activates rescue equipment and personnel, and formulates and pushes rescue action plans based on real-time decision-making guidance information.

[0019] The original data records of the scene include video images, on-site audio, geographic information and climate information. The key morphological feature data include image color depth analysis results, image edge detection records, sound waveform analysis results, sound tone recognition results and timestamp matching results. The real-time decision-making guidance information includes three-dimensional spatial position information, time series analysis results and accident range mapping results. The rescue action plan includes resource allocation, route designation, equipment activation and personnel scheduling.

[0020] See also Figure 2 , the steps to obtain the original data records of the scene are: Install video and audio equipment at the rescue site to capture real-time images and sounds of the rescue site and generate audio-visual data sets; Install video and audio equipment to capture real-time images and sounds at the scene and generate audio-visual data sets, which will serve as first-hand information for rescue operations. The installation of video and audio equipment should choose angles that can cover the most critical areas of the rescue scene, and ensure that the equipment has sufficient recording time and storage space. The installation location and angle of the equipment should be tested in advance through simulated rescue scenarios to ensure that all key activities in emergency situations can be effectively captured. At the same time, the equipment also needs to have stable power support and an emergency backup power system to ensure that it can still work normally under the influence of extreme climate or other unpredictable environmental factors.

[0021] Collect GPS location data and environmental sensor data from the rescue site in real time, build a comprehensive geographic and climate data set, and perform preliminary data association with the audio-visual data set to obtain a correlated data set; In rescue operations, synchronously collected GPS and environmental sensor data are crucial. Specifically, the GPS module needs to capture geographic coordinates, including latitude and longitude, in real time to determine the precise location of the rescue team, thereby guiding the planning of rescue routes and the delivery of rescue materials. In addition, environmental sensors record meteorological conditions including temperature, humidity, wind speed, etc. in real time. These data are essential for assessing the on-site environment, safety risks, and formulating rescue plans. In order to achieve real-time synchronization of data, high-speed data transmission technology is used to ensure that all data is immediately transmitted back to the command center to support the command center in making quick decisions. At the same time, all collected data must be sorted and associated through the data integration platform, and various types of data must be cross-validated and analyzed to eliminate data anomalies and errors. Through a rigorous data processing process, the command center can obtain a comprehensive, real-time updated on-site situation map, which greatly improves the efficiency and safety of rescue. In addition, all equipment is equipped with an emergency power supply system to ensure that it can continue to operate in extreme weather or other unforeseen circumstances, so as to avoid interruptions in data collection that affect the efficiency and safety of rescue operations.

[0022] To synchronize the time of the associated data sets, we use the formula: Calculate the average value of each data type at the same time point , get the original scene data record, where Representative Class data in time The observed value of is the total number of data categories; In data processing, all inputs (such as video images, audio waveforms, GPS coordinates, temperature readings, etc.) need to be converted into digital format. , there are three data types, whose data values ​​are 20, 30 and 50 respectively, then the formula is calculated as: Indicates at time The average value for all data types is 33.33, which is helpful in understanding the overall status of the data collected at that point in time.

[0023] See also Figure 3 , the image and sound processing steps are: Based on the original scene data record, image data and sound data are extracted, image preprocessing is performed, preliminary waveforms of sound data are obtained synchronously, and image preprocessing results and preliminary sound waveform data are generated; Image data and sound data are extracted from the original data set. First, high-pass filtering is performed on the image data to remove low-frequency noise. Standard image processing libraries (such as OpenCV) are used to sharpen the image and enhance the edge information of the image for subsequent color depth analysis and edge detection. For sound data, the window function method is used to segment the sound signal into several small blocks. Fast Fourier transform (FFT) is applied to each block to obtain frequency domain data. This method is used to analyze the changes in fundamental frequency and pitch. The formula is: ,in, represents the sound signal in the time domain, represents the sound signal in the frequency domain, is the frequency, This step not only cleans the signal, but also lays the foundation for subsequent advanced analysis, generating denoised image and sound data ready for more in-depth feature extraction.

[0024] Based on the image preprocessing results and preliminary sound waveform data, the image preprocessing results are subjected to color depth analysis, the image edges are extracted using edge detection, the waveform of the sound data is synchronously analyzed, the fundamental frequency and pitch of the sound are identified through spectrum analysis, and image depth edge data and sound spectrum analysis results are generated; When performing color depth analysis on an image, color histogram equalization is used to enhance the contrast of the image, making the colors of different depths in the image more obvious, which is convenient for subsequent edge detection; edge detection uses the Sobel operator to calculate the amplitude and direction of the image gradient. The formula is: ,in and are the gradients of the image in the horizontal and vertical directions, Represents the edge strength at a certain point in the image. Through this calculation, the edge information in the image can be accurately located. For sound data, its spectral characteristics are further analyzed, and the Hanning window is used to reduce spectrum leakage. The calculated frequency response helps determine the characteristic frequency of the sound, so that different sounds can be classified or labeled. This not only improves the accuracy of data analysis, but also makes subsequent data integration more accurate. The generated image depth and edge data and sound spectrum analysis results will be used for the next step of synchronous processing.

[0025] Integrate image depth edge data and sound spectrum analysis results, compare the features of image and sound data, identify the visual and auditory information of the scene, and generate visual and auditory data analysis results; When integrating the processing results of image and sound, a synchronization method based on time series analysis is used to align the timestamps of the edge detection results of the image and the spectrum data of the sound. The formula is: ,in is the time series of image edge data, is the time series of sound spectrum data, This synchronization not only takes into account the temporal matching of visual and auditory information, but also increases the complexity and accuracy of multimodal data processing. The final visual and auditory data analysis results provide strong data support for subsequent applications, such as dynamic event detection and scene understanding in complex environments.

[0026] See also Figure 4 , the steps to obtain key morphological feature data are: Extract timestamps based on the results of visual and auditory data analysis, confirm that the visual data and auditory data are aligned on the time axis, and generate image and sound sequence data with synchronized timestamps; Execute the extraction of timestamp data for image and sound sequences from synchronized data. The process first involves the use of high-precision time synchronization, such as using GPS or Network Time Protocol (NTP) to calibrate the equipment to ensure the consistency of time for all recording devices. Next, each frame of the image and each sound sample is timestamped. This timestamp is automatically added during the data acquisition process to ensure that each data point has an accurate time reference. In addition, the time data needs to be preprocessed, including filtering outliers and filling in missing timestamps to enhance the integrity and availability of the data. The work at this stage provides a solid foundation for subsequent time series analysis, so that further image and sound data analysis can accurately correspond to specific time points, thereby ensuring the accuracy and effectiveness of subsequent analysis.

[0027] Based on the image and sound sequence data with synchronized timestamps, the image sequence and the sound sequence are compared using the formula, Calculate the edge change of image data and the spectrum change value of sound data at each time point , generate dynamic change feature data of images and sounds, where represents the edge strength of the image sequence, represents the frequency response value of the sound sequence, is the minimum cumulative distance of the previous state; like and , =10, then: The results show that the image and sound are The time difference value is 5.92, which helps us understand the changing characteristics of the two data types at this moment. It can identify the characteristics of dynamic changes and use them for further analysis.

[0028] Analyze the dynamic change feature data of images and sounds, classify the features, analyze the movement trajectory of objects in the rescue scene and the change pattern of sounds, and construct key morphological feature data; After dynamic feature recognition, support vector machine (SVM) is used to classify and analyze the features of image and sound data. First, the training set for machine learning is extracted from the dynamic change feature data obtained in the previous step. The data includes but is not limited to the edge change information of the image and the frequency change information of the sound, and each data point is labeled to correspond to its role in the dynamic sequence. Next, a suitable kernel function (such as radial basis function or polynomial kernel function) is selected to transform the original feature space so that the nonlinearly separable data can be separable in a high-dimensional space. The optimal regularization parameter C and kernel function parameters are selected by grid search and cross-validation to balance the complexity of the model and the degree of fit of the training data. In addition, in order to improve the generalization ability of the model, a series of model validation tests are implemented, and performance tests are performed through control data sets in laboratory environments and actual scene data to evaluate the performance of the model on unknown data. Finally, a machine learning model that can accurately identify and predict dynamic morphological features in image and sound data is generated, providing strong data support for further practical applications.

[0029] See also Figure 5 , the mapping steps of three-dimensional space are: Based on the key morphological feature data, the spatial coordinates and timestamps of each feature are extracted, coordinate mapping is performed and time calibration is completed, the spatial position and time series information are integrated, and a preliminary data set is generated; Based on the key morphological feature data of the accident scene, the spatial coordinates and timestamps of each feature are extracted. The spatial coordinate information is first collected from the monitoring equipment and on-site sensors. The spatial position data of each monitoring point is parsed through the data acquisition module, and all monitoring positions are converted into standardized three-dimensional coordinates. The spatial distribution of the feature points is initially obtained. At the same time, the occurrence time and duration of each feature point are obtained, and the time data is sorted in chronological order to ensure the consistency of time calibration. After completing the preliminary collection, the coordinate point and timestamp data are checked to filter out points with large errors and deviations from the spatial distribution characteristics, remove abnormal data and update the preliminary data set. After the data cleaning is completed, the processed spatial coordinate data and timestamps are unified to form a preliminary data set, which contains the three-dimensional coordinates and corresponding timestamp information of each key point, ensuring that the time and space information in the subsequent analysis have good data consistency.

[0030] The coordinates and time series of the preliminary data set were synchronized, and the time and space points of key morphological features were matched and integrated to obtain a complete three-dimensional time and space data set of the accident; The time and space information generated in the preliminary data set are processed synchronously. First, the center point of the accident is selected as the coordinate reference in the spatial information. The projection coordinate transformation is performed on each spatial coordinate point in the preliminary data set. Each feature point is mapped to a position consistent with the reference coordinate system of the accident area, so that the spatial coordinate point has an accurate spatial reference in the same coordinate system, thereby forming standard spatial position data. At the same time, the time series information is calibrated according to the time reference of the accident. Each timestamp data point is synchronously adjusted through the time deviation calibration method to correct the deviation from the time of the accident to ensure the consistency of the time sequence. After the calibration is completed, the time of all feature points is calibrated on the same time axis of the accident, completing the dual synchronization of time and space. The three-dimensional time and space data set formed after the final integration can be used for the comprehensive analysis of the spatial distribution and time evolution of subsequent accidents.

[0031] According to the complete accident three-dimensional time-space data set, the formula is used: Calculate the The spatial location of key morphological features , generate the three-dimensional spatial mapping result of the accident range, where For the The coordinates of the locations, is the base coordinate of the center point, is the timestamp, Indicates the number of positions; The following values ​​are collected: Coordinate point 1 , timestamp .

[0032] Coordinate point 2 , timestamp .

[0033] Coordinate point 3 , timestamp .

[0034] Benchmark .

[0035] Calculate the distance from each point to the reference point: Multiply the distance of each point by its timestamp weight: After summing, calculate the weighted average: The results show that the calculated weighted average distance That is the The three-dimensional spatial mapping position of the key morphological features is used to identify the spatial distribution position within the accident range.

[0036] See also Figure 6 ,The steps for obtaining real-time decision-making guidance information are: Based on the three-dimensional spatial mapping results of the accident range, the key affected areas of the accident area are identified, and the spatial distribution characteristics and damage degree of each area are analyzed item by item. The formula is used. Calculate the impact index for each region , compare the impact index of each region and get the regional priority ranking table, where, Indicates the damaged area of ​​the region, Indicates the density of people affected in the area, Indicates the start time of the event. Indicates the current time of the event; The damaged area of ​​the representative area, in square meters, is obtained through data collection. square meters; Represents the density of people affected in the area, in units of people per square meter. People / square meter; and Represents the start time of the event and the current time, in hours. The event occurs at Hours, the current time is Hours. The calculation process is as follows: The results show that the current impact index for , indicating that at the current moment, the impact on the area is no longer urgent, probably because some time has passed since the incident occurred and emergency rescue and treatment have been partially or completely completed.

[0037] According to the regional priority table, call the real-time resource inventory information, compare the emergency demand and inventory, and get the actual demand gap in each region. Use the formula: Compute resource allocation priority , generate a resource configuration plan, where represents the impact index of the region, Indicates the emergency resource demand of the region, Indicates the current available resource inventory; The impact index of the representative region is (derived from the above calculation results); represents the emergency resource demand, unit; Represents the current available resource inventory, which is Unit. The calculation process is: The results show that the current resource allocation priority for , which means that based on the current impact index and resource needs, there is no need to allocate more resources to this area for the time being.

[0038] Based on the resource allocation plan and combined with the response data of the rescue path, the formula is adopted. Calculate resource arrival time for each priority zone , output real-time decision-making guidance information, where represents the path length, Indicates the current rescue vehicle speed. represents the average speed of the area, Indicates waiting time; represents the path length, which is kilometer; Indicates the current rescue vehicle speed, km / h; is the average speed of the area, km / h; represents the waiting time, Minutes, converted to hours Hours. The calculation process is: The results showed that rescue vehicles were expected to arrive at approx. Arrive at the accident scene within hours and instruct rescue teams on how to adjust travel speed and route to optimize response time.

[0039] See also Figure 7 , the steps to obtain the rescue action plan are: Based on real-time decision-making guidance information, analyze the urgency of the rescue mission and the availability of resources, make a preliminary match between the rescue team and equipment, and obtain a preliminary rescue mission matching plan; Based on real-time decision-making guidance information, the urgency of the rescue mission and the availability of resources are analyzed, and preliminary matching of rescue teams and equipment is carried out. The process includes comprehensive consideration of the current status of all rescue resources and mission requirements, and the use of resource matching algorithms to ensure that each resource is reasonably allocated to where it is most needed. The process of refining resource allocation also needs to consider the transportation time and cost of each resource, and prioritize each resource to ensure that the most urgent needs can be met to the greatest extent possible when resources are limited. Ultimately, this strategy ensures that the resources for the rescue operation are optimally configured.

[0040] According to the preliminary rescue mission matching plan, rescue route planning is carried out using the formula: Calculate the shortest effective time from each rescue point to the resource assembly point , get the optimal rescue route plan, where Represents the priority of the rescue point. Represents the emergency demand of the rescue point, Represents the distance between the resource and the rescue point; The priority of a rescue point 5 (based on the severity of the rescue mission, 1-10 indicates increasing severity, 5 indicates the current rescue point output is medium severity), emergency demand 20 units (based on site assessment demand), distance is 2 km. According to the formula, we first calculate the square root of the product of demand and priority, that is , and then divide it by the distance to get Therefore, the shortest effective time is 5 time units. This result shows that under given priority and emergency demand, resources should reach the rescue point at a minimum speed of 5 time units, which provides a calculation basis for resource allocation and rescue route planning.

[0041] Use the optimal rescue route plan to activate the required rescue equipment and personnel, formulate a rescue action plan, and push it to the execution unit in real time to generate a rescue action plan; Using the optimal rescue route plan, activate the required rescue equipment and personnel, formulate the final action plan, and push it to the execution unit in real time to generate a rescue action execution plan. This step ensures the real-time update and maximum efficiency of all plans and resources. The rescue command center transmits the action plan to all relevant personnel and equipment in real time through the advanced communication system, including detailed schedules and staffing, to ensure that each team can reach the designated location in the shortest time. At the same time, the system can also provide real-time feedback on the execution status and adjust the plan to deal with possible emergencies. The implementation of this process ultimately forms a rescue operation network that is responsive and efficient.

[0042] An AI-based intelligent supervision method for emergency rescue scenarios includes the following steps: S1: Install video and audio equipment to capture real-time images and sounds of the rescue scene, synchronously collect GPS location data and environmental sensor data of the rescue scene in real time, perform time synchronization processing, and obtain the original data record of the scene; S2: Based on the original scene data record, perform synchronous processing of image data and sound data, use edge detection to extract image edges, identify the fundamental frequency and tone of the sound through spectrum analysis, identify the visual and auditory information of the scene, and generate visual and auditory data analysis results; S3: Based on the results of visual and auditory data analysis, extract timestamps, compare image sequences and sound sequences, classify features, analyze the movement trajectory of objects in the rescue scene and the change pattern of sound, and construct key morphological feature data; S4: Based on the key morphological feature data, coordinate mapping is performed and time calibration is completed, coordinate and time series are synchronized, the spatial position of the key morphological features is calculated and mapped, and a three-dimensional spatial mapping result of the accident range is generated; S5: Based on the three-dimensional spatial mapping results of the accident range, identify the key affected areas in the accident area, prioritize the areas, compare the emergency demand with the inventory, combine the response data of the rescue path, calculate the resource arrival time for each priority area, and output real-time decision-making guidance information; S6: Based on real-time decision-making guidance information, analyze the urgency and resource availability of the rescue mission, plan the rescue route, activate the required rescue equipment and personnel, and push it to the execution unit in real time to generate a rescue action plan.

[0043] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. An AI-based intelligent monitoring system for emergency rescue scenarios, characterized by: The system comprises: The scene data acquisition module captures images and sounds through the rescue scene video and audio equipment, synchronously collects geographic and climate data from GPS and environmental sensors, synchronizes the data in time, and outputs the original scene data records; A data feature analysis module performs color depth and edge detection processing on the image, waveform analysis and tone recognition on the sound based on the original data record of the scene, and synchronously compares the timestamp data of the image sequence and the sound sequence according to the image and sound processing results, identifies dynamic change features, and constructs key morphological feature data; An intelligent decision support module analyzes the spatial position and time series information in the key morphological feature data, performs three-dimensional spatial mapping of the accident range, evaluates rescue priorities in real time based on the mapping results, determines resource requirements, optimizes resource allocation and response time, and outputs real-time decision guidance information; The rescue action execution module adjusts the rescue resource allocation, specifies the optimal rescue route, activates rescue equipment and personnel, and formulates and pushes the rescue action plan according to the real-time decision guidance information.

2. The AI-based emergency rescue scene intelligent supervision system according to claim 1 is characterized by: The steps for obtaining the original scene data record are: Install video and audio equipment at the rescue site to capture real-time images and sounds of the rescue site and generate audio-visual data sets; Collect GPS location data and environmental sensor data of the rescue site in real time, construct a comprehensive geographical and climatic data set, and perform preliminary data association with the audio-visual data set to obtain an associated data set; The associated data set is time synchronized using the formula: Calculate the average value of each data type at the same time point , get the original scene data record, where Representative Class data in time The observed value of is the total number of data categories.

3. The AI-based emergency rescue scene intelligent supervision system according to claim 2 is characterized by: The image and sound processing steps are as follows: Based on the original data record of the scene, extract image data and sound data, perform image preprocessing, synchronously obtain preliminary waveforms of sound data, and generate image preprocessing results and preliminary sound waveform data; Based on the image preprocessing result and the preliminary sound waveform data, color depth analysis is performed on the image preprocessing result, edge detection is used to extract image edges, the waveform of the sound data is synchronously analyzed, the fundamental frequency and pitch of the sound are identified through spectrum analysis, and image depth edge data and sound spectrum analysis results are generated; The image depth edge data and the sound spectrum analysis results are integrated, the features of the image and sound data are compared, the visual and auditory information of the scene is identified, and the visual and auditory data analysis results are generated.

4. The AI-based emergency rescue scene intelligent supervision system according to claim 3 is characterized by: The steps to obtain key morphological feature data are as follows: Extracting timestamps based on the visual and auditory data analysis results, confirming that the visual data and the auditory data are aligned on the time axis, and generating image and sound sequence data with synchronized timestamps; Based on the image and sound sequence data of the synchronization timestamp, the image sequence and the sound sequence are compared, and the formula is used, Calculate the edge change of image data and the spectrum change value of sound data at each time point , generate dynamic change feature data of images and sounds, where represents the edge strength of the image sequence, represents the frequency response value of the sound sequence, is the minimum cumulative distance of the previous state; The dynamic change feature data of the image and sound are analyzed, the features are classified, the motion trajectory of objects in the rescue scene and the change pattern of the sound are analyzed, and key morphological feature data are constructed.

5. The AI-based emergency rescue scene intelligent supervision system according to claim 4 is characterized in that: The mapping steps of three-dimensional space are: Based on the key morphological feature data, the spatial coordinates and timestamps of each feature are extracted, coordinate mapping is performed and time calibration is completed, and the spatial position and time series information are integrated to generate a preliminary data set; Synchronizing the coordinates and time series of the preliminary data set, matching and integrating the time and space points of key morphological features, and obtaining a complete three-dimensional time-space data set of the accident; According to the complete accident three-dimensional time-space data set, the formula is used: Calculate the The spatial location of key morphological features , generate the three-dimensional spatial mapping result of the accident range, where For the The coordinates of the locations, is the base coordinate of the center point, is the timestamp, Indicates the number of positions.

6. The AI-based emergency rescue scene intelligent supervision system according to claim 5 is characterized by: The steps for obtaining real-time decision-making guidance information are as follows: Based on the three-dimensional spatial mapping results of the accident range, the key affected areas of the accident area are identified, and the spatial distribution characteristics and damage degree of each area are analyzed item by item. The formula is used. Calculate the impact index for each region , compare the impact index of each region and get the regional priority ranking table, where, Indicates the damaged area of ​​the region, Indicates the density of people affected in the area, Indicates the start time of the event. Indicates the current time of the event; According to the regional priority ranking table, call the real-time resource inventory information, compare the emergency demand and inventory, and get the actual demand gap of each region. Use the formula: Compute resource allocation priority , generate a resource configuration plan, where represents the impact index of the region, Indicates the emergency resource demand of the region, Indicates the current available resource inventory; Based on the resource allocation scheme and combined with the response data of the rescue path, the formula is adopted: Calculate resource arrival time for each priority zone , output real-time decision-making guidance information, where represents the path length, Indicates the current rescue vehicle speed. represents the average speed of the area, Indicates the waiting time.

7. The AI-based emergency rescue scene intelligent supervision system according to claim 6 is characterized by: The steps to obtain a rescue action plan are: Based on the real-time decision guidance information, the urgency and resource availability of the rescue mission are analyzed, and a preliminary matching of rescue teams and equipment is performed to obtain a preliminary rescue mission matching plan; According to the preliminary rescue mission matching plan, rescue route planning is carried out using the formula: Calculate the shortest effective time from each rescue point to the resource assembly point , get the optimal rescue route plan, where Represents the priority of the rescue point. Represents the emergency demand of the rescue point, Represents the distance between the resource and the rescue point; The optimal rescue route plan is used to activate the required rescue equipment and personnel, formulate a rescue action plan, and push it to the execution unit in real time to generate a rescue action plan.

8. An AI-based intelligent supervision method for emergency rescue scenarios, characterized in that: The AI-based emergency rescue scene intelligent supervision system according to any one of claims 1 to 7 comprises the following steps: Install video and audio equipment to capture real-time images and sounds of the rescue scene, collect GPS location data and environmental sensor data of the rescue scene in real time, perform time synchronization processing, and obtain original scene data records; Based on the original data record of the scene, synchronous processing of image data and sound data is performed, edge detection is used to extract image edges, fundamental frequency and pitch of the sound are identified through spectrum analysis, visual and auditory information of the scene is identified, and visual and auditory data analysis results are generated; According to the visual and auditory data analysis results, extract timestamps, compare image sequences and sound sequences, classify features, analyze the movement trajectories of objects in the rescue scene and the change patterns of sounds, and construct key morphological feature data; Based on the key morphological feature data, coordinate mapping is performed and time calibration is completed, coordinate and time series synchronization is performed, the spatial position of the key morphological features is calculated and mapped, and a three-dimensional spatial mapping result of the accident range is generated; Based on the three-dimensional spatial mapping results of the accident range, identify the key affected areas in the accident area, perform regional priority sorting, compare emergency demand and inventory, combine the response data of the rescue path, calculate the resource arrival time of each priority area, and output real-time decision-making guidance information; Based on the real-time decision-making guidance information, the urgency and resource availability of the rescue mission are analyzed, the rescue route is planned, the required rescue equipment and personnel are activated, and the information is pushed to the execution unit in real time to generate a rescue action plan.

Images