Intelligent triggering system for emergency broadcast based on multi-source seismic data fusion

Abstract

The present invention provides an emergency broadcast intelligent triggering system based on multi-source seismic data fusion, which relates to the field of earthquake monitoring and early warning technology, including: a data receiving module for receiving multi-source earthquake early warning information in real time, including source parameter data, strong vibration observation data and real-time early warning parameters, and outputting standardized multi-source seismic data through a built-in communication protocol; a localized calculation module for performing real-time analysis of the standardized multi-source seismic data, extracting P-wave initial parameters, seismic acceleration prediction values and source rupture characteristics, dynamically calibrating the analysis results in combination with the measured data of the built-in seismometer, and outputting a multi-dimensional seismic data stream verified by the seismometer. The present invention realizes the generation and efficient dissemination of earthquake early warning information through multi-source seismic data fusion, real-time analysis and calibration, dynamic risk modeling and intelligent broadcast scheduling, improves the timeliness and pertinence of emergency response, and reduces the risk of earthquake disasters.

Description

Technical Field

[0001] The present invention relates to the technical field of earthquake monitoring and early warning, and in particular to an emergency broadcast intelligent triggering system based on multi-source earthquake data fusion. Background Art

[0002] Existing technologies face serious limitations due to the conflict between static triggering mechanisms and the dynamic evolution of earthquakes. Traditional systems rely on pre-set, fixed trigger conditions (such as a single magnitude or intensity threshold) and are unable to dynamically adjust to the real-time changes in the earthquake rupture process. This leads to the following key issues:

[0003] During an earthquake rupture, source parameters (such as magnitude, epicenter location, and rupture direction) may change as the rupture propagates. For example, in the early stages of an earthquake, due to limited station data, the system may underestimate the magnitude. If the subsequent rupture propagation causes the actual magnitude to increase significantly, traditional systems, unable to dynamically adjust the trigger threshold, will continue to use the initial low threshold, resulting in high-risk areas not being covered in a timely manner. Conversely, if the initial magnitude is misjudged as too high, the system will continue to send redundant warnings to low-risk areas, causing a decline in public trust.

[0004] Existing broadcast systems often generate warnings based on fixed templates, without dynamic adaptation based on local building structural characteristics, population density, and real-time intensity data. For example, earthquake mitigation strategies for high-rise buildings differ significantly from those for low-rise brick-concrete structures, yet AI technology cannot automatically adapt these guidance to specific needs. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide an intelligent triggering system for emergency broadcasting based on multi-source seismic data fusion, which automatically generates risk avoidance prompts adapted to the target area and improves the efficiency of information delivery.

[0006] In order to solve the above technical problems, the technical solutions of the present invention are as follows:

[0007] First, the intelligent triggering system for emergency broadcasting based on multi-source seismic data fusion includes:

[0008] The data receiving module is used to receive multi-source earthquake early warning information in real time, including source parameter data, strong motion observation data and real-time early warning parameters, and output standardized multi-source earthquake data through the built-in communication protocol;

[0009] The localized calculation module is used to perform real-time analysis of standardized multi-source seismic data, extracting initial P-wave parameters, predicted ground acceleration values, and source rupture characteristics. The module dynamically calibrates the analysis results based on the measured data from the built-in seismometer and outputs a multi-dimensional seismic data stream verified by the seismometer.

[0010] The data processing module is used to input the multi-dimensional seismic data stream verified by the seismometer into a preset geographic grid coordinate system. By jointly analyzing P-wave parameters, predicted ground acceleration values, and source rupture characteristics, it generates a dynamically updated magnitude-epicenter location probability distribution matrix, calculates dynamic trigger thresholds, and generates graded warning trigger instructions.

[0011] The emergency broadcast module is used to match the earthquake avoidance guidance strategy of the target area based on the graded warning trigger instructions and the emergency response plan library preset by the AI chip, and generate multimedia warning information including earthquake countdown parameters, intensity distribution prediction and AI-optimized risk avoidance prompts;

[0012] The collaborative broadcast control module is used to distribute the multimedia warning information to voice broadcast terminals, digital display terminals and public information screens through the IP network protocol, realize the priority transmission of voice broadcasts and layered rendering of dynamic information, and optimize the AI chip's plan matching logic and dynamic trigger threshold judgment conditions based on real-time feedback from terminal response data.

[0013] Furthermore, the standardized multi-source seismic data is analyzed in real time to extract the initial P-wave parameters, predicted ground acceleration values, and source rupture characteristics. The analysis results are dynamically calibrated with the measured data from the built-in seismometer, and a multi-dimensional seismic data stream verified by the seismometer is output, including:

[0014] The seismic waveform data is scanned in real time using the sliding window analysis method. Based on the preset waveform slope threshold and energy change rate threshold, the initial moment of the P wave is identified and the initial parameters of the P wave are extracted.

[0015] Based on the three-dimensional coordinates of the earthquake source and the spatial distribution of geographic grid cells in the initial P-wave parameters, combined with the physical laws of earthquake motion attenuation and historical statistical relationships, the basic predicted value of earthquake acceleration for each grid cell is calculated.

[0016] The localized calculation module uses a built-in intensity meter to obtain real-time ground acceleration data. The basic prediction value is dynamically compared with the measured data. If the deviation exceeds the preset error threshold, the magnitude-distance attenuation parameter and geological medium correction factor are corrected to generate a calibrated ground acceleration prediction value range.

[0017] According to the spatial distribution characteristics of the calibrated earthquake acceleration prediction value interval and the temporal variation law of the initial P-wave parameters, the earthquake source rupture direction, rupture velocity and rupture length parameters are extracted.

[0018] The calibrated earthquake acceleration prediction value interval, P-wave initial parameters and source rupture characteristics are integrated, and the intensity meter verification mark is added to generate a multi-dimensional seismic data stream.

[0019] Furthermore, based on the three-dimensional coordinates of the earthquake source in the initial P-wave parameters and the spatial distribution of geographic grid cells, combined with the physical laws of earthquake attenuation and historical statistical relationships, the basic predicted value of earthquake acceleration for each grid cell is calculated, including:

[0020] The length of the seismic wave propagation path is calculated based on the spatial distance between the three-dimensional coordinates of the earthquake source and the center point of the geographic grid unit, and the geological density and elastic modulus parameters are extracted by combining the distribution data of the geological medium type corresponding to the path;

[0021] Based on geological density and elastic modulus parameters, combined with the physical laws of earthquake motion attenuation, a magnitude-distance attenuation relationship model is constructed. The magnitude-distance attenuation relationship model defines the logarithmic attenuation characteristics of the acceleration attenuation baseline value under different earthquake magnitudes as a function of the propagation path length.

[0022] Based on the historical earthquake motion statistical data set, the attenuation benchmark value is statistically calibrated to generate a mapping table of acceleration attenuation benchmark values under different magnitude and distance combinations;

[0023] The acceleration attenuation benchmark value mapping table is bound to the spatial distribution of the geographic grid coordinate system to generate the basic predicted value of seismic acceleration for each grid cell.

[0024] Furthermore, the localized calculation module uses a built-in intensity meter to obtain real-time earthquake acceleration data. The basic prediction value is dynamically compared with the measured data. If the deviation exceeds the preset error threshold, the magnitude-distance attenuation parameter and geological medium correction factor are corrected to generate a calibrated earthquake acceleration prediction value range, including:

[0025] Based on the spatial distribution of geographic grid cells, the real-time ground acceleration measured data are mapped to the corresponding grid cells;

[0026] The predicted earthquake acceleration value of each grid cell is compared point by point with the measured data of the same grid cell, and the predicted deviation rate is calculated. If the predicted deviation rate exceeds the preset error threshold, the magnitude-distance attenuation parameter and geological medium correction factor are adjusted according to the deviation direction to obtain the corrected attenuation parameter and correction factor.

[0027] Based on the corrected attenuation parameters and correction coefficients, the predicted seismic acceleration values are recalculated to generate a calibrated predicted value interval containing a confidence interval range.

[0028] Furthermore, the multi-dimensional seismic data stream verified by the intensity meter is input into a preset geographic grid coordinate system. By jointly analyzing P-wave parameters, predicted ground acceleration values, and source rupture characteristics, a dynamically updated magnitude-epicenter location probability distribution matrix is generated. The dynamic trigger threshold is calculated and a graded warning trigger instruction is generated, including:

[0029] Based on the geographic grid coordinate system, the distance of the seismic wave propagation path from the earthquake source to each grid cell is calculated, and the geological density and elastic modulus data are extracted in combination with the geological medium distribution parameters corresponding to the path;

[0030] Based on the calibrated earthquake acceleration prediction value interval and the source rupture direction and expansion velocity parameters, the spatiotemporal constraints of the magnitude and epicenter location are constructed. The spatiotemporal constraints include the magnitude upper limit, rupture expansion range, and time window.

[0031] Based on geological parameters and spatiotemporal constraints, the Monte Carlo sampling method is used to iteratively integrate the earthquake source probability distribution, the prior probability of earthquake motion intensity, and the spatiotemporal constraints to update the magnitude-epicenter location joint probability density of each grid cell and generate a dynamically updated magnitude-epicenter location probability distribution matrix.

[0032] From the dynamically updated magnitude-epicenter location probability distribution matrix, a set of grid cells whose magnitude probability exceeds a preset intensity threshold is extracted and marked as a high-probability area;

[0033] Calculate the regional comprehensive risk coefficient based on the population density level and building seismic resistance level parameters corresponding to the high-probability areas in the external database;

[0034] Real-time monitoring of dynamic trigger thresholds and magnitude probability growth rates. When the trigger threshold reaches the preset critical value and the growth rate exceeds the growth rate threshold, a graded warning trigger instruction containing the target area code and warning level is generated.

[0035] Furthermore, based on geological parameters and spatiotemporal constraints, the Monte Carlo sampling method is used to iteratively integrate the earthquake source probability distribution, the prior probability of earthquake motion intensity, and the spatiotemporal constraints. The magnitude-epicenter location joint probability density of each grid cell is updated to generate a dynamically updated magnitude-epicenter location probability distribution matrix, including:

[0036] According to the magnitude upper limit and rupture extension range in the temporal and spatial constraints, the random generation interval of the Monte Carlo sampling parameters is defined;

[0037] Random sampling is performed within the parameter random generation interval to generate multiple sets of candidate parameter combinations of magnitude, epicenter location and seismic intensity;

[0038] For each candidate parameter combination, calculate the probability value in the earthquake source probability distribution, the probability value of the earthquake intensity, and the compliance weight under the spatiotemporal constraints;

[0039] The source probability value, the earthquake intensity probability value and the compliance weight are integrated to obtain the joint probability density of the corresponding parameter combination;

[0040] The joint probability density of all candidate parameter combinations is statistically analyzed, the mean probability density in each grid cell is calculated, and the mean is mapped to the corresponding grid cell in the geographic grid coordinate system. The value of the magnitude-epicenter location probability distribution matrix is updated. When the number of Monte Carlo sampling reaches the preset threshold, the updated magnitude-epicenter location probability distribution matrix is output.

[0041] Furthermore, based on the graded warning trigger instructions and the pre-set AI chip's emergency response plan library, the target area's earthquake avoidance guidance strategy is matched to generate multimedia warning information containing earthquake countdown parameters, intensity distribution predictions, and AI-optimized risk avoidance prompts, including:

[0042] According to the warning level in the graded warning trigger instruction, the corresponding response strategy template is matched from the emergency response plan library to determine the priority, language style and geographical coverage of the broadcast content;

[0043] Based on the seismic wave propagation velocity parameters and the updated magnitude-epicenter location probability distribution matrix, the propagation time of the seismic wave to each grid cell within the geographical coverage area is calculated, and regionalized countdown parameters synchronized with the time axis are generated;

[0044] The intensity distribution prediction data of high-probability areas are spatially correlated and matched with the road network topology and shelter coordinates in the external database to generate intensity distribution spatial data superimposed with road traffic status and shelter resource identification;

[0045] Based on the database of building structural characteristics in the target area, the AI chip analyzes the building type and seismic resistance. Combined with the intensity level predicted from the spatial intensity distribution data, it dynamically generates seismic avoidance guidance content adapted to the building type.

[0046] The regional countdown parameters, intensity distribution spatial data and vibration avoidance guidance content are synchronously integrated according to timestamps and spatial positions to generate multimedia warning information including voice broadcasts, text prompts and dynamic visualization elements.

[0047] Furthermore, the intensity distribution prediction data of high-probability areas are spatially correlated and matched with the road network topology and shelter coordinates in the external database to generate intensity distribution spatial data with superimposed road traffic status and shelter resource identification, including:

[0048] Based on the geographic grid coordinate system, the intensity distribution prediction data of high-probability areas, road network topology and shelter coordinates are spatially aligned to generate an associated dataset in a unified coordinate system.

[0049] Based on the road network topology in the linked dataset and combined with real-time traffic data, the traffic status of each road segment in the high-probability area is analyzed to generate road traffic status data containing road grade, congestion level, and availability identification. Based on the coordinates of the shelters in the linked dataset and combined with the capacity database, the real-time available capacity of each shelter is marked to generate shelter resource availability identification data.

[0050] The intensity distribution prediction data, road traffic status data and refuge resource availability identification data are fused according to the spatial location of the geographic grid unit to generate intensity distribution spatial data with superimposed road traffic status and refuge resource identification.

[0051] Furthermore, the multimedia warning information is distributed to voice broadcast terminals, digital display terminals, and public information screens via the IP network protocol, enabling priority transmission of voice broadcasts and layered rendering of dynamic information. Simultaneously, based on real-time feedback from terminal response data, the AI chip's plan matching logic and dynamic trigger threshold determination conditions are optimized, including:

[0052] Assign transmission priority to voice broadcast data through IP network protocol, and dynamically adjust the rendering level of text prompts and dynamic information elements based on the transmission status of voice broadcast;

[0053] Receive the broadcast completion status of voice broadcast terminals, rendering delay time of digital display terminals, and network load data of public information screens in real time to generate terminal response data sets;

[0054] Based on the broadcast delay and rendering efficiency data in the terminal response dataset, the matching weight of the emergency response plan in the AI chip is adjusted, the generation logic of the shock avoidance guidance strategy is optimized, and based on the network load fluctuation data in the terminal response dataset, the judgment conditions of the dynamic trigger threshold are dynamically corrected.

[0055] In a second aspect, a computer-readable storage medium stores a program, which implements the system when executed by a processor.

[0056] The above solution of the present invention includes at least the following beneficial effects:

[0057] Dynamic calibration and analysis of field data from the built-in intensity meter corrects for prediction errors caused by differences in geological conditions or propagation path errors (for example, calibration reduces acceleration prediction errors by over 30% in soft soil areas), ensuring the reliability of output data. By extracting initial P-wave parameters and source rupture characteristics, combined with a geographic grid coordinate system, a dynamically updated magnitude-epicenter location probability distribution matrix is generated, reflecting the spatiotemporal changes in earthquake risk in real time. For example, during the propagation of the source rupture, the matrix is updated every two seconds, buying valuable time for emergency response.

[0058] A comprehensive risk factor is calculated based on parameters such as regional population density and building seismic resistance ratings. Alert triggering conditions are dynamically adjusted based on the magnitude probability growth rate, reducing false alarms while prioritizing high-risk areas (e.g., the trigger threshold is lowered by 20% in densely populated areas). The emergency broadcast module automatically matches emergency plan templates based on the alert level and generates personalized evacuation strategies. For example, in areas with dense masonry structures, the priority is to "take shelter in the corner between load-bearing walls"; in high-rise building complexes, the emphasis is on "avoiding elevators and seeking shelter nearby," improving the scenario-specific adaptability of the guidance strategy. Dynamic alert information is generated, including an earthquake countdown, intensity distribution map, and evacuation resource identification. This information is presented multimodally through voice, text, and visual layers to meet the needs of diverse users (e.g., hearing-impaired individuals can receive alerts through visual information).

[0059] Voice broadcasts are assigned the highest transmission priority through the IP network protocol, ensuring that critical information such as "earthquake countdown" and "emergency evacuation instructions" are delivered even during network congestion (for example, voice data transmission delay is controlled within 500 milliseconds). Based on terminal feedback data (such as broadcast delay and screen load), the emergency plan matching logic is dynamically adjusted. For example, dynamic map rendering is automatically simplified in areas with poor network connectivity to enhance information accessibility. Trigger thresholds are also adjusted to provide early warning to address transmission delays. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 It is a schematic diagram of an intelligent triggering system for emergency broadcasting based on multi-source seismic data fusion provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0061] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0062] like Figure 1 As shown, an embodiment of the present invention proposes an emergency broadcast intelligent triggering system based on multi-source seismic data fusion, including:

[0063] The data receiving module is used to receive multi-source earthquake early warning information in real time, including source parameter data, strong motion observation data and real-time early warning parameters, and output standardized multi-source earthquake data through the built-in communication protocol;

[0064] The localized calculation module is used to perform real-time analysis of standardized multi-source seismic data, extracting initial P-wave parameters, predicted ground acceleration values, and source rupture characteristics. The module dynamically calibrates the analysis results based on the measured data from the built-in seismometer and outputs a multi-dimensional seismic data stream verified by the seismometer.

[0065] The data processing module is used to input the multi-dimensional seismic data stream verified by the seismometer into a preset geographic grid coordinate system. By jointly analyzing P-wave parameters, predicted ground acceleration values, and source rupture characteristics, it generates a dynamically updated magnitude-epicenter location probability distribution matrix, calculates dynamic trigger thresholds, and generates graded warning trigger instructions.

[0066] The emergency broadcast module is used to match the earthquake avoidance guidance strategy of the target area based on the graded warning trigger instructions and the emergency response plan library preset by the AI chip, and generate multimedia warning information including earthquake countdown parameters, intensity distribution prediction and AI-optimized risk avoidance prompts;

[0067] The collaborative broadcast control module is used to distribute the multimedia warning information to voice broadcast terminals, digital display terminals and public information screens through the IP network protocol, realize the priority transmission of voice broadcasts and layered rendering of dynamic information, and optimize the AI chip's plan matching logic and dynamic trigger threshold judgment conditions based on real-time feedback from terminal response data.

[0068] In an embodiment of the present invention, multiple types of data, such as source parameters and strong vibration observations, are collected in real time and output in a standardized manner. This avoids the limitations of a single data source and enables the grasp of earthquake information from multiple dimensions. For example, by combining source parameters and strong vibration observation data, the earthquake intensity and impact range can be more accurately judged. Multi-source data is analyzed in real time, key parameters are extracted, and dynamic calibration is performed using the measured data of the built-in intensity meter. For example, in complex geological areas, calibration can correct prediction deviations caused by geological conditions, ensuring that the output multi-dimensional earthquake data stream is authentic and reliable, providing strong support for accurate early warning.

[0069] By jointly analyzing multi-dimensional data, a dynamically updated magnitude-epicenter probability distribution matrix is generated and dynamic trigger thresholds are calculated. This in-depth data analysis approach comprehensively considers multiple factors to more scientifically assess earthquake risk. Compared to traditional fixed-threshold warnings, this approach can reduce false alarms and missed alarms, making warnings more tailored to actual earthquake conditions. Based on the hierarchical warning trigger instructions and emergency response plan library, personalized multimedia warning information is generated, taking into account the actual conditions of the target area. For example, customized earthquake avoidance guidance is provided for different building types and population distribution areas, allowing the public to quickly obtain risk avoidance information tailored to their circumstances and improving the effectiveness of risk avoidance measures. Plan matching and threshold determination are optimized based on terminal response data. For example, in areas of network congestion, voice broadcasts are prioritized and warning trigger strategies are adjusted to ensure efficient transmission of warning information, improving the adaptability and reliability of the entire warning system.

[0070] In a preferred embodiment of the present invention, standardized multi-source seismic data is analyzed in real time to extract initial P-wave parameters, predicted ground acceleration values, and source rupture characteristics. The analysis results are dynamically calibrated with measured data from a built-in seismometer, and a multi-dimensional seismic data stream verified by the seismometer is output, which may include:

[0071] The seismic waveform data is scanned in real time using the sliding window analysis method. Based on the preset waveform slope threshold and energy change rate threshold, the initial moment of the P wave is identified and the initial parameters of the P wave are extracted.

[0072] Based on the three-dimensional coordinates of the earthquake source in the initial P-wave parameters and the spatial distribution of geographic grid cells, combined with the physical laws of earthquake motion attenuation and historical statistical relationships, the basic predicted value of earthquake acceleration for each grid cell is calculated, specifically including:

[0073] The length of the seismic wave propagation path is calculated based on the spatial distance between the three-dimensional coordinates of the earthquake source and the center point of the geographic grid unit, and the geological density and elastic modulus parameters are extracted by combining the distribution data of the geological medium type corresponding to the path;

[0074] Based on geological density and elastic modulus parameters, combined with the physical laws of earthquake motion attenuation, a magnitude-distance attenuation relationship model is constructed. The magnitude-distance attenuation relationship model defines the logarithmic attenuation characteristics of the acceleration attenuation baseline value under different earthquake magnitudes as a function of the propagation path length.

[0075] Based on the historical earthquake motion statistical data set, the attenuation benchmark value is statistically calibrated to generate a mapping table of acceleration attenuation benchmark values under different magnitude and distance combinations;

[0076] Bind the acceleration attenuation benchmark value mapping table to the spatial distribution of the geographic grid coordinate system to generate the basic predicted value of seismic acceleration for each grid cell;

[0077] The localized calculation module uses a built-in intensity meter to obtain real-time ground acceleration data. The basic prediction value is dynamically compared with the measured data. If the deviation exceeds the preset error threshold, the magnitude-distance attenuation parameter and geological medium correction factor are corrected to generate a calibrated ground acceleration prediction value range. Specifically, the following are included:

[0078] Based on the spatial distribution of geographic grid cells, the real-time ground acceleration measured data are mapped to the corresponding grid cells;

[0079] The predicted earthquake acceleration value of each grid cell is compared point by point with the measured data of the same grid cell, and the predicted deviation rate is calculated. If the predicted deviation rate exceeds the preset error threshold, the magnitude-distance attenuation parameter and geological medium correction factor are adjusted according to the deviation direction to obtain the corrected attenuation parameter and correction factor.

[0080] Based on the corrected attenuation parameter and correction coefficient, the predicted value of the earthquake acceleration is recalculated to generate a calibrated predicted value interval containing a confidence interval range;

[0081] According to the spatial distribution characteristics of the calibrated earthquake acceleration prediction value interval and the temporal variation law of the initial P-wave parameters, the earthquake source rupture direction, rupture velocity and rupture length parameters are extracted.

[0082] The calibrated earthquake acceleration prediction value interval, P-wave initial parameters and source rupture characteristics are integrated, and the intensity meter verification mark is added to generate a multi-dimensional seismic data stream.

[0083] In the embodiment of the present invention, the seismic monitoring point collects data at a frequency of 50-200 times per second to form a time series waveform. The sliding window method is used to scan the waveform data with a window width of 1 second and a sliding step of 0.1 second. In each 1-second window, the waveform slope is obtained by calculating the amplitude difference between adjacent data points. For example, if the amplitude of the first data point in a window is 2 and the amplitude of the 100th data point is 5, the slope is calculated as At the same time, the waveform amplitudes are squared and summed, and the energy sum changes in adjacent windows are compared to obtain the energy change rate. Based on historical seismic data, the waveform slope threshold is set to 0.03 and the energy change rate threshold is set to 1.5. When the waveform slope within a window exceeds 0.03 and the energy change rate reaches 1.5, the start of that window is determined as the P-wave initial moment. After determining the initial moment, parameters such as the timestamp (accurate to the millisecond), amplitude (e.g., 3.2 μGal), and frequency component (e.g., main frequency 8 Hz) are extracted.

[0084] After obtaining the three-dimensional coordinates of the earthquake source (longitude 110°, latitude 30°, depth 10km) and the coordinates of the center point of the geographic grid unit (longitude 110.1°, latitude 30.1°, surface), the calculation of the length of the earthquake wave propagation path begins. First, the longitude and latitude coordinates must be converted into three-dimensional rectangular coordinates. This is like converting an "address" on the Earth's surface into "precise coordinates" in three-dimensional space. A spatial rectangular coordinate system is established with the center of the Earth as the origin. According to the geographic coordinate conversion rules, the longitude and latitude values of the earthquake source and the center point of the geographic grid unit, as well as the depth of the earthquake source (the surface depth is considered to be 0), are substituted into the rules for calculation to obtain their corresponding X, Y, and Z coordinate values in the three-dimensional rectangular coordinate system.

[0085] After obtaining the three-dimensional rectangular coordinates, the three-dimensional distance formula is used to calculate the straight-line distance between the earthquake source and the center point of the geographic grid cell in three-dimensional space. This distance is like the length of a straight tunnel dug inside the Earth, connecting the earthquake source and the grid center point. However, in reality, seismic waves propagate along the surface of the Earth, not in a straight line through the interior, so this straight-line distance is not the final propagation path length. Next, the straight-line distance just calculated is corrected based on the Earth's average radius of 6371 km. Because the Earth is a sphere, the actual shortest path between the earthquake source and the grid center point on the Earth's surface is different from the straight-line distance in three-dimensional space. Using the principles and methods of spherical trigonometry, a spherical triangle is formed with the Earth's center, the earthquake source, and the grid center point. The straight-line distance is adjusted based on the relationship between the sides and angles of this triangle, ultimately obtaining a more realistic earthquake wave propagation path length of approximately 14.1 km.

[0086] While calculating the propagation path length, the geological regions it traverses are determined based on the path's trajectory on the Earth's surface and underground. This process is like performing a "geological scan" of the propagation path, determining the types of strata it passes through. If the propagation path is determined to pass through a granite area, the corresponding geological parameters of the area are extracted from the geological database, such as the granite density of 2700 kg / m³ and the elastic modulus of 60 GPa.

[0087] Constructing a decay relationship model:

[0088] Based on geological density and elastic modulus, and incorporating seismic attenuation patterns from seismological studies, a magnitude-distance relationship is established. For example, the initial acceleration baseline for a magnitude 6 earthquake at 10 km is set at 200 Gal. With each additional km of distance, the acceleration decays logarithmically, with the magnitude of the decay being related to geological parameters. By determining corresponding initial baseline values and decay rates for different magnitudes, a comprehensive attenuation relationship model framework is constructed.

[0089] A historical dataset containing 5,000 earthquake records from the past 10 years was used to compare the model-calculated acceleration baseline values for different magnitude-distance combinations with the actual measured acceleration values. For example, the model calculated a magnitude 6 earthquake at 200 Gal at a distance of 10 km, while the actual measured acceleration values were mostly between 180 and 220 Gal. Model parameters were adjusted based on these discrepancies, and after repeated calibration, a mapping table of acceleration attenuation baseline values was generated, covering earthquakes of magnitudes 3 to 8 and distances of 1 to 100 km. The mapping table was linked to a geographic grid coordinate system, with each grid cell corresponding to a unique magnitude-distance combination. For example, a grid cell corresponding to a magnitude 5 earthquake at a distance of 20 km could be assigned an attenuation baseline value of 120 Gal by querying the mapping table. This attenuation baseline value was used as the baseline prediction for the ground motion acceleration for that grid cell. The seismometer collects acceleration data at a sampling frequency of 100 Hz and assigns the data to the corresponding geographic grid cell based on the latitude and longitude coordinates of the monitoring point. For example, data from a monitoring point at coordinates (110.2°, 30.2°) is assigned to the grid cell corresponding to the coordinates. Compare the basic prediction value of each grid unit with the measured data point by point and calculate the prediction deviation rate. For example, if the basic prediction value of a grid unit is 150Gal and the measured average value is 160Gal, the deviation rate is ≈6.25%. If the deviation rate exceeds the preset 5% threshold, the model parameters are adjusted based on the direction of the deviation. If the actual value is greater than the predicted value, the magnitude-distance attenuation parameter is increased, and the geological medium correction factor is fine-tuned. For example, the elastic modulus correction factor for granite areas is adjusted from 1.0 to 1.05.

[0090] Based on the revised parameters, the predicted acceleration values for each grid cell were recalculated and the confidence intervals were determined using statistical methods. Simulations were performed 100 times for the same grid cell, and the range of values at a 95% confidence level was used. For example, if the calculated result was 155 Gal and the confidence interval was ±5 Gal, the calibrated predicted value interval would be [150, 160] Gal. Analyze the spatial distribution of the calibrated acceleration predictions. If a significant increase in the predicted values for multiple adjacent grid cells in a certain direction is observed, for example, from 100 Gal to 200 Gal from east to west, and the P-wave onset in that direction is delayed by 0.5-1 second, based on the seismic wave propagation velocity (e.g., 6 km / s), the rupture direction of the earthquake source can be inferred to be east-to-west, with a rupture velocity of approximately 6 km / s. The rupture length can be estimated by calculating the distance corresponding to the range of predicted value variations in that direction: rupture length = rupture velocity × predicted value variation time window. The rupture velocity is the speed at which the earthquake source rupture surface expands (in km / s). Here, 6 km / s indicates that the rupture surface expands 6 kilometers eastward and westward per second. The predicted value variation time window is the time interval (in seconds) from the initial change to the stabilization of the earthquake prediction parameter (e.g., acceleration). Once this time window is determined using monitoring data, multiplying it by the rupture velocity yields the distance the rupture surface extends in that direction, i.e., the rupture length (estimated to be approximately 30 km). The calibrated acceleration prediction value interval (such as [150, 160]Gal), P-wave initial parameters (time, amplitude, frequency) and source rupture characteristics (direction, velocity, length) are integrated, and the intensity meter verification mark is added to form a multi-dimensional data stream containing time (such as 2024-01-0110:00:00.000), location (grid coordinates), intensity prediction, and source characteristics.

[0091] By extracting initial P-wave parameters through sliding window analysis and threshold judgment, P-waves can be accurately identified. By combining historical data to construct and calibrate an attenuation relationship model, and dynamically calibrating predicted values using measured data, the accuracy of seismic acceleration predictions is effectively improved and errors are reduced. Seismic acceleration predictions are individually calculated based on the geological medium parameters and spatial distribution of different geographic grid cells, and parameters can be modified in real time based on measured data, enabling better adaptation to diverse geological conditions and earthquake scenarios, and improving prediction reliability. Extracting source rupture characteristics not only predicts seismic intensity but also provides insights into source rupture conditions, providing more comprehensive and valuable information for earthquake research and emergency decision-making.

[0092] In a preferred embodiment of the present invention, a multi-dimensional seismic data stream verified by a seismometer is input into a preset geographic grid coordinate system. By jointly analyzing P-wave parameters, predicted ground acceleration values, and source rupture characteristics, a dynamically updated magnitude-epicenter location probability distribution matrix is generated, and a dynamic trigger threshold is calculated to generate a graded warning trigger instruction, which may include:

[0093] Based on the geographic grid coordinate system, the distance of the seismic wave propagation path from the earthquake source to each grid cell is calculated, and the geological density and elastic modulus data are extracted in combination with the geological medium distribution parameters corresponding to the path;

[0094] Based on the calibrated earthquake acceleration prediction value interval and the source rupture direction and expansion velocity parameters, the spatiotemporal constraints of the magnitude and epicenter location are constructed. The spatiotemporal constraints include the magnitude upper limit, rupture expansion range, and time window.

[0095] Based on geological parameters and spatiotemporal constraints, the Monte Carlo sampling method is used to iteratively integrate the earthquake source probability distribution, the prior probability of earthquake motion intensity, and the spatiotemporal constraints. The magnitude-epicenter location joint probability density of each grid cell is updated to generate a dynamically updated magnitude-epicenter location probability distribution matrix. Specifically, the following steps are performed:

[0096] According to the magnitude upper limit and rupture extension range in the temporal and spatial constraints, the random generation interval of the Monte Carlo sampling parameters is defined;

[0097] Random sampling is performed within the parameter random generation interval to generate multiple sets of candidate parameter combinations of magnitude, epicenter location and seismic intensity;

[0098] For each candidate parameter combination, calculate the probability value in the earthquake source probability distribution, the probability value of the earthquake intensity, and the compliance weight under the spatiotemporal constraints;

[0099] The source probability value, the earthquake intensity probability value and the compliance weight are integrated to obtain the joint probability density of the corresponding parameter combination;

[0100] The joint probability density of all candidate parameter combinations is statistically analyzed, the mean probability density within each grid cell is calculated, and the mean is mapped to the corresponding grid cell in the geographic grid coordinate system. The values of the magnitude-epicenter location probability distribution matrix are updated. When the number of Monte Carlo sampling reaches the preset threshold, the updated magnitude-epicenter location probability distribution matrix is output;

[0101] From the dynamically updated magnitude-epicenter location probability distribution matrix, a set of grid cells whose magnitude probability exceeds a preset intensity threshold is extracted and marked as a high-probability area;

[0102] Calculate the regional comprehensive risk coefficient based on the population density level and building seismic resistance level parameters corresponding to the high-probability areas in the external database;

[0103] Real-time monitoring of dynamic trigger thresholds and magnitude probability growth rates. When the trigger threshold reaches the preset critical value and the growth rate exceeds the growth rate threshold, a graded warning trigger instruction containing the target area code and warning level is generated.

[0104] In an embodiment of the present invention, the surface of the earth and the underground space are divided according to a unified standard to construct a three-dimensional geographic grid coordinate system. In order to balance the calculation accuracy and efficiency, each grid unit is set as a cube with a side length of 1 km. Taking a certain earthquake as an example, assuming that the three-dimensional coordinates of the earthquake source are longitude 110 °, latitude 30 °, and depth 10 km, the hundreds of thousands of grid cells covering the study area will be processed one by one. When calculating the distance from the earthquake source to the center point of the grid unit, the first step is coordinate conversion. With the center of the earth as the origin, the method of converting geographic coordinates to rectangular coordinates is adopted to substitute the longitude, latitude and depth information of the earthquake source and the center point of the grid unit. For example, for the earthquake source point, its longitude and latitude are converted into X, Y, and Z coordinate values in a three-dimensional rectangular coordinate system through specific conversion rules. The characteristics of the earth's approximate sphere will be taken into account during the conversion process to ensure the accuracy of the coordinates.

[0105] After the coordinate conversion is completed, the plane distance formula cannot be directly used because the earth is not a perfect sphere. The coordinates are corrected twice using the earth's average radius of 6371 km.

[0106] The specific operation is to fine-tune the converted rectangular coordinates based on the spherical curvature of the Earth to make the coordinate values more consistent with the actual shape of the Earth. Subsequently, the straight-line distance from the earthquake source to the center point of each grid cell is calculated using the three-dimensional space distance formula. However, seismic waves actually propagate along the surface of the Earth, so the straight-line distance must be further converted into the actual propagation path along the surface of the Earth using the principles of spherical trigonometry. For example, by calculating the length of the great circle arc between two points on the sphere, a more accurate propagation distance value can be obtained.

[0107] While calculating the propagation path distance, corresponding parameters are extracted along the propagation path from a professional geological database. This database integrates data from multiple sources, including geological exploration and geophysical surveys, and details the geological type, density, and elastic modulus of different regions. If the propagation path passes through a granite area, the density of the granite (between 2600-2800 kg / m³, 2700 kg / m³ is used here) and the elastic modulus (approximately 60-70 GPa, 60 GPa is used here) are accurately extracted. If the propagation path passes through a sedimentary rock area, the corresponding sedimentary rock parameters are obtained.

[0108] Constructing spatiotemporal constraints:

[0109] Based on the previously calibrated predicted range of seismic acceleration values, combined with the known focal rupture direction (e.g., east to west) and propagation velocity (6 km / s), we began to construct spatial and temporal constraints for the magnitude and epicenter location. To determine the upper magnitude limit, we conducted in-depth analysis of historical earthquake data and professional seismological theory. For example, using a specific seismic zone as an example, we retrieved earthquake records from the past century in that region, analyzed the maximum magnitude, accumulated tectonic stress, and combined them with factors such as plate motion trends to determine the upper magnitude limit for this earthquake. Based on this assessment, if the maximum historical magnitude is 7.8 and the current geological stress state shows no significant anomalies, we set the upper magnitude limit at 8.0, while retaining a safety margin. To determine the rupture extension range, we calculated the focal rupture velocity and time. Assuming that 10 seconds after the earthquake, at a rupture velocity of 6 km / s, the focal rupture surface would theoretically extend 60 kilometers in an east-to-west direction. However, in actual calculations, the hindering or promoting effects of geological structural differences on rupture are taken into consideration. For example, when encountering a hard rock layer, the rupture speed may decrease, thereby correcting the expansion range and delineating a more realistic rupture impact area.

[0110] The time window is determined based on factors such as seismic wave propagation velocity and source rupture velocity. By analyzing historical seismic wave propagation velocity statistics for the region and combining them with the characteristics of the earthquake's source, a 30-second window is established from the moment of the earthquake. During this 30-second window, the system considers the impact of seismic wave propagation and source rupture on the surrounding area to be in a critical phase of change. Beyond this time, the impact stabilizes or undergoes other complex changes, and is no longer included in the core analysis period.

[0111] Define the sampling interval:

[0112] Based on spatiotemporal constraints, the random generation interval for Monte Carlo sampling parameters was determined. For magnitude, considering the upper limit of 8.0 and the decreasing probability of an earthquake with increasing magnitude, the random generation interval was set to [5.0, 8.0], covering most possible magnitudes. The random generation range for epicenter location was strictly limited to the previously determined rupture propagation area to ensure that the sampling location was meaningful. The random generation interval for seismic intensity was set to a reasonable range, such as [20Gal, 300Gal], based on historical data on the distribution of seismic intensity in the region at different magnitudes, distances, and geological conditions, combined with current geological parameters.

[0113] Within a defined interval, large-scale random sampling is initiated, with each sampling generating a set of parameter combinations for magnitude, epicenter location, and ground motion intensity. For example, a single sampling might yield a combination of a magnitude of 6.5, an epicenter located in a grid cell (longitude 110.2°, latitude 30.1°), and a ground motion intensity of 120 Gal. Continuous sampling generates a vast number of candidate parameter combinations.

[0114] Calculate the probability value and weight:

[0115] For each candidate parameter combination, three key values are calculated:

[0116] Source Probability: This value assesses the likelihood of an earthquake occurring at the epicenter of a given parameter combination, based on the source probability distribution determined by historical earthquake data and geological structures. For example, near geological fault zones, the source probability value is higher, assigning a probability of 0.7-0.9; whereas in geologically stable areas, the probability value is lower, at 0.1-0.3.

[0117] Seismic intensity probability values are based on a constructed seismic attenuation model that comprehensively considers the physical laws of seismic wave propagation. In this model, earthquake magnitude directly determines the initial energy of seismic waves. For example, with each increase in magnitude, the energy carried by seismic waves increases exponentially. This means that, under the same propagation conditions, earthquakes with higher magnitudes will produce a higher baseline seismic intensity in the target area. The propagation path of seismic waves is determined by calculating the actual propagation distance from the earthquake source to the target grid cell. Areas closer to the earthquake source experience relatively less energy loss in received seismic waves, resulting in higher seismic intensity. Conversely, areas farther away experience continuous energy loss during propagation, resulting in lower seismic intensity. For example, consider a propagation path through different geological regions: For example, assume the propagation path passes through granite strata, which have a geological density of approximately 2700 kg / m³ and an elastic modulus of 60 GPa. This type of rock has a dense structure, which absorbs and scatters seismic waves less effectively, resulting in faster seismic wave propagation and slower energy attenuation. If the seismic waves of a magnitude 5 earthquake propagate 10 km through this area, the predicted ground motion intensity in this area would be 120 Gal, based on a ground motion attenuation model. Analysis of historical earthquakes of the same magnitude, distance, and similar geological conditions reveals that actual ground motion intensities ranged from 110 to 130 Gal 70% of the time. Therefore, the ground motion intensity probability value for this scenario is 0.7, which is relatively high. If the propagation path passes through a silty soil layer, with a density of approximately 1800 kg / m³ and an elastic modulus of only 5 GPa, the soft soil easily distorts the waveform and dissipates energy during seismic wave propagation, significantly reducing the propagation velocity and causing rapid energy attenuation. Similarly, a magnitude 5 earthquake propagating 10 km in this area might only produce 60 Gal. Analysis of historical earthquake data reveals that actual ground motion intensities ranged from 50 to 70 Gal 30% of the time. Therefore, the ground motion intensity probability value for this scenario is set at 0.3, significantly lower than that for areas with solid rock.

[0118] The analysis process is repeated across all target grid cells, taking into account each specific combination of magnitude, epicenter location, propagation distance, and geological conditions. Statistical analysis of extensive historical earthquake data, combined with theoretical calculations based on seismic attenuation models, comprehensively determines the probability of seismic intensity occurring within a certain range for each combination, ultimately yielding a seismic intensity probability value ranging from 0.2 to 0.8.

[0119] Compliance Weight: This is used to determine whether a parameter combination meets the spatiotemporal constraints. If the magnitude exceeds the upper limit, the epicenter is outside the rupture extension range, or the time is outside the set time window, the weight is set to 0. For combinations that fully meet the conditions, the weight ranges from 0.8 to 1.0, depending on the degree of compliance with the constraints. For example, a combination located exactly in the center of the rupture extension range and with a magnitude close to the average expectation receives a weight of 1.0; a combination near the edge of the range receives a weight of 0.8.

[0120] The fusion obtains the joint probability density:

[0121] The source probability value, seismic intensity probability value, and compliance weight are combined. For example, by weighted multiplication, the three values are combined to obtain the joint probability density for each parameter combination. Assuming a parameter combination with a source probability value of 0.7, a seismic intensity probability value of 0.6, and a compliance weight of 0.9, the joint probability density is 0.7 × 0.6 × 0.9 = 0.378, which directly reflects the probability of the earthquake scenario occurring.

[0122] Update the probability distribution matrix:

[0123] The joint probability density of all candidate parameter combinations is counted, and the mean of the joint probability density of all combinations in each grid cell is calculated. For example, a grid cell contains 1,000 sets of candidate parameter combinations. The joint probability density of these 1,000 sets is added and divided by 1,000 to get the mean. Subsequently, the mean is mapped to the grid cell corresponding to the geographic grid coordinate system, and the magnitude-epicenter probability distribution matrix value is updated. The system continuously repeats the sampling, calculation, and update process. When the number of samples reaches the preset 100,000 times, the calculation stops and the final updated probability distribution matrix is output, which clearly shows the probability of earthquake occurrence in each grid cell at different magnitudes.

[0124] Identification of high-probability areas and calculation of risk factors:

[0125] From the updated probability distribution matrix, grid cells with a probability of earthquake magnitude exceeding a preset intensity threshold (e.g., 0.6) are selected and marked as high-probability areas. Once high-probability areas are identified, the corresponding population density and building seismic rating parameters are retrieved from an external database. Population density is categorized as high (over 5,000 people per square kilometer), medium (1,000-5,000 people), and low (less than 1,000 people). Building seismic ratings are categorized as Class A (strong seismic design), Class B (medium seismic design), and Class C (standard design) based on structural type and construction standards. A weighted approach is used to calculate the comprehensive regional risk factor. For example, a weight of 0.6 is assigned to population density and a weight of 0.4 is assigned to building seismic rating. For a high-probability area with a high population density (corresponding to a value of 3) and a building seismic rating of C (corresponding to a value of 1), the comprehensive risk factor is 3 × 0.6 + 1 × 0.4 = 2.2. A higher risk factor indicates a greater risk to people and buildings in the area during an earthquake.

[0126] Real-time monitoring of the dynamic trigger threshold and magnitude probability growth rate. The dynamic trigger threshold is dynamically adjusted based on the region's comprehensive risk factor, historical earthquake data, and emergency response needs. In densely populated areas with weak buildings and high earthquake risk, the dynamic trigger threshold might be set at 0.7; in relatively safe areas, the threshold might be set at 0.8. The magnitude probability growth rate reflects the rate of change in probability over time, with an increase of 0.01 per second, for example, considered a unit of change. When the dynamic trigger threshold reaches a preset critical value (e.g., 0.7) and the magnitude probability growth rate exceeds the growth rate threshold (e.g., 0.02 per second), the earthquake risk is determined to have met the warning criteria. At this point, a hierarchical warning trigger command is generated, containing the target area code (e.g., a unique code based on the grid cell number) and the warning level. Warning levels are categorized into three levels: Level 1 (extremely high risk) requires immediate evacuation; Level 2 (high risk) urges preparation for evacuation; and Level 3 (moderate risk) urges attention to subsequent developments, ensuring that warning information is accurately delivered to the appropriate areas to guide emergency response efforts.

[0127] By taking into account propagation path distance, geological parameters, and spatiotemporal constraints, combined with Monte Carlo sampling, a multi-factor probability distribution matrix is calculated for magnitude and epicenter location. Compared to single-factor analysis, this more accurately reflects the likelihood and impact of an earthquake, reducing prediction errors. The probability distribution matrix and dynamic trigger thresholds can be updated in real time, allowing for timely adjustments to early warning strategies based on actual earthquake development and newly acquired data, adapting to the uncertainty and complexity of the earthquake process and improving the timeliness and effectiveness of the early warning system. By calculating a comprehensive regional risk factor and generating graded early warning trigger instructions, this system provides more targeted decision-making for emergency management departments. Based on the different warning levels, appropriate emergency measures can be implemented, enabling the rational allocation of rescue resources, improving emergency response efficiency, and minimizing losses caused by earthquake disasters.

[0128] In a preferred embodiment of the present invention, based on the hierarchical warning trigger instructions and the emergency response plan library preset in the AI chip, the earthquake avoidance guidance strategy of the target area is matched to generate multimedia warning information containing earthquake countdown parameters, intensity distribution prediction and AI-optimized risk avoidance prompts, which may include:

[0129] According to the warning level in the graded warning trigger instruction, the corresponding response strategy template is matched from the emergency response plan library to determine the priority, language style and geographical coverage of the broadcast content;

[0130] Based on the seismic wave propagation velocity parameters and the updated magnitude-epicenter location probability distribution matrix, the propagation time of the seismic wave to each grid cell within the geographical coverage area is calculated, and regionalized countdown parameters synchronized with the time axis are generated;

[0131] The intensity distribution prediction data of high-probability areas are spatially correlated and matched with the road network topology and shelter coordinates in the external database to generate intensity distribution spatial data with superimposed road traffic status and shelter resource identification. Specifically, the data includes:

[0132] Based on the geographic grid coordinate system, the intensity distribution prediction data of high-probability areas, road network topology and shelter coordinates are spatially aligned to generate an associated dataset in a unified coordinate system.

[0133] Based on the road network topology in the linked dataset and combined with real-time traffic data, the traffic status of each road segment in the high-probability area is analyzed to generate road traffic status data containing road grade, congestion level, and availability identification. Based on the coordinates of the shelters in the linked dataset and combined with the capacity database, the real-time available capacity of each shelter is marked to generate shelter resource availability identification data.

[0134] The intensity distribution prediction data, road traffic status data and refuge resource availability identification data are integrated according to the spatial location of the geographic grid unit to generate intensity distribution spatial data with superimposed road traffic status and refuge resource identification;

[0135] Based on the database of building structural characteristics in the target area, the AI chip analyzes the building type and seismic resistance. Combined with the intensity level predicted from the spatial intensity distribution data, it dynamically generates seismic avoidance guidance content adapted to the building type.

[0136] The regional countdown parameters, intensity distribution spatial data and vibration avoidance guidance content are synchronously integrated according to timestamps and spatial positions to generate multimedia warning information including voice broadcasts, text prompts and dynamic visualization elements.

[0137] In an embodiment of the present invention, the emergency response plan library pre-classifies and stores policy templates according to the warning level (level one, level two, and level three). Taking the level one warning as an example, after receiving the graded warning trigger instruction containing "level one warning", the corresponding template is retrieved from the plan library. The template has preset the broadcast content priority: first broadcast the earthquake countdown and core risk avoidance instructions, and then scroll through the intensity distribution and evacuation instructions; the language style is set to quick and concise short sentences, such as "Immediately hide in the corner of the load-bearing wall!"; the geographical coverage range is based on the target area code in the instruction, locking all grid cells in the high-probability area to ensure that the information coverage has no blind spots. For the level two warning, the template will adjust the content priority, first explain the earthquake risk level, and then gradually guide risk avoidance. The language style is relatively gentle, and the coverage range can be narrowed to the grid cells with the most concentrated risks.

[0138] Regionalized countdown parameter generation:

[0139] Based on the average propagation velocity parameter of seismic waves (e.g., 6 km / s) and the updated magnitude-epicenter probability distribution matrix, the propagation time is calculated for each grid cell within the geographic coverage area. The specific steps are as follows:

[0140] The distance from the earthquake source corresponding to each grid cell is obtained from the probability distribution matrix (for example, a grid cell is 15 km away from the earthquake source). The distance is divided by the propagation speed to obtain the time it takes for the earthquake wave to reach the grid cell (15 km ÷ 6 km / s = 2.5 seconds). The countdown end time for the grid cell is generated (14:30:02.5) by adding the propagation time, taking the time of the earthquake as the starting point (for example, 2025-05-21 14:30:00). The countdown time of all grid cells is aggregated by geographical region to generate regional countdown parameters with a time axis interval of 1 second, for example: "In the next 2 seconds, area A (grids 1-10) will be affected by the earthquake; in the next 3 seconds, area B (grids 11-20) will enter the warning range."

[0141] Using a geographic grid coordinate system (such as WGS84), we spatially align the predicted intensity distribution data for high-probability areas (e.g., grid cells 1-5 predicted intensity VII), road network topology data (e.g., grid cell 3 contains main road X1), and shelter coordinate data (e.g., grid cell 4 contains shelter S1 at longitude 110.1° and latitude 30.1°). Using a coordinate transformation algorithm, we unify data from different sources into the same grid cell. For example, we map the coordinates of main road X1 and shelter S1 to the corresponding grid cells 3 and 4, forming a linked dataset containing locations and attributes.

[0142] Combined with real-time traffic data (such as vehicle density monitored by radar), the road network topology in the associated dataset is analyzed:

[0143] For arterial road X1, if the real-time traffic density exceeds 80 vehicles / km, it is considered congested, marked in red, and the availability indicator is "Travel with Caution";

[0144] For the secondary arterial road Y1, the traffic density is less than 30 vehicles / km, which is considered to be unobstructed, marked green, and the availability indicator is "priority passage";

[0145] Finally, the road traffic status data including road grade (main road / secondary road), congestion level (red / yellow / green), and availability mark is generated. For example, the main road X1 in grid unit 3 is displayed as "main road-red-pass with caution".

[0146] The shelter capacity database (e.g., Shelter S1 has a total capacity of 500 people) is retrieved and combined with real-time registration data (e.g., 200 people already accommodated) to calculate the real-time available capacity (500 - 200 = 300 people) and label it as "Shelter S1 - Available Capacity 300 People." This process is repeated for all shelters to generate shelter resource availability data. For example, Shelter S1 in grid cell 4 will be displayed as "Blue Icon - 300 People Available." The intensity distribution prediction data (e.g., grid cells 3-4 are at Level VII), road traffic status data (e.g., main road X1 is congested), and shelter resource availability data (e.g., Shelter S1 has 300 people available) are spatially fused by grid cell. For example, in grid cell 3, a superimposed visualization is generated: the background color is orange, representing Level VII, the main road X1 is marked with a red congestion line, and the nearby Shelter S1 displays a blue available icon and capacity value, forming a visual representation of the intensity distribution.

[0147] The target area building structural characteristics database stores seismic capacity data of different building types, such as:

[0148] Masonry structure (seismic fortification intensity VI): In the predicted intensity VII area, the AI chip determines that the building has a high risk of collapse;

[0149] Frame structure (seismic fortification intensity VIII): In the predicted intensity VII area, the structure is judged to be basically safe but attention should be paid to secondary disasters.

[0150] Combined with the spatial data of intensity distribution, the AI chip dynamically generates adaptation guidance content:

[0151] To users of masonry structures: "Your building is a masonry structure. The predicted intensity is Level VII. Please take shelter immediately in the corner of the load-bearing wall and use tables and chairs to protect your head!";

[0152] To users of frame structures: "The building you are in is a frame structure, and the predicted intensity is Level VII. Please avoid windows, hide under the desk, and wait for evacuation instructions!"

[0153] Using timestamps and spatial locations as indexes, regional countdown parameters, intensity distribution spatial data, and earthquake avoidance guidance are synchronously integrated:

[0154] Timeline synchronization: When the countdown reaches 2 seconds, a voice message will be broadcast: "The earthquake will reach area A in 2 seconds!" At the same time, the grid cells in area A on the map will flash orange (indicating a VIIth intensity), and a text prompt will pop up indicating the corresponding masonry structure's earthquake avoidance action.

[0155] Spatial location synchronization: In the grid unit 3-4 area, a dynamic visualization layer displays the congested main road X1 (red line) and the available shelter S1 (blue icon). At the same time, a voice broadcast announces, "In the grid unit 3-4 area, please avoid the main road X1 and go to the shelter S1 (available for 300 people) first!"

[0156] The final generated multimedia warning information includes:

[0157] Voice broadcast: emergency instructions (such as countdown, evasive action) played in order of priority;

[0158] Text prompt: Intensity level and evacuation instructions for each area (e.g., "Grid cells 1-5: Intensity VII, it is recommended to go to Shelter S2");

[0159] Dynamic visualization: real-time updated intensity distribution map, road traffic status layer, and refuge resource signs.

[0160] By matching warning levels to emergency plan templates, a "tiered response and targeted policy implementation" system is implemented, shortening information generation time and ensuring that high-risk areas receive emergency instructions first. For example, the delay in broadcasting core information under a Level 1 warning can be controlled to within 1 second. Regionalized countdown parameters are synchronized in real time with the seismic wave propagation path, allowing users to intuitively perceive the arrival time of the threat (e.g., "Earthquake will arrive in 3 seconds"). Combined with dynamic intensity maps, this system provides dual-dimensional "time + space" risk avoidance guidance, improving evacuation decision-making efficiency. By integrating intensity forecasts, road accessibility, and evacuation resource data, users can quickly obtain comprehensive information on "where danger is, which roads are accessible, and where to evacuate." For example, in high-intensity areas, congested roads can be automatically avoided, and the shortest safe evacuation route can be recommended. AI-adapted prompts based on building type avoid "one-size-fits-all" recommendations and improve the scientific nature of risk avoidance measures. For example, users with masonry structures receive targeted reinforcement recommendations, while those with frame structures receive guidance on secondary disaster prevention.

[0161] In a preferred embodiment of the present invention, the multimedia warning information is distributed to voice broadcast terminals, digital display terminals, and public information screens via the IP network protocol, achieving priority transmission of voice broadcasts and layered rendering of dynamic information. Simultaneously, based on real-time feedback from terminal response data, the AI chip's plan matching logic and dynamic trigger threshold determination conditions are optimized, which may include:

[0162] Assign transmission priority to voice broadcast data through IP network protocol, and dynamically adjust the rendering level of text prompts and dynamic information elements based on the transmission status of voice broadcast;

[0163] Receive the broadcast completion status of voice broadcast terminals, rendering delay time of digital display terminals, and network load data of public information screens in real time to generate terminal response data sets;

[0164] Based on the broadcast delay and rendering efficiency data in the terminal response dataset, the matching weight of the emergency response plan in the AI chip is adjusted, the generation logic of the shock avoidance guidance strategy is optimized, and based on the network load fluctuation data in the terminal response dataset, the judgment conditions of the dynamic trigger threshold are dynamically corrected.

[0165] In this embodiment, multimedia warning information, including voice broadcast data, text prompts, and dynamic information elements (such as map layers and shelter signs), is categorized and encapsulated via IP network protocols (e.g., TCP / IP). Voice broadcast data is assigned the highest transmission priority, similar to the "emergency channel" mechanism in network communications. For example, voice data is marked as "priority 1" in the IP packet header, ensuring it receives priority bandwidth even in network congestion.

[0166] During transmission, the system monitors the transmission status of voice announcements (e.g., latency, packet loss) in real time. If the delay exceeds a preset threshold (e.g., 500 milliseconds), it automatically adjusts the rendering level of text prompts and dynamic information elements. For example, it temporarily reduces the refresh rate of dynamic maps (from 5 to 2 per second) or simplifies text content (e.g., omitting non-critical evacuation instructions) to free up network resources and ensure the continuity of voice data. Conversely, if voice transmission is stable, it restores the rendering level of the full information to ensure users receive complete dynamic information.

[0167] Receive feedback data from various terminals in real time:

[0168] Voice broadcast terminal: Feedback "broadcast completion status", that is, the timestamp difference between each voice command from the time it is sent to the time it is played (for example, command A is sent at 14:30:02 and played at 14:30:03, which takes 1 second);

[0169] Digital display terminals: Feedback on "rendering latency," which is the time it takes for text or images to be fully displayed after being received (e.g., a grid cell intensity map takes 800 milliseconds to load);

[0170] Public information screen: Feedback "network load data", including the number of currently connected terminals and bandwidth usage (for example, if a public information screen in a certain area is connected to 500 devices, the bandwidth usage reaches 70%).

[0171] Integrate the above data into a terminal response dataset, for example:

[0172] Voice terminal A: Command ID-001, broadcasting time takes 1.2 seconds, delay is normal;

[0173] Digital Terminal B: Map rendering takes 1.5 seconds, exceeding the threshold by 0.5 seconds;

[0174] Public information screen C: Bandwidth utilization is 85%, approaching the congestion threshold (90%).

[0175] AI chip optimization and threshold correction:

[0176] Analyze "broadcast delay" and "rendering efficiency" in terminal response data. For example, if digital display terminals in a certain area experience widespread high rendering delays (e.g., an average rendering delay exceeding 1 second), this indicates that the network environment in that area is poor and cannot support the real-time display of complex dynamic information. In this case, the AI chip will automatically reduce the weight of "dynamic map rendering" in the emergency response plan for that area, prioritize the generation of concise text instructions (such as "Go to the nearest shelter immediately"), and increase the number of repetitions of voice broadcasts (e.g., changing from a single broadcast to a three-time loop broadcast) to ensure the transmission of key information.

[0177] Based on network load fluctuation data from public information screens, the warning triggering logic is dynamically adjusted. For example, when the network load rate in a certain area is consistently above 80%, it indicates a decline in terminal reception capacity, potentially leading to delayed warning information. In this case, the system will preemptively adjust the dynamic triggering threshold, lowering the "magnitude probability growth rate threshold" from the default "0.02 per second" to "0.015 per second." This allows the warning to be triggered earlier, reserving more time for network transmission and avoiding warning failures due to transmission delays.

[0178] A voice-prioritized transmission mechanism ensures that core information, such as earthquake countdowns and evacuation instructions, can be delivered quickly even during network congestion. For example, in high-load scenarios like subways and shopping malls, voice broadcast delays can be kept to within one second. The system adjusts the information rendering level in real time based on terminal feedback, automatically simplifying non-critical content (e.g., reducing map accuracy) in areas with poor network connectivity to prevent user information loss due to data transmission failures. For example, in remote mountainous areas with limited network bandwidth, the system can automatically switch to a "pure voice + simple text" mode to ensure warning coverage. The AI plan matching logic is optimized based on terminal response data, ensuring that evacuation strategies are more tailored to actual transmission conditions. For example, in areas with a high proportion of elderly people, if the voice terminal broadcast completion rate reaches 95%, the system will automatically increase the detail of voice instructions. In areas with a high concentration of young people, if the digital terminal rendering efficiency is high, dynamic maps and AR evacuation guidance features will be enhanced. The system dynamically adjusts the trigger threshold based on network load data to avoid delayed warnings due to transmission delays. For example, in scenarios with temporary high loads, such as large event venues, the system can lower the trigger threshold in advance, accelerating the issuance of warnings by 2-3 seconds, buying valuable time for evacuation. From information distribution to terminal feedback to strategy optimization, a complete emergency response closed loop is formed to improve the system's adaptability and reliability, especially in complex network environments or sudden traffic peak scenarios, to ensure the stability and effectiveness of the early warning system.

[0179] The embodiment of the present invention further provides a computer-readable storage medium storing instructions, which, when executed on a computer, causes the computer to execute the system described above. All implementations in the above system embodiments are applicable to this embodiment and can achieve the same technical effects.

[0180] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. Emergency broadcast intelligent triggering system based on multi-source seismic data fusion, characterized by: include: The data receiving module is used to receive multi-source earthquake early warning information in real time, including source parameter data, strong motion observation data and real-time early warning parameters, and output standardized multi-source earthquake data through the built-in communication protocol; The localized calculation module is used to perform real-time analysis of standardized multi-source seismic data, extracting initial P-wave parameters, predicted ground acceleration values, and source rupture characteristics. The module dynamically calibrates the analysis results based on the measured data from the built-in seismometer and outputs a multi-dimensional seismic data stream verified by the seismometer. The data processing module is used to input the multi-dimensional seismic data stream verified by the seismometer into a preset geographic grid coordinate system. By jointly analyzing P-wave parameters, predicted ground acceleration values, and source rupture characteristics, it generates a dynamically updated magnitude-epicenter location probability distribution matrix, calculates dynamic trigger thresholds, and generates graded warning trigger instructions. The emergency broadcast module is used to match the earthquake avoidance guidance strategy of the target area based on the graded warning trigger instructions and the emergency response plan library preset by the AI chip, and generate multimedia warning information including earthquake countdown parameters, intensity distribution prediction and AI-optimized risk avoidance prompts; The collaborative broadcast control module is used to distribute the multimedia warning information to voice broadcast terminals, digital display terminals and public information screens through the IP network protocol, realize the priority transmission of voice broadcasts and layered rendering of dynamic information, and optimize the AI chip's plan matching logic and dynamic trigger threshold judgment conditions based on real-time feedback from terminal response data.

2. The intelligent triggering system for emergency broadcast based on multi-source seismic data fusion according to claim 1 is characterized in that: The system performs real-time analysis on standardized multi-source seismic data, extracting initial P-wave parameters, predicted ground acceleration values, and source rupture characteristics. The system dynamically calibrates the analysis results based on the measured data from the built-in seismometer, and outputs a multi-dimensional seismic data stream verified by the seismometer, including: The seismic waveform data is scanned in real time using the sliding window analysis method. Based on the preset waveform slope threshold and energy change rate threshold, the initial moment of the P wave is identified and the initial parameters of the P wave are extracted. Based on the three-dimensional coordinates of the earthquake source and the spatial distribution of geographic grid cells in the initial P-wave parameters, combined with the physical laws of earthquake motion attenuation and historical statistical relationships, the basic predicted value of earthquake acceleration for each grid cell is calculated. The localized calculation module uses a built-in intensity meter to obtain real-time ground acceleration data. The basic prediction value is dynamically compared with the measured data. If the deviation exceeds the preset error threshold, the magnitude-distance attenuation parameter and geological medium correction factor are corrected to generate a calibrated ground acceleration prediction value range. According to the spatial distribution characteristics of the calibrated earthquake acceleration prediction value interval and the temporal variation law of the initial P-wave parameters, the earthquake source rupture direction, rupture velocity and rupture length parameters are extracted. The calibrated earthquake acceleration prediction value interval, P-wave initial parameters and source rupture characteristics are integrated, and the intensity meter verification mark is added to generate a multi-dimensional seismic data stream.

3. The intelligent triggering system for emergency broadcast based on multi-source seismic data fusion according to claim 2 is characterized in that: Based on the three-dimensional coordinates of the earthquake source in the initial P-wave parameters and the spatial distribution of geographic grid cells, combined with the physical laws of earthquake attenuation and historical statistical relationships, the basic predicted value of earthquake acceleration for each grid cell is calculated, including: The length of the seismic wave propagation path is calculated based on the spatial distance between the three-dimensional coordinates of the earthquake source and the center point of the geographic grid unit, and the geological density and elastic modulus parameters are extracted by combining the distribution data of the geological medium type corresponding to the path; Based on geological density and elastic modulus parameters, combined with the physical laws of earthquake motion attenuation, a magnitude-distance attenuation relationship model is constructed. The magnitude-distance attenuation relationship model defines the logarithmic attenuation characteristics of the acceleration attenuation baseline value under different earthquake magnitudes as a function of the propagation path length. Based on the historical earthquake motion statistical data set, the attenuation benchmark value is statistically calibrated to generate a mapping table of acceleration attenuation benchmark values under different magnitude and distance combinations; The acceleration attenuation benchmark value mapping table is bound to the spatial distribution of the geographic grid coordinate system to generate the basic predicted value of seismic acceleration for each grid cell.

4. The intelligent triggering system for emergency broadcast based on multi-source seismic data fusion according to claim 3 is characterized in that: The localized calculation module uses a built-in intensity meter to obtain real-time ground acceleration data. The basic prediction value is dynamically compared with the measured data. If the deviation exceeds the preset error threshold, the magnitude-distance attenuation parameter and geological medium correction factor are corrected to generate a calibrated ground acceleration prediction value range, including: Based on the spatial distribution of geographic grid cells, the real-time ground acceleration measured data are mapped to the corresponding grid cells; The predicted earthquake acceleration value of each grid cell is compared point by point with the measured data of the same grid cell, and the predicted deviation rate is calculated. If the predicted deviation rate exceeds the preset error threshold, the magnitude-distance attenuation parameter and geological medium correction factor are adjusted according to the deviation direction to obtain the corrected attenuation parameter and correction factor. Based on the corrected attenuation parameters and correction coefficients, the predicted seismic acceleration values are recalculated to generate a calibrated predicted value interval containing a confidence interval range.

5. The intelligent triggering system for emergency broadcast based on multi-source seismic data fusion according to claim 4 is characterized in that: The multi-dimensional seismic data stream verified by the seismometer is input into the preset geographic grid coordinate system. By jointly analyzing P-wave parameters, predicted ground acceleration values, and source rupture characteristics, a dynamically updated magnitude-epicenter location probability distribution matrix is generated. The dynamic trigger threshold is calculated and a graded warning trigger instruction is generated, including: Based on the geographic grid coordinate system, the distance of the seismic wave propagation path from the earthquake source to each grid cell is calculated, and the geological density and elastic modulus data are extracted in combination with the geological medium distribution parameters corresponding to the path; Based on the calibrated earthquake acceleration prediction value interval and the source rupture direction and expansion velocity parameters, the spatiotemporal constraints of the magnitude and epicenter location are constructed. The spatiotemporal constraints include the magnitude upper limit, rupture expansion range, and time window. Based on geological parameters and spatiotemporal constraints, the Monte Carlo sampling method is used to iteratively integrate the earthquake source probability distribution, the prior probability of earthquake motion intensity, and the spatiotemporal constraints to update the magnitude-epicenter location joint probability density of each grid cell and generate a dynamically updated magnitude-epicenter location probability distribution matrix. From the dynamically updated magnitude-epicenter location probability distribution matrix, a set of grid cells whose magnitude probability exceeds a preset intensity threshold is extracted and marked as a high-probability area; Calculate the regional comprehensive risk coefficient based on the population density level and building seismic resistance level parameters corresponding to the high-probability areas in the external database; Real-time monitoring of dynamic trigger thresholds and magnitude probability growth rates. When the trigger threshold reaches the preset critical value and the growth rate exceeds the growth rate threshold, a graded warning trigger instruction containing the target area code and warning level is generated.

6. The intelligent triggering system for emergency broadcast based on multi-source seismic data fusion according to claim 5 is characterized in that: Based on geological parameters and spatiotemporal constraints, the Monte Carlo sampling method is used to iteratively integrate the earthquake source probability distribution, the prior probability of earthquake motion intensity, and the spatiotemporal constraints. The joint probability density of magnitude and epicenter location of each grid cell is updated to generate a dynamically updated magnitude-epicenter location probability distribution matrix, including: According to the magnitude upper limit and rupture extension range in the temporal and spatial constraints, the random generation interval of the parameters of Monte Carlo sampling is defined; Random sampling is performed within the parameter random generation interval to generate multiple sets of candidate parameter combinations of magnitude, epicenter location and ground motion intensity; For each candidate parameter combination, calculate the probability value in the earthquake source probability distribution, the probability value of the earthquake intensity, and the compliance weight under the spatiotemporal constraints; The earthquake source probability value, the earthquake intensity probability value and the compliance weight are integrated to obtain the joint probability density of the corresponding parameter combination; The joint probability density of all candidate parameter combinations is statistically analyzed, the mean probability density in each grid cell is calculated, and the mean is mapped to the corresponding grid cell in the geographic grid coordinate system. The value of the magnitude-epicenter location probability distribution matrix is updated. When the number of Monte Carlo sampling reaches the preset threshold, the updated magnitude-epicenter location probability distribution matrix is output.

7. The intelligent triggering system for emergency broadcast based on multi-source seismic data fusion according to claim 6 is characterized in that: Based on the graded warning trigger instructions and the pre-set AI chip's emergency response plan library, the system matches the target area's earthquake avoidance guidance strategy and generates multimedia warning information containing earthquake countdown parameters, intensity distribution predictions, and AI-optimized risk avoidance prompts, including: According to the warning level in the graded warning trigger instruction, the corresponding response strategy template is matched from the emergency response plan library to determine the priority, language style and geographical coverage of the broadcast content; Based on the seismic wave propagation velocity parameters and the updated magnitude-epicenter location probability distribution matrix, the propagation time of the seismic wave to each grid cell within the geographical coverage area is calculated, and regionalized countdown parameters synchronized with the time axis are generated; The intensity distribution prediction data of high-probability areas are spatially correlated and matched with the road network topology and shelter coordinates in the external database to generate intensity distribution spatial data superimposed with road traffic status and shelter resource identification; Based on the database of building structural characteristics in the target area, the AI chip analyzes the building type and seismic resistance. Combined with the intensity level predicted from the spatial intensity distribution data, it dynamically generates seismic avoidance guidance content adapted to the building type. The regional countdown parameters, intensity distribution spatial data and vibration avoidance guidance content are synchronously integrated according to timestamps and spatial positions to generate multimedia warning information including voice broadcasts, text prompts and dynamic visualization elements.

8. The intelligent triggering system for emergency broadcast based on multi-source seismic data fusion according to claim 7 is characterized in that: The intensity distribution prediction data of high-probability areas are spatially correlated and matched with the road network topology and shelter coordinates in the external database to generate intensity distribution spatial data with superimposed road traffic status and shelter resource identification, including: Based on the geographic grid coordinate system, the intensity distribution prediction data of high-probability areas, road network topology and shelter coordinates are spatially aligned to generate an associated dataset in a unified coordinate system. Based on the road network topology in the linked dataset and combined with real-time traffic data, the traffic status of each road segment in the high-probability area is analyzed to generate road traffic status data containing road grade, congestion level, and availability identification. Based on the coordinates of the shelters in the linked dataset and combined with the capacity database, the real-time available capacity of each shelter is marked to generate shelter resource availability identification data. The intensity distribution prediction data, road traffic status data and refuge resource availability identification data are fused according to the spatial location of the geographic grid unit to generate intensity distribution spatial data with superimposed road traffic status and refuge resource identification.

9. The intelligent triggering system for emergency broadcast based on multi-source seismic data fusion according to claim 8, characterized in that: The multimedia warning information is distributed to voice broadcast terminals, digital display terminals, and public information screens via the IP network protocol, enabling voice broadcast priority transmission and dynamic information layered rendering. Simultaneously, based on real-time feedback from terminal response data, the AI chip's plan matching logic and dynamic trigger threshold determination conditions are optimized, including: Assign transmission priority to voice broadcast data through IP network protocol, and dynamically adjust the rendering level of text prompts and dynamic information elements based on the transmission status of voice broadcast; Receive the broadcast completion status of voice broadcast terminals, rendering delay time of digital display terminals, and network load data of public information screens in real time to generate terminal response data sets; Based on the broadcast delay and rendering efficiency data in the terminal response dataset, the matching weight of the emergency response plan in the AI chip is adjusted, the generation logic of the shock avoidance guidance strategy is optimized, and based on the network load fluctuation data in the terminal response dataset, the judgment conditions of the dynamic trigger threshold are dynamically corrected.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a program, which, when executed by a processor, implements the system according to any one of claims 1 to 9.

Images