Earthquake early warning method and device, electronic equipment and computer readable storage medium

Abstract

This application provides an earthquake early warning method, device, electronic device, and computer-readable storage medium, relating to the field of earthquake early warning technology. The earthquake early warning method includes: acquiring real-time observation data from at least one seismic station within a monitoring area; determining a reference intensity value for the corresponding seismic station based on the real-time observation data; for any target early warning location: if the reference intensity value of an existing seismic station meets preset early warning conditions, assigning the reference intensity value to the target early warning location within its spatial influence range to obtain a first predicted intensity value; furthermore, considering all seismic stations within the monitoring area as virtual sub-sources, and for the same virtual sub-source, determining a second predicted intensity value at the target early warning location based on ground motion attenuation relationships; and fusing the first and second predicted intensity values ​​to generate a final early warning intensity field for the target early warning location. This method addresses the potential for missed or false alarms that may occur with traditional source parameter estimation methods, without considering source parameters.

Description

Technical Field

[0001] This application relates to the field of earthquake early warning technology, and more specifically, to an earthquake early warning method, device, electronic equipment, and computer-readable storage medium. Background Technology

[0002] In existing earthquake early warning technologies, intensity field prediction methods based on source parameter estimation and attenuation models are prone to significant deviations in source parameter inversion under strong earthquakes or complex rupture scenarios, leading to missed or false alarms. To overcome this deficiency, the industry has developed various prediction methods that do not rely on source parameters. Among them, real-time earthquake ground motion prediction methods based on radiative transfer theory simulate the propagation and attenuation of ground motion energy through the random diffusion of a large number of particles, which requires high computational resources and computing power, increasing the difficulty of deployment in real-time early warning systems. Early warning methods based on the assumption of local attenuation-free propagation directly use the intensity observed by stations to provide warnings to the surrounding areas, effectively avoiding missed and false alarms caused by uncertainties in source parameter estimation. However, their warning time is limited, usually only a few seconds, which is difficult to meet the needs of longer-term warnings. Intensity field prediction methods based on deep learning are currently in the initial exploratory stage, and the maturity and reliability of the models still need to be verified. Current on-site ground motion prediction methods rely solely on P-wave information received from a single seismic station to estimate the maximum potential ground motion. Early statistical models exhibit significant uncertainties. While the introduction of artificial intelligence has improved prediction accuracy, combining it with regional early warning methods to extend warning time still increases the risk of false alarms due to these uncertainties. Overall, existing earthquake early warning methods independent of source parameters remain insufficient in terms of warning timeliness, necessitating the development of new methods that balance longer warning times with lower false alarm risks. Summary of the Invention

[0003] In view of the above, the purpose of this application is to provide an earthquake early warning method, device, electronic device and computer-readable storage medium to improve the above-mentioned problems existing in the prior art.

[0004] In a first aspect, this application provides an earthquake early warning method, comprising: acquiring real-time observation data from at least one seismic station within a monitoring area; determining a reference intensity value corresponding to the seismic station based on the real-time observation data; wherein the reference intensity value includes an estimated maximum intensity value predicted based on P-wave information and / or the measured intensity value of the seismic station; for any target early warning location: if the reference intensity value of the seismic station meets a preset early warning condition, assigning the reference intensity value to the target early warning location within the spatial influence range, centered on the seismic station, based on a preset spatial influence range, to obtain a first predicted intensity value; furthermore, considering all seismic stations within the monitoring area as virtual sub-sources, for the same virtual sub-source, using the reference intensity value to back-calculate the intensity parameters of the virtual sub-source, and determining a second predicted intensity value at the target early warning location based on the ground motion attenuation relationship; and fusing the first predicted intensity value and the second predicted intensity value to generate a final early warning intensity field at the target early warning location.

[0005] In the above implementation process, on the one hand, by directly assigning reference intensity values ​​to the warning locations of nearby targets with seismic stations as the center, the maximum intensity measured by the stations or estimated by P-waves can be quickly reflected, while avoiding large blind spots or distortions caused by simply relying on spatial interpolation in sparsely populated areas. On the other hand, by treating all seismic stations as virtual sub-sources, the intensity parameters of each virtual sub-source are calculated using the reference intensity values, and the second predicted intensity value of the target warning location is determined based on the ground motion attenuation relationship. This allows for continuous intensity prediction of any spatial location based on independent estimates from multiple stations in the early stage of warning before accurate source location is obtained, thus expanding the effective warning coverage and shortening the response time from triggering the station to generating warning information for the target location. Furthermore, by integrating the first and second predicted intensity values, the deviation caused by inaccurate back-calculation of intensity parameters or unreasonable setting of spatial influence range in a single prediction channel can be effectively suppressed, thereby improving the accuracy, stability, and spatial continuity of the final early warning intensity field. This provides more reliable and targeted earthquake early warning information for different regions, buys valuable time for emergency decisions such as personnel evacuation and emergency control of facilities, and enhances the effectiveness of disaster prevention and mitigation against destructive earthquakes.

[0006] Based on the first aspect, acquiring real-time observation data from at least one seismic station within the monitoring area and determining the reference intensity value corresponding to the seismic station based on the real-time observation data includes: acquiring real-time observation data within multiple increasing time windows after the P-wave is triggered at the seismic station, wherein the real-time observation data includes at least the triaxial composite peak velocity. Peak displacement and cumulative absolute velocity Substitute the real-time observation data into the statistical regression relationship to obtain the intensity prediction value corresponding to each parameter; for the current time window length, select the intensity prediction values ​​corresponding to the two parameters with smaller regression standard deviations in the real-time observation data, and take their average value as the estimated maximum intensity value under this time window.

[0007] In the above implementation process, within multiple increasing time windows following the triggering of the P-wave at the seismic station, the composite peak velocity in the three directions is continuously acquired. Peak displacement and cumulative absolute velocity Real-time observation data, including those from earthquakes, are used to obtain intensity prediction values ​​for each parameter through statistical regression relationships. This enables rapid capture and dynamic updating of seismic ground motion information, effectively balancing the timeliness of early warning and the stability of intensity estimation. As the time window length increases, richer seismic ground motion information is incorporated, improving the reliability of intensity estimation. Furthermore, for the current time window length, the intensity prediction values ​​corresponding to the two parameters with smaller regression standard deviations are selected, and their average is taken as the estimated maximum intensity value for that time window. This effectively avoids the problem of large regression dispersion and significant prediction bias of a single parameter under specific magnitude, distance, or site conditions, reducing the risk of false alarms or missed alarms caused by inappropriate parameter selection, improving the accuracy and robustness of the predicted intensity, and providing a more reliable input basis for the subsequent generation of the first and second predicted intensity values.

[0008] Based on the first aspect, the step of acquiring real-time observation data from at least one seismic station within the monitoring area and determining a reference intensity value corresponding to the seismic station based on the real-time observation data further includes: comparing each of the real-time observation data obtained by calculating the current time window length with a preset parameter threshold corresponding to the target warning intensity threshold; and selecting the estimated maximum intensity value as the reference intensity value if any of the real-time observation data reaches or exceeds its corresponding preset parameter threshold.

[0009] In the aforementioned implementation process, seismic stations can quickly lock onto reference intensity values ​​and enter subsequent early warning procedures within a very short time after the arrival of the P-wave, once they capture key parameter characteristics that can corroborate a potentially destructive earthquake. This eliminates the need to wait for calculation results over a longer time window or for all parameters to reach their thresholds, significantly shortening the time for issuing the first early warning information in the event of a strong earthquake and providing valuable lead time for emergency response in the vicinity of the seismic source area. Simultaneously, the judgment logic based on multi-parameter threshold cross-validation effectively avoids false triggering caused by accidental fluctuations in a single parameter or non-destructive interference, ensuring that high-intensity early warnings are activated only when the observed data truly reflects a strong seismic motion level. This improves the system's response speed while enhancing the reliability and anti-interference capability of early warning decisions, providing a more reliable and timely input for the subsequent fusion and generation of the first and second predicted intensity values.

[0010] Based on the first aspect, wherein the preset parameter threshold includes the triaxial synthesis peak velocity. Threshold, peak displacement Threshold and cumulative absolute velocity Threshold; the target warning intensity threshold is preset. For a given time window length, strong motion record samples from multiple stations in historical earthquakes are obtained. Each sample contains the real-time observation data under the time window length and the maximum intensity value finally measured by the seismic station. A spatial region defined by three thresholds is determined in the three-dimensional parameter space composed of the real-time observation data. The boundary values ​​corresponding to the spatial region on the real-time observation data are used as the preset parameter thresholds.

[0011] In the aforementioned implementation process, based on historical strong ground motion records and the final measured maximum intensity value, the threshold boundaries of the three-dimensional composite peak velocity (Pv), peak displacement (Pd), and cumulative absolute velocity (CAV) are jointly determined in the three-dimensional parameter space. Real-time observation data is considered as a whole for collaboratively identifying destructive ground motions, and its boundary delineation directly anchors the statistical correlation between the target warning intensity threshold and the achievement of the measured maximum intensity. Furthermore, the threshold determination process is independent of specific real-time warning calculations, allowing for offline fine-tuning and regional adaptation using abundant historical earthquake examples. This facilitates adjustments to the exceedance probability control underreporting level based on the actual seismic network performance and disaster prevention objectives in various regions.

[0012] Based on the first aspect, the step of assigning the reference intensity value to a target warning location within a preset spatial influence range, centered on the seismic station and based on a preset spatial influence range, to obtain a first predicted intensity value, includes: determining whether the reference intensity value reaches or exceeds the target warning intensity threshold; determining that the reference intensity value meets the preset warning conditions if the reference intensity value reaches or exceeds the target warning intensity threshold; and assigning the reference intensity value to all target warning locations less than or equal to the spatial influence distance threshold from the seismic station as the first predicted intensity value.

[0013] In the above implementation process, the reference intensity value is compared with the target warning intensity threshold. When conditions are met, the reference intensity value is directly assigned to the surrounding target warning location as the first predicted intensity value, centered on the seismic station and based on the spatial influence distance threshold. This achieves rapid definition and efficient coverage of the near-field warning range. In the early warning stage, before the source location is precisely determined, once a station observes a strong earthquake that may cause damage, the intensity estimation result can be immediately issued to the target warning location within a preset range around the station. This minimizes the processing and transmission delay of near-source area warning information, providing extremely valuable early warning time for actions such as personnel evacuation and emergency control of critical facilities in the near-station area. At the same time, by setting a clear spatial influence distance threshold to constrain the range of the first predicted intensity value assignment, false alarms in distant areas and excessive expansion of the warning area caused by the uncontrolled spread of high intensity values ​​from a single station are effectively avoided, ensuring the spatial conservatism and reliability of the first predicted intensity field. Using the target warning intensity threshold as an activation prerequisite further filters out a large number of invalid warnings caused by non-destructive micro-earthquakes or accidental interference, ensuring that radial influence assignment is only initiated when the ground motion level truly poses a potential hazard, significantly improving the relevance of the warning information and the robustness of the system operation.

[0014] Based on the first aspect, the step of treating all seismic stations within the monitoring area as virtual sub-sources, and for the same virtual sub-source, using the reference intensity value to back-calculate the intensity parameter of the virtual sub-source, and determining the second predicted intensity value at the target warning location based on the seismic motion attenuation relationship, includes: using the reference intensity value of the seismic station as the virtual epicenter intensity; using a pre-established regional instrumental seismic intensity attenuation relationship, setting the distance to zero, substituting the virtual epicenter intensity, and inversely solving to obtain the magnitude corresponding to the virtual sub-source, which is used as the intensity parameter; wherein, the attenuation relationship uses magnitude and distance as independent variables and instrumental seismic intensity as the dependent variable.

[0015] In the aforementioned implementation process, without waiting for joint location and magnitude determination by multiple stations, each station triggering an early warning can be independently transformed into a potential virtual sub-source. This reduces the dependence on obtaining complete source information and the time lag effect, enabling the construction of a set of usable virtual earthquake events within a very short time after P-wave triggering. This lays the foundation for subsequently calculating the second predicted intensity value for target early warning locations at any distance based on ground motion attenuation relationships. Simultaneously, each seismic station, as an independent virtual sub-source, can generate intensity estimates for the same target early warning location. The estimation results from multiple virtual sub-sources implicitly contain multi-perspective constraints on the actual source location and attenuation patterns, which helps reduce abnormal deviations caused by local site effects or reference intensity value errors of a single virtual sub-source through subsequent fusion processes. The resulting second predicted intensity value can achieve continuous spatial coverage, effectively filling the gaps in the first predicted intensity value outside the spatial influence range of the station, broadening the effective coverage of early warning information, and enabling target early warning locations far from the station to quickly obtain intensity estimates based on physical attenuation relationships. This enhances the spatial integrity and service fairness of large-scale earthquake early warning.

[0016] Based on the first aspect, the method of considering all seismic stations within the monitoring area as virtual sub-sources, and for the same virtual sub-source, using the reference intensity value to back-calculate the intensity parameters of the virtual sub-source, and determining the second predicted intensity value at the target warning location based on the ground motion attenuation relationship, further includes: calculating the distance from the target warning location to the seismic station serving as the virtual sub-source; substituting the distance and the magnitude obtained by inverse solving into the attenuation relationship to calculate the predicted intensity value of the virtual sub-source at the target warning location; and when the target warning location is simultaneously within the range of action of multiple virtual sub-sources, taking the maximum value among the predicted intensity values ​​calculated by each virtual sub-source as the second predicted intensity value.

[0017] In the aforementioned implementation process, the computational load is small and the response is fast. It can quickly generate a continuous intensity estimation surface covering the entire monitoring area in the early stage of the warning, without spatial blind spots. This allows target warning locations far from all stations to obtain physically based intensity predictions, significantly expanding the spatial service range of the warning information. When the target warning location is simultaneously within the range of multiple virtual sub-sources, the maximum value of the predicted intensity calculated from each virtual sub-source is taken as the second predicted intensity value. From the perspective of disaster prevention and mitigation, this ensures that the estimation of virtual sub-sources that may cause the most adverse impact is not missed, effectively reducing the risk of underestimation caused by the directional effect of the seismic source mechanism, local site anomalies at the station, or back-calculation deviations of the intensity parameters of individual virtual sub-sources. This makes the second predicted intensity value have an overall envelope and safety redundancy for the true hazard. The resulting second predicted intensity value has good spatial gradation and physical interpretability under the constraint of the seismic motion attenuation relationship, complementing the strong constraint characteristics of the first predicted intensity value in the near-source region. The fusion of the two will further enhance the comprehensive performance of the final warning intensity field in terms of accuracy, coverage integrity, and safety of disaster prevention decisions.

[0018] Secondly, embodiments of this application provide an earthquake early warning device, comprising: a reference intensity determination module, a first prediction module, a second prediction module, and a fusion module; the reference intensity determination module is used to acquire real-time observation data from at least one seismic station within a monitoring area, and determine the reference intensity value corresponding to the seismic station based on the real-time observation data; wherein, the reference intensity value includes an estimated maximum intensity value predicted based on P-wave information and / or the measured intensity value of the seismic station; the first prediction module is used, for any target early warning location, if the reference intensity value of the seismic station satisfies a preset... Under the condition of warning, taking the seismic station as the center and based on the preset spatial influence range, the reference intensity value is assigned to the target warning location within the spatial influence range to obtain the first predicted intensity value; the second prediction module is used to regard all seismic stations in the monitoring area as virtual sub-sources, and for the same virtual sub-source, the intensity parameters of the virtual sub-source are calculated back using the reference intensity value, and the second predicted intensity value at the target warning location is determined according to the ground motion attenuation relationship; the fusion module is used to fuse the first predicted intensity value and the second predicted intensity value to generate the final warning intensity field at the target warning location.

[0019] In the above implementation process, the reference intensity determination module continuously extracts and updates the reference intensity values ​​of each seismic station from real-time observation data, providing a unified and dynamically updated input basis for subsequent prediction channels; the first prediction module uses stations that meet the early warning conditions and their spatial influence range to quickly assign intensity values ​​to near-field target early warning locations, ensuring the immediacy of response in high-intensity areas; the second prediction module virtualizes all stations in the region into multiple sub-sources, and back-calculates and extrapolates based on attenuation relationships, giving the device the ability to continuously estimate intensity over large areas or even far-field regions without accurately locating the seismic source; the fusion module organically integrates the two complementary predicted intensity fields, combining the authenticity of near-field observations with the physical completeness of far-field attenuation extrapolation, and finally generates a spatially continuous and intensity value-reliable target early warning location final early warning intensity field.

[0020] Thirdly, embodiments of this application also provide an electronic device, which includes a memory and a processor. The memory stores program instructions, and when the processor reads and runs the program instructions, it executes the steps in any of the above implementation methods.

[0021] This application also provides a computer-readable storage medium storing computer program instructions, which are read and executed by a processor to perform the steps in any of the above implementations. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a first schematic diagram of an earthquake early warning method provided in an embodiment of this application; Figure 2 This is a second schematic diagram of the earthquake early warning method provided in the embodiments of this application; Figure 3 Historical strong vibration acceleration records provided for the application embodiments; Figure 4 A third schematic diagram of the earthquake early warning method provided in the embodiments of this application; Figure 5 A fourth schematic diagram of the earthquake early warning method provided in the embodiments of this application; Figure 6 A comparison chart of three-parameter classification thresholds under different exceedance probabilities with a time window length of 5 s, provided for embodiments of this application; Figure 7 A comparison chart of model performance evaluation metrics under different exceedance probability conditions provided in the embodiments of this application; Figure 8 The fifth schematic diagram is for an embodiment of the earthquake early warning method provided in this application; Figure 9 A sixth schematic diagram of the earthquake early warning method provided in the embodiments of this application; Figure 10 A first schematic diagram of a regional intensity field prediction method using stations as sub-sources provided in an embodiment of this application; Figure 11 A second schematic diagram of a regional intensity field prediction method using stations as sub-sources provided in an embodiment of this application; Figure 12 The seventh schematic diagram is a representation of the earthquake early warning method provided in this application embodiment; Figure 13 The epicenter location map of the selected earthquake provided for the embodiments of this application; Figure 14 The evaluation index provided in this application is a curve showing the change of intensity threshold. Figure 15 The Kumamoto seismic instrument intensity contour map provided in this application embodiment; Figure 16 A comparison chart of the intensity predictions for the Kumamoto earthquake provided in this application embodiment and the PLUM method; Figure 17 Instrumental intensity map of the 6.2 magnitude earthquake in Jishishan, Gansu Province, provided for embodiments of this application; Figure 18 A comparison chart of the predicted intensity and real-time observed intensity of the 6.2 magnitude earthquake in Jishishan, Gansu Province, provided for embodiments of this application; Figure 19 A schematic diagram of an earthquake early warning device provided in an embodiment of this application; Figure 20 This is a block diagram of an electronic device provided in an embodiment of this application.

[0024] Icons: 010-Reference Intensity Determination Module; 011-Data Acquisition Unit; 012-Prediction Calculation Unit; 013-Threshold Judgment Unit; 020-First Prediction Module; 030-Second Prediction Module; 040-Fusion Module; 100-Electronic Device; 111-Memory; 112-Memory Controller; 113-Processor; 114-Peripheral Interface; 115-Input / Output Unit; 116-Display Unit. Detailed Implementation

[0025] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of the embodiments of this application.

[0026] In view of the above, the purpose of this application is to provide an earthquake early warning method, device, electronic device and computer-readable storage medium to improve the above-mentioned problems existing in the prior art.

[0027] In one aspect, this application provides an earthquake early warning method applied to a server, which can be an electronic device with logical computing functions, such as a personal computer (PC), tablet computer, smartphone, or personal digital assistant (PDA).

[0028] Please see Figure 1 , Figure 1 This is a first schematic diagram of an earthquake early warning method provided in an embodiment of this application.

[0029] The earthquake early warning method provided in this application includes: acquiring real-time observation data from at least one seismic station within a monitoring area, and determining a reference intensity value for the corresponding seismic station based on the real-time observation data; wherein the reference intensity value includes an estimated maximum intensity value predicted based on P-wave information and / or a measured intensity value from the seismic station; for any target early warning location: if the reference intensity value of a seismic station meets a preset early warning condition, with the seismic station as the center and based on a preset spatial influence range, the reference intensity value is assigned to the target early warning location within the spatial influence range to obtain a first predicted intensity value; furthermore, all seismic stations within the monitoring area are regarded as virtual sub-sources, and for the same virtual sub-source, the intensity parameters of the virtual sub-source are calculated using the reference intensity value, and a second predicted intensity value at the target early warning location is determined based on the ground motion attenuation relationship; the first predicted intensity value and the second predicted intensity value are fused to generate a final early warning intensity field for the target early warning location.

[0030] In the above implementation process, real-time observation data of seismic stations within the monitoring area are acquired and reference intensity values ​​are determined. The reference intensity values ​​include the estimated maximum intensity value predicted based on P-wave information and / or the measured intensity value of the station. This allows the maximum intensity that the station may experience to be predicted in advance using P-wave information before the measured maximum intensity of the station is reached, thereby gaining more time for subsequent early warning decisions and improving the timeliness of early warning. Meanwhile, for any target warning location, on the one hand, if the reference intensity value of the station meets the preset warning conditions, the reference intensity value is assigned to the target warning location within the preset spatial influence range, centered on the station, to obtain the first predicted intensity value, thus achieving rapid local warning without relying on source parameter estimation; on the other hand, all seismic stations in the monitoring area are regarded as virtual sub-sources, and the intensity parameters of the virtual sub-sources are calculated using the reference intensity value, and the second predicted intensity value at the target warning location is determined according to the ground motion attenuation relationship, so that the warning range can be dynamically adjusted according to the actual observed intensity of the station, avoiding the limitation of warning timeliness on the fixed warning range; finally, the final warning intensity field is generated by fusing the first and second predicted intensity values, effectively balancing the coverage and timeliness of the warning, and the entire warning process does not rely on source parameter estimation, which can solve the problem of missed or false alarms caused by source parameter estimation deviations in traditional source parameter estimation methods under large earthquakes and complex multi-earthquake conditions.

[0031] In one specific embodiment of this application, real-time observation data from at least one seismic station within the monitoring area is acquired. Multiple seismic stations are deployed within the monitoring area, each collecting ground motion data in real time and transmitting it to a server via a communication network. The real-time observation data includes, but is not limited to, acceleration records, velocity records, or displacement records. When a seismic station detects the arrival of a P-wave, the server begins receiving the real-time observation data uploaded by that station. A reference intensity value for the corresponding seismic station is determined based on the real-time observation data. The reference intensity value has two sources: one is the estimated maximum intensity value predicted based on P-wave information, and the other is the measured intensity value from the seismic station. Either one can be used alone, or both can be used in combination. Spatial intensity prediction is performed for any target warning location. The target warning location can be a pre-divided grid point within the monitoring area, an administrative division unit, or the geographical location of the user terminal.

[0032] Spatial intensity prediction is performed through the following two parallel sub-steps: First, determine whether any seismic station's reference intensity value meets the preset early warning conditions (e.g., the reference intensity value reaches or exceeds the preset target early warning intensity threshold). If so, using the seismic station as the center, and based on a preset spatial influence range (e.g., a circular area with a radius of 30 kilometers), assign the reference intensity value to all target early warning locations within the spatial influence range to obtain the first predicted intensity value. When a target early warning location is simultaneously within the spatial influence range of multiple seismic stations that meet the conditions, take the maximum value among all corresponding reference intensity values ​​as its first predicted intensity value.

[0033] Second, each seismic station within the monitoring area is considered a virtual sub-source. For each virtual sub-source, the reference intensity value of that station is used as the virtual epicenter intensity. Using a pre-established regional instrumental seismic intensity attenuation relationship (with magnitude and distance as independent variables and instrumental seismic intensity as the dependent variable), the distance is set to zero and substituted into the inverse kinematics solution to obtain the magnitude corresponding to the virtual sub-source as the intensity parameter. Then, the distance from the target warning location to the virtual sub-source is calculated. The distance and the magnitude obtained from the inverse kinematics solution are substituted into the attenuation relationship to calculate the predicted intensity value of the virtual sub-source at the target warning location. When the target warning location is simultaneously within the range of multiple virtual sub-sources, the maximum value among the predicted intensity values ​​calculated from each virtual sub-source is taken as the second predicted intensity value.

[0034] Based on the execution results of two parallel sub-steps, for the same target warning location, the first predicted intensity value and the second predicted intensity value are fused to generate the final warning intensity field for that target warning location. The fusion method can be to take the larger of the two values, or other fusion strategies such as weighted averaging can be used.

[0035] Optionally, the above steps can be executed once every preset time interval (e.g., 1 second) after the first trigger. As the number of triggering stations increases and the time window length increases, the spatial warning area is continuously updated to achieve real-time dynamic prediction of the seismic intensity field.

[0036] In one embodiment of this application, the specific method for fusing the first and second predicted intensity values ​​to generate the final warning intensity field of the target warning location can be as follows: When the number of triggering stations in the monitoring area is small, the data basis for back-calculating the virtual sub-source intensity is insufficient, and the reliability of the second predicted intensity value is low. In this case, the first predicted intensity value dominates the fusion result. As the number of triggering stations gradually increases, the stability and reliability of the virtual sub-source back-calculation improve, and the weight of the second predicted intensity value increases accordingly, gradually becoming the main contributor to the fusion result. At the same time, the distance between the target warning location and the nearest triggering station is also taken into consideration. The greater the distance, the worse the spatial representativeness of the direct assignment by a single station, and the weight of the first predicted intensity value decreases accordingly. By combining the station quantity factor and the distance attenuation factor, a dynamically adjusted weight coefficient is formed, and finally, the two intensity values ​​are weighted and synthesized.

[0037] In one embodiment of this application, the specific method for generating the final warning intensity field of the target warning location by fusing the first predicted intensity value and the second predicted intensity value can be as follows: starting from the fundamental need to prevent missed earthquake warnings, without introducing additional parameters, directly select the larger of the first and second predicted intensity values ​​as the final warning intensity. The methodological basis for this is that the first predicted intensity value relies on a single P-wave information or measured data for rapid response and has a strong ability to capture near the epicenter; the second predicted intensity value relies on multiple inversion and attenuation relationships, has better regional continuity, but may lead to a lower local intensity estimate due to inversion errors or attenuation model deviations. Taking the envelope value of both ensures that at any target location, at least the highest intensity estimate given by both methods is obtained, thereby maximizing the safety of the warning.

[0038] In one embodiment of this application, the specific method for generating the final warning intensity field of the target warning location by fusing the first predicted intensity value and the second predicted intensity value may be as follows: the degree of difference between the first predicted intensity value and the second predicted intensity value at the same target location is determined: if the difference between the two is small, it indicates that the conclusions obtained by the two independent methods are mutually corroborated, and the average of the two is taken to reduce the random error that may be introduced by a single method; if the difference between the two is significantly greater than the preset tolerance range, it is determined that at least one of them has an abnormal estimate, and the second predicted intensity value with stronger physical constraints and a broader data base is trusted first, while reserving room for upward correction to ensure that the actual ground motion intensity is not underestimated due to method differences.

[0039] In one embodiment of this application, the specific method for generating the final warning intensity field of the target warning location by fusing the first and second predicted intensity values ​​can be as follows: Two key distance thresholds are defined to divide the spatial relationship between the target warning location and the nearest triggering station into three regions: a near-field confidence region, a transition region, and a far-field extrapolation region. Within the near-field confidence region, the first predicted intensity value is directly used because the result of a single station's assignment has high local representativeness. In the far-field extrapolation region, the spatial influence range of a single station no longer has reasonable extrapolation capability, and the second predicted intensity value calculated based on the attenuation relationship of multiple stations is relied upon entirely. In the transition region between the two, a smooth weighting is performed based on the relative distance between the target location and the two boundaries, so that the results of the two methods are naturally connected spatially, avoiding abrupt changes in intensity values ​​at the boundary of the zones.

[0040] Please see Figure 2 , Figure 2 This is a second schematic diagram of an earthquake early warning method provided in an embodiment of this application.

[0041] Based on the first aspect, real-time observation data from at least one seismic station within the monitoring area is acquired, and the reference intensity value of the corresponding seismic station is determined based on the real-time observation data. This includes: acquiring real-time observation data within multiple increasing time windows after the P-wave is triggered at the seismic station, wherein the real-time observation data includes at least the triaxial composite peak velocity. Peak displacement and cumulative absolute velocity Substitute the real-time observation data into the statistical regression relationship to obtain the intensity prediction value corresponding to each parameter; for the current time window length, select the intensity prediction values ​​corresponding to the two parameters with smaller regression standard deviations in the real-time observation data, and take their average value as the estimated maximum intensity value under this time window.

[0042] Understandably, obtaining velocity and displacement time histories is fundamental for performing seismic motion analysis. In a specific embodiment of this application, the acceleration record is first integrated to obtain the velocity time history. Then, the velocity record is integrated again to obtain the displacement time history, and a 0.075 Hz fourth-order Butterworth high-pass filter is applied to the displacement record to eliminate low-frequency interference.

[0043] In a specific embodiment of this application, the three-component composite acceleration, velocity, and displacement within a 3-second time window after the initial arrival of the P-wave are calculated using the following formula:

[0044] in, , , These are the time histories of the east-west component acceleration, the north-south component acceleration, and the vertical component acceleration (cm / s).2 ), The time history of the three-component composite acceleration is given.

[0045]

[0046] in, , , These are the time histories of the east-west component velocity, the north-south component velocity, and the vertical component velocity (cm / s), respectively. The three-component composite velocity time history is given.

[0047]

[0048] in, , , These are the time histories of the east-west component displacement, the north-south component displacement, and the vertical component displacement (cm), respectively. This is the time history of the three-component composite displacement.

[0049] Among them, peak speed The formula for calculation is: Peak displacement The formula for calculation is: Cumulative absolute acceleration The calculation formula is .

[0050] In the above implementation process, real-time observation data, including the triaxial composite peak velocity Pv, peak displacement Pd, and cumulative absolute velocity CAV, are acquired within multiple incremental time windows after the P-wave is triggered at the seismic station. The intensity prediction values ​​corresponding to each parameter are obtained using statistical regression relationships. For the current time window length, the average of the prediction values ​​corresponding to the two parameters with smaller regression standard deviations is selected as the estimated maximum intensity value. On the one hand, this can provide an estimate of the maximum intensity that the station may suffer in the future within a very short time after the arrival of the P-wave, thus gaining valuable time for early warning decisions. On the other hand, since the prediction accuracy of different characteristic parameters varies under different time window lengths, averaging the two parameters with smaller standard deviations dynamically can effectively reduce the uncertainty of single parameter prediction and improve the accuracy and robustness of the estimated maximum intensity value.

[0051] In one specific embodiment of this application, please refer to Figure 3 , Figure 3 Historical strong ground motion acceleration records are provided for the embodiments of this application, and Table 1 is attached, which is a table of coefficients for the on-site intensity prediction model.

[0052] Table 1. Coefficients of the On-Site Intensity Prediction Model

[0053] In this embodiment, when a P-wave trigger is detected at a seismic station, real-time observation data of the station is continuously acquired within multiple incremental time windows (e.g., 3 seconds, 4 seconds, 5 seconds, etc.) starting from the trigger time. The real-time observation data includes at least the three-dimensional composite peak velocity Pv, peak displacement Pd, and cumulative absolute velocity CAV. Specifically, the three-dimensional composite peak velocity Pv is the maximum value of the velocity time history synthesized from the east-west, north-south, and vertical components within a given time window; the peak displacement Pd is the maximum value of the displacement time history synthesized from the three directional components within a given time window; and the cumulative absolute velocity CAV is the integral value of the absolute acceleration of the synthesized three directional components within a given time window. The three characteristic parameters mentioned above are substituted into a pre-established statistical regression relationship, which is obtained by fitting a large number of station records from historical earthquake events. Specifically, using historical strong ground motion data, for each time window length, a first regression equation is established with the logarithm of Pv as the independent variable and the measured maximum instrumental seismic intensity as the dependent variable; a second regression equation is established with the logarithm of Pd as the independent variable and the measured maximum instrumental seismic intensity as the dependent variable; and a third regression equation is established with the logarithm of CAV as the independent variable and the measured maximum instrumental seismic intensity as the dependent variable. The standard deviation of each regression equation is recorded. This statistical regression relationship is pre-stored in the server for real-time retrieval.

[0054] Optionally, based on the peak velocity of different time window lengths after P-wave triggering ( ), peak displacement ( ) and cumulative absolute velocity ( The statistical relationship between seismic intensity and measured seismic intensity is as follows:

[0055] The first regression equation is: The second regression equation is: The third regression equation is: . Earthquake intensity measured by Chinese instruments; , , These represent the peak velocity (cm / s), peak displacement (cm), and cumulative absolute velocity (cm / s) of the three components under a specific time window length. , , , , , The regression coefficients are for a specific time window length. The standard deviation is denoted as .

[0056] For the current time window length, calculate respectively , , The intensity prediction values ​​corresponding to real-time observation data are obtained, and the standard deviations of the three regression equations for that time window are acquired. The magnitudes of the three standard deviations are compared, and the intensity prediction values ​​corresponding to the two parameters with smaller standard deviations are selected. Their arithmetic mean is taken as the estimated maximum intensity value for that time window. For example, if the regression standard deviations of Pd and CAV are less than the regression standard deviation of Pv in the current time window, then the average of the intensity prediction values ​​corresponding to Pd and CAV is taken as the final estimated maximum intensity value.

[0057] In this way, the server can continuously update the estimated maximum intensity value after the P wave is triggered as the time window length increases. Each update uses the two parameter combinations with the best prediction accuracy under the current time window, thereby improving the prediction accuracy as much as possible while ensuring the timeliness of the prediction.

[0058] Please see Figure 4 , Figure 4 This is a third schematic diagram of the earthquake early warning method provided in the embodiments of this application.

[0059] Based on the first aspect, acquiring real-time observation data from at least one seismic station within the monitoring area and determining the reference intensity value of the corresponding seismic station based on the real-time observation data further includes: comparing each real-time observation data calculated from the current time window length with a preset parameter threshold corresponding to the target warning intensity threshold; and selecting the estimated maximum intensity value as the reference intensity value if any real-time observation data reaches or exceeds its corresponding preset parameter threshold.

[0060] In the above implementation process, each real-time observation data calculated from the current time window length is compared with a preset parameter threshold corresponding to the target warning intensity threshold. When any real-time observation data reaches or exceeds its corresponding threshold, the estimated maximum intensity value is selected as the reference intensity value. On the one hand, this avoids the problem of "overestimation of small earthquakes" caused by the uncertainty of the local intensity prediction model. That is, by using the measured waveform parameters as a pre-filtering condition, the estimated intensity value is only accepted when the actual seismic energy received by the station is large enough, thereby reducing the risk of false alarms in the system. On the other hand, this discrimination mechanism allows that only one of the three characteristic parameters needs to meet the standard to pass, avoiding missed judgments due to the insensitivity of a single parameter. While suppressing false alarms, it retains as much effective warning information as possible, achieving a balance between the accuracy and timeliness of the warning.

[0061] In one specific embodiment of this application, a parameter threshold corresponding to the target warning intensity threshold is preset. The target warning intensity threshold is the minimum intensity standard for triggering a warning (e.g., 6 degrees of seismic intensity according to Chinese instruments). For this target warning intensity threshold and each time window length, the Pv threshold, Pd threshold, and CAV threshold are pre-determined using statistical methods and stored in a server for real-time retrieval. During real-time warning, for the current time window length, the calculated real-time observation data of Pv, Pd, and CAV are compared with the preset corresponding thresholds. The comparison uses an "OR" logic, i.e., when Pv... Pv threshold, or Pd Pd threshold, or CAV Under the CAV threshold, it is determined that the discrimination condition is met:

[0062] in, This is the measured value of CAV. The CAV threshold; This is the measured value of Pv. Pv threshold; This is the measured value of Pd. The threshold value for Pd is [value].

[0063] At this point, the calculated estimated maximum intensity value can be used as a reference intensity value for the station and proceed to the subsequent early warning judgment process.

[0064] In a specific embodiment of this application, if the real-time observation data of Pv, Pd, and CAV do not reach their respective thresholds, it is determined that the discrimination condition is not met. In this case, the estimated maximum intensity value under that time window is rejected and not used as a reference intensity value. At this time, one of the following two technical paths is selected according to actual needs: First, the server can wait for the next longer time window length and repeat the above steps until the estimated maximum intensity value under a certain time window passes the discrimination; Second, if the measured intensity value of the station has reached the usable standard, the measured intensity value is directly used as the reference intensity value.

[0065] By using the above method, the predicted intensity value based on P-wave information is only accepted when the actual earthquake motion amplitude observed by the seismic station reaches a certain intensity. This effectively suppresses the overestimation of intensity that may occur under small earthquake conditions due to statistical models, reduces the overall false alarm rate of the system, and since the discrimination condition is that only one of three can pass, the timeliness benefits brought by the on-site intensity prediction are preserved to the greatest extent while controlling false alarms.

[0066] Please see Figure 5 , Figure 5 This is a fourth schematic diagram of the earthquake early warning method provided in the embodiments of this application.

[0067] Based on the first aspect, the preset parameter thresholds include the three-dimensional synthesis peak velocity. Threshold, peak displacement Threshold and cumulative absolute velocity Threshold; a preset target warning intensity threshold is obtained. For a given time window length, strong motion record samples from multiple stations in historical earthquakes are acquired. Each sample contains real-time observation data within the time window length and the maximum intensity value finally measured by the seismic station. A spatial region defined by three thresholds is determined in the three-dimensional parameter space composed of real-time observation data. The boundary values ​​corresponding to this spatial region on the real-time observation data are used as preset parameter thresholds.

[0068] In the aforementioned implementation process, strong ground motion records from multiple stations during historical earthquakes are used to statistically determine preset parameter thresholds within a three-dimensional parameter space composed of real-time observation data. This approach ensures that the threshold settings have clear physical meaning and statistical basis; that is, the thresholds are not empirically set but objectively determined based on the distribution patterns of measured intensity reaching the target warning intensity from a large amount of historical data, thus improving the reliability and interpretability of the discrimination criteria. Furthermore, by defining spatial regions bounded by three thresholds within the three-dimensional parameter space, three characteristic parameters with different physical dimensions are incorporated into a unified discrimination framework, avoiding the one-sidedness of single-parameter discrimination. Moreover, the boundary values ​​of the real-time observation data thresholds directly derive from the statistical distribution of measured data, accurately reflecting the differences in the sensitivity of different ground motion characteristic parameters to the target intensity.

[0069] In one embodiment of this application, a target warning intensity threshold is preset. This threshold is an intensity limit value used to determine whether a warning message needs to be sent to the public, such as a China instrumental seismic intensity of 6 degrees. This threshold can be set according to actual warning needs and pre-stored in a server. For a given time window length (e.g., 3 seconds, 4 seconds, 5 seconds, etc.), strong ground acceleration data recorded by multiple seismic stations during historical earthquake events are collected as statistical samples. Each sample includes three real-time observation data points recorded by the station within that time window: the three-dimensional composite peak velocity Pv, peak displacement Pd, and cumulative absolute velocity CAV, as well as the maximum instrumental seismic intensity value ultimately measured by the station during the entire earthquake process. Based on the relationship between the measured maximum intensity value and the target warning intensity threshold, all samples are divided into two categories: samples whose measured maximum intensity value reaches or exceeds the target warning intensity threshold are positive samples, and the rest are negative samples. A three-dimensional parameter space is constructed with Pv, Pd, and CAV as the three coordinate axes, and each sample is mapped to a data point in this space according to its corresponding real-time observation data value. In this three-dimensional parameter space, a spatial region (e.g., a cuboid region with its origin at the zero point and each boundary value representing an intercept on one of the three coordinate axes) is defined by three boundary values: Pv threshold, Pd threshold, and CAV threshold. Samples falling outside this region are those that satisfy any one of the following conditions: Pv reaches or exceeds the Pv threshold, Pd reaches or exceeds the Pd threshold, or CAV reaches or exceeds the CAV threshold.

[0070] The percentage of positive samples among all samples falling outside the designated area is calculated. This percentage is then adjusted by changing the values ​​of the three threshold boundaries to achieve a predetermined requirement (e.g., the exceedance probability value of 60% selected in the paper, meaning 60% of the samples falling outside the designated area are positive samples). The corresponding boundary values ​​on the three coordinate axes are then the Pv threshold, Pd threshold, and CAV threshold for that time window and the corresponding warning intensity threshold for that target.

[0071] The aforementioned three-parameter threshold combination is associated with the corresponding time window length and target warning intensity threshold and stored in the server. During real-time warning, the corresponding threshold combination is directly invoked for judgment based on the current time window length and target warning intensity threshold.

[0072] In the above manner, the process of determining the preset parameter threshold makes full use of the statistical regularity of historical earthquake data, so that the judgment criteria can objectively reflect the physical fact that "when the local ground motion observation parameters reach what level, the station is likely to encounter the target intensity sufficiently".

[0073] Meanwhile, since thresholds are established for different time window lengths, the judgment criteria are dynamically adjusted as P-wave information accumulates, ensuring that reasonable judgments that match the amount of available information can be made at different warning stages.

[0074] In one embodiment of this application, to avoid false alarms caused by overestimation of local intensity, the measured intensity of 6 degrees is used as the classification boundary. Historical strong ground motion observation data are used to statistically analyze the exceedance probability of intensity 6 degrees and its corresponding three-parameter threshold combination. Please refer to Table 2, which shows the three-parameter thresholds corresponding to the 60% exceedance probability under different time window lengths.

[0075] Table 2. Thresholds for the 60% exceedance probability under different time window lengths (corresponding to 3 parameters).

[0076] In this embodiment, the on-site intensity prediction result is accepted only if the measured three parameters meet the three-parameter threshold discrimination condition corresponding to a certain exceedance probability; otherwise, the prediction result is not accepted. Please refer to [link to relevant documentation]. Figure 6 , Figure 6 The comparison chart of the three-parameter classification thresholds under different exceedance probabilities with a time window length of 5 s provided in the embodiments of this application shows the three-parameter thresholds with an exceedance probability of 90% for a degree of 6 under a time window length of 5 s. That is, when the proportion of samples with a measured intensity greater than or equal to 6 degrees among all samples falling outside the gray cube is 90%, the maximum value intercepted by the gray cube on the three coordinate axes is the three thresholds with an exceedance probability of 90%.

[0077] Optionally, an exceedance probability value is set, and a spatial region defined by three thresholds is determined in the three-dimensional parameter space composed of the real-time observation data, such that among all samples falling outside the region, the proportion of samples whose measured maximum intensity value reaches or exceeds the target warning intensity threshold is equal to the exceedance probability value; the exceedance probability value is used to control the balance between accuracy and timeliness when the estimated maximum intensity value is used as the reference intensity value.

[0078] In the above implementation process, an exceedance probability value is set, and a spatial region defined by three thresholds is determined in the three-dimensional parameter space composed of real-time observation data. The proportion of samples whose measured maximum intensity value reaches or exceeds the target warning intensity threshold among all samples falling outside the region is equal to the exceedance probability value. On the one hand, the trade-off between accuracy and timeliness of the warning system is transformed into a quantifiable and adjustable statistical parameter. The higher the exceedance probability value, the higher the proportion of samples falling outside the region that actually reach the target intensity, i.e., the higher the confidence in accepting the estimated intensity value and the lower the risk of false alarm. However, at the same time, the number of samples to be judged is reduced and the conditions are more stringent, which may delay some situations that could have been warned earlier, and the average warning time is shortened accordingly. Conversely, the lower the exceedance probability value, the more lenient the judgment conditions and the higher the timeliness of the warning, but the risk of false alarm increases accordingly. On the other hand, by introducing the exceedance probability value as a single control parameter, the system can flexibly adjust the warning under different application scenarios. A higher exceedance probability value can be used for scenarios with low tolerance for false alarms, and a lower exceedance probability value can be used for scenarios with high timeliness requirements, without having to re-formulate the entire set of judgment rules.

[0079] In a specific embodiment of this application, an exceedance probability value is set. This exceedance probability value represents the conditional probability that, in the three-dimensional parameter space, when any parameter of a sample reaches or exceeds the corresponding threshold (i.e., the sample falls outside the spatial region defined by the threshold), the measured maximum intensity value of the sample truly reaches or exceeds the target warning intensity threshold. For example, an exceedance probability value of 60% means that among all samples that satisfy the condition "Pv ≥ Pv threshold, or Pd ≥ Pd threshold, or CAV ≥ CAV threshold", 60% of the samples' measured maximum intensity does indeed reach the target warning intensity threshold.

[0080] In a three-dimensional parameter space composed of real-time observation data (Pv, Pd, CAV), three threshold boundaries are set along the Pv, Pd, and CAV axes, starting from the origin. These three threshold boundaries collectively define a cuboid space region. For samples outside this region, the proportion of samples whose measured maximum intensity value reaches or exceeds the target warning intensity threshold is calculated. By iteratively adjusting the three threshold boundary values, this proportion is made equal to the set exceedance probability value. During the adjustment process, the relative proportion of the real-time observation data thresholds can be fixed (e.g., using the quantiles of each parameter's distribution in the positive samples as a reference), and the threshold boundaries can be scaled overall until the proportion requirement is met. At this point, the boundary values ​​on the three coordinate axes are the Pv, Pd, and CAV thresholds corresponding to the exceedance probability value. In practical applications, the selection of the exceedance probability value needs to comprehensively consider the timeliness and accuracy requirements of the warning system.

[0081] Please see Figure 7 , Figure 7A comparison chart of model performance evaluation metrics under different exceedance probability conditions provided in the embodiments of this application.

[0082] In one embodiment of this application, through retrospective analysis of historical data, the average warning time, false alarm rate, and missed alarm rate under different exceedance probability values ​​were statistically analyzed. Finally, an exceedance probability value of 60% was selected as the practical discrimination threshold. This value ensures high warning timeliness while keeping the false alarm rate and missed alarm rate at a low level. In actual deployment, an appropriate exceedance probability value can be selected based on factors such as seismic activity, seismic network density, and social tolerance in the specific warning area. The corresponding determined three-parameter thresholds are pre-stored in the system for real-time retrieval.

[0083] When the platform is If the maximum intensity observed in an earthquake is greater than or equal to 6 degrees, then the time by which the predicted intensity of 6 degrees is predicted based on the information from the station a few seconds prior to the actual measured intensity reaching 6 degrees is ( ). The warning time (in seconds) is defined as follows: if the predicted intensity remains below 6 degrees until the measured intensity of 6 degrees is reached, the warning time is 0. For all records, the average warning time is... It can be represented as:

[0084] in, This represents the number of stations with an observed intensity greater than or equal to 6 degrees in all records. For the first Warning time (s) for each station; The average warning time is in seconds.

[0085] A false alarm is defined as an earthquake in which the maximum measured seismic intensity at a station is less than 6 degrees, but the predicted intensity is greater than 6 degrees. The false alarm rate is defined as the percentage of records with a maximum observed intensity less than 6 degrees where the predicted intensity is greater than 6 degrees.

[0086] In this way, the exceedance probability value provides a unified and interpretable adjustment parameter for the three-parameter threshold discrimination mechanism, enabling the early warning system to make quantitative strategy choices between "preferring false alarms to early warnings" and "preferring missed alarms to avoid false alarms" according to actual needs, thereby enhancing the system's practicality and flexible adaptability.

[0087] in, False alarm rate; This refers to the number of stations where the intensity of observation is less than 6 degrees. As a false alarm criterion, among all stations with a maximum observed intensity of less than 6 degrees, when the first... The value is 1 if the predicted intensity of a station is greater than or equal to 6 degrees, and 0 otherwise. For the first Maximum observation intensity at each station; For the first Maximum predicted intensity for each station.

[0088] Missed reports are defined as the maximum instrumented seismic intensity measured at station j during an earthquake. Greater than 6 degrees, but predicted intensity A reading less than 6 degrees indicates a false alarm. The false negative rate is defined as the percentage of records with a predicted intensity of less than 5 degrees out of all observed maximum intensity greater than 6 degrees.

[0089] in, The false negative rate; To identify underreporting criteria, among all records with observed intensity greater than or equal to 6 degrees, when the first... The value is 1 if the predicted intensity of a station is less than 6 degrees, and 0 otherwise.

[0090] according to Figure 7 The higher the exceedance probability threshold, the shorter the average warning time, but the higher the false alarm rate; the false alarm rate, however, does not show a clear pattern. Furthermore, without a threshold, the average warning time actually decreases because the increase in false alarm samples reduces the number of correct warning samples that can provide a longer warning time. In this application's embodiment, considering the optimal balance between high timeliness, low false alarm rate, and low false alarm rate, an exceedance probability of 60% was selected as the threshold for practical on-site intensity prediction.

[0091] Please see Figure 8 , Figure 8 This is a fifth schematic diagram of the earthquake early warning method provided in the embodiments of this application.

[0092] Based on the first aspect, when the reference intensity value of an existing seismic station meets the preset early warning conditions, taking the seismic station as the center and based on the preset spatial influence range, the reference intensity value is assigned to the target early warning location within the spatial influence range to obtain the first predicted intensity value. This includes: determining whether the reference intensity value reaches or exceeds the target early warning intensity threshold; if the reference intensity value reaches or exceeds the target early warning intensity threshold, determining that the reference intensity value meets the preset early warning conditions; and setting a spatial influence distance threshold, assigning the reference intensity value to all target early warning locations that are less than or equal to the spatial influence distance threshold from the seismic station as the first predicted intensity value.

[0093] In the above implementation process, it is determined whether the reference intensity value reaches or exceeds the target warning intensity threshold. When the condition is met, the reference intensity value is assigned to all target warning locations within the spatial influence range based on the earthquake station as the center and a preset spatial influence distance threshold, serving as the first predicted intensity value. This achieves rapid local warning without relying on source parameter estimation. It eliminates the need to wait for estimations of parameters such as the earthquake epicenter location and magnitude. Warning information can be provided to the surrounding area of ​​a single station based solely on the measured or predicted intensity value, avoiding the problems of missed or false alarms caused by source parameter estimation errors in traditional methods under large earthquakes and complex multi-seismic conditions. The use of a preset spatial influence distance threshold (e.g., 30 kilometers) to define the warning range is based on the physical assumption that local propagation of ground motion has no attenuation. It is assumed that the attenuation of ground motion within this distance range is negligible. Therefore, the intensity observed by the station can represent the intensity level of its surrounding area, thus achieving reliable regional warning with low computational complexity.

[0094] In a specific embodiment of this application, the reference intensity values ​​of each seismic station determined by the aforementioned steps are obtained. These reference intensity values ​​may be the estimated maximum intensity values ​​after being filtered by three-parameter threshold discrimination, or they may be the measured intensity values ​​of the stations. The intensity values ​​that are available and reliable at the current time are selected.

[0095] The reference intensity values ​​of each station are compared with a preset target warning intensity threshold, which is the minimum intensity standard for triggering a warning (e.g., 6 degrees according to Chinese instrument seismic intensity). If the reference intensity value of a station reaches or exceeds this threshold, the station is deemed to meet the preset warning conditions and is marked as a warning triggering station. A preset spatial influence distance threshold is set based on the assumption of no attenuation in local ground motion propagation (e.g., 30 kilometers), indicating that ground motion attenuation within this distance range centered on the station is negligible, and the intensity at the station is approximately equal to the intensity in the surrounding area.

[0096] For each target warning location within the monitoring area, the spatial distance to each warning triggering station is calculated. The spatial distance can be calculated using the great circle distance formula or the simplified planar distance formula. If the distance from the target warning location to a certain warning triggering station is less than or equal to the spatial influence distance threshold, the reference intensity value of that station is assigned to the target warning location as a candidate first predicted intensity value. When a target warning location is simultaneously within the spatial influence range of multiple warning triggering stations, based on the principle of safety priority in earthquake early warning, the most conservative estimate is taken to maximize personnel safety; therefore, the maximum value among all assigned candidate intensity values ​​is taken as the final first predicted intensity value for that target warning location. Through this method, the generation of the first predicted intensity value requires no source parameter information, relying only on the station's own observed or estimated intensity value and the preset spatial influence distance. The algorithm is simple in logic and highly efficient, making it suitable for rapid deployment and operation in real-time earthquake early warning systems. Furthermore, since the warning triggering condition is directly based on the target warning intensity threshold, the timing of warning activation can be effectively controlled, avoiding public fatigue caused by frequent warnings triggered by low-intensity events.

[0097] Please see Figure 9 , Figure 9 The sixth schematic diagram is a diagram illustrating the earthquake early warning method provided in the embodiments of this application.

[0098] Based on the first aspect, all seismic stations within the monitoring area are considered as virtual sub-sources. For the same virtual sub-source, the intensity parameters of the virtual sub-source are calculated using reference intensity values. The second predicted intensity value at the target warning location is determined based on the seismic motion attenuation relationship. This includes: using the reference intensity values ​​of the seismic stations as the virtual epicenter intensity; using the pre-established regional instrumental seismic intensity attenuation relationship, setting the distance to zero, substituting the virtual epicenter intensity, and inversely solving to obtain the magnitude corresponding to the virtual sub-source, which is used as the intensity parameter; wherein, the attenuation relationship uses magnitude and distance as independent variables and instrumental seismic intensity as the dependent variable.

[0099] In the above implementation process, the reference intensity values ​​of seismic stations are used as virtual epicenter intensity. Utilizing a pre-established regional instrument seismic intensity attenuation relationship, the distance is set to zero and substituted into the inverse solution of the virtual epicenter intensity to obtain the magnitude corresponding to the virtual sub-source as the intensity parameter. This allows each station with a reliable reference intensity value to independently function as the epicenter of a virtual seismic event, eliminating the need to rely on estimations of real source parameters (such as epicenter location, magnitude, rupture scale, and strike) as in traditional methods. This fundamentally avoids intensity field prediction errors caused by biases in source parameter estimation. Furthermore, by mapping the station's observed intensity to a virtual magnitude through the inverse attenuation relationship, mature seismic intensity attenuation laws can be utilized to extend the local observation information of a single station to a wider spatial range. This allows the warning coverage to be adaptively adjusted according to the actual observed intensity of the station. The higher the observed intensity of a station, the higher the virtual magnitude calculated, and the larger the predicted high-intensity impact range through the attenuation relationship. This solves the problem of insufficient timeliness in warnings of high-intensity events using fixed warning range methods.

[0100] In a specific embodiment of this application, reference intensity values ​​for each seismic station determined through the aforementioned steps are obtained. To ensure the reliability of the second prediction, stations selected as virtual sub-sources are typically those whose reference intensity values ​​have passed the three-parameter threshold discrimination screening or have been measured to reach a certain level. For each seismic station selected as a virtual sub-source, its reference intensity value is used as the virtual epicenter intensity I0 of that virtual sub-source. The virtual epicenter intensity indicates that, assuming an earthquake occurred at the location of the station, the instrumental seismic intensity at that location is I0.

[0101] The pre-established regional instrumental seismic intensity attenuation relationship is retrieved. This attenuation relationship is an empirical formula obtained by statistical regression of historical earthquake data in the target warning area, with magnitude M and distance R as independent variables and instrumental seismic intensity I as the dependent variable.

[0102] Where M is the magnitude, R is the epicentral distance or line source distance (in km), and I is the seismic intensity measured by Chinese instruments.

[0103] Setting the distance parameter R = 0 and substituting it into the attenuation relation, the attenuation relation degenerates into a single-variable equation containing only the magnitude M. Substituting the virtual epicenter intensity I0 into the left side of the equation, the magnitude M is obtained by inverse solving. For example, setting R = 0 in the above attenuation relation, we get:

[0104] By solving this quadratic equation (taking a positive root within a reasonable range), the magnitude M corresponding to the virtual sub-source is obtained. This magnitude M is the intensity parameter of the virtual sub-source. The intensity parameter comprehensively reflects the energy level of the ground motion at the station and is the basis for subsequent calculations of spatial attenuation.

[0105] Please see Figure 10 , Figure 10 This is a first schematic diagram of a regional intensity field prediction method using stations as sub-sources, provided in an embodiment of this application.

[0106] The regional intensity field prediction method using stations as sub-sources assumes that each station is the epicenter of an earthquake (sub-source). It calculates the magnitude of the sub-source using station observation data and predicts the intensity at grid points without stations by combining attenuation relationships. The specific process is as follows: For any triggered station i, let its current sub-source intensity be denoted as . In the embodiments of this application, The intensity of the epicenter of the sub-source is determined by the current observed intensity at the station and the intensity prediction result accepted after dynamic threshold discrimination. That is, the observed intensity or the predicted maximum intensity (the predicted intensity after threshold discrimination criterion) after the P-wave is triggered at the station.

[0107] If the predicted intensity at a certain moment passes the threshold discrimination, the larger value between the predicted value and the current observed intensity is used as the equivalent sub-source intensity of the station; otherwise, only the current observed intensity is used.

[0108] In other words, for the same sub-source, the intensity of grid points without stations is predicted using this attenuation relationship and the magnitude term, with the epicentral distance used. When the observed intensity from multiple stations exceeds 6 degrees, for the same grid point, the maximum estimated intensity from multiple stations at that point is taken as the estimated intensity of that grid point.

[0109] Through the above method, the inverse calculation process of the intensity parameters of the virtual sub-source establishes a quantitative mathematical relationship between the intensity of the seismic station and its attenuation. This allows each effective station to independently generate an intensity prediction field that attenuates outward from its own location, laying the foundation for subsequent multi-source fusion prediction. This process avoids the limitation of traditional methods that require determining the actual source location first, and is particularly suitable for rapid parallel computation in situations with high seismic network density and simultaneous triggering by multiple stations.

[0110] The aforementioned method uses the seismic intensity attenuation relation of Chinese instruments to quantify the change in intensity with distance, and its expression is as follows:

[0111] Where I represents the seismic intensity measured by Chinese instruments; M represents the magnitude; and R represents the epicentral distance (km).

[0112] Define the magnitude term as At distance In this case, the intensity at the source location can be determined. The magnitude term corresponding to this sub-source is derived as follows: .

[0113] Furthermore, for any target grid point x, let its distance from sub-source i be . Then the intensity contribution of this sub-source to grid point x is:

[0114] If a grid point is simultaneously affected by multiple sub-sources, its final predicted intensity is controlled by the maximum value of the contributions from each sub-source, i.e.:

[0115] in, Let x be the predicted intensity of the earthquake early warning method provided in the embodiments of this application, obtained at grid point x. This represents the number of triggered sub-sources participating in the calculation at the current moment.

[0116] Specifically, please combine Figure 10 See Figure 11 , Figure 11 This is a second schematic diagram of a regional intensity field prediction method using stations as sub-sources, provided in an embodiment of this application.

[0117] Two triggered station sub-sources, denoted as the first sub-source and the second sub-source, are distributed on a spatially regular grid, along with two grid points k and j to be observed. The predicted intensity corresponding to the first sub-source is... The predicted intensity corresponding to the second sub-source is .

[0118] The distances from the two sub-sources to the grid points to be measured are denoted as R1, R2, R3, and R4, respectively.

[0119] The magnitude term obtained after source intensity inversion is: and .

[0120] We can obtain the attenuation relationship of the first sub-source with respect to grid points k and j as follows: and .

[0121] Similarly, the attenuation relationship of the second sub-source for grid points k and j is as follows: and .

[0122] Extrapolating spatially through the attenuation relationship and taking the maximum value among multiple sub-sources yields the predicted intensity of grid points k and j. and .

[0123] Please see Figure 12 , Figure 12 The seventh schematic diagram is a representation of the earthquake early warning method provided in the embodiments of this application.

[0124] Based on the first aspect, all seismic stations within the monitoring area are considered as virtual sub-sources. For the same virtual sub-source, the intensity parameters of the virtual sub-source are calculated using the reference intensity value. The second predicted intensity value at the target warning location is determined based on the ground motion attenuation relationship. This also includes: calculating the distance from the target warning location to the seismic station that serves as the virtual sub-source; substituting the distance and the magnitude obtained from the inverse solution into the attenuation relationship to calculate the predicted intensity value of the virtual sub-source at the target warning location; and when the target warning location is simultaneously within the range of action of multiple virtual sub-sources, taking the maximum value among the predicted intensity values ​​calculated by each virtual sub-source as the second predicted intensity value.

[0125] In the above implementation process, the distance from the target warning location to the seismic station serving as a virtual sub-source is calculated. The distance and the magnitude obtained from the inverse solution are substituted into the attenuation relationship to calculate the predicted intensity value of the virtual sub-source at the target warning location. When the target warning location is simultaneously within the influence range of multiple virtual sub-sources, the maximum value among the predicted intensity values ​​is taken as the second predicted intensity value. This allows each virtual sub-source to estimate a continuous spatial intensity distribution field that decreases with distance from its own location based on mature seismic motion attenuation laws. This transforms the local intensity information of a single station into a spatially continuous regional intensity estimate, effectively compensating for the deficiency of the assumption of no intensity attenuation within a fixed spatial influence range in the first prediction. For stations with higher observed intensity, the inverse-calculated virtual magnitude is larger, and the high-intensity influence range can expand outward beyond the fixed distance limit. For stations with lower observed intensity, the inverse-calculated virtual magnitude is smaller, and the influence range naturally shrinks, realizing the adaptive dynamic adjustment of the warning range according to the measured intensity of the station. When a target warning location is simultaneously within the coverage area of ​​multiple virtual sub-sources, fusing the maximum predicted intensity of each virtual sub-source can fully utilize the observation information from multiple stations for cross-validation and mutual supplementation. In densely networked areas, multiple stations estimate the intensity of the same target location from different orientations. The strategy of taking the maximum value not only conforms to the conservative principle of prioritizing safety in earthquake early warning, but also effectively compensates for the potential underestimation of intensity caused by differences in local site conditions at a single station, thereby improving the spatial continuity and reliability of the early warning results.

[0126] In a specific embodiment of this application, for each target warning location within the monitoring area, the spatial distance R from it to each virtual sub-source (i.e., each effective seismic station) is calculated. The distance calculation can use the great circle distance formula or a simplified planar distance formula for local areas. The specific calculation method is the same as the method for calculating the distance from the target warning location to the station, and the same distance calculation module can be used.

[0127] For each virtual sub-source, the calculated distance R and the virtual magnitude M obtained from the inverse kinematics are substituted into the pre-established regional instrumental seismic intensity attenuation formula I = f(M, R) to calculate the predicted intensity value of the virtual sub-source at the target warning location. This predicted intensity value represents the instrumental seismic intensity level expected to be experienced at the target warning location, R kilometers away from the station where the virtual sub-source is located, assuming an earthquake of magnitude M has occurred at that station.

[0128] For each target warning location, the predicted intensity values ​​calculated by all virtual sub-sources at that location are aggregated. If a target warning location is within the effective coverage area of ​​only one virtual sub-source, the predicted intensity value of that virtual sub-source is directly used as the second predicted intensity value for that target warning location.

[0129] If a target warning location falls within the effective coverage area of ​​multiple virtual sub-sources, the predicted intensity values ​​calculated by each virtual sub-source at that location are compared, and the maximum value is taken as the second predicted intensity value for that target warning location. The physical basis for this maximum value strategy is that earthquake early warning prioritizes personnel safety and should employ the most conservative estimate. Simultaneously, in strong motion observations, due to varying site conditions at different stations, some stations may observe higher intensities due to local site amplification effects. Taking the maximum value from multiple source predictions can, to some extent, reflect the ground motion level under the most unfavorable site conditions, ensuring the safety of the warning results.

[0130] In this way, the generation of the second predicted intensity value integrates the isolated observation information of multiple single stations into a spatially continuous intensity prediction field. Its warning range is adaptively adjusted according to the magnitude of the actual observed intensity at the station, which effectively makes up for the lack of flexibility of the fixed warning radius method when dealing with events of different magnitudes, and improves the adaptability of the warning system to large earthquakes, distant earthquakes and complex multi-earthquake conditions.

[0131] To evaluate the performance of the method provided in this application embodiment, a retrospective analysis was conducted using 83 earthquakes from the Japan K-NET network. The analysis process is as follows: First, the space is divided into grids with equal latitude and longitude intervals centered on the epicenter, with a grid spacing of 0.01° × 0.01°. To save computation time, the spatial grid division range for earthquakes with magnitude 4.0 and below is set at ±1° of the epicenter latitude and longitude, ±2° for 4.1–6.0, ±3° for 6.1–7.9, and ±3.5° for 8.0 and above. Second, all records are sorted in ascending order of P-wave absolute arrival time. Starting 3 seconds after the first earthquake, intensity prediction is performed for all spatial grid points using the information observed at the current moment, with a prediction performed every 1 second. Finally, the intensity of the spatial grid points is interpolated using the maximum intensity of all observed records (cubic spline interpolation) to form a spatial intensity field, which is considered as the final observed intensity of each grid point for comparison and evaluation of the method's performance.

[0132] in, Figure 13 An epicenter location map of the selected earthquake provided for an embodiment of this application.

[0133] To comprehensively evaluate the performance of the method, this application defines four evaluation metrics: average relative warning time, accuracy, false alarm rate, and false negative rate. The definitions of each metric are as follows: First, the average relative warning time is the average of the relative warning times of all grid points where both the observed intensity and predicted intensity exceed the threshold. In an earthquake, for a grid point in space where the observed intensity exceeds a given threshold (e.g., 5 degrees), the time when the estimated intensity at that grid point first exceeds the threshold using the warning method based on the local no-attenuation assumption is the actual arrival time of the threshold at that grid point. The time when the predicted intensity at that grid point first exceeds the threshold using the method provided in this application is the predicted time of the threshold at that grid point. Therefore, the relative warning time of the method provided in this application at that grid point is the time difference between the predicted time and the arrival time. Since it is difficult to calculate the true arrival time of the threshold intensity at grid points without observation stations in space, this method can be used to evaluate the advance warning time of the method provided in this application relative to the warning method based on the local no-attenuation assumption; therefore, it is the average relative warning time.

[0134] Assume there is This earthquake was numbered 1 to 1. In each earthquake, the measured intensity exceeded the threshold. The grid points are (of which) (Representing the earthquake number), then the measured value exceeds... The total number of grid points is Assume that any measured intensity exceeds... The relative warning time of the grid points is ( This represents the earthquake sequence number. (Representing the grid point number), then the average relative warning time for:

[0135] Second, in an earthquake, if the actual final observed intensity at a certain spatial grid point is less than the intensity threshold... However, the predicted intensity is greater than The false alarm rate is defined as the percentage of false alarm grid points out of the total number of grid points, where the system has generated one false alarm at that grid point. This definition of false alarm takes into account the uncertainty of instrument intensity; a difference of less than 1 degree is not considered a false alarm.

[0136] Assume the first The number of grid points that were falsely reported in this earthquake was The false alarm rate is:

[0137] Third, in an earthquake, if the actual final observed intensity at a certain grid point in space is greater than the intensity threshold... However, the measured intensity reached the intensity threshold. Previously, the predicted values ​​had not exceeded The false negative rate is defined as the percentage of the total number of false negative grid points out of the total number of grid points.

[0138] Assume the first The number of grid points missed in this earthquake was The accuracy rate is:

[0139] Please see Figure 14 , Figure 14The evaluation index provided in this application embodiment varies with the intensity threshold. Retrospective analysis results show that, with an intensity alarm threshold of 5 degrees, the average relative warning time of the method provided in this application embodiment is 3 seconds, which is an average improvement of 3 seconds compared to the warning method based on the local attenuation assumption (PLUM). With an intensity alarm threshold of 6 degrees, the average relative warning time of the method provided in this application embodiment is 1 second, which is an average improvement of 1 second compared to the warning method based on the local attenuation assumption. The higher the alarm threshold, the shorter the average relative warning time. With an intensity alarm threshold of 5 degrees, the false alarm rate of the method provided in this application embodiment is 1.4%, while the false alarm rate of the warning method based on the local attenuation assumption is approximately 1%. With an intensity alarm threshold of 6 degrees, the false alarm rate of the method provided in this application embodiment is 0.6%, comparable to the warning method based on the local attenuation assumption. If the intensity alarm threshold is 5 degrees, the false alarm rate of the method provided in this application embodiment is 0.0015%, and the false alarm rate of the early warning method based on the assumption of no local attenuation is about 0.004%; if the intensity alarm threshold is 6 degrees, the false alarm rate of the method provided in this application embodiment is 0%, and the false alarm rate of the early warning method based on the assumption of no local attenuation is 0.0027%.

[0140] To more intuitively demonstrate the performance of the method provided in this application embodiment in actual earthquakes, and to compare the difference between the method provided in this application embodiment and the early warning method based on the assumption of no local attenuation, this application embodiment selected a magnitude 7.3 earthquake that occurred on April 16, 2016, in Kumamoto Prefecture, Japan (32.753°N, 130.762°E), with a focal depth of 12 km, for retrospective comparison. The spatial intensity field distribution of this earthquake was calculated and interpolated using strong motion observation records, as shown below. Figure 15 As shown, Figure 15 The Kumamoto seismic instrument intensity contour map provided for embodiments of this application. Please also refer to... Figure 16 , Figure 16 A comparison chart of the intensity predictions for the Kumamoto earthquake provided in this application embodiment and the PLUM method is shown. Figure 16 This paper presents a comparison of the predicted intensity of the method provided in this application and the early warning method based on the assumption of no local attenuation, starting from 3 seconds after the first earthquake was triggered and continuing to 24 seconds after the first earthquake was triggered. As can be seen from the figure, because the method provided in this application takes into account intensity attenuation, the range of predicted intensity can be adjusted according to the magnitude of the observed intensity, thus achieving a faster prediction speed.

[0141] Since the method was established using strong-motion observation data from Japan and K-NET network density conditions, a retrospective earthquake case study was conducted to further test its performance under strong-motion observation data and network conditions in my country. At 23:59 Beijing time on December 18, 2023, a magnitude 6.2 earthquake occurred in Jishishan County, Linxia Prefecture, Gansu Province (35.70°N, 102.79°E), with a focal depth of 10 km. It was a reverse-thrust earthquake with a left-lateral strike-slip pattern, causing extensive damage to buildings. Please refer to [link / reference]. Figure 17 , Figure 17 For the instrument intensity map of the 6.2 magnitude earthquake in Jishishan, Gansu Province provided in this application embodiment, the National Strong Motion Observation Center acquired a total of 741 sets of three-component strong motion observation acceleration records within a distance of 200 km from the epicenter, 239 sets within a distance of 100 km from the epicenter, with a minimum epicenter distance of 6 km and a maximum observation intensity of 9.5 degrees.

[0142] For the 239 sets of strong ground motion observation records within a 100 km radius of the epicenter, please combine them with... Figure 18 , Figure 18 The comparison chart of predicted intensity and real-time observed intensity for the 6.2 magnitude earthquake in Jishishan, Gansu Province, provided in this embodiment of the application, firstly involved manual extraction of the P-wave first arrival to ensure accurate P-wave arrival time, and then the method provided in this embodiment was used to start prediction from 3 seconds after the first earthquake was triggered. Figure 18 It is known that 3 seconds after the first trigger, an intensity of 7 degrees or higher was observed in the epicenter. If 5 degrees is used as the warning threshold, the method provided in this application embodiment can provide warnings for an area with a radius of about 60 km; 9 seconds after the first trigger, warnings can be provided for the entire 5-degree zone.

[0143] Secondly, this application provides an earthquake early warning device, please refer to [link to relevant documentation]. Figure 19 , Figure 19 A schematic diagram of an earthquake early warning device provided in an embodiment of this application.

[0144] The earthquake early warning device includes: a reference intensity determination module 010, a first prediction module 020, a second prediction module 030, and a fusion module 040. The reference intensity determination module 010 acquires real-time observation data from at least one seismic station within the monitoring area and determines the reference intensity value of the corresponding seismic station based on the real-time observation data. The reference intensity value includes the estimated maximum intensity value predicted based on P-wave information and / or the measured intensity value from the seismic station. The first prediction module 020, for any target early warning location, determines the early warning value based on a preset early warning condition provided that the reference intensity value of the existing seismic station meets the pre-defined early warning conditions. In the case of a seismic station as the center, based on a preset spatial influence range, a reference intensity value is assigned to the target warning location within the spatial influence range to obtain a first predicted intensity value; the second prediction module 030 is used to regard all seismic stations within the monitoring area as virtual sub-sources, and for the same virtual sub-source, the intensity parameters of the virtual sub-source are calculated using the reference intensity value, and the second predicted intensity value at the target warning location is determined according to the ground motion attenuation relationship; the fusion module 040 is used to fuse the first predicted intensity value and the second predicted intensity value to generate the final warning intensity field at the target warning location.

[0145] In the above implementation process, a reference intensity determination module is set up to acquire real-time observation data of seismic stations within the monitoring area and determine reference intensity values. These reference intensity values ​​include the estimated maximum intensity value predicted based on P-wave information and / or the measured intensity value at the station. This allows the device to predict the maximum intensity a station might experience using P-wave information before the measured maximum intensity is reached, providing more decision-making time for the subsequent early warning module. A first prediction module is also set up. When the reference intensity value at a station meets the preset early warning conditions, the reference intensity value is assigned to the target early warning location within the preset spatial influence range, centered on the station, to obtain the first predicted intensity value. This achieves rapid local early warning without relying on source parameter estimation, effectively avoiding false alarms or missed alarms caused by deviations in source parameter estimation. A second prediction module is set up, treating all seismic stations within the monitoring area as virtual sub-sources. It uses reference intensity values ​​to calculate the intensity parameters of these virtual sub-sources and determines the second predicted intensity value at the target warning location based on ground motion attenuation relationships. This allows the warning range to be adaptively and dynamically adjusted according to the actual observed intensity at the stations, compensating for the insufficient timeliness of a fixed warning range for high-intensity events. A fusion module is also included, fusing the first and second predicted intensity values ​​to generate the final warning intensity field. This effectively balances the coverage and timeliness of the warning. The overall architecture is clear, with each module having a defined and relatively independent function, facilitating deployment and parallel computation within a real-time earthquake early warning system.

[0146] In one specific embodiment of this application, the reference intensity determination module 010 is used to acquire real-time observation data from at least one seismic station within the monitoring area, and determine the reference intensity value of the corresponding seismic station based on the real-time observation data. The real-time observation data includes, but is not limited to, acceleration records, velocity records, or displacement records. The reference intensity value has two sources: one is the estimated maximum intensity value predicted based on P-wave information, and the other is the measured intensity value from the seismic station. Either one can be used alone, or both can be used in combination. The reference intensity determination module 010 may further include a data acquisition unit, a prediction calculation unit, and a threshold discrimination unit.

[0147] In a specific embodiment of this application, the data acquisition unit 011 is used to acquire real-time observation data within multiple incremental time windows after the P-wave is triggered by the seismic station. The real-time observation data includes at least the triaxial composite peak velocity Pv, peak displacement Pd, and cumulative absolute velocity CAV. The data acquisition unit is communicatively connected to the seismic station within the monitoring area, receives strong motion observation data uploaded by the station in real time, and extracts corresponding waveform data segments according to the P-wave trigger time and incremental time window lengths (e.g., 3 seconds, 4 seconds, 5 seconds, etc.), calculating the three characteristic parameters Pv, Pd, and CAV for each time window length.

[0148] In a specific embodiment of this application, the prediction unit 012 is connected to the data acquisition unit 011. It is used to substitute the real-time observation data output by the data acquisition unit into a pre-established statistical regression relationship to obtain the intensity prediction value corresponding to each parameter. For the current time window length, it selects the intensity prediction values ​​corresponding to the two parameters with smaller regression standard deviations in the real-time observation data, and takes their average as the predicted maximum intensity value for that time window. The statistical regression relationship is pre-stored in the device and specifically includes a first regression equation with the logarithm of Pv as the independent variable and the maximum instrumental seismic intensity as the dependent variable. The second regression equation with the logarithm of d as the independent variable and the maximum instrumented seismic intensity as the dependent variable. The third regression equation with the logarithm of CAV as the independent variable and the maximum instrumented seismic intensity as the dependent variable. The coefficients and standard deviations of the regression equations were obtained through regression statistics of historical strong vibration records for a specific time window length.

[0149] In a specific embodiment of this application, the threshold discrimination unit 013 is connected to the pre-estimation calculation unit 012. It compares each real-time observation data calculated by the data acquisition unit 011 within the current time window with a preset parameter threshold corresponding to the target warning intensity threshold. If any real-time observation data reaches or exceeds its corresponding preset parameter threshold, the estimated maximum intensity value output by the pre-estimation calculation unit 012 is selected as the reference intensity value. The preset parameter thresholds include the Pv threshold, the Pd threshold, and the CAV threshold, corresponding to the three characteristic parameters Pv, Pd, and CAV, respectively. The preset parameter thresholds are predetermined and stored in the device as follows: a preset target warning intensity threshold is established; for a given time window length, strong motion record samples from multiple stations in historical earthquakes are acquired, each sample containing Pv, Pd, CAV under that time window length, and the final measured maximum intensity value of that station; an exceedance probability value is set, and a spatial region defined by the three thresholds is determined in the three-dimensional parameter space composed of Pv, Pd, and CAV, such that the proportion of samples falling outside this region whose measured maximum intensity value reaches or exceeds the target warning intensity threshold is equal to the exceedance probability value; the boundary values ​​corresponding to this spatial region on the Pv axis, Pd axis, and CAV axis, respectively, are used as the Pv threshold, Pd threshold, and CAV threshold. The threshold discrimination unit uses "OR" logic for comparison, that is, when Pv... Pv threshold, or Pd Pd threshold, or CAV When the CAV threshold is reached, it is determined that the discrimination is passed, and the estimated maximum intensity value is accepted as the reference intensity value and output to the first prediction module 020 and the second prediction module 030; when the real-time observation data does not reach their respective thresholds, the estimated maximum intensity value under that time window is rejected, and the next time window is waited for re-discrimination.

[0150] In the above implementation process, the data acquisition unit 011 continuously acquires various ground motion characteristic parameters within multiple incremental time windows after the P-wave is triggered, providing a rich data foundation with different sensitivity characteristics for subsequent intensity prediction; the prediction calculation unit 012 dynamically selects the two parameter combinations with the best prediction accuracy under the current time window and averages them, effectively reducing the uncertainty of single parameter prediction and improving the accuracy and robustness of the predicted maximum intensity value; the threshold discrimination unit 013 performs pre-filtering with three-parameter "OR" logic, which suppresses the risk of false alarms of "overestimation of small earthquakes" caused by the uncertainty of statistical models, while retaining as much effective early warning information as possible, achieving a balance between early warning accuracy and timeliness.

[0151] In a specific embodiment of this application, the first prediction module 020 is connected to the reference intensity determination module 010 and receives the reference intensity value output by the reference intensity determination module 010. For any target warning location within the monitoring area, the first prediction module 020 first determines whether there is a seismic station whose reference intensity value meets the preset warning conditions (i.e., the reference intensity value reaches or exceeds the preset target warning intensity threshold). If the reference intensity value meets the preset warning conditions, then, taking the station as the center and based on a preset spatial influence range (e.g., a circular area with a radius of 30 kilometers), the reference intensity value is assigned to all target warning locations within the spatial influence range to obtain the first predicted intensity value. When a target warning location is simultaneously within the spatial influence range of multiple stations that meet the warning triggering conditions, the maximum value among all corresponding reference intensity values ​​is taken as its first predicted intensity value.

[0152] In a specific embodiment of this application, the second prediction module 030 is connected to the reference intensity determination module 010 and receives the reference intensity value output by the reference intensity determination module 010. The second prediction module 030 treats each of the seismic stations within the monitoring area as a virtual sub-source. For each virtual sub-source, the reference intensity value of that station is used as the virtual epicenter intensity. Using a pre-established regional instrumental seismic intensity attenuation relationship (which uses magnitude and distance as independent variables and instrumental seismic intensity as the dependent variable), the distance is set to zero and substituted into the inverse solution of the virtual epicenter intensity to obtain the magnitude corresponding to the virtual sub-source as the intensity parameter. Then, the distance from the target warning location to the virtual sub-source is calculated, and the distance and the magnitude obtained from the inverse solution are substituted into the attenuation relationship to calculate the predicted intensity value of the virtual sub-source at the target warning location. When the target warning location is simultaneously within the range of multiple virtual sub-sources, the maximum value among the predicted intensity values ​​is taken as the second predicted intensity value.

[0153] The fusion module 040 is connected to the first prediction module 020 and the second prediction module 030, respectively, and receives the first predicted intensity value and the second predicted intensity value. For the same target warning location, the fusion module 040 fuses the two predicted intensity values ​​to generate the final warning intensity value for that target warning location. The fusion method can be to take the larger of the two values, or other fusion strategies such as weighted averaging can be used.

[0154] The above modules can be implemented in software on the server, and together with the communication interface, they can receive real-time data from the stations and execute them in a rolling manner at preset time intervals (e.g., 1 second) to continuously update the spatial warning area.

[0155] Thirdly, embodiments of this application also provide an electronic device, which includes a memory and a processor. The memory stores program instructions, and when the processor reads and runs the program instructions, it executes the steps in any of the above implementation methods.

[0156] Please see Figure 20 , Figure 20 This is a block diagram illustrating an electronic device according to an embodiment of this application. The electronic device 100 may include a memory 111, a memory controller 112, a processor 113, a peripheral interface 114, an input / output unit 115, and a display unit 116. Those skilled in the art will understand that... Figure 20 The structure shown is for illustrative purposes only and does not limit the structure of the electronic device 100. For example, the electronic device 100 may also include components that are more... Figure 20 The more or fewer components shown, or having the same Figure 20 The different configurations shown.

[0157] The aforementioned memory 111, memory controller 112, processor 113, peripheral interface 114, input / output unit 115, and display unit 116 are electrically connected directly or indirectly to each other to achieve data transmission or interaction. For example, these components can be electrically connected to each other through one or more communication buses or signal lines. The aforementioned processor 113 is used to execute executable modules stored in the memory.

[0158] The memory 111 can be, but is not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), etc. The memory 111 stores programs, and the processor 113 executes these programs upon receiving execution instructions. The methods executed by the electronic device 100 as defined in any embodiment of this application can be applied to the processor 113, or implemented by the processor 113.

[0159] The aforementioned processor 113 may be an integrated circuit chip with signal processing capabilities. The processor 113 may be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it may also be a digital signal processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor may be a microprocessor or any conventional processor.

[0160] The peripheral interface 114 described above couples various input / output devices to the processor 113 and the memory 111. In some embodiments, the peripheral interface 114, the processor 113, and the memory controller 112 can be implemented on a single chip. In other instances, they can be implemented on separate chips.

[0161] The input / output unit 115 described above is used to provide user input data. The input / output unit 115 may be, but is not limited to, a mouse and keyboard, etc.

[0162] The aforementioned display unit 116 provides an interactive interface (e.g., a user interface) between the electronic device 100 and the user, or displays image data for the user's reference. In this embodiment, the display unit can be a liquid crystal display (LCD) or a touch display. If it is a touch display, it can be a capacitive touchscreen or a resistive touchscreen that supports single-point and multi-point touch operations. Supporting single-point and multi-point touch operations means that the touch display can sense touch operations generated simultaneously from one or more locations on the touch display and pass the sensed touch operations to the processor for calculation and processing.

[0163] This application also provides a computer-readable storage medium storing computer program instructions, which are read and executed by a processor to perform steps in an earthquake early warning method.

[0164] In summary, the earthquake early warning method provided in this application includes: acquiring real-time observation data from at least one seismic station within a monitoring area, and determining a reference intensity value for the corresponding seismic station based on the real-time observation data; wherein the reference intensity value includes an estimated maximum intensity value predicted based on P-wave information and / or a measured intensity value from the seismic station; for any target early warning location: if the reference intensity value of a seismic station meets preset early warning conditions, with the seismic station as the center, based on a preset spatial influence range, the reference intensity value is assigned to the target early warning location within the spatial influence range to obtain a first predicted intensity value; furthermore, all seismic stations within the monitoring area are considered as virtual sub-sources, and for the same virtual sub-source, the intensity parameters of the virtual sub-source are calculated using the reference intensity value, and a second predicted intensity value at the target early warning location is determined based on the ground motion attenuation relationship; the first predicted intensity value and the second predicted intensity value are fused to generate a final early warning intensity field for the target early warning location. Without considering source parameters, the intensity can be predicted in a small area by relying on real-time intensity observations from stations, thereby solving the problems of missed and false alarms that may occur in traditional source parameter estimation methods under large earthquakes and complex multi-seismic conditions.

[0165] In the several embodiments provided in this application, it should be understood that the disclosed device can also be implemented in other ways. The device embodiments described above are merely illustrative; for example, the block diagrams in the accompanying drawings illustrate the possible architecture, functions, and operations of the device according to various embodiments of this application. In this regard, each block in the block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagram, and combinations of block diagrams, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0166] In addition, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0167] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0168] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application. It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0169] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

[0170] It should be noted that, in the embodiments of this application, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

Claims

1. An earthquake early warning method, characterized in that, The earthquake early warning method includes: Acquire real-time observation data from at least one seismic station within the monitoring area, and determine a reference intensity value corresponding to the seismic station based on the real-time observation data; wherein, the reference intensity value includes the estimated maximum intensity value predicted based on P-wave information and / or the measured intensity value of the seismic station. For any target warning location: If the reference intensity value of the earthquake station meets the preset early warning conditions, the reference intensity value is assigned to the target early warning location within the preset spatial influence range, with the earthquake station as the center, to obtain the first predicted intensity value. Furthermore, all seismic stations within the monitoring area are considered as virtual sub-sources. For the same virtual sub-source, the intensity parameters of the virtual sub-source are calculated using the reference intensity value, and the second predicted intensity value at the target warning location is determined based on the seismic motion attenuation relationship. By combining the first predicted intensity value and the second predicted intensity value, the final warning intensity field of the target warning location is generated.

2. The method according to claim 1, characterized in that, The step of acquiring real-time observation data from at least one seismic station within the monitoring area, and determining the reference intensity value corresponding to the seismic station based on the real-time observation data, includes: Real-time observation data is acquired within multiple increasing time windows following P-wave triggering at the seismic station. This real-time observation data includes at least the triaxial composite peak velocity. Peak displacement and cumulative absolute velocity ; Substitute the real-time observation data into the statistical regression relationship to obtain the intensity prediction value corresponding to each parameter; For the current time window length, select the intensity prediction values ​​corresponding to the two parameters with smaller regression standard deviations in the real-time observation data, and take their average value as the estimated maximum intensity value under this time window.

3. The method according to claim 2, characterized in that, The step of acquiring real-time observation data from at least one seismic station within the monitoring area and determining the reference intensity value corresponding to the seismic station based on the real-time observation data further includes: Each of the real-time observation data obtained by calculating the current time window length is compared with a preset parameter threshold corresponding to the target warning intensity threshold. If any of the real-time observation data reaches or exceeds its corresponding preset parameter threshold, the estimated maximum intensity value is selected as the reference intensity value.

4. The method according to claim 3, characterized in that, in, The preset parameter threshold includes the peak velocity of the triaxial synthesis. Threshold, peak displacement Threshold and cumulative absolute velocity Threshold; The target warning intensity threshold is preset. For a given time window length, strong motion record samples from multiple stations in historical earthquakes are obtained. Each sample contains the real-time observation data under the time window length and the maximum intensity value finally measured by the earthquake station. In the three-dimensional parameter space formed by the real-time observation data, a spatial region defined by three thresholds is determined, and the boundary values ​​corresponding to the spatial region on the real-time observation data are used as the preset parameter thresholds.

5. The method according to claim 3, characterized in that, When the reference intensity value of the existing seismic station meets the preset early warning conditions, the reference intensity value is assigned to a target early warning location within the preset spatial influence range, centered on the seismic station, to obtain a first predicted intensity value. This includes: Determine whether the reference intensity value reaches or exceeds the target warning intensity threshold; If the reference intensity value reaches or exceeds the target warning intensity threshold, it is determined that the reference intensity value meets the preset warning condition. A preset spatial influence distance threshold is used to assign the reference intensity value to all target warning locations that are less than or equal to the spatial influence distance threshold from the seismic station, which is then used as the first predicted intensity value.

6. The method according to claim 1, characterized in that, The step of treating all seismic stations within the monitoring area as virtual sub-sources, and for the same virtual sub-source, using the reference intensity value to back-calculate the intensity parameters of the virtual sub-source, and determining the second predicted intensity value at the target warning location based on the seismic motion attenuation relationship, includes: The reference intensity value of the earthquake station is used as the virtual epicenter intensity. Using a pre-established regional instrumental seismic intensity attenuation relationship, setting the distance to zero, and substituting the virtual epicenter intensity, the magnitude corresponding to the virtual sub-source is obtained by inverse solving, and used as the intensity parameter; The attenuation relationship is described with magnitude and distance as independent variables and instrumental seismic intensity as dependent variable.

7. The method according to claim 6, characterized in that, The method of treating all seismic stations within the monitoring area as virtual sub-sources, and for the same virtual sub-source, using the reference intensity value to back-calculate the intensity parameters of the virtual sub-source, and determining the second predicted intensity value at the target warning location based on the seismic motion attenuation relationship, further includes: Calculate the distance from the target early warning location to the seismic station that serves as the virtual sub-source; Substituting the distance and the magnitude obtained by inverse kinematics into the attenuation relationship, the predicted intensity value of the virtual sub-source at the target warning location is calculated. When the target warning location is simultaneously within the effective range of multiple virtual sub-sources, the maximum value among the predicted intensity values ​​calculated by each virtual sub-source is taken as the second predicted intensity value.

8. An earthquake early warning device, characterized in that, The earthquake early warning device includes: a reference intensity determination module, a first prediction module, a second prediction module, and a fusion module; The reference intensity determination module is used to acquire real-time observation data from at least one seismic station within the monitoring area, and determine the reference intensity value corresponding to the seismic station based on the real-time observation data; wherein, the reference intensity value includes the estimated maximum intensity value predicted based on P-wave information and / or the measured intensity value of the seismic station. The first prediction module is used to, for any target warning location, if the reference intensity value of the seismic station meets the preset warning conditions, assign the reference intensity value to the target warning location located within the spatial influence range, with the seismic station as the center and based on the preset spatial influence range, to obtain the first predicted intensity value. The second prediction module is used to regard all seismic stations in the monitoring area as virtual sub-sources. For the same virtual sub-source, the intensity parameters of the virtual sub-source are calculated using the reference intensity value, and the second predicted intensity value at the target warning location is determined based on the ground motion attenuation relationship. The fusion module is used to fuse the first predicted intensity value and the second predicted intensity value to generate the final warning intensity field of the target warning location.

9. An electronic device, characterized in that, The electronic device includes a memory and a processor. The memory stores program instructions, and when the processor executes the program instructions, it performs the steps of the method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions that, when executed by a processor, perform the steps of the method according to any one of claims 1-7.

Images