Early warning special-purpose terminal data acquisition and earthquake event detection processing method

Abstract

The invention relates to the technical field of earthquake monitoring, and discloses an early warning special terminal data acquisition and earthquake event detection processing method, which comprises the following steps of: acquiring background noise of an accelerometer, a microphone and a barometer, judging a current coupling state, and obtaining a dynamic trigger threshold value and a coupling coefficient; monitoring the acceleration signal with a dynamic trigger threshold to capture an event data window; performing cross-domain coherence analysis on the multi-modal signals in the window, calculating a comprehensive coherence score, and comparing the comprehensive coherence score with an identification threshold to judge and discard pseudo signals; if it is determined that the seismic event is a high-confidence-coefficient seismic event, the coupling coefficient is used for carrying out inverse deduction on the measured peak acceleration, and normalized equivalent ground motion parameters are calculated; and finally, encapsulating and reporting an event information packet containing the normalization parameter, the coherence score and the coupling state. According to the method, the false alarm rate is reduced through cross-domain identification, and the parameter accuracy is improved through coupling state correction.

Description

Technical Field

[0001] This invention relates to the field of earthquake monitoring technology, specifically to a method for data acquisition and earthquake event detection and processing using a dedicated early warning terminal. Background Technology

[0002] With the widespread adoption of smartphones, their integrated sensors, such as accelerometers, microphones, and barometers, have enabled the construction of low-cost, high-density crowdsourced earthquake monitoring networks. These terminals can capture local vibrations and environmental changes, providing new data sources for early earthquake warnings and rapid intensity reporting.

[0003] The existing earthquake detection methods for dedicated early warning terminals mainly rely on accelerometers. These methods set a fixed vibration trigger threshold on the terminal. When the real-time acceleration signal detected by the terminal (e.g., its amplitude or the output of a specific algorithm) exceeds the preset threshold, the method determines it as a suspected event. Subsequently, the terminal captures vibration waveform data before and after the suspected event and reports it to the central server. The server then uses spatial correlation analysis of data from multiple devices to finally confirm whether there is a real earthquake event.

[0004] However, existing technologies have significant shortcomings in practical applications. First, the mechanism relying solely on accelerometer threshold triggering is highly susceptible to interference from non-seismic signals. Vibrations generated by smart terminals during daily use (such as carrying, placing, tapping, or in vehicles) have characteristics similar to seismic P-waves, leading to frequent false alarms. This not only consumes communication bandwidth and server resources but also reduces the overall reliability of the method. Second, existing methods ignore the impact of the coupling state between the smart terminal and the environment on the measurement data. The way the terminal is placed can distort its response to ground motion, causing the reported peak acceleration and other parameters to fail to reflect the true ground motion, thus affecting subsequent assessments. Therefore, this invention proposes a data acquisition and seismic event detection processing method for a dedicated early warning terminal to address the shortcomings of existing technologies. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method for data acquisition and earthquake event detection and processing using a dedicated early warning terminal. This method solves the problems of being unable to effectively distinguish between real earthquake signals and non-earthquake interference generated by daily activities, resulting in a high false alarm rate, and the inability to effectively distinguish between the coupling state of the smart terminal and the environment, which causes sensor response distortion and makes it impossible to accurately invert the real ground motion parameters.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for data acquisition and earthquake event detection and processing using a dedicated early warning terminal. This method is applied to an intelligent terminal equipped with an accelerometer, microphone, and barometer, and includes the following steps: The environmental coupling state quantification step involves continuously acquiring background noise signals from multiple sensors within a sliding background time window. By analyzing the characteristics of this background noise, the current coupling state of the smart terminal is determined. Based on the determined current coupling state, the method retrieves or generates a matching dynamic trigger threshold and coupling coefficient. The significance of this step is to enable subsequent detection and correction to have context-aware capabilities.

[0007] After obtaining the dynamic parameters, the system proceeds to the dynamic threshold triggering and event window capture step. This step continuously monitors the real-time acceleration signal. When the L2 norm of the real-time acceleration signal (representing the instantaneous total energy of the signal) is greater than the acquired dynamic triggering threshold, the system determines that a suspected event has occurred. The design of this dynamic threshold is intended to suppress conventional noise caused by specific coupling states (such as handheld operation) and is only sensitive to abnormal vibrations that exceed the baseline of this state. When triggered, an event data window containing multimodal synchronization data will be captured.

[0008] After capturing the event data window, a cross-domain coherence identification step is performed to determine the authenticity of the event. The core principle of this step is that real earthquake P waves will simultaneously generate signals with the same physical origin and high correlation in time and frequency characteristics in ground vibration (accelerometer), infrasound (microphone), and atmospheric pressure disturbance (barometer). However, most non-seismic interferences (such as local knocking) do not have this cross-domain correlation. In the identification process, the comprehensive coherence score is first calculated, and then this score is compared with a preset identification threshold. If the comprehensive coherence score is less than or equal to the identification threshold, it indicates that the event is uncorrelated in multiple physical domains, is judged as a false signal, and the event data window is discarded, terminating the processing. If the score is greater than the identification threshold, it is judged as a high-confidence earthquake event, and subsequent steps are executed.

[0009] For windows identified as high-confidence seismic events, a normalized ground motion parameter back-calculation step is performed. The purpose of this step is to eliminate the influence of coupling state on the measurement results. The method extracts the measured peak acceleration from the event data window and uses the coupling coefficient obtained in the environmental coupling state quantification step to calculate the normalized equivalent ground motion parameters.

[0010] Finally, the method performs a standardized earthquake event information reporting step, which encapsulates an event information package containing key information, including: the normalized equivalent ground motion parameters, the comprehensive coherence score, and the current coupling state. Subsequently, this event information package is reported to the cloud-based early warning server to provide high-quality data input for subsequent earthquake early warning or intensity rapid reporting.

[0011] Preferably, the environmental coupling state quantification step can be further refined. After the system synchronously collects the background noise signal, it extracts the state feature vector. The state feature vector may include, but is not limited to: the L2 norm variance of the accelerometer triaxial signal (reflecting the stability of the vibration environment), the infrasound energy ratio of the microphone signal (distinguishing between indoor and outdoor or wind noise), and the short-time variance of the barometer signal (reflecting air pressure stability). Subsequently, the state feature vector is classified using a preset state recognition model (e.g., support vector machine, neural network, or decision tree), and the current coupling state is output.

[0012] To enable rapid retrieval of parameters in the environmental coupling state quantification step, the system can pre-store a parameter lookup table in memory. The parameter lookup table establishes a mapping relationship between the coupling state and the dynamic trigger threshold and coupling coefficient. When the environmental coupling state quantification step determines the current coupling state, the system directly retrieves and outputs the corresponding dynamic trigger threshold and coupling coefficient from the parameter lookup table.

[0013] Preferably, in the dynamic threshold triggering and event window capture step, the time range of the event data window is precisely defined. The starting point of the time range is defined as the triggering time minus a preset pre-trigger duration, and the ending point is defined as the triggering time plus a preset post-trigger duration. This ensures that the captured data window completely includes the initial arrival portion of the seismic wave (especially the P wave).

[0014] Preferably, the cross-domain coherence identification step, before performing coherence analysis, first preprocesses the signal to extract the most relevant signal components. This preprocessing includes: Extract the vibration reference signal (e.g., vertical component or principal vibration direction component) from the accelerometer data in the event data window.

[0015] Apply a bandpass filter (e.g., 0.1Hz to 20Hz) to the microphone data in the event data window to extract the infrasound signal and filter out human voices and high-frequency noise.

[0016] Apply a high-pass filter or differential processing to the barometer data in the event data window to extract pressure disturbance signals and filter out slow pressure drift.

[0017] Preferably, after preprocessing the signal, the cross-domain coherence analysis can be performed using wavelet coherence analysis. The wavelet coherence analysis method calculates the wavelet coherence spectrum between the vibration reference signal and the infrasound signal, and between the vibration reference signal and the pressure disturbance signal, within a key P-wave time-frequency window (e.g., a specific frequency range and time range). Subsequently, by integrating or averaging the wavelet coherence spectrum within the key P-wave time-frequency window, the accelerometer microphone coherence index and the accelerometer barometer coherence index are calculated respectively.

[0018] To obtain the final cross-domain coherence identification criteria, the system calculates the comprehensive coherence score through a weighted fusion model. For example, the comprehensive coherence score can be calculated as the sum of the product of the accelerometer microphone coherence index and the microphone weight coefficient, plus the product of the accelerometer barometer coherence index and the barometer weight coefficient.

[0019] Preferably, in the step of reverse calculation of normalized ground motion parameters, the method for extracting the measured peak acceleration is: calculating the maximum L2 norm of the accelerometer data in the event data window over the entire time range of the event data window.

[0020] Preferably, in the standardized earthquake event information reporting step, the event information package, in addition to including the core parameters defined in the aforementioned standardized earthquake event information reporting step, may also include: the P-wave arrival timestamp and the terminal geographic coordinates, to provide complete spatiotemporal information.

[0021] Preferably, to ensure the timeliness and reliability of the standardized earthquake event information reporting steps, this step can adopt a concurrent reporting mechanism or a rapid fault-tolerant strategy. For example, when a smart terminal is simultaneously connected to Wi-Fi and cellular networks (such as 4G or 5G), the system can report the event information packet simultaneously through both network interfaces, or immediately report through a backup network (such as cellular network) after the preferred network (such as Wi-Fi) times out or fails to report.

[0022] This invention provides a method for data acquisition and earthquake event detection and processing using a dedicated early warning terminal. It has the following beneficial effects: 1. This invention utilizes the physical coherence correlation of real earthquake P waves in vibration (accelerometer), infrasound (microphone), and pressure disturbance (barometer) through the cross-domain coherence identification step. However, non-seismic interference generated by daily activities (such as knocking and carrying) usually do not have this cross-domain consistency. Therefore, by calculating the comprehensive coherence score and comparing it with the identification threshold, this method can filter out false signals and improve the accuracy of high-confidence earthquake events.

[0023] 2. The present invention uses environmental coupling state quantization in the environmental coupling state quantization step to first determine the coupling relationship between the smart terminal and the environment and obtain the coupling coefficient for correction. In the normalized ground motion parameter back-calculation step, the coupling coefficient is used to back-calculate the measured peak acceleration, eliminating the sensor response distortion caused by different placement methods (such as desktop, handheld), thereby obtaining normalized equivalent ground motion parameters that are closer to the real ground motion.

[0024] 3. This invention determines the current coupling state through an environmental coupling state quantification step and retrieves or generates a matching dynamic trigger threshold. This dynamic threshold is applied in the dynamic threshold triggering and event window capture steps, enabling the method to maintain high sensitivity in capturing weak signals in quiet environments (such as desktop placement) and automatically increase the threshold in noisy environments (such as handheld or vehicle use), suppressing background noise interference and preventing frequent false triggers caused by routine activities. Attached Figure Description

[0025] Figure 1 This is an overall flowchart of the present invention; Figure 2 This is a schematic diagram of the environmental coupling state quantification process of the present invention; Figure 3 This is a schematic diagram of the dynamic threshold triggering and event window capture process of the present invention; Figure 4 This is a schematic diagram of the cross-domain coherence identification process of the present invention; Figure 5 This is a schematic diagram of the reverse calculation process for normalized ground motion parameters according to the present invention; Figure 6 This is a schematic diagram of the standardized earthquake event information reporting process of the present invention. Detailed Implementation

[0026] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] This invention provides a method for data acquisition and earthquake event detection and processing using a smart terminal. The method is applied to a smart terminal, which is equipped with a processor, a memory, a communication unit, and a multimodal sensor. The multimodal sensor includes at least an accelerometer, a microphone, and a barometer.

[0028] In one embodiment, the method of the present invention is loaded and executed by the processor of a smart terminal to implement a series of functional modules that work together to complete the accurate identification and data correction of earthquake events.

[0029] The environmental coupling state quantification module is used to quantify the physical environment coupling state of the smart terminal in real time under non-event-triggered background conditions. This module continuously and synchronously collects background noise signals from multimodal sensors, including accelerometer background signals. Microphone background signal and barometer background signal The environment coupling state quantization module operates within a sliding background time window. Within this process, state feature vectors are extracted from the collected background noise signals. State feature vector At least including: L2 norm variance of accelerometer triaxial signals This characterizes the micro-vibration stability of the terminal.

[0030] ; in, It is a sliding background time window Number of samples within; Discrete time The three-axis acceleration vector; It is its L2 norm; Is it the L2 norm in the sliding background time window? The mean within; To calculate the standard coefficients of the sample variance (not the population variance), This indicates all time windows belonging to the sliding background. discrete time Perform summation.

[0031] Accelerometer frequency domain characteristics The frequency domain characteristics of an accelerometer reflect the frequency properties of background vibrations, such as the zero-crossing rate. It is used to help distinguish between high-frequency jitter and low-frequency displacement.

[0032] Infrasound energy percentage of microphone signal First, the microphone background signal Perform bandpass filtering (e.g.) Infrasound signal obtained .

[0033] ; in, Instantaneous energy of infrasound signals; Total energy of infrasound; Total energy across the entire frequency band; Total energy of infrasound.

[0034] Short-time variance of barometer signal It characterizes the stability of the atmospheric pressure environment.

[0035] ; in, Discrete time air pressure value; It is in the sliding background time window The mean within;

[0036] The environmental coupling state quantization module utilizes a preset state recognition model. (e.g., Gaussian mixture models or decision trees), for the extracted state feature vectors Classify and determine the current coupling state of the terminal. .

[0037] ; in, Belongs to a predefined set of discrete states ,For example This represents a hard surface left to stand still. This represents a soft surface left to stand.

[0038] For discrete state sets Each state in The memory stores two corresponding parameters: Dynamic trigger threshold : Acceleration trigger sensitivity applicable to this state.

[0039] Coupling coefficient : A dimensionless coefficient, representing the state The transmission relationship between the actual ground motion amplitude and the sensor measurement amplitude is determined by the environmental coupling state quantization module. Real-time retrieval and output and .

[0040] The dynamic triggering and capture module is used to dynamically adjust the event triggering sensitivity according to the current coupling state and capture event data. The dynamic triggering and capture module obtains the dynamic triggering threshold from the environmental coupling state quantization module. .

[0041] The dynamic triggering and capture module continuously monitors the real-time accelerometer background signal. When the trigger condition is met When this occurs, it is determined to be a suspected event triggered, and the trigger time is recorded. .

[0042] Upon triggering, the dynamic triggering and capture module immediately... Capture a preset time window before and after the time. Synchronization data for all modalities within the window, generating an event data window. .

[0043] The cross-domain coherence identification module is used to identify the physical source of suspected events and distinguish between high-confidence earthquake events and pseudo-signals. The identification principle of the cross-domain coherence identification module is based on the fact that real earthquake P-waves (compression waves) will simultaneously generate co-source and synchronous disturbances in three physical domains: solid (crust), gas (atmosphere), and sound waves (infrasound); while pseudo-signals (such as falling objects) mainly generate mechanical vibrations and lack this cross-domain physical coherence.

[0044] The cross-domain coherence identification module first preprocesses the event data window to extract signals for coherence analysis: Vibration signal For example, extract vertical component .

[0045] Infrasound signal :right Perform bandpass filtering on a specific frequency band.

[0046] Pressure disturbance signal :right Perform high-pass filtering or differential processing.

[0047] Cross-domain coherence identification module calculation and Between, and and Between, within the key time-frequency window (scale) of the P-wave ,time Cross-domain coherence within )

[0048] In a preferred embodiment, wavelet coherence (WTC) is used. : ; in, and These are signals and Continuous wavelet transform; It is a cross-wavelet spectrum; It is a time-frequency smoothing operator; Smoothed cross-spectral power; The product of smoothed autospectral power.

[0049] Through the In the key time-frequency region of P-wave , Intra-integration yields the accelerometer microphone coherence parameters. Coherence parameters of accelerometer and barometer .

[0050] Finally, the overall coherence score is calculated. : ; in, and These are preset weighting coefficients.

[0051] The cross-domain coherence identification module will Compared with the preset identification threshold Comparison: like The suspected event was determined to be a high-confidence earthquake event, and the data packet was passed to the next module.

[0052] like If the signal is identified as a spurious signal, the event handling process is terminated and the data packet is discarded.

[0053] The normalized parameter inverse calculation module is used to eliminate the influence of coupling state on the measurement amplitude after an event is identified as a high-confidence seismic event, and to inversely calculate the normalized ground motion parameters.

[0054] This module first starts from the event data window. In the process, extract the measured peak acceleration. : ; Subsequently, the normalization parameter inverse calculation module obtains the coupling coefficient corresponding to the occurrence of the event from the environment coupling state quantization module. Based on physical model (in (Based on the actual ground motion amplitude), perform inverse calculations to obtain normalized equivalent ground motion parameters. : ; in, It is a standardized estimate of ground motion intensity that is decoupled from the specific placement of the terminal.

[0055] The standardized information reporting module is used to encapsulate and report the identified and corrected standardized data. This module extracts the raw, high-dimensional waveform data into concise, high-value event information packages. The event information packet must contain at least the following fields: P-wave arrival timestamp (For example Terminal geographic coordinates Event confidence Normalized equivalent ground motion parameters and the coupling state when the event occurs. .

[0056] Finally, the standardized information reporting module transmits the information packet through the communication unit of the smart terminal. The data is sent to the cloud-based early warning server for subsequent aggregation analysis and early warning dissemination.

[0057] See attached document Figure 1 The method is executed in the smart terminal environment of the first part, and is completed through the processor calling and executing a series of functional modules in a coordinated manner. The overall process of the method includes: Environmental Coupling State Quantization: This step is continuously executed in the background by the environmental coupling state quantization module. This step is a preliminary step of the method of this invention, and its purpose is to enable the smart terminal to obtain prior information about its physical environment before any event occurs. The environmental coupling state quantization module continuously monitors the background noise of multimodal sensors (accelerometer, microphone, barometer), and as described in Part I (Invention Summary), by analyzing the background noise signal ( Extracting feature vectors ( ), and input the state recognition model ( ), to determine the current coupling state of the terminal in real time. .

[0058] The key output of this step is two AND... The bound, dynamically updated parameters are provided for use in subsequent steps: Dynamic trigger threshold This parameter is passed to the dynamic threshold triggering and event window capture steps to adaptively adjust the sensitivity of event triggering.

[0059] Coupling coefficient This parameter is stored and passed to the normalized ground motion parameter back-calculation step, used to correct the measurement amplitude after the event is confirmed to be real.

[0060] Dynamic threshold triggering and event window capture: This step is performed by the dynamic triggering and capture module, which continuously receives dynamic trigger thresholds provided by the environment coupling state quantization module. Meanwhile, the dynamic triggering and capture module monitors the accelerometer background signal in real time. .

[0061] When satisfied When conditions are met, determine if a suspected event has been triggered and record the trigger time. The triggering condition here is state-adaptive; for example, when When left to stand on a hard surface Lower to maintain high sensitivity; when When in a vehicle environment, To avoid frequent false triggering caused by continuous vibration.

[0062] Upon triggering, the module immediately captures a... Centered on, with a preset length of Multimodal synchronization data window The event data window is then passed to the cross-domain coherence identification step for identification.

[0063] Cross-domain coherence identification: This step is performed by the cross-domain coherence identification module. This step is the core technology gate of the method of this invention, used to distinguish high-confidence seismic events from pseudo-signals at the physical source. The cross-domain coherence identification module receives the event data window captured by the dynamic threshold triggering and event window capture steps, and performs acceleration ( ), infrasound ( and pressure disturbance The signal is subjected to cross-domain coherence analysis (e.g., wavelet coherence WTC) to calculate the comprehensive coherence score. .

[0064] This step includes a crucial conditional branch: Condition A (determined to be a spurious signal): If If the threshold for identification is reached, the cross-domain coherence identification module determines that the suspected event is a "false signal" (e.g., a drop or collision whose physical characteristics are not correlated with vibration in the acoustic or pressure domains). The event data window... If discarded, the processing flow terminates, and the method returns to the background monitoring state of the environmental coupling state quantification step.

[0065] Condition B (determined as a high-confidence event): If If so, the event is determined to be a high-confidence earthquake event, and the event data window and The results are passed to the normalized ground motion parameter back-calculation step for subsequent parameter back-calculation.

[0066] Normalized ground motion parameter backpropagation: This step is performed by the normalized parameter backpropagation module and is only activated when the cross-domain coherence identification step determines it to be a high-confidence seismic event.

[0067] The normalized ground motion parameter backpropagation step performs a crucial data correction process that depends on prior states. This module receives two inputs: Event data window from the cross-domain coherence identification step that has passed the identification process. .

[0068] Coupling coefficients derived from the environmental coupling state quantization step, corresponding to the occurrence of the event. .

[0069] Back-engineering normalized ground motion parameters starts from Extracting the measured peak acceleration Subsequently, the environmental coupling state quantization step provides... Perform inverse calculations to obtain normalized equivalent ground motion parameters. .

[0070] Standardized earthquake event information reporting: This step is performed by the standardized information reporting module, which collects the processing results of all preceding steps and encapsulates them into standardized event information packages. .

[0071] It contains the result, not the original waveform, and includes at least: the P-wave first arrival timestamp. (Source: Terminal geographic coordinates Event confidence Normalized equivalent ground motion parameters and coupling state After encapsulation, the standardized earthquake event information reporting process is completed via the communication unit of the smart terminal. The message is sent to the cloud-based early warning server. After the report is completed, the method returns to the background monitoring status of the environmental coupling state quantification step, awaiting the next event.

[0072] See attached document Figure 2 Environmental coupling state quantification is a preliminary step of the method of this invention, which is continuously executed in the background of the smart terminal. This step runs before the suspected event is triggered. Its purpose is to use the background noise data of the multimodal sensor to determine and quantify the physical environment coupling state of the terminal itself, thereby providing dynamically updated prior parameters for the subsequent dynamic threshold triggering step and normalization parameter back-calculation step.

[0073] In one embodiment, the specific execution process of the environmental coupling state quantization step includes: First, in the background state (i.e., the normal operating state where the dynamic triggering conditions defined in the dynamic threshold triggering step are not met), the processor of the smart terminal controls the multimodal sensor within a sliding background time window. Within the system, background noise signals are acquired synchronously. The signals include at least: Triaxial background signal of accelerometer (ACC) ; Background sound pressure signal of microphone (MIC) ; Background atmospheric pressure signal from the barometer (BARO) .

[0074] Next, the environmental coupling state quantization module extracts features from the collected background noise signal to construct a state feature vector describing the current physical state. In one embodiment, It should include at least the following components: Accelerometer stability characteristics Accelerometer stability characteristics reflect the micro-vibration level and stability of the platform on which the terminal is located. This is achieved by calculating the accelerometer background signal. The L2 norm in the sliding background time window The variance within the range is obtained.

[0075] Accelerometer frequency domain characteristics The frequency domain characteristics of an accelerometer reflect the frequency properties of background vibrations, such as the zero-crossing rate. It is used to help distinguish between high-frequency jitter and low-frequency displacement.

[0076] Infrasound energy ratio The proportion of infrasound energy is used to identify the presence of specific low-frequency noise sources (such as vehicle engines). First, the microphone signal... Perform bandpass filtering (e.g.) Infrasound signal obtained Then calculate its energy percentage.

[0077] Pressure stability characteristics The air pressure stability characteristic reflects the stability of the atmospheric environment and is used to distinguish between indoor (stable) and outdoor (wind disturbance).

[0078] Finally, the state feature vectors are combined into .

[0079] Subsequently, the environmental coupling state quantization step is based on Determine the current coupling state of the terminal. Predefine a set of discrete, mutually exclusive environmental coupling states. The set of environmental coupling states is predetermined through experimental calibration, and each state... Corresponding to a specific set Feature range.

[0080] For example, Corresponding to the static state of a hard surface, its It is characterized by extremely low and ; For example, For soft surfaces in a static state (such as a sofa), its Features and Similar, but its (discussed later) coupling coefficient and Clearly different; For example, Corresponding to the vehicle's driving status, its Performance is consistently high and specific frequency bands; For example, Corresponding to whether it is held in hand or in a pocket, its Manifestation It exhibits periodic characteristics of walking or shaking.

[0081] The environmental coupling state quantization step uses a pre-trained state recognition model. (Such as Gaussian Mixture Model (GMM), decision tree, or support vector machine (SVM)) to perform this classification task, state recognition model The input is The output is The index, thus determining .

[0082] Finally, the environmental coupling state quantization step prepares and outputs the necessary prior parameters for subsequent steps. A parameter lookup table (LUT) is pre-stored in the smart terminal's memory, and this LUT stores each coupling state... Mapped to a set of parameters, which at least include: Dynamic trigger threshold The dynamic trigger threshold is based on the state. Typical background noise levels (e.g.) It is set by adding a safety margin, for example, (Hard surfaces) are set very low to maintain high sensitivity, while The (vehicle-mounted) settings are set too high to avoid misinterpreting road bumps as suspected incidents.

[0083] Coupling coefficient The coupling coefficient is a dimensionless gain value, representing the state... Below, real ground motion Compared with the measured value of the terminal sensor The transfer function relationship between them can be approximated as follows: , It is achieved by testing each type on a controlled vibration table. The state is obtained through calibration experiments, for example. The value for (hard surfaces) is close to 1.0, while The value for (soft surfaces) is significantly less than 1.0 because of its vibration attenuation.

[0084] In each execution cycle of the environment coupling state quantization step, the environment coupling state quantization module determines... Then, immediately retrieve the corresponding result from the lookup table. and . The output is used in the dynamic threshold triggering and event window capture steps. The output (or temporary storage) is used for the back-calculation step of normalized ground motion parameters.

[0085] The dynamic threshold triggering and event window capture step is a key connecting step in this invention. The purpose of the dynamic threshold triggering and event window capture step is to use the dynamic parameters provided by the environmental coupling state quantization step to adaptively monitor the vibration signal and capture a complete, multimodal synchronous data window for subsequent identification when a suspected event occurs.

[0086] See attached document Figure 3 In one embodiment, the specific execution process of the dynamic threshold triggering and event window capture step includes: the dynamic threshold triggering and event window capture step continuously receives dynamically updated threshold values ​​from the environment coupling state quantization step. Dynamic trigger threshold As the terminal's current coupling state It adapts and adjusts itself to changes.

[0087] Simultaneously, the dynamic threshold triggering and event window capture steps monitor the triaxial acceleration signals collected by the accelerometer of the smart terminal in real time. The dynamic threshold triggering and event window capture steps continue to... The L2 norm (i.e., the total amplitude of vibration) and the received A comparison is performed. A suspected event is determined when the following triggering conditions are met: ; in, yes The L2 norm, with a state-adaptive triggering condition, ensures that in low-noise states (such as...) It exhibits high sensitivity when placed on a hard, static surface, and high sensitivity under high noise conditions (such as...). When mounted in a vehicle, it exhibits high anti-interference capabilities, avoiding false triggering caused by background noise.

[0088] When the above triggering conditions are met, the dynamic triggering and capture module immediately records the current moment as the trigger moment. In Record Then, the dynamic triggering and capture module immediately executes the event window capture operation, which extracts a preset time length from the sensor data buffer of the smart terminal. The data window, the preset time length of the data window. The time range is defined as: ; in, It is a preset pre-trigger duration used to ensure that P-arrival information of the vibration signal is captured before it crosses the threshold; It is a preset trigger duration used to ensure that the main energy of the P wave and subsequent wave phases are captured.

[0089] This window capture operation is multimodal synchronous; the dynamic triggering and capture module synchronizes all relevant sensor data acquired based on the same clock, ensuring that the captured dataset is included. Within the time window, the dataset is encapsulated as an event data window. .

[0090] yes Accelerometer data in the window, yes Microphone data within the window, yes The barometer data in the window.

[0091] Finally, the dynamic triggering and capturing module will encapsulate the event data window. As its output, it is passed to the cross-domain coherence identification step for subsequent physical source identification.

[0092] See attached document Figure 4 The cross-domain coherence identification step is executed by the cross-domain coherence identification module. This step receives the event data window of suspected events transmitted by the dynamic threshold triggering and event window capture steps. .

[0093] The purpose of the cross-domain coherence identification step is to identify the physical source of suspected events and distinguish between high-confidence seismic events and pseudo-signals. The physical principle underlying the identification process is that high-confidence seismic events and pseudo-signals have fundamental differences in energy coupling and co-origin characteristics across multiple physical domains.

[0094] The P-wave of a high-confidence seismic event is a compression wave that propagates through the Earth's crust. When the P-wave reaches the location of a smart terminal, physical effects simultaneously produce measurable, homogeneous responses in three different physical domains: Solid vibration domain: P-waves, as crustal vibrations, directly cause mechanical motion on the Earth's surface (and smart terminals). This mechanical motion is captured by accelerometers (ACC), which is what appears in the event data window. .

[0095] Sound wave domain: Rapid vertical motion of the Earth's surface (caused by P-waves) acts like a piston, compressing the air medium above it and radiating infrasound pulses (usually with frequencies below Hz) of the same origin as the surface motion. These infrasound pulses are captured by a microphone (MIC), i.e.,... .

[0096] Atmospheric pressure domain: Similarly, the atmospheric compression or thinning caused by ground uplift or subsidence due to P waves directly creates localized, rapid atmospheric pressure disturbances. These disturbances are captured by barometers (BARO). .

[0097] Since the three signals mentioned above (vibration, infrasound, and pressure disturbance) are all driven by the same physical event (P-wave), they must exhibit high physical homology and phase correlation, i.e., high cross-domain coherence, within the specific time-frequency window of the P-wave's initial arrival.

[0098] In contrast, the physical characteristics of spurious signals are different, taking a typical terminal drop or collision as an example: Solid vibration domain: Drops or collisions generate violent mechanical shocks, causing the accelerometer (ACC) to capture high-amplitude vibration signals. .

[0099] Sound wave domain and atmospheric pressure domain: This type of mechanical impact is mainly transmitted within the solid. Unlike P-waves, it does not generate strictly homologous, phase-locked infrasound pulses or atmospheric pressure disturbances through the surface-atmosphere coupling mechanism. Although falling objects also produce sound signals, the time-frequency and phase relationships between the sound signals and the impact vibration signals are completely different from the physical characteristics of P-waves.

[0100] Therefore, the identification principle of the cross-domain coherence identification step is: through the event data window... , and (After preprocessing) cross-domain coherence analysis is performed to quantitatively calculate their physical homology (i.e., If the calculated coherence is high ( If the result is positive, it indicates that the physical characteristics of the "suspected event" are consistent with the cross-domain coupling physical model of P-waves, and it is determined to be a "high-confidence earthquake event".

[0101] If the calculated coherence is low ( If the result is negative, it indicates that the energy of the event is mainly concentrated in the mechanical vibration domain and lacks cross-domain physical coupling. Its characteristics are consistent with the physical model of the pseudo signal, so it is judged as a pseudo signal and discarded.

[0102] The cross-domain coherence identification step is performed by the cross-domain coherence identification module. Before performing coherence analysis, the cross-domain coherence identification module first needs to process the raw event data window provided by the dynamic threshold triggering and event window capture steps. Perform signal preprocessing.

[0103] The purpose of the preprocessing step is to extract specific signal components that are directly related to the physical effects of P-waves from the raw, broadband, noisy sensor data. This ensures that the subsequent coherence calculations are performed on physically related frequency bands, thereby improving the accuracy of identification.

[0104] In one embodiment, the specific execution process of the signal preprocessing step includes: Extracting vibration reference signals Extracting vibration reference signals from vibration signals Extracted from the P-wave, it serves as the vibration benchmark for cross-domain coherence analysis. Since the P-wave longitudinal characteristics generate a significant vertical (Z-axis) motion component at the Earth's surface, in a preferred embodiment, it is directly extracted... Vertical axis acceleration data As a vibration reference signal: ; In another embodiment, to address the issue of arbitrary terminal orientation, triaxial data can be used. Perform principal component analysis (PCA) to extract the signal of the first principal component (i.e., the direction of maximum variance) as... .

[0105] Extracting infrasound signals Extracting infrasound signals from microphone data Extracting from the P-wave coupling, the main energy of the sound wave is concentrated in the infrasound frequency band (e.g., below Hz). To extract this component and suppress irrelevant interference such as human voice and high-frequency noise, the cross-domain coherence identification module... A bandpass filter is applied, and the bandpass filter is set to the key frequency band of the infrasound, for example... ,in It can be 0.5Hz. It can be 20Hz.

[0106] ; in, This represents a bandpass filtering operation.

[0107] Extract pressure disturbance signal Extracting pressure disturbance signals from barometer data Extracted from air pressure data. It includes rapid pressure disturbances caused by P-waves, as well as slow background pressure drifts caused by weather changes. To extract the rapid disturbance components related to P-waves, the cross-domain coherence identification module... A high-pass filter is applied to remove low-frequency background drift. The cutoff frequency of the high-pass filter is... Set to a lower value, for example =0.1Hz or 0.5Hz.

[0108] ; In another embodiment, the high-pass filtering operation can be calculated. First-order difference To approximate the implementation, that is ,in It is discrete time.

[0109] After completing the above preprocessing, the cross-domain coherence identification module obtains a set of aligned analysis signals with irrelevant frequency band interference filtered out. ,in .

[0110] The cross-domain coherence analysis step is performed after the signal preprocessing step and is completed by the cross-domain coherence discrimination module. The input to the cross-domain coherence analysis step is the preprocessed, time-aligned analysis signal set from the signal preprocessing step. ,in The purpose of the cross-domain coherence analysis step is to quantitatively calculate the physical homology, i.e., coherence, of these signals in the key time-frequency domain.

[0111] In a preferred embodiment, the cross-domain coherence identification module employs wavelet coherence (WTC) analysis, which evaluates the phase locking and energy correlation between two signals in the time-frequency domain.

[0112] The analysis process first calculates the coherence of the two signal pairs separately: Signal pair 1: Vibration reference signal With infrasound signals .

[0113] Signal pair 2: Vibration reference signal With pressure disturbance signal .

[0114] With signal pair 1 and For example, the calculation process of the wavelet coherence analysis method includes: First of all, and Perform continuous wavelet transforms on each to obtain their time-frequency (time-frequency) representations. ,scale Complex wavelet coefficients in the domain and Next, calculate and Cross wavelet spectrum : ; in, yes .

[0115] Subsequently, wavelet coherence is calculated. , It is a value normalized in the time-frequency domain, and its range is... , where 1 represents at a specific time and scale The formula for calculating complete coherence is as follows: ; in, Indicates taking the modulus of a complex number; It is a smoothing operator that performs in both time and scale; It is a scale factor used for energy normalization.

[0116] The calculation result is a time-frequency coherence spectrum. To obtain the scalar value used for judgment, the cross-domain coherence discrimination module operates within a predefined key time-frequency window (time window) corresponding to the physical characteristics of the P-wave. ,scale Inside, for Perform integration or calculate the mean. lie in nearby, Corresponding to and The preprocessing frequency band (e.g., 0.5Hz) Hz).

[0117] Therefore, the coherence index of the accelerometer microphone is calculated. : ; in, and These represent the measures (length or range) of the time window and the scale window, respectively. That is and Mean coherence within the critical window of the P-wave; Double integral symbol; All values ​​falling within the critical time-frequency window of the P-wave are summed up.

[0118] In exactly the same way, for signal pairs and Performing the above calculations, we obtain and wavelet coherence between and in the same critical time-frequency window Integrating within the range, the coherence parameters of the accelerometer and barometer are calculated. : ; The final output of this step is two scalar values: and It is passed to the comprehensive score and event identification steps to calculate the final comprehensive coherence score.

[0119] The comprehensive score and event identification step follows the cross-domain coherence analysis step and is completed by the cross-domain coherence identification module. The core function of this step is to fuse multiple calculated cross-domain coherence indicators to form a single identification score, and then determine the nature of the suspected event based on this score. In one embodiment, the specific execution process of the comprehensive score and event identification step includes: Overall coherence score Calculation: The cross-domain coherence discrimination module receives two quantization metrics: accelerometer-microphone coherence metrics. Coherence parameters of accelerometer and barometer To comprehensively reflect the cross-domain coupling characteristics of events, the cross-domain coherence identification module calculates a unified comprehensive coherence score through a weighted fusion model. This score is the final quantitative representation of the physical homology of events, and the calculation formula is as follows: ; in, These are the weighting coefficients assigned to the coherence metrics of the accelerometer microphone; These are the weighting coefficients assigned to the coherence parameters of the accelerometer and barometer. Weighting coefficient and It is a pre-defined non-negative real number that satisfies the normalization condition. The determination of these weighting coefficients is based on statistical analysis of actual calibration data on sensor response characteristics, background noise levels in the environment, and the sensitivity of different coupled physical channels to seismic P-waves. For example, under certain specific conditions, the barometer's instantaneous pressure disturbance response to P-waves is more stable and predictable than the microphone's response to infrasound. The value can be set to greater than .

[0120] The range of values ​​is A score closer to 1 indicates stronger synchronicity and phase consistency of the event across the three physical domains of acceleration, infrasound, and air pressure, and is more consistent with the physical model of earthquake P-waves.

[0121] Event nature identification and determination: In calculation The cross-domain coherence identification module will then compare this score with a preset identification threshold. By comparing the results, a final determination of the nature of the event is made, and a threshold for identification is set. This is a key parameter whose value is determined through training and validation on a large dataset of real earthquake events and various pseudo-signals (such as falling, knocking, traffic vibrations, etc.). This threshold aims to achieve a balance between high recall for real earthquake events and low false alarm rate for pseudo-signals.

[0122] The overall score and event identification steps include a logical decision branch based on the comparison results: Condition 1 (Determination of High-Confidence Earthquake Events): If The cross-domain coherence identification module then determined the suspected event to be a high-confidence seismic event. This determination indicates that the event not only exhibits significant characteristics in mechanical vibration but also shows highly coherent responses to the vibration signal in both infrasound and atmospheric pressure disturbances. This synchronous coupling characteristic across multiple physical domains verifies that the event originates from a physical process capable of simultaneously exciting responses in solid and gaseous media, consistent with the propagation characteristics of seismic P-waves. After making this determination, the event data window... And the calculated The results will be passed to the normalized ground motion parameter back-calculation step for subsequent measurement data correction.

[0123] Condition 2 (Determination of False Signals): If If the cross-domain coherence identification module determines that the suspected event is a false signal, this determination indicates that although the sensor detects a vibration of a certain amplitude, the vibration lacks high time-frequency coherence with infrasound or atmospheric pressure disturbances. For example, everyday activities such as dropping, colliding, or picking up a terminal can cause obvious mechanical vibration, but the sound or pressure changes generated in the air medium do not have the same source coupling relationship as P waves. In this case, the event data window will be discarded by the cross-domain coherence identification module, the event processing flow will be terminated, and the system will return to the background monitoring state shown in the environmental coupling state quantification step and the dynamic threshold triggering and event window capture step, waiting for the next potential event to occur.

[0124] Through the above-mentioned comprehensive score calculation and identification mechanism, the method of the present invention realizes the identification of the physical nature of earthquake events, thereby suppressing false alarms caused by non-earthquake sources and improving the reliability of intelligent terminal earthquake early warning systems.

[0125] See attached document Figure 5The normalized ground motion parameter back-calculation step is executed by the normalized parameter back-calculation module. The normalized ground motion parameter back-calculation step is only activated and executed after the suspected event is determined to be a high-confidence earthquake event in the cross-domain coherence identification step. The purpose of the normalized ground motion parameter back-calculation step is to eliminate the measurement amplitude distortion caused by different terminal placement (coupling) states and obtain standardized ground motion parameters.

[0126] The normalized ground motion parameter back-calculation step first performs a measurement parameter extraction step: In one embodiment, the specific execution process of the measurement parameter extraction step includes: The normalized parameter inversion module receives the authenticated event data window from the cross-domain coherence identification step. The normalized parameter inversion module first locates the accelerometer data within the data window. .

[0127] The task of the normalized parameter back-calculation module is to... From the waveform data, scalar values ​​are extracted. These scalar values ​​represent the state of the terminal. The maximum vibration amplitude measured within the time window is defined as the peak acceleration. To ensure that the measurement value is not affected by the spatial orientation of the terminal device, Calculated as Throughout The maximum L2 norm (i.e., the maximum vector composite magnitude) within the window. It contains a series of discrete moments The triaxial acceleration vector The calculation formula is as follows: ; in, It is a complete event time window; At discrete time The L2 norm of the triaxial acceleration vector is calculated as follows: ; The only output of the cross-domain coherence identification step is the calculated scalar value. scalar value It is then passed to the next coupled correction and parameter back-calculation step of the normalized ground motion parameter back-calculation as input for performing normalization correction.

[0128] The coupling correction and parameter inverse calculation steps are executed immediately after the measurement parameter extraction steps and are performed by the normalized parameter inverse calculation module.

[0129] The purpose of the coupling correction and parameter inverse estimation steps is to use pre-quantified environmental prior information to correct the original measurement values ​​extracted in the measurement parameter extraction steps, thereby eliminating measurement biases introduced by the smart terminal due to different placement states, and calculating standardized ground motion estimates independent of the coupling state. In one embodiment, the execution process of the coupling correction and parameter inverse estimation steps includes: First, the normalized parameter inverse calculation module receives the output from the measurement parameter extraction step, namely the measured peak acceleration. .

[0130] Meanwhile, the normalized parameter back-calculation module retrieves or receives the coupling coefficient corresponding to the occurrence of this "high-confidence earthquake event". ; It refers to the physical coupling state of the terminal at the time the event occurs (e.g., Let it stand on a hard surface. (Let it sit on a soft surface).

[0131] In the environmental coupling state quantification step, according to The dimensionless coefficients are retrieved from a pre-stored parameter lookup table. These dimensionless coefficients quantitatively characterize the... In this state, real ground motion Compared with the measured value of the terminal sensor The magnitude transfer gain between them can be approximated by the physical model as follows: ; The normalized parameter inverse calculation module performs a reverse calculation operation based on the above physical model to solve for the normalized ground motion parameters. The normalized parameter inverse calculation module obtains the received data... Divide by its corresponding A corrected, standardized parameter was calculated, which was defined as the normalized equivalent ground motion parameter. : ; By reverse calculation The value eliminates the effects of different coupling states, for example, when the terminal is in (soft surface, its) When it is significantly less than 1.0, its Will be Division amplification; when the terminal is in (Hard surfaces, their) When it approaches 1.0, its The basics remain unchanged.

[0132] therefore, It is an estimate that is decoupled from the specific placement location of the terminal and can be standardized to reflect the actual ground motion intensity. The final output of the coupling correction and parameter inverse calculation steps is a scalar value. scalar value Together with the confidence level from the cross-domain coherence identification module in the signal preprocessing step This information, along with other data, is transmitted to the standardized earthquake event information reporting process for subsequent event information encapsulation.

[0133] See attached document Figure 6 The standardized earthquake event information reporting step is the last step in the processing flow of the method of the present invention on the smart terminal side. It is executed by the standardized information reporting module and is activated after the normalization parameter back-calculation step has completed the calculation.

[0134] The purpose of the standardized earthquake event information reporting procedure is to encapsulate and report the highly condensed and standardized event results obtained from the terminal side (i.e., the edge side), rather than the raw waveform data, to the cloud early warning server. The key technical advantage of this design is that it reduces the occupation of network uplink bandwidth and ensures that event information can still be reported with a high success rate and low latency during earthquake disasters (when the communication network is congested or interrupted).

[0135] The standardized earthquake event information reporting process first performs a data packet encapsulation step. In one embodiment, the specific execution process of the data packet encapsulation step includes: the standardized information reporting module, acting as a data aggregation unit, first collects key calculation results from all preceding steps. The standardized information reporting module does not process the raw waveforms but only collects summary information. The collected information fields include at least: Event trigger timestamp The event trigger timestamp originates from the trigger time recorded by the dynamic triggering and capturing module. The arrival time of the P-wave, after being corrected by a P-wave first arrival algorithm (such as STA / LTA or a more complex picking algorithm), has high precision (e.g., millisecond level) and is one of the core inputs for cloud servers to perform joint inversion positioning of multiple stations.

[0136] Terminal geographic coordinates The terminal's geographic coordinates are provided by the smart terminal's positioning unit (such as GPS, BeiDou, or network positioning), and are usually expressed as latitude and longitude. .

[0137] Event confidence Event confidence is derived from the cross-domain coherence identification module. This value quantifies the physical authenticity of the event, and the cloud server can use it as a weighting factor.

[0138] Normalized motion parameters The normalized motion parameters are derived from the normalized parameter inverse calculation module. This value is a standardized estimate of ground motion amplitude that has eliminated local coupling effects and is the core input for the cloud server to assess earthquake intensity.

[0139] Environmental coupling state : Environmental coupling state (e.g.) , (etc.) originates from the environmental coupling state quantification module and is used as an auxiliary metadata report for quality assessment or model iteration in the cloud.

[0140] Terminal Unique Identifier Used to identify the source of data in the cloud.

[0141] After collecting all the above fields, the standardized information reporting module encapsulates these fields according to a predefined, lightweight data protocol format to generate a standardized event information package. This encapsulation format is designed to be highly compact to minimize the byte size of data packets, for example, It is a structured message containing the above six key fields. The final output of the data packet encapsulation step is a fully encapsulated event information packet ready to be sent. Event information packet It is then passed to the communication reporting step and sent to the cloud early warning server through the terminal's communication unit (such as 4G / 5G / Wi-Fi).

[0142] The data reporting step follows immediately after the data packet encapsulation step and is performed by the standardized information reporting module. In one embodiment, the specific execution process of the data reporting step includes: First, the standardized information reporting module receives the event information packet that has been encapsulated in the previous step. .

[0143] The standardized information reporting module immediately invokes the communication unit of the smart terminal, such as a 5G / 4G cellular network module or a Wi-Fi module, to transmit the event information packet through one or more available network interfaces. Send to one or more pre-configured cloud alert servers via network addresses (such as IP addresses or domain names).

[0144] The reporting process has the following key technical features to ensure high reliability and low latency in disaster scenarios: High priority and concurrent reporting strategy: This reporting task is given high priority in the operating system of the smart terminal to ensure that it can access the communication hardware immediately. In a preferred embodiment, in order to deal with the interruption of specific networks (such as local Wi-Fi routing) caused by earthquakes, the standardized information reporting module implements a concurrent reporting mechanism. If the terminal is simultaneously connected to Wi-Fi and cellular networks, the standardized information reporting module can initiate reporting requests through these two independent network interfaces at the same time. As long as any request is successfully delivered to the server, it is considered as a successful report.

[0145] Fast fault tolerance and limited retries: In another embodiment, the standardized information reporting module adopts a fast fault tolerance strategy. It first attempts to use a preferred network (e.g., Wi-Fi) and sets an extremely short connection and response timeout threshold (e.g., 0 milliseconds). If the preferred network fails to report successfully within this threshold (e.g., connection failure or failure to receive application layer confirmation from the server), the standardized information reporting module will immediately and seamlessly switch to a backup network (e.g., 5G) for reporting without long waiting or retries. The standardized information reporting module only executes a limited, aggressive retry strategy (e.g., retrying a maximum of 3 times within a total of 2 seconds) to avoid unnecessary resource consumption when the network is completely interrupted.

[0146] Congestion resistance based on lightweight packaging: The core advantage of the data reporting step lies in its reported... It is a standardized summary information packet with a very small data volume (e.g., only a few hundred bytes to a few thousand bytes), compared to the reported raw multimodal sensor waveform data. It will be in the megabyte range, through only reporting. The method of this invention consumes almost zero uplink bandwidth. This means that even when an earthquake causes severe congestion in communication networks (especially cellular networks) and uplink bandwidth is greatly compressed, this small data packet still has a very high probability of penetrating the congestion and successfully reaching the cloud-based early warning server. The final output of the data reporting step is... Once the report is successfully sent to the cloud server (whether successful or not), the local processing flow of the method of this invention on the smart terminal side ends, and the system returns to the continuous background monitoring state defined by the environmental coupling state quantification step and the dynamic threshold triggering and event window capture step, waiting for the next suspected event to be triggered.

Claims

1. A method for data acquisition and earthquake event detection and processing using a dedicated early warning terminal, characterized in that, The method is applied to a smart terminal equipped with an accelerometer, microphone, and barometer, and includes the following steps: S1. Environmental Coupling State Quantization: Continuously collect background noise signals from the accelerometer, the microphone, and the barometer within a sliding background time window, determine the current coupling state of the smart terminal, and retrieve or generate dynamic trigger thresholds and coupling coefficients based on the current coupling state. S2, Dynamic Threshold Triggering and Event Window Capture: Continuously monitor the real-time acceleration signal. When the L2 norm of the real-time acceleration signal is greater than the dynamic trigger threshold, it is determined as a suspected event and an event data window containing multimodal synchronization data is captured. S3. Cross-domain coherence identification: Perform cross-domain coherence analysis on the acceleration signal, microphone signal, and barometer signal in the event data window, calculate the comprehensive coherence score, and compare the comprehensive coherence score with a preset identification threshold. If the comprehensive coherence score is less than or equal to the identification threshold, it is determined to be a spurious signal and the event data window is discarded; if the comprehensive coherence score is greater than the identification threshold, it is determined to be a high-confidence earthquake event and subsequent steps are executed. S4. Back-dive of normalized ground motion parameters: When a high-confidence earthquake event is identified, the measured peak ground acceleration is extracted from the event data window, and the normalized equivalent ground motion parameters are calculated using the coupling coefficient obtained in step S1. S5. Standardized earthquake event information reporting: Encapsulate an event information packet containing the normalized equivalent ground motion parameters, the comprehensive coherence score, and the current coupling state, and report the event information packet to the cloud early warning server.

2. The method for data acquisition and earthquake event detection and processing using a dedicated early warning terminal according to claim 1, characterized in that, The S1 step specifically includes: The background noise signal is acquired synchronously within the sliding background time window; Extract state feature vectors from the background noise signal. The state feature vectors include at least: the L2 norm variance of the accelerometer triaxial signal, the infrasound energy ratio of the microphone signal, and the short-time variance of the barometer signal. The state feature vector is classified using a preset state recognition model to determine the current coupling state.

3. The method for data acquisition and earthquake event detection and processing using a dedicated early warning terminal according to claim 2, characterized in that, The S1 step further includes: pre-storing a parameter lookup table in the memory, wherein the parameter lookup table maps the coupling state to the corresponding dynamic trigger threshold and coupling coefficient; Based on the current coupling state determined in step S1, the dynamic trigger threshold and the coupling coefficient are retrieved from the parameter lookup table and output.

4. The method for data acquisition and earthquake event detection and processing using a dedicated early warning terminal according to claim 1, characterized in that, In step S2, the time range of the event data window is defined as from the trigger time minus a preset pre-trigger duration to the trigger time plus a preset post-trigger duration.

5. The method for data acquisition and earthquake event detection and processing using a dedicated early warning terminal according to claim 1, characterized in that, Step S3, prior to coherence analysis, also includes: Extract the vibration reference signal from the accelerometer data in the event data window; A bandpass filter is applied to the microphone data in the event data window to extract the infrasound signal; Apply a high-pass filter or differential processing to the barometer data in the event data window to extract the pressure disturbance signal.

6. The method for data acquisition and earthquake event detection and processing using a dedicated early warning terminal according to claim 5, characterized in that, The cross-domain coherence analysis specifically includes: The wavelet coherence analysis method is used to calculate the wavelet coherence spectrum between the vibration reference signal and the infrasound signal, and between the vibration reference signal and the pressure disturbance signal, within the key time-frequency window of the P-wave. The accelerometer microphone coherence index and the accelerometer barometer coherence index are calculated by integrating or averaging the wavelet coherence spectrum within the key time-frequency window of the P-wave.

7. The method for data acquisition and earthquake event detection and processing using a dedicated early warning terminal according to claim 6, characterized in that, The calculation of the overall coherence score includes: The comprehensive coherence score is calculated using a weighted fusion model. The comprehensive coherence score is equal to the accelerometer microphone coherence index multiplied by the microphone weight coefficient, plus the accelerometer barometer coherence index multiplied by the barometer weight coefficient.

8. The method for data acquisition and earthquake event detection and processing using a dedicated early warning terminal according to claim 1, characterized in that, In step S4, the extraction and measurement of peak acceleration includes: Calculate the maximum L2 norm of the accelerometer data in the event data window over the entire time range of the event data window.

9. The method for data acquisition and earthquake event detection and processing using a dedicated early warning terminal according to claim 1, characterized in that, In step S5, the event information package further includes: the P-wave arrival timestamp and the terminal geographic coordinates.

10. The method for data acquisition and earthquake event detection and processing using a dedicated early warning terminal according to claim 1, characterized in that, The S5 step also includes: By employing a concurrent reporting mechanism or a fast fault-tolerance strategy, when the smart terminal simultaneously accesses both Wi-Fi and cellular networks, it simultaneously or immediately after the preferred network times out, reports the event information packet through both network interfaces.

Images