Multi-stage dynamic triggered avalanche infrasound monitoring and warning method

Abstract

The application relates to the technical field of natural disaster monitoring and early warning, and discloses a multi-stage dynamic triggering avalanche infrasound monitoring and early warning method, which comprises the following steps: collecting infrasound data by using an infrasound array, and acquiring meteorological monitoring data and an avalanche risk level of a target monitoring area; determining a monitoring operation mode of the infrasound array; when running in a deep hibernation stage mode, controlling the infrasound array to perform low-frequency intermittent collection; when running in a pre-triggering standby stage mode, controlling the infrasound array to start full-channel continuous collection, and switching to a full-amount identification stage mode when the infrasound signal amplitude meets the pre-triggering condition; when running in the full-amount identification stage mode, if it is determined that an effective avalanche event occurs, switching to a staged early warning stage mode; and when running in the staged early warning stage mode, issuing an avalanche early warning signal of a corresponding level. By using the application, the power consumption of the equipment in the field during a non-high-risk period can be reduced, and the reliability of avalanche event identification and the adaptability to complex environments can be improved.

Description

Technical Field

[0001] This invention relates to the field of natural disaster monitoring and early warning technology, specifically to a multi-level dynamically triggered avalanche infrasound monitoring and early warning method. Background Technology

[0002] Avalanches are one of the most sudden natural disasters in high-altitude mountainous areas worldwide, threatening settlements, infrastructure, and tourism activities. Current avalanche forecasting systems primarily rely on fixed-point monitoring of meteorological elements and snow conditions to assess the probability of avalanches in a region; however, real-time monitoring data of actual avalanche events is the core basis for capturing early signs of snow instability and a key trigger for initiating emergency response.

[0003] During avalanche movement, turbulent powder clouds disturb the atmosphere, generating infrasound vibrations, which are natural infrasound sources. Therefore, infrasound technology is also an important technical direction for avalanche monitoring. Existing infrasound avalanche monitoring technologies still have some problems and cannot meet the operational needs of high-altitude mountainous areas: (1) The existing mainstream scheme adopts a continuous acquisition mode. In remote high-altitude mountainous areas without power grid coverage, relying on solar power, the equipment may shut down due to insufficient power supply on cloudy or snowy days in winter. The simple intermittent acquisition scheme is very easy to lose avalanche signals that last only tens of seconds, which cannot meet the reliability requirements of disaster monitoring; (2) Existing triggering schemes mostly use fixed thresholds. However, the background noise of mountainous environments has significant diurnal variations, seasonal variations, and weather-related fluctuations. The fixed threshold is either set too low, causing frequent false triggers due to wind noise and other interference, or set too high, causing low-energy avalanche events to be missed, which cannot adapt to the complex and ever-changing field environment. How to effectively solve the above difficulties is an urgent problem to be solved. Summary of the Invention

[0004] This invention provides a multi-level dynamically triggered avalanche infrasound monitoring and early warning method, which can reduce the power consumption of equipment during non-high-risk periods in the field, improve the reliability of avalanche event identification and adaptability to complex environments, and meet the operational monitoring and disaster prevention and early warning needs of avalanches in remote mountainous areas at high altitudes.

[0005] Therefore, the present invention provides the following technical solution: A multi-level dynamically triggered avalanche infrasound monitoring and early warning method, the method comprising: Step 1: Use an infrasound array to collect infrasound data and obtain meteorological monitoring data and avalanche risk level of the target monitoring area; Step 2: Determine the monitoring and operation mode of the infrasound array. The monitoring and operation mode includes deep sleep mode, pre-trigger monitoring mode, full recognition mode, and graded early warning mode. Step 3: When the infrasound array is running in the deep sleep mode, control the infrasound array to perform low-frequency intermittent acquisition. When snow accumulation, snowfall events, or avalanche hazard level increases are detected, switch to the pre-trigger guard mode. Step 4: When the infrasound array is running in the pre-trigger guard mode, control the infrasound array to start full-channel continuous acquisition, and switch to the full-quantity recognition mode when the infrasound signal amplitude meets the pre-trigger condition. Step 5: When the infrasound array is running in the full recognition level mode, determine whether it is a valid avalanche event; if it is determined to be a valid avalanche event, switch to the graded early warning level mode. Step 6: When the infrasound array is operating in the graded warning level mode, it issues an avalanche warning signal of the corresponding level.

[0006] Optionally, in step 2, the initial operating mode of the infrasound array is the deep dormancy mode. Based on the snow cover period determination, avalanche risk level and meteorological conditions, the deep dormancy mode, the pre-triggered monitoring mode, the full-volume identification mode and the graded early warning mode are triggered step by step, and the acquisition strategy and processing logic are adjusted progressively.

[0007] Optionally, in step 4, when the infrasound array is running in the pre-triggered guard mode, the infrasound array is controlled to start full-channel continuous acquisition, and the original infrasound data for a preset duration is retained through a circular rolling buffer mechanism. At the same time, the dynamic adaptive trigger threshold is calculated in real time. When the infrasound signal amplitude exceeds the dynamic adaptive trigger threshold, the infrasound signal amplitude meets the pre-triggered condition, and the system switches to the full-volume recognition mode. When it is a non-snow accumulation period or the avalanche hazard level is reduced, the system returns to the deep dormancy mode.

[0008] Optionally, in step 4, the dynamic adaptive trigger threshold It is calculated by multiplying the multidimensional correction coefficient and the basic threshold. The calculation formula is as follows:

[0009] in, The site's basic noise threshold. This is the real-time environmental noise correction factor. This is a correction factor for avalanche risk. This is a seasonal correction factor; The circular rolling buffer mechanism is as follows: in the pre-triggered guard level mode, the most recent 60 seconds of full-channel raw infrasound data are continuously retained in the local edge buffer, and data exceeding the time limit is automatically rolled over; when the pre-triggered condition is met, the complete signal segment from 30 seconds before the trigger to real-time acquisition after the trigger is locked and extracted to ensure the integrity of the avalanche infrasound signal.

[0010] Optionally, in step 5, when the infrasound array is running in the full recognition level mode, the complete infrasound signal segment in the circular rolling buffer is retrieved. Based on the synchronous timing signal segments of each channel of the infrasound array, the generalized cross-correlation-phase transformation method is used to perform array cross-correlation processing, detection clustering, and feature parameter calculation. The preset multi-threshold constraint conditions are used to determine whether it is a valid avalanche event. If it is determined to be a non-avalanche false alarm event, the process returns to the pre-triggered guard level mode. If it is determined to be a valid avalanche event, the process switches to the graded early warning level mode.

[0011] Optionally, in step 5, the constraint rules for cluster detection are: the time interval between two adjacent infrasound detections is <10s, the difference in back azimuth angle is <10°, and the detection clusters generated by clustering satisfy: the number of effective infrasound detections is ≥15, and the total duration is ≥20s; The calculated characteristic parameters include apparent velocity change, apparent velocity change trend, apparent velocity-correlation coefficient ratio, and detection density; The multi-threshold constraint is specifically: the number of effective infrasound detections within the detection cluster. Duration: 20 seconds D <90s; Maximum apparent velocity v max <450m / s; average apparent velocity 300m / s< v m <425m / s; apparent velocity change -30m / s< <0m / s; apparent velocity variation trend ≤-1; average correlation coefficient >0.75; apparent velocity to correlation coefficient ratio <530m / s; detection density >0.85s -1 ; If a detection cluster simultaneously satisfies all of the above multi-threshold constraints, it is determined to be a valid avalanche event.

[0012] Optionally, in step 6, when the infrasound array is operating in the graded early warning mode, it issues an avalanche early warning signal of the corresponding level based on the identified valid avalanche events. At the same time, based on the identification results of valid avalanche events and changes in avalanche risk level, it dynamically optimizes the monitoring strategy and the calculation parameters of the dynamic adaptive trigger threshold. After the early warning ends, it returns to the pre-trigger guard mode.

[0013] Optionally, in step 6, avalanche warning signals of corresponding levels are issued. The rules for graded warnings are as follows: if there are ≥1 valid avalanche events in a single day, a primary warning is issued, indicating that the snow layer in the area is showing signs of instability; if there are ≥2 valid avalanche events in a single day, a medium warning is issued, indicating that avalanche activity in the area has significantly increased; if there are ≥3 valid avalanche events in a single day, a high-level warning is issued, indicating that the area has entered a period of high avalanche risk. When dynamically optimizing the monitoring strategy, based on the signal characteristics and corresponding environmental parameters of the identified valid avalanche events and non-false avalanche events, the calculation rules of each correction coefficient of the dynamic adaptive trigger threshold and the parameter range of the multi-threshold constraint conditions are optimized through self-learning iteration, thereby reducing the false alarm rate and missed detection rate in the monitoring process.

[0014] Optionally, in step 6, the self-learning iterative optimization method is as follows: based on historically accumulated positive samples (i.e., confirmed as valid avalanche events) and negative samples (i.e., confirmed as non-avalanche false alarm events), and the environmental parameters corresponding to the positive and negative samples, a batch or incremental learning method is adopted, using logistic regression, decision tree, quantile analysis, or gradient descent algorithm, to periodically adjust the calculation rules or value range of each correction coefficient in the dynamic adaptive trigger threshold, as well as the upper and lower limits of each parameter in the multi-threshold constraint conditions, in order to minimize the false alarm rate and the false negative rate.

[0015] A computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor to perform the steps of the multi-level dynamically triggered avalanche infrasound monitoring and early warning method.

[0016] This invention provides a multi-level dynamically triggered avalanche infrasound monitoring and early warning method. Addressing the limitations of existing avalanche monitoring technologies, such as limited coverage and insufficient all-weather monitoring capabilities, as well as the high power consumption during continuous data acquisition and the difficulty in balancing false alarm and missed detection rates with fixed thresholds, this invention designs a four-level progressive dynamic triggering architecture comprising a deep sleep mode, a pre-triggered monitoring mode, a full-scale recognition mode, and a graded early warning mode. Through a multi-source linked dynamic adaptive triggering threshold mechanism, a balance between low-power monitoring and high-reliability monitoring is achieved. Simultaneously, a dedicated recognition algorithm is constructed based on avalanche dynamics characteristics to accurately determine avalanche events and provide graded early warnings. Furthermore, a self-learning closed-loop system enables dynamic optimization of the monitoring strategy. Compared with existing technologies, this invention has the following technical advantages: (1) The four-level progressive dynamic triggering architecture designed in this invention can reduce the pressure on solar power supply in the deep dormancy mode during the non-snow accumulation period and adapt to long-term operation in remote mountainous areas with no power grid coverage; at the same time, through the ring rolling cache mechanism, it ensures that the complete avalanche signal is not lost when triggered, and achieves a balance between low power consumption and full protection during high-risk periods.

[0017] (2) The present invention has a multi-source linkage dynamic adaptive trigger threshold mechanism that dynamically adjusts the trigger threshold by combining real-time environmental noise, avalanche risk level and seasonal characteristics. When the wind noise is high, the threshold is automatically raised to reduce false alarms, and when the avalanche risk is high, the threshold is automatically lowered to avoid missed detection.

[0018] (3) Based on historical event identification results and false alarm feedback, the present invention can automatically optimize the trigger threshold and identification parameters. As the running time increases, the false alarm rate and false alarm rate of the algorithm continue to decrease, and the reliability of long-term operation continues to improve. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the embodiments will be briefly described below. Obviously, the drawings described below are merely some embodiments of the present invention, and those skilled in the art can obtain other drawings based on these drawings without any creative effort.

[0020] Figure 1 This is a flowchart of a multi-level dynamically triggered avalanche infrasound monitoring and early warning method in a specific embodiment of the present invention; Figure 2 This is a state switching logic diagram for a four-level monitoring operation mode in a specific embodiment of the present invention. Detailed Implementation

[0021] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0022] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 should fall within the scope of protection of the present invention.

[0023] like Figure 1 As shown, Figure 1 This is a flowchart of a multi-level dynamically triggered avalanche infrasound monitoring and early warning method according to a specific embodiment of the present invention. The multi-level dynamically triggered avalanche infrasound monitoring and early warning method includes: Step 1: Collect infrasound data using the deployed array of infrasound acquisition instruments, and simultaneously acquire meteorological monitoring data and regional avalanche risk level data for the target monitoring area; Step 2: Determine the monitoring and operation mode of the secondary array. The monitoring and operation modes include deep dormancy mode, pre-triggered monitoring mode, full-scale identification mode, and graded early warning mode. The initial operating mode of the infrasound array is the deep dormancy mode, based on snow cover period determination, avalanche risk level, and meteorological conditions, such as... Figure 2 As shown, the four modes are triggered step by step, progressively adjusting the acquisition strategy and processing logic; The rules for determining the snow cover period are as follows: if the on-site meteorological station monitors that the snow depth is ≥5cm for more than 3 consecutive days, it is determined to be the snow cover period; if there is no snow cover for 7 consecutive days, it is determined to be the non-snow cover period.

[0024] Step 3: When the infrasound array is operating in deep sleep mode, it is controlled to perform low-frequency sub-intermittent data acquisition, only conducting background noise baseline checks and judging pre-avalanche meteorological conditions. In deep sleep mode, the low-frequency sub-intermittent acquisition strategy is implemented, retaining only the core clock and basic data reception functions. The infrasound sensor, data acquisition unit, and edge computing unit are in deep sleep most of the time, with no high-power computational consumption and minimal power consumption. When snow accumulation, snowfall events, or an increase in avalanche hazard level are detected, it is woken up and switched to pre-triggered standby mode. Step 4: When the infrasound array is operating in pre-triggered standby mode, control the infrasound array to start continuous acquisition across all channels. A circular rolling buffer mechanism is used to retain raw infrasound data for a preset duration. Pre-triggered standby mode is the system's normal operating mode, performing only lightweight signal amplitude judgment and dynamic threshold calculation. The circular rolling buffer provides temporary data storage, without running high-performance array processing and recognition algorithms, keeping power consumption at a normal level. Simultaneously, a dynamic adaptive trigger threshold is calculated in real-time. When the infrasound signal amplitude meets the pre-triggered condition, the system immediately switches to full-volume recognition mode. When it is a non-snowy period or the avalanche risk level decreases, the system returns to deep sleep mode. The dynamic adaptive trigger threshold is calculated by multiplying a multi-dimensional correction coefficient by a base threshold. The calculation formula is as follows:

[0025] in, The site's basic noise threshold. This is the real-time environmental noise correction factor. This is a correction factor for avalanche risk. This is the seasonal adjustment factor.

[0026] The noise correction coefficient The calculation formula is:

[0027] in, For the pre-triggered guard level mode, the average amplitude of the ambient background noise in the 2-10Hz frequency band within the sliding window in the most recent 10 minutes; To monitor the historical average background noise amplitude of the site under non-snowfall and windless conditions; The value range is limited to 1.0~3.0.

[0028] avalanche risk correction factor The calculation formula is:

[0029] in, This is the base correction factor corresponding to the regional avalanche risk level; This is a snowfall correction factor calculated based on the on-site snow depth and 24-hour cumulative snowfall. This is a temperature correction factor calculated based on the temperature change range over the past 24 hours; The value range is limited to 0.8~2.0.

[0030] The seasonal correction factor The rule for determining the value is: during the snow cover period =1.0, during the non-snow cover period =3.0~5.0.

[0031] The ring-shaped rolling buffer mechanism is as follows: In the pre-trigger guard level mode, the most recent 60 seconds of full-channel raw infrasound data are continuously retained in the local edge buffer, and data exceeding the time limit is automatically rolled over; when the pre-trigger condition is met, the complete signal segment from 30 seconds before the trigger to the real-time acquisition after the trigger is immediately locked and extracted, and transmitted to the full-volume recognition level mode processing stage to ensure the integrity of the avalanche infrasound signal.

[0032] Step 5: When the infrasound array is running in full recognition mode, the complete infrasound signal segment in the circular buffer is retrieved. Based on the synchronous timing signal segments of each channel of the infrasound array, the generalized cross-correlation-phase transform (GCC-PHAT) method is used to perform array cross-correlation processing, detection clustering, and feature parameter calculation. The full recognition mode is an event-triggered temporary high-computing-power mode, which is only started for a short time after the pre-triggering condition is met. It runs the complete array cross-correlation processing, detection clustering, feature parameter calculation and multi-threshold joint judgment process. The edge computing unit is in a full-load operation state, and the power consumption is significantly increased. The system determines whether an event is a valid avalanche event based on preset multi-threshold constraints. If the event is determined to be non-avalanche, it reverts to the pre-triggered guard mode. If the event is determined to be a valid avalanche event, it switches to the graded early warning mode. The constraint rules for cluster detection are: the time interval between two adjacent infrasound detections is <10s, the difference in back azimuth angle is <10°, and the detection clusters generated by clustering satisfy: the number of valid infrasound detections is ≥15, and the total duration is ≥20s.

[0033] The formulas for calculating the characteristic parameters are as follows: Apparent velocity change The calculation formula is:

[0034] Among them, the detection clusters are divided into initial segments according to time. it Middle section mt Final section ft , The average apparent velocity of the initial segment. The average apparent velocity of the final segment; Apparent speed change trend The calculation formula is:

[0035] Here, sgn() is the sign function; it outputs -1 when the value inside the parentheses is negative, +1 when it is positive, and 0 when it is 0. The average apparent velocity of the middle segment; Apparent velocity-correlation coefficient ratio The calculation formula is:

[0036] in, To detect the average apparent velocity of the cluster, To detect the average cross-correlation number of clusters; Detection density The calculation formula is:

[0037] in, To detect the total number of effective infrasound detectors within the cluster, D The total duration of the detected cluster is measured in seconds (s).

[0038] The specific multi-threshold constraint conditions are as follows: Number of tests ; Duration: 20 seconds D <90s; Maximum apparent velocity v max <450m / s; Average apparent velocity 300 m / s v m <425m / s; Apparent velocity change -30m / s< <0m / s; Apparent speed change trend ≤-1; Average correlation coefficient >0.75; Apparent velocity to correlation coefficient ratio <530m / s; Detection density >0.85s -1 ; If a detection cluster satisfies all of the above constraints, it is determined to be a valid avalanche event.

[0039] Step 6: The tiered early warning mode operates at the highest power consumption. Based on full recognition, it simultaneously performs early warning signal transmission and self-learning parameter iterative optimization. The wireless communication module and storage module operate at full power, consuming the maximum power. In the tiered early warning mode, based on the identified valid avalanche events, corresponding avalanche early warning signals are issued. Simultaneously, based on the event recognition results and changes in avalanche risk, the monitoring strategy and the calculation parameters of the dynamic adaptive trigger threshold are dynamically optimized. After the early warning ends, it returns to the pre-trigger duty mode, completing the closed-loop operation of monitoring-recognition-early warning.

[0040] The rules for tiered early warning are as follows: If there is ≥1 valid avalanche event in a single day, a primary warning will be issued, indicating that the snow cover in the area is showing signs of instability. If there are ≥2 valid avalanche events in a single day, a medium-level warning will be issued, indicating a significant increase in avalanche activity in the region; If there are ≥3 valid avalanche events in a single day, a high-level warning will be issued, indicating that the area has entered a period of high avalanche risk.

[0041] The self-learning iterative optimization method is as follows: based on historically accumulated positive samples (confirmed as valid avalanche events) and negative samples (confirmed as non-avalanche false alarm events) and their corresponding environmental parameters (real-time noise amplitude, avalanche risk level, snow depth, snowfall, temperature difference, etc.), a batch or incremental learning method is used, employing algorithms such as logistic regression, decision trees, quantile analysis, or gradient descent, to periodically adjust the correction coefficients in the dynamically adaptive trigger threshold. , , The calculation rules or value range of ) and the upper and lower limits of each parameter in the multi-threshold constraint conditions are determined to minimize the false alarm rate and the false negative rate.

[0042] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0043] The present invention also provides a storage medium, which is a computer-readable storage medium storing a computer program thereon, the computer program being executable when it runs. Figure 1 The method shown may include some or all of the steps. The storage medium may include read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk, etc. The storage medium may also include non-volatile memory or non-transitory memory, etc.

[0044] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data provider to another website, computer, server, or data provider via wired or wireless means.

[0045] The embodiments of the present invention have been described in detail above. Specific implementation methods have been used to illustrate the present invention. The descriptions of the embodiments above are only for the purpose of helping to understand the method of the present invention, and are merely some, not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention, and the content of this specification should not be construed as a limitation of the present invention. Therefore, any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A multi-level dynamically triggered avalanche infrasound monitoring and early warning method, characterized in that, The method includes: Step 1: Use an infrasound array to collect infrasound data and obtain meteorological monitoring data and avalanche risk level of the target monitoring area; Step 2: Determine the monitoring and operation mode of the infrasound array. The monitoring and operation mode includes deep sleep mode, pre-trigger monitoring mode, full recognition mode, and graded early warning mode. Step 3: When the infrasound array is running in the deep sleep mode, control the infrasound array to perform low-frequency intermittent acquisition. When snow accumulation, snowfall events, or avalanche hazard level increases are detected, switch to the pre-trigger guard mode. Step 4: When the infrasound array is running in the pre-trigger guard mode, control the infrasound array to start full-channel continuous acquisition, and switch to the full-quantity recognition mode when the infrasound signal amplitude meets the pre-trigger condition. Step 5: When the infrasound array is running in the full recognition level mode, determine whether it is a valid avalanche event; if it is determined to be a valid avalanche event, switch to the graded early warning level mode. Step 6: When the infrasound array is operating in the graded warning level mode, it issues an avalanche warning signal of the corresponding level.

2. The multi-level dynamically triggered avalanche infrasound monitoring and early warning method according to claim 1, characterized in that, In step 2, the initial operating mode of the infrasound array is the deep dormancy mode. Based on the snow cover period determination, avalanche risk level and meteorological conditions, the deep dormancy mode, the pre-triggered duty mode, the full-volume identification mode and the graded early warning mode are triggered step by step, and the acquisition strategy and processing logic are adjusted progressively.

3. The multi-level dynamically triggered avalanche infrasound monitoring and early warning method according to claim 2, characterized in that, In step 4, when the infrasound array is running in the pre-trigger guard mode, the infrasound array is controlled to start full-channel continuous acquisition, and the original infrasound data for a preset duration is retained through a ring rolling buffer mechanism. At the same time, the dynamic adaptive trigger threshold is calculated in real time. When the infrasound signal amplitude exceeds the dynamic adaptive trigger threshold, the infrasound signal amplitude meets the pre-trigger condition, and the system switches to the full-volume recognition mode. When it is a non-snowy period or the avalanche hazard level is reduced, return to the deep hibernation mode.

4. The multi-level dynamically triggered avalanche infrasound monitoring and early warning method according to claim 3, characterized in that, In step 4, the dynamic adaptive trigger threshold It is calculated by multiplying the multidimensional correction coefficient and the base threshold, and the calculation formula is as follows: in, The site's basic noise threshold. This is the real-time environmental noise correction factor. This is a correction factor for avalanche risk. This is a seasonal correction factor; The circular rolling buffer mechanism is as follows: in the pre-triggered guard level mode, the most recent 60 seconds of full-channel raw infrasound data are continuously retained in the local edge buffer, and data exceeding the time limit is automatically rolled over; when the pre-triggered condition is met, the complete signal segment from 30 seconds before the trigger to real-time acquisition after the trigger is locked and extracted to ensure the integrity of the avalanche infrasound signal.

5. The multi-level dynamically triggered avalanche infrasound monitoring and early warning method according to claim 3, characterized in that, In step 5, when the infrasound array is running in the full recognition level mode, the complete infrasound signal segment in the circular rolling buffer is retrieved. Based on the synchronous timing signal segments of each channel of the infrasound array, the generalized cross-correlation-phase transformation method is used to perform array cross-correlation processing, detection clustering, and feature parameter calculation. The preset multi-threshold constraint conditions are used to determine whether it is a valid avalanche event. If it is determined to be a non-avalanche false alarm event, it returns to the pre-triggered guard level mode. If it is determined to be a valid avalanche event, it switches to the graded early warning level mode.

6. The multi-level dynamically triggered avalanche infrasound monitoring and early warning method according to claim 5, characterized in that, In step 5, the constraint rules for cluster detection are: the time interval between two adjacent infrasound detections is <10s, the difference in back azimuth angle is <10°, and the detection clusters generated by clustering satisfy: the number of effective infrasound detections is ≥15, and the total duration is ≥20s; The calculated characteristic parameters include apparent velocity change, apparent velocity change trend, apparent velocity-correlation coefficient ratio, and detection density; The multi-threshold constraint is specifically: the number of effective infrasound detections within the detection cluster. Duration: 20 seconds D <90s; Maximum apparent velocity v max <450m / s; average apparent velocity 300m / s< v m <425m / s; apparent velocity change -30m / s< <0m / s; apparent velocity variation trend ≤-1; average correlation coefficient >0.75; apparent velocity to correlation coefficient ratio <530m / s; detection density >0.85s -1 ; If a detection cluster simultaneously satisfies all of the above multi-threshold constraints, it is determined to be a valid avalanche event.

7. The multi-level dynamically triggered avalanche infrasound monitoring and early warning method according to claim 6, characterized in that, In step 6, when the infrasound array is operating in the graded early warning mode, it issues an avalanche early warning signal of the corresponding level based on the identified valid avalanche events. At the same time, based on the identification results of valid avalanche events and changes in avalanche risk level, it dynamically optimizes the monitoring strategy and the calculation parameters of the dynamic adaptive trigger threshold. After the early warning ends, it returns to the pre-trigger guard mode.

8. The multi-level dynamically triggered avalanche infrasound monitoring and early warning method according to claim 7, characterized in that, In step 6, an avalanche warning signal of the corresponding level is issued. The rule for graded warning is: if there are ≥1 valid avalanche events in a single day, a primary warning is issued, indicating that the snow layer in the area is showing signs of instability. If there are ≥2 valid avalanche events in a single day, a medium-level warning will be issued, indicating a significant increase in avalanche activity in the region; If there are ≥3 valid avalanche events in a single day, a high-level warning will be issued, indicating that the area has entered a high-risk period for avalanches. When dynamically optimizing the monitoring strategy, based on the signal characteristics and corresponding environmental parameters of the identified valid avalanche events and non-false avalanche events, the calculation rules of each correction coefficient of the dynamic adaptive trigger threshold and the parameter range of the multi-threshold constraint conditions are optimized through self-learning iteration, thereby reducing the false alarm rate and missed detection rate in the monitoring process.

9. The multi-level dynamically triggered avalanche infrasound monitoring and early warning method according to claim 8, characterized in that, In step 6, the self-learning iterative optimization method is as follows: based on historically accumulated positive samples (confirmed as valid avalanche events) and negative samples (confirmed as non-avalanche false alarm events), and the environmental parameters corresponding to the positive and negative samples, a batch or incremental learning method is adopted. Through logistic regression, decision tree, quantile analysis, or gradient descent algorithm, the calculation rules or value range of each correction coefficient in the dynamic adaptive trigger threshold, as well as the upper and lower limits of each parameter in the multi-threshold constraint conditions, are periodically adjusted to minimize the false alarm rate and the missed detection rate.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is run by the processor, it executes the steps of the multi-level dynamically triggered avalanche infrasound monitoring and early warning method as described in any one of claims 1 to 9.

Images