A data processing method and system for geological disaster early warning

Abstract

This invention provides a data processing method and system for geological disaster early warning, belonging to the field of data processing technology. The method includes: comparing environmental state sequences collected by multiple nodes with safety boundaries, and outputting a basic early warning prompt when a boundary is exceeded; extracting the initial node experiencing a sudden change, and calculating the stress wave and lag time based on its physical distance from the node to be tested and the geological stress transmission rate; controlling the node to be tested to start high-frequency sampling when entering the lag time to extract an encrypted displacement sequence; determining the main slip orientation based on the displacement vector of the initial node, and selecting vertically distributed reference nodes; performing differential processing on the encrypted displacement sequence using the reference displacement sequence collected by the reference nodes, and generating a risk status based on the sequence after filtering out common-mode interference. This invention acquires high-density deformation data and filters out environmental interference while ensuring low-power operation of the sensor network, thus improving the accuracy of early warning classification.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a data processing method and system for geological disaster early warning. Background Technology

[0002] Currently, geological disaster early warning applications typically require the deployment of numerous sensor nodes within field monitoring areas to collect data on stratum water content, slope inclination angle, and displacement deformation. Since field equipment often relies on independent batteries, nodes are frequently set to a fixed low-frequency dormant acquisition mode to ensure long-term network operation. When localized structural instability occurs deep within the Earth's surface, the stress propagation process exhibits specific spatial propagation delay characteristics. If surrounding nodes maintain their original low-frequency sampling rhythm, it becomes difficult to capture the instantaneous deformation evolution details caused by stress propagation in a timely manner. Conversely, activating global high-frequency continuous sampling accelerates overall equipment power consumption, resulting in a trade-off between the node's operational lifespan and the high-density data sampling requirements during sudden disasters.

[0003] Field sensors are also susceptible to interference from objective environmental factors such as wind resistance and micro-seismic waves during operation. These natural activities can synchronously trigger common background displacement fluctuations at adjacent nodes. These background fluctuations will superimpose with the objective displacement of geological bodies during actual data collection. In the system comprehensive assessment phase, if calculations are performed directly based on displacement sequences containing these associated waveforms, external environmental fluctuations will cause aliasing interference in the calculation of the absolute amount of macroscopic slippage. This will lead to a certain degree of calculation bias in the final disaster warning level assessment, affecting the objectivity of risk status assessment and the accuracy of data calculation. Summary of the Invention

[0004] The purpose of this invention is to provide a data processing method and system for geological disaster early warning, in order to solve the technical problems of existing geological disaster monitoring networks, which have difficulty in balancing the high-frequency sampling requirements of sudden deformation during stress transmission with the long-term low-power operation of nodes, and the fact that the collected displacement data is easily affected by common fluctuations in the natural environment in the field, thus leading to inaccurate risk status assessment.

[0005] In a first aspect, the present invention provides a data processing method for geological disaster early warning, comprising:

[0006] By comparing the environmental state sequence collected by multiple nodes deployed in the monitoring area with the safety boundary, a basic early warning prompt is output when it is determined that the environmental state sequence exceeds the safety boundary;

[0007] Extract the first node where the environmental state sequence undergoes a sudden change, and combine the spatial distribution topology to extract the physical distance between the first node and the adjacent nodes to be tested;

[0008] By combining the current geological stress transmission rate with the physical distance, the stress wave lag time from the initial node to the node to be tested is calculated;

[0009] The node under test is controlled to start a high-frequency sampling process when it enters the stress wave and hysteresis period to extract the encrypted displacement sequence.

[0010] The main slip direction is determined based on the displacement vector output by the first node, and reference nodes that are perpendicular to the main slip direction are selected from the plurality of nodes;

[0011] The reference displacement sequence collected by the reference node is obtained, and the encrypted displacement sequence is differentially processed using the reference displacement sequence. An updated risk state is generated based on the encrypted displacement sequence after filtering out common-mode interference.

[0012] Optionally, the comparison between the environmental state sequence collected by multiple nodes deployed within the monitoring area and the safety boundary, and the output of a basic early warning when it is determined that the environmental state sequence exceeds the safety boundary, includes:

[0013] The safety boundary is formed by extracting the pre-set upper limit of moisture content tolerance and the upper limit of tilt angle tolerance;

[0014] When the latest formation water content in the environmental state sequence is greater than the water content tolerance limit, or the latest slope inclination angle is greater than the inclination angle tolerance limit, the anomaly determination logic is triggered.

[0015] When the aforementioned anomaly detection logic occurs, a basic early warning prompt containing the identifier of the abnormal node and the mutation value is generated and pushed to the remote monitoring platform.

[0016] Optionally, after triggering the exception detection logic, the following may also be included:

[0017] Retrieve the frequency of false alarms recorded in the historical early warning database;

[0018] If the frequency of false alarms exceeds a set frequency threshold, the numerical range of the upper limit of moisture content tolerance and the upper limit of tilt angle tolerance will be expanded according to a set expansion ratio to dynamically update the safety boundary.

[0019] Optionally, before comparing the environmental state sequences collected by multiple nodes deployed within the monitoring area with the security boundary, the following steps are also included:

[0020] Drive all the nodes to continuously collect formation water content and slope inclination angle at a preset low-frequency dormancy cycle;

[0021] The water content of the formation and the slope inclination angle are processed using median filtering logic to filter out transient extreme noise.

[0022] The formation water content and slope inclination angle, which are from different time dimensions and have had transient extreme noise eliminated, are spliced ​​together in a time sequence to generate the environmental state sequence.

[0023] Optionally, the step of calculating the stress wave duration and hysteresis time from the initial node to the node to be tested, by combining the current geological stress transmission rate with the physical distance, includes:

[0024] Divide the physical distance by the geological stress conduction rate to obtain the foundation conduction time;

[0025] Extract the current rainfall level of the monitoring area, and perform duration compensation scaling on the basic conduction time based on the rainfall level to obtain the compensated conduction time;

[0026] Using the timestamp of the mutation at the first node as the starting point, and adding the compensated propagation time, the expected impact time is obtained.

[0027] Centered on the expected impact time, the stress impact hysteresis period is generated by extending the allowable fault tolerance span forward and backward respectively.

[0028] Optionally, controlling the node under test to initiate a high-frequency sampling process and extract an encrypted displacement sequence when entering the stress wave and hysteresis period includes:

[0029] Before the current time reaches the start time of the stress wave and hysteresis period, the node to be tested is controlled to maintain a low-power sleep mode.

[0030] When the current time enters the stress wave and hysteresis period, the node to be tested is controlled to increase the sampling frequency to a predetermined high frequency value, and the deformation displacement data is continuously collected and combined into the encrypted displacement sequence.

[0031] When the current time crosses the end of the stress wave and hysteresis period, the node to be tested is controlled to return to the low-power sleep mode.

[0032] Optionally, determining the main slip orientation based on the displacement vector output by the initial node, and selecting reference nodes that are perpendicular to the main slip orientation from among the plurality of nodes, includes:

[0033] Extract the displacement and offset values ​​of the first node in all axes in the three-dimensional coordinate system;

[0034] The displacement offset values ​​of all axes are vector synthesized to obtain the synthetic displacement vector of the first node, and the spatial ray pointed to by the synthetic displacement vector is defined as the main slip direction;

[0035] Calculate the spatial direction vector of the line connecting the first node and the candidate nodes;

[0036] Calculate the inner product of the spatial direction vector and the composite displacement vector, and select candidate nodes whose absolute values ​​of the corresponding inner product values ​​are within the zero tolerance range, and mark them as the reference nodes.

[0037] Optionally, after being marked as the reference node, the following is also included:

[0038] Continuously collect the displacement data fluctuation variance of all the reference nodes, and calculate the group fluctuation average of the displacement data fluctuation variance;

[0039] If the variance of displacement data fluctuation of a specific reference node exceeds the average value of the group fluctuation, it is determined that the associated reference node is affected by a local independent disturbance.

[0040] Associated reference nodes affected by local independent disturbances are removed from the common-mode reference pool, and the remaining reference nodes are retained to provide the reference displacement sequence.

[0041] Optionally, the step of acquiring the reference displacement sequence collected by the reference node, performing differential processing on the encrypted displacement sequence using the reference displacement sequence, and generating an updated risk state based on the encrypted displacement sequence after filtering out common-mode interference includes:

[0042] Calculate the spatial linear intervals between the node to be tested and all the reference nodes;

[0043] Using the reciprocal of the spatial linear interval as a weight, the reference displacement sequences output by all the reference nodes are weighted and fused to generate the estimated common-mode interference sequence of the location of the node to be examined;

[0044] Subtract the estimated common-mode interference sequence from the encrypted displacement sequence to obtain the net deformation displacement sequence;

[0045] Based on the cumulative change magnitude and acceleration mapping of the net deformation displacement sequence, the corresponding disaster alarm level is output as the updated risk status.

[0046] Secondly, the present invention provides a data processing system for geological disaster early warning, characterized in that it includes:

[0047] The basic monitoring module is configured to compare the environmental state sequence collected by multiple nodes deployed in the monitoring area with the safety boundary, and output a basic early warning prompt when it is determined that the environmental state sequence exceeds the safety boundary;

[0048] The spacing extraction module is configured to extract the first node where the environmental state sequence undergoes a sudden change, and to extract the physical spacing between the first node and the adjacent nodes to be examined by combining the spatial distribution topology.

[0049] The time window calculation module is configured to combine the current geological stress transmission rate with the physical distance to calculate the stress wave lag time from the first node to the node to be tested;

[0050] A high-frequency scheduling module is configured to control the node under test to start a high-frequency sampling process when it enters the stress wave and hysteresis period, and extract the encrypted displacement sequence.

[0051] The benchmark screening module is configured to determine the main slip direction based on the displacement vector output by the first node, and select reference nodes that are perpendicular to the main slip direction from the plurality of nodes;

[0052] The differential risk assessment module is configured to acquire the reference displacement sequence collected by the reference node, perform differential processing on the encrypted displacement sequence using the reference displacement sequence, and generate an updated risk status based on the encrypted displacement sequence after filtering out common-mode interference.

[0053] The present invention has achieved the following beneficial effects:

[0054] This invention extracts the initial node where a sudden change in environmental state occurs, and calculates the stress wave and lag time to adjacent nodes under test by combining physical distance and geological stress transmission rate. This allows the nodes under test to initiate high-frequency sampling and acquire encrypted displacement sequences within the expected delay interval. This data processing logic selectively records high-frequency data of transient deformation while maintaining the monitoring equipment's normal low-power operation, balancing the objective requirements of long-term sensor network endurance and detailed data capture of sudden events. Simultaneously, this invention determines the main slip orientation based on the synthetic displacement vector of the initial node and selects reference nodes with vertical spatial distribution within the monitoring area. The extracted reference displacement sequences are then used to perform differential filtering on the high-frequency acquired encrypted displacement sequences. This computational method suppresses common-mode interference components from the combined superposition of wind resistance and microseismic events in the field, separating out net deformation parameters that objectively reflect the actual relative sliding of the strata. Disaster risk assessment based on these net deformation parameters reduces computational errors caused by external background fluctuations, improving the objectivity and accuracy of geological disaster early warning results and system risk status output.

[0055] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0056] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0057] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0058] Figure 1 This is a main flowchart of a data processing method for geological disaster early warning provided in an embodiment of the present invention;

[0059] Figure 2 A flowchart illustrating a method for dynamically updating security boundaries provided in an embodiment of the present invention;

[0060] Figure 3 A flowchart illustrating a method for measuring stress waves and hysteresis time provided in an embodiment of the present invention;

[0061] Figure 4 A flowchart illustrating a method for filtering reference nodes and removing local independent disturbances, provided in an embodiment of the present invention;

[0062] Figure 5 A flowchart illustrating a method for differential risk assessment and risk status generation provided in an embodiment of the present invention;

[0063] Figure 6 This is a structural block diagram of a data processing system for geological disaster early warning provided in an embodiment of the present invention. Detailed Implementation

[0064] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0065] It is understood that this embodiment provides a data processing method for geological disaster early warning. This method is applied to a computing device configured with concurrent data processing capabilities, including a data center monitoring server, an edge computing gateway, or a distributed computing system comprised of these. The computing device establishes a data communication link with a sensor network deployed within a preset monitoring area via a communication network. The sensor network includes multiple nodes distributed within the monitoring area. Each node is internally configured with a processor, memory, communication components, and environmental condition sensing components. The environmental condition sensing components include a moisture content sensor, a tilt sensor, and a three-dimensional displacement sensor.

[0066] like Figure 1As shown, in step S10, before comparing the environmental state sequence collected by multiple nodes deployed in the monitoring area with the safety boundary, all the nodes are driven to continuously collect the formation water content and slope inclination angle with a preset low-frequency sleep cycle; the formation water content and slope inclination angle are processed using median filtering logic to filter out transient extreme value noise; the formation water content and slope inclination angle that have eliminated transient extreme value noise and belong to different time dimensions are spliced ​​together in time to generate the environmental state sequence.

[0067] Specifically, the computing device sends configuration commands to each node via the downlink to set the preset low-frequency sleep cycle parameters for the nodes. During sleep, the nodes cut off the power supply to the environmental state sensing components to reduce the overall system power consumption. When the timer count inside the node reaches the preset low-frequency sleep cycle, a wake-up signal is triggered. In response to this wake-up signal, the node processor connects the power supply circuits to the water content sensor and the tilt sensor. The node performs multi-cycle continuous sampling of the analog signals output by each sensor through the analog-to-digital conversion channel to obtain discrete values ​​reflecting the physical state, and maps them to the formation water content and slope tilt angle at the current trigger time, respectively.

[0068] Furthermore, the node processor establishes independent one-dimensional sliding window data structures in memory for formation water content and slope inclination angle, with the depth dimension of each sliding window configured to have an odd number of recording units. Each time a new discrete value is acquired, it is stored in the corresponding sliding window, and the oldest historical data within the window is removed. The node processor sorts all values ​​within the current sliding window according to their magnitude, extracts the value at the median position after sorting, and uses this median filtering logic to eliminate transient extreme value noise caused by random physical collisions or electromagnetic interference.

[0069] The preset low-frequency sleep cycle is set based on the balance point between the conventional consolidation creep rate of natural strata and the chemical power consumption characteristics of field node batteries, and is fixed in the internal timer to a configuration of 4 to 8 hours. Meanwhile, the depth of the one-dimensional sliding window is limited to 5 or 7 recording units. If the window depth is reduced to less than 3 units, the algorithm will lack sufficient statistical samples, making it difficult to effectively filter out short-term wind resistance oscillations under extreme convective weather; if expanded to more than 9 units, it is easy to introduce arithmetic phase hysteresis spanning tens of hours in the data cleaning process, causing a delay in the system's response to real landslide precursors. Through the dual limitations of the sleep clock and the storage stack depth, the link of non-geological extreme values ​​to the subsequent safety boundary comparison logic is effectively blocked.

[0070] While outputting valid data, the node processor reads the current absolute timestamp, binds the absolute timestamp with the corresponding valid data, and encapsulates it to generate a structured data frame. The node appends and splices multiple data frames generated continuously across multiple sleep cycles in chronological order to generate the environmental state sequence, and reports it to the computing device through the communication component.

[0071] Step S20: Compare the environmental state sequence collected by multiple nodes deployed in the monitoring area with the safety boundary, and output a basic early warning prompt when it is determined that the environmental state sequence exceeds the safety boundary.

[0072] The safety boundary is constructed by extracting pre-set upper limits for moisture content tolerance and tilt angle tolerance. The upper limit for moisture content tolerance sets the maximum allowable moisture parameter of a geological unit before it experiences a complete loss of effective shear strength; the upper limit for tilt angle tolerance sets the maximum allowable deflection angle of the surface geological structure before macroscopic sliding shear fracture occurs. The computing device receives the environmental state sequence reported by each node, extracts the data frame with the timestamp closest to the current system time, and separates the latest formation moisture content and the latest slope tilt angle carried by that data frame.

[0073] The processor of the computing device compares the latest formation moisture content with the upper limit of moisture content tolerance, and compares the latest slope inclination angle with the upper limit of slope angle tolerance. Specifically, the initial value of the upper limit of moisture content tolerance is based on the liquid limit moisture content of the soil in the monitoring area. Determined, value is The calibration logic is as follows: when the node exploration records entered by the system show that there are historical tensile cracks on the surrounding surface, the value is set to... If the surface structure is intact and without cracks, then the value is taken as follows: The initial value of the upper limit of the tilt angle tolerance is based on the designed stable slope angle of the slope where the monitoring point is located. Determined, value is ,in This is a preset critical offset, with a value range of [value range missing]. .

[0074] When the latest formation water content in the environmental state sequence exceeds the upper limit of water content tolerance, or the latest slope inclination angle exceeds the upper limit of inclination angle tolerance, an anomaly detection logic is triggered. Upon the occurrence of this anomaly detection logic, a basic early warning message containing the anomaly node identifier and the mutation value is generated and pushed to the remote monitoring platform. The computing device obtains the hardware address or network interface identifier of the node that triggered the anomaly detection logic and converts it into the anomaly node identifier. Simultaneously, the actual value of the physical parameter that triggered the limit violation is extracted as the mutation value. The computing device encapsulates the anomaly node identifier, the mutation value, and the system absolute timestamp at the time of triggering the detection into a data packet of a specific format, generates the basic early warning message, and sends it to the remote monitoring platform for storage and logging via the communication network interface.

[0075] like Figure 2 As shown in step S30, after triggering the anomaly judgment logic, the frequency of false alarms recorded in the historical early warning database is retrieved; if the frequency of false alarms is greater than the set frequency threshold, the numerical range of the upper limit of moisture content tolerance and the upper limit of tilt angle tolerance are expanded according to the set expansion ratio to dynamically update the safety boundary.

[0076] Specifically, the computing device uses the identifier of the abnormal node as a search keyword to access the historical early warning database and extract historical early warning records associated with the abnormal node within a preset time span. The computing device iterates through the extracted record set, filtering out records marked as having natural baseline drift without actual sliding displacement, and counts the number of such records as the false alarm frequency. The computing device compares the false alarm frequency with a set frequency threshold. When the false alarm frequency exceeds the set frequency threshold, it determines that the current application's security boundary requires parameter updates. The computing device retrieves the corresponding set expansion ratio from the configuration matrix. A multiplication operation is performed, multiplying the current moisture content tolerance upper limit parameter by one and adding a coefficient generated by the set expansion ratio to calculate the updated moisture content tolerance upper limit. Simultaneously, the same multiplication operation is performed on the current tilt angle tolerance upper limit to obtain the updated tilt angle tolerance upper limit. The computing device overwrites the calculated new upper limit value to the parameter configuration area in system memory, completing the dynamic adaptive update processing of the security boundary.

[0077] The system employs quantitative calibration criteria for mapping false alarm frequency to a set amplification ratio. A frequency threshold is set and evaluated using a time-sliding window, with a normalized configuration of three cumulative drift alarms without actual displacement within a single monthly cycle. The amplification ratio is set based on dual constraints of sensor hardware drift tolerance and seasonal climate extremes, and is fixed at 5%. Limiting the iteration step size of this amplification ratio effectively prevents the upper limit value from being excessively amplified due to frequent triggers by severe weather, thus promoting false alarm convergence while maintaining early warning sensitivity to minor slip precursors.

[0078] Step S40: Extract the first node where the environmental state sequence undergoes a sudden change, and extract the physical distance between the first node and the adjacent nodes to be tested by combining the spatial distribution topology.

[0079] When the computing device identifies multiple nodes reporting data exceeding the safety boundary within a set, similar time window in the receive buffer queue, it extracts the set of node identifiers that triggered the anomaly detection logic and reads the absolute timestamps recorded in each data frame. Through comparison and data sorting operations on the timeline, it selects the single node with the smallest time coordinate value and marks it as the initial node. The physical coordinates of the initial node map to the starting geometric origin where geological deformation initially broke the unstable mechanical equilibrium.

[0080] The computing device loads spatial distribution topology data reflecting the three-dimensional coordinate positions of nodes within the monitoring network. Using the three-dimensional coordinates of the initial node as the search reference point, other online nodes within the preset influence envelope radius that have not yet triggered abnormal boundary conditions are extracted and uniformly classified into a set of adjacent nodes to be examined. The set of similar time windows for determining multi-node bursts is configured as a physical time width of 1.5 to 2.0 seconds. This span covers the communication queuing delay of the underlying low-power wide area network multiplexing and the stress wave travel time difference of the near-field grid. Data frames that exceed this time limit are dynamically set as mutually independent geological rupture events. In addition, the preset influence envelope radius range is controlled by the exponential decay physical equation of Rayleigh surface waves in heterogeneous soil layers. The system extracts the nominal noise floor lower limit of the three-dimensional displacement sensor as the ultimate cutoff condition, substitutes it into the decay equation, and solves inversely to obtain the spatial geometric boundary of the oscillation energy dissipation to an indistinguishable state. Under typical loess or strongly weathered mudstone conditions, this boundary is solved and mapped to a fixed spherical radius of 80 to 120 meters. It should be noted that the 80-120 meter range mentioned above is only a statistical distribution engineering reference domain for the algorithm under typical geological conditions; during actual scheduling and execution, the range will be dynamically calculated based on the aforementioned attenuation formula for each iteration. The value is set as an absolute preset to affect the range of the envelope radius, without using a fixed truncation.

[0081] For each of the adjacent nodes to be tested in the set, the processor of the computing device retrieves the three-dimensional position vector of the initial node and the three-dimensional position vector of the adjacent node to be tested. The difference between the two position vectors in the three orthogonal coordinate components is calculated, each difference is squared, and then summed. The sum is then subjected to a square root operation. Through the above geometric calculations, the absolute scalar length of the three-dimensional spatial line from the initial node to the adjacent node to be tested is obtained and configured as the physical distance for data storage.

[0082] The preset influence envelope radius range is solved in reverse by the following attenuation equation: ;in Distance from the starting node Amplitude at that point The initial amplitude of the fluctuation at the initial node is extracted as follows: the synthetic magnitude of the three-dimensional displacement vector of the initial node at the moment the anomaly judgment logic is triggered is obtained, and the transient peak value of this magnitude is defined as... , The formation absorption attenuation coefficient (in this embodiment, the value is determined based on the formation hardness). Its dimensional unit is The specific mapping criterion is: based on the Protodyakonov rock firmness coefficient. When addressing is retrieved, (extremely soft soil and rock) The value is 0.08, when (Soft rock) time The value is 0.05, when (For harder rock) The value is 0.02; let The distance is equal to the nominal noise floor lower limit of the three-dimensional displacement sensor. That is, the envelope radius.

[0083] like Figure 3 As shown, in step S50, the stress wave and lag time period from the first node to the node to be tested is calculated by combining the current geological stress transmission rate with the physical distance.

[0084] Dividing the physical distance by the geological stress conduction rate yields the foundation conduction time. The computing device reads the geological stress conduction rate constant of the current monitoring area from the environmental parameter configuration file. The processor executes the division instruction, using the physical distance as the dividend and the geological stress conduction rate as the divisor. The quotient output by this operation represents the standard time required for the stress wave to cross the physical distance under a homogeneous static model, i.e., the foundation conduction time.

[0085] The current rainfall level in the monitored area is extracted, and the basic conduction time is scaled and compensated based on the rainfall level to obtain the compensated conduction time. The computing device obtains the cumulative rainfall depth parameter of the target area through the meteorological data interface, determines the numerical boundary interval it falls into, and outputs the corresponding rainfall level. The data mapping relationship table of rainfall level and duration compensation multiplier stored in the system memory is queried to extract the corresponding duration compensation multiplier. The processor performs a multiplication operation to multiply the basic conduction time by the duration compensation multiplier to calculate the compensated conduction time that incorporates the effect of water infiltration wave velocity hysteresis.

[0086] The system's built-in geological stress transmission rate constant is typically set within the typical soil shear wave velocity range of 150 m / s to 300 m / s. Rainfall infiltration leads to an increase in pore water pressure in the soil and rock, weakening the effective stress in the soil and thus causing physical hysteresis in stress wave transmission. The pre-stored data mapping table between rainfall levels and duration compensation multipliers is constructed based on this nonlinear damping law. The processor divides the values ​​according to the downlink meteorological data: when the 24-hour cumulative rainfall depth is less than or equal to 10 mm, surface water has not formed effective infiltration, and the duration compensation multiplier is directly set to 1.0; when the cumulative rainfall depth is between 10 mm and 50 mm, the shallow soil pores are gradually filled with water, the wave velocity damping increases, and the duration compensation multiplier is interpolated to 1.15 to 1.25; when the cumulative rainfall depth exceeds 50 mm, the infiltration of deep gravity water significantly hinders the stress wave velocity, and the duration compensation multiplier is correspondingly calibrated to 1.35 to 1.50. By looking up the multiplier in the table and scaling it, the rainfall damping is quantified into a time-compensated increment.

[0087] The specific multiplier values ​​within the aforementioned defined intervals are all calculated using linear interpolation. For example, when the cumulative rainfall depth... When the time is in the range of 10 mm to 50 mm, the duration compensation multiplier The calculation formula is When the cumulative rainfall exceeds 50 mm, the linear mapping is applied similarly, or the value is capped at 1.50. Specifically, when the cumulative rainfall depth... When the thickness is in the range of 50 mm to 100 mm, the linear mapping formula is as follows: When the cumulative rainfall depth For values ​​greater than 100 mm, the multiplier is fixed at 1.50. It should be noted that the numerical constants (10, 50, 100) in the above difference formula are all in millimeters (mm).

[0088] Starting from the timestamp of the abrupt change at the initial node, the expected arrival time is obtained by adding the compensated propagation time. Extending the expected arrival time forward and backward by a reserved tolerance span generates the stress wave hysteresis period. The computing device extracts the absolute timestamp of the environmental parameter exceeding the limit at the initial node as the starting data value. Adding the starting point to the compensated propagation time, the expected arrival time representing the estimated arrival of the stress wave at the node to be verified is calculated. Based on the numerical baseline of the expected arrival time, a fixed natural time constant variable, i.e., the reserved tolerance span, is subtracted and added to define the start and end boundary markers of the time window. The reserved tolerance span is used to enclose the waveform dispersion time difference caused by the heterogeneity of the geological medium, as well as the hardware wake-up delay of the sensor switching from extremely low-power sleep mode to full-power high-frequency sampling. The processor sets the reserved tolerance span as a dynamically calculated value, and the calculation equation is configured as: an adaptive time span for the compensated propagation time, superimposed with a constant circuit wake-up hysteresis parameter of 0.5 seconds. The adaptive time span is determined based on the rainfall level: 10% for rainfall less than or equal to 10 mm, 12.5% ​​for 10 to 50 mm, and 15% for rainfall greater than 50 mm. Assuming the calculated post-compensation propagation time is 6.0 seconds, the processor extracts 10% of this (0.6 seconds), adds 0.5 seconds, and obtains a reserved fault tolerance span of 1.1 seconds. The processor extends this 1.1-second interval on the absolute time axis before and after the expected wave arrival time, generating a bidirectional extended sampling window. This dynamic widening mechanism ensures that the node under test can accurately capture the stress wave leading edge and the main oscillation tail, avoiding the situation where the deformation peak data is not collected within the wake-up blind zone due to an excessively narrow time window setting. The continuous time interval completely defined by these two boundary markers on the absolute time series is configured as the stress wave hysteresis period.

[0089] Step S60: Control the node to be tested to start a high-frequency sampling process when it enters the stress wave and hysteresis period to extract the encrypted displacement sequence.

[0090] Before the current time reaches the start time of the stress wave and hysteresis period, the node under test is controlled to maintain a low-power sleep mode. The front-end control program of the node under test periodically compares the local system clock counter with the issued start time parameter. When the clock value is before the start time, the node cuts off the power supply channel of the three-dimensional displacement sensor and maintains the low-power sleep mode.

[0091] When the current time enters the stress wave and hysteresis period, the node under test is controlled to increase the sampling frequency to a predetermined high-frequency value and continuously collect deformation displacement data to form the encrypted displacement sequence. When the clock value reaches the starting time marker, the node processor controls the power supply of the three-dimensional displacement sensor to be turned on, overwrites the local control register stack, and modifies the sampling period parameter of the analog-to-digital conversion channel to the predetermined high-frequency value. The predetermined high-frequency value is limited by both the acoustic radio frequency band of geological micro-fractures and the Nyquist-Shannon sampling theorem. The dominant frequency of transient stress waves excited by the penetration of micro-fractures in shallow strata is usually in the range of 5 Hz to 20 Hz. In order to reconstruct the complete time-domain envelope of deformation evolution without aliasing distortion, the node processor overwrites the control register and sets the lower limit of the predetermined high-frequency value to more than four times the highest characteristic frequency, that is, the hard interrupt executes an analog-to-digital conversion cycle of 80 Hz to 100 Hz. This asymmetric clock frequency adjustment, with a significantly different relatively long sleep cycle, allows for the continuous acquisition of at least four discrete displacement vectors within each high-frequency oscillation cycle of the stress wave. Driven by this predetermined high-frequency value, the three-dimensional displacement sensor continuously acquires digital signals of the three-dimensional physical displacement changes in the space where the node is located. The node sequentially appends the continuously output deformation displacement data slices into a data buffer structure according to the absolute sampling time order, combining them to generate the encrypted displacement sequence of the structured array entity.

[0092] When the current time crosses the end time of the stress wave and hysteresis period, the node under test is controlled to return to the low-power sleep mode. When the local system clock value is greater than the end time boundary, the node logic confirms that the stress wave disturbance has passed through the node area, the node processor cuts off the power supply to the three-dimensional displacement sensor, resets the sampling frequency parameters, and controls the node under test to exit the continuous sampling process and return to the low-power sleep mode.

[0093] like Figure 4 As shown, in step S70, the main sliding orientation is determined based on the displacement vector output by the first node, and reference nodes that are perpendicular to the main sliding orientation are selected from the plurality of nodes.

[0094] The displacement offset values ​​of the initial node in all axes in the three-dimensional coordinate system are extracted; the displacement offset values ​​in all axes are vector synthesized to obtain the synthetic displacement vector of the initial node, and the spatial ray pointed to by the synthetic displacement vector is defined as the main slip orientation. The computing device analyzes the deformation data output by the three-dimensional displacement sensor of the initial node in three orthogonal physical axes to obtain the displacement offset values. The processor performs spatial vector superposition matrix operation on the displacement offset values ​​in the above three dimensions to calculate the synthetic displacement vector containing spatial direction characteristics, and fixes the spatial ray containing this vector as the main slip orientation representing the central sliding thrust trajectory.

[0095] The spatial direction vector connecting the initial node and the candidate nodes is calculated. The inner product of the spatial direction vector and the composite displacement vector is calculated, and candidate nodes whose absolute values ​​of the inner product are within the zero tolerance range are selected and marked as reference nodes. The computing device identifies the remaining nodes in the monitoring network that are normally online and have not issued any warnings as candidate nodes, extracts the three-dimensional coordinate parameters of the candidate nodes and the initial node, and performs a subtraction operation on the coordinates of the two nodes to generate a spatial direction vector. The processor extracts the corresponding coordinate components of the spatial direction vector and the composite displacement vector, performs a product operation on the coaxial components, and sums them to calculate the scalar value of the inner product. The computing device uses a logic comparator to compare whether the absolute value of the inner product falls within the preset zero tolerance range. Due to the rugged terrain and assembly tilt tolerance during node pre-installation, if the candidate nodes are required to be strictly orthogonal to the main sliding orientation in three-dimensional space, the number of usable reference nodes is significantly reduced. The processor establishes a dynamic judgment boundary based on the cosine theorem of the angle between spatial vectors in analytical geometry. The system fixes the lower limit of the zero-value tolerance range to zero, and the upper limit threshold is calculated in real time as the product of the Euclidean magnitude of the initial node's composite displacement vector and the Euclidean magnitude of the candidate node's spatial direction vector, multiplied by a constant 0.087. This constant corresponds to the approximate absolute value of the cosine function cos(85°), which essentially relaxes the geometric deflection tolerance for orthogonality determination to 5 degrees. As long as the spatial angle between the candidate node's connection line and the principal sliding plane remains within the range of 85 to 95 degrees, the corresponding inner product scalar will fall within this calculation upper limit. Since the inner product value approaching zero indicates that the two spatial vectors have an orthogonal geometric angle relationship, the computing device includes candidate nodes that meet the tolerance conditions into the common-mode reference pool and marks them as the reference nodes.

[0096] Step S80, after marking the reference node, further includes: continuously collecting the displacement data fluctuation variance of all reference nodes and calculating the group fluctuation average of the displacement data fluctuation variance; if the displacement data fluctuation variance of a specific reference node exceeds the group fluctuation average, it is determined that the associated reference node is affected by a local independent disturbance; the associated reference node affected by the local independent disturbance is removed from the common mode reference pool, and the remaining reference nodes are retained to provide the reference displacement sequence.

[0097] The computing device synchronously acquires continuous baseline displacement data sequences of each reference node in the common-mode reference pool within a set time interval. This set time interval is fixed by using a time sliding window to extract historical data from 300 to 600 seconds prior to the current moment, ensuring coverage of the complete natural environmental fluctuation cycle. The statistical algorithm module calculates the expected mean of the data sequence for each individual node, and performs a squared cumulative calculation based on the floating-point differences of discrete data points from the central mean to obtain the displacement data fluctuation variance, which characterizes the discrete characteristics of the data distribution. The displacement data fluctuation variance values ​​of all reference nodes are summed and divided by the total number of reference nodes to calculate the group fluctuation average, which measures the overall normal background fluctuation of the region.

[0098] The displacement data fluctuation variance of each reference node is numerically compared with the group fluctuation average and the set tolerance constant. When it is determined that the displacement data fluctuation variance of a specific reference node is greater than the preset upper limit of the group fluctuation average, it is determined that the surrounding area of ​​the associated reference node has been subjected to occasional physical load interference, i.e., affected by local independent disturbance. The preset upper limit for isolation determination here does not use static empirical values, but is based on a statistical architecture that jointly solves the dynamic environment basis and hardware noise floor. The processor fixes the nominal sensor static zero-bias drift variance into the set tolerance constant, configured as 0.02 square millimeters. The computing channel extracts the group fluctuation average, which represents the common-mode interference level of the macro environment, and multiplies it by a fixed confidence amplification multiplier. The system initially defaults to the confidence amplification multiplier being 2.5. When the computing device detects that the average wind speed in the monitoring area exceeds level 6 through the meteorological interface, the multiplier is dynamically switched and locked to 3.0, and then summed with the above tolerance constant to generate the preset upper limit for the current solution.

[0099] Background noise generated by natural wind loads or micro-earthquakes exhibits stationary continuity within the local grid space. When a specific measuring rod experiences a physical impact, its variance distortion increases significantly and exceeds this Gaussian confidence envelope. Relying on this joint baseline control, the system can automatically relax the population tolerance in extreme weather conditions, while effectively eliminating interference from single-point mechanical impacts on the common-mode reference sequence. The computing device performs an object removal operation, deleting the identifier pointers of the affected associated reference nodes from the common-mode reference pool record, and the reference displacement sequence is provided by the remaining reference nodes that conform to the stationary background characteristics.

[0100] like Figure 5 As shown, in step S90, the reference displacement sequence collected by the reference node is obtained, the encrypted displacement sequence is differentially processed using the reference displacement sequence, and an updated risk state is generated based on the encrypted displacement sequence after filtering out common-mode interference.

[0101] The spatial linear interval between the node to be tested and all the reference nodes is calculated. Using the reciprocal of the spatial linear interval as a weight, the reference displacement sequences output by all the reference nodes are weighted and fused to generate an estimated common-mode interference sequence for the location of the node to be tested. The computing device reads the 3D coordinates of the node to be tested and the 3D coordinates of each reference node, and calculates the absolute spatial linear length constant between each pair of measurement points, i.e., the spatial linear interval. The processor performs a reciprocal operation on each spatial linear interval and performs normalization function processing to calculate the weight coefficients corresponding to each reference node sequence. The underlying logic of the normalization function processing strictly follows the inverse distance weighted attenuation criterion. The processor extracts the reciprocal of the spatial linear interval of a single reference node as an independent weight factor, and simultaneously accumulates the reciprocals of the spatial linear intervals of all available reference nodes in the common-mode reference pool within the current calculation cycle to construct a global normalization denominator. The independent weight factor of a single node is divided by this global normalization denominator to output the corresponding weight coefficient. This arithmetic division operation normalizes the sum of the weight coefficients of all reference nodes in the global domain to 1. This operation objectively fits the mechanical laws governing the attenuation of environmental wind resistance and surface microseismic energy with physical distance in three-dimensional space. The closer the physical distance is to the reference point of the node under test, the higher the weighted fusion ratio is when reconstructing the estimated common-mode interference sequence. Under the condition of strictly aligned timestamp indices, the values ​​of each reference displacement sequence under the same timestamp slice are multiplied with their respective weight coefficients, and the product results are submitted to the bus accumulator for summation. The aforementioned weighted fusion data array, continuously generated on the time axis, fits and reconstructs the estimated common-mode interference sequence used to characterize the background fluctuations of the geographical location of the node under test.

[0102] The net deformation displacement sequence is obtained by subtracting the estimated common-mode interference sequence from the encrypted displacement sequence. The processor's processing channel extracts the encrypted displacement sequence and the estimated common-mode interference sequence, which are synchronized with the time parameters, and executes a subtraction operation instruction for data at the same index position. This subtraction operation cancels out the common environmental displacement influence components, including macroscopic meteorological wind resistance fluctuations and microseismic background, retains the deformation mechanical displacement variables that objectively characterize the plastic relative sliding of the geological structure, and outputs the net deformation displacement sequence.

[0103] Based on the disaster alarm level corresponding to the mapping between the cumulative change amplitude and acceleration of the net deformation displacement sequence, the updated risk status is output. The processor performs mathematical integral superposition calculation on the floating-point values ​​of the net deformation displacement sequence within the discrete time interval to calculate the absolute displacement scalar of the macroscopic slip of the strata, i.e., the cumulative change amplitude. Simultaneously, the processor solves the data deformation trend rate through the first-order and second-order difference partial derivative operation interface to extract the constant index of acceleration. Before solving for the acceleration, the processor first performs low-pass filtering downsampling processing on the net deformation displacement sequence. Specifically, the moving average algorithm is used to convert and aggregate the transient deformation data collected at predetermined high frequencies into a macroscopic displacement feature sequence with an hourly time step, and then substitutes it into the discrete second-order difference formula to calculate the macroscopic acceleration scalar with millimeters per square day. The sliding window size of the moving average algorithm is configured to be equivalent to the length of 1 hour of this time step, that is, the arithmetic mean of all high-frequency discrete sampling points within each 1-hour slice interval is used as the macroscopic feature benchmark point for that hour.

[0104] Specifically, the changing acceleration is calculated using the discrete second-order difference formula: ;in For the present acceleration at any moment These represent the displacement values ​​of the current and the two previous sampling points in the net deformation displacement sequence, respectively. This is the time step for the encryption sampling period. A multi-dimensional state level threshold constraint matrix is ​​pre-configured in the system memory.

[0105] The processor uses the calculated cumulative change amplitude and acceleration as input parameters, performs conditional size logic matching and addressing within the multidimensional state level threshold constraint matrix, and extracts the disaster alarm level for the corresponding interval. The multidimensional state level threshold constraint matrix constructs a decision logic tree that links displacement and acceleration as dual parameters, aiming to identify the complete rheological stages of a geological body from elastic deformation, plastic yielding to shear failure. The matrix has built-in discretization addressing mapping rules: When the cumulative change amplitude is less than 5 mm and the change acceleration is in the steady-state fluctuation zone of -0.5 mm / s² to 0.5 mm / s², the system determines that the weak displacement of the strata originates from natural consolidation settlement or sensor temperature drift, and the addressing output is a normal inspection status; when the cumulative change amplitude reaches the creep range of 5 mm to 15 mm, or when the cumulative change amplitude does not exceed the limit but the change acceleration suddenly rises to 0.5 mm / s² to 2.0 mm / s², it indicates that the shear resistance inside the soil begins to decay and is accompanied by microscopic slow displacement, and the addressing is matched with a yellow alert status; when the cumulative change amplitude reaches the 15 mm threshold and the change acceleration increases synchronously to more than 2.0 mm / s², the significant increase in the first-order partial derivative of deformation dynamically confirms that the landslide entity has cut off the sliding surface and entered the accelerated fracturing period, and the system directly skips the low-order condition judgment, triggers and outputs a red critical status.

[0106] Furthermore, for boundary conditions beyond the aforementioned basic judgments, the multi-dimensional state level threshold constraint matrix incorporates this fallback logic mapping: when the cumulative change amplitude reaches 15 mm or more, but the change acceleration is less than or equal to 2.0 mm per square day, the stratum is determined to have entered a period of continuous slip and creep, and an orange alert status is output; when the cumulative change amplitude is less than 5 mm, but the change acceleration is greater than 2.0 mm per square day, the system is determined to have experienced a sudden change due to strong surface disturbance, and a yellow attention status is output and a data verification command is triggered; when the change acceleration is less than -0.5 mm per square day, the stratum is determined to be in a stage of local stress release or rebound unloading, and the routine inspection status is maintained.

[0107] The computing device encapsulates the alarm level into the updated risk status message, records it on disk, and outputs it to the external network interface.

[0108] It is understood that this application provides a data processing system for geological disaster early warning. This data processing system, as a software or hardware-software collaborative entity architecture for implementing data processing, scheduling, and control, can be deployed within a monitoring server or edge computing gateway device. Figure 6 As shown, the data processing system includes a basic monitoring module, a spacing extraction module, a time window calculation module, a high-frequency scheduling module, a benchmark screening module, and a differential risk assessment module.

[0109] The basic monitoring module is configured to compare environmental state sequences collected by multiple nodes deployed within the monitoring area with safety boundaries. When the environmental state sequence exceeds the safety boundary, a basic early warning is output. The basic monitoring module's parsing program decomposes the data packets uploaded by the nodes, extracting time-series parameters of formation water content and slope inclination angle. It retrieves the configured upper limits for water content and slope inclination angle from the parameter table, performing numerical subtraction and sign bit detection. If the parameter value is determined to be greater than the corresponding safety boundary scalar, the basic monitoring module assembles a basic early warning data packet containing the node's underlying hardware identifier and the exceeded value, and distributes it to the event queue. Simultaneously, based on the false alarm frequency variable recorded in the polling log, the module calculates and updates the safety boundary configuration in the cache according to a set expansion ratio constant.

[0110] The spacing extraction module is configured to extract the first node where a sudden change occurs in the environmental state sequence, and extract the physical spacing between the first node and adjacent nodes to be tested based on the spatial distribution topology. The spacing extraction module sorts the timestamp fields of concurrent abnormal alarms in ascending order along the one-dimensional coordinate axis, locking the first node with the first minimum time value. It extracts the three-dimensional mesh coordinate parameters of the first node and adjacent nodes to be tested covered by the envelope range, and obtains the scalar physical spacing between node pairs by performing the sum of squares and square root operations of the differences between the orthogonal vector components.

[0111] The time window calculation module is configured to combine the current geological stress conduction rate with the physical distance to calculate the stress wave lag time from the initial node to the node to be tested. The time window calculation module calls a division operation to divide the physical distance by the geological stress conduction rate constant to calculate the basic conduction time. Based on the obtained rainfall level variable, it addresses and matches the corresponding duration compensation scaling multiplier to calculate the compensated conduction time. By superimposing the abrupt change time start scalar of the initial node with the compensated conduction time, it generates the expected wave time parameter, and extends the front and rear boundaries along the time axis using a reserved tolerance span constant to generate a stress wave lag time parameter package.

[0112] A high-frequency scheduling module is configured to control the node under test to initiate a high-frequency sampling process when entering the stress wave and hysteresis period, and to extract an encrypted displacement sequence. The high-frequency scheduling module's running state machine continuously compares the hardware system timestamp with the time boundary of the measurement interval. When the starting boundary is reached, a control frame signaling is constructed to overwrite the register of the node under test, and a sampling cycle of a predetermined high-frequency value is configured to obtain the deformation data stream and reassemble it into an encrypted displacement sequence; when the ending boundary is crossed, a zeroing parameter is issued to restore the sleep mode.

[0113] The benchmark selection module is configured to determine the main slip orientation based on the displacement vector output by the initial node, and select reference nodes that are perpendicular to the main slip orientation from among the multiple nodes. The module performs vector synthesis analysis on the three physical displacement axial offsets of the initial node to establish the main slip orientation vector, and calculates the spatial direction vectors of the lines connecting the remaining candidate nodes to the initial node. It calculates the inner product scalar of the two through processor hardware-level multiplication, and filters out a set of orthogonal candidate nodes whose absolute values ​​fall within the zero tolerance range. Subsequently, it calls statistical functions to calculate the variance of displacement data fluctuation for each node, identifies and removes nodes deemed to be affected by local independent disturbances due to abnormal data fluctuations, and obtains a reference node queue composed of nodes with stable background characteristics.

[0114] The differential risk assessment module is configured to acquire reference displacement sequences collected by the reference nodes, perform differential processing on the encrypted displacement sequence using the reference displacement sequences, and generate an updated risk state based on the encrypted displacement sequences after filtering out common-mode interference. The differential risk assessment module solves for the reciprocal parameter of the spatial linear interval of each reference node to generate normalized ratio weight coefficients. Multiplication and weighted fusion calculations are performed on each reference displacement sequence of the same time slice to fit an estimated common-mode interference sequence for the coordinate position of the node to be tested. Through arithmetic matrix subtraction, the encrypted displacement sequence is subtracted from the estimated common-mode interference sequence to eliminate background noise sequence segments, obtaining a net deformation displacement sequence object array. Finally, based on the cumulative change amplitude and acceleration parameters obtained from the integral and differential operations of the net deformation displacement sequence, matching segments are searched in the multidimensional state level threshold constraint matrix judgment tree, and the corresponding disaster alarm level enumeration variables are extracted and output.

[0115] The specific processing steps and logic executed by each module of the data processing system in this embodiment are consistent with the corresponding process objects in the aforementioned method embodiments.

[0116] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A data processing method for geological disaster early warning, characterized in that, include: By comparing the environmental state sequence collected by multiple nodes deployed in the monitoring area with the safety boundary, a basic early warning prompt is output when it is determined that the environmental state sequence exceeds the safety boundary; Extract the first node where the environmental state sequence undergoes a sudden change, and combine the spatial distribution topology to extract the physical distance between the first node and the adjacent nodes to be tested; By combining the current geological stress transmission rate with the physical distance, the stress wave lag time from the initial node to the node to be tested is calculated; The node under test is controlled to start a high-frequency sampling process when it enters the stress wave and hysteresis period to extract the encrypted displacement sequence. The main slip direction is determined based on the displacement vector output by the first node, and reference nodes that are perpendicular to the main slip direction are selected from the plurality of nodes; The reference displacement sequence collected by the reference node is obtained, and the encrypted displacement sequence is differentially processed using the reference displacement sequence. An updated risk state is generated based on the encrypted displacement sequence after filtering out common-mode interference. 2.The data processing method for geological disaster warning according to claim 1, characterized in that, The comparison involves examining the environmental state sequences collected by multiple nodes deployed within the monitoring area against the safety boundary. When the environmental state sequence exceeds the safety boundary, a basic early warning is output, including: The safety boundary is formed by extracting the pre-set upper limit of moisture content tolerance and the upper limit of tilt angle tolerance; When the latest formation water content in the environmental state sequence is greater than the water content tolerance limit, or the latest slope inclination angle is greater than the inclination angle tolerance limit, the anomaly determination logic is triggered. When the aforementioned anomaly detection logic occurs, a basic early warning prompt containing the identifier of the abnormal node and the mutation value is generated and pushed to the remote monitoring platform. 3.The data processing method for geological disaster warning according to claim 2, characterized in that, After the exception detection logic is triggered, it also includes: Retrieve the frequency of false alarms recorded in the historical early warning database; If the frequency of false alarms exceeds a set frequency threshold, the numerical range of the upper limit of moisture content tolerance and the upper limit of tilt angle tolerance will be expanded according to a set expansion ratio to dynamically update the safety boundary. 4.The data processing method for geological disaster warning according to claim 1, characterized in that, Before comparing the environmental state sequences collected by multiple nodes deployed within the monitoring area with the security boundary, the process also includes: Drive all the nodes to continuously collect formation water content and slope tilt angle at a preset low-frequency dormancy cycle; The water content of the formation and the slope inclination angle are processed using median filtering logic to filter out transient extreme noise. The formation water content and slope inclination angle, which are from different time dimensions and have had transient extreme noise eliminated, are spliced ​​together in a time sequence to generate the environmental state sequence. 5.The data processing method for geological disaster warning according to claim 1, characterized in that, The calculation of the stress wave latency period from the initial node to the node to be tested, by combining the current geological stress transmission rate with the physical distance, includes: Divide the physical distance by the geological stress conduction rate to obtain the foundation conduction time; Extract the current rainfall level of the monitoring area, and perform duration compensation scaling on the basic conduction time based on the rainfall level to obtain the compensated conduction time; Using the timestamp of the mutation at the first node as the starting point, and adding the compensated propagation time, the expected impact time is obtained. Centered on the expected impact time, the stress impact hysteresis period is generated by extending the allowable fault tolerance span forward and backward respectively. 6.The data processing method for geological disaster warning according to claim 5, characterized in that, The control of the node under test to initiate a high-frequency sampling process when entering the stress wave and hysteresis period, and to extract the encrypted displacement sequence, includes: Before the current time reaches the start time of the stress wave and hysteresis period, the node to be tested is controlled to maintain a low-power sleep mode. When the current time enters the stress wave and hysteresis period, the node to be tested is controlled to increase the sampling frequency to a predetermined high frequency value, and the deformation displacement data is continuously collected and combined into the encrypted displacement sequence. When the current time crosses the end of the stress wave and hysteresis period, the node to be tested is controlled to return to the low-power sleep mode. 7.The data processing method for geological disaster warning according to claim 1, characterized in that, The process of determining the main slip orientation based on the displacement vector output by the initial node, and selecting reference nodes that are perpendicular to the main slip orientation from among the plurality of nodes, includes: Extract the displacement and offset values ​​of the first node in all axes in the three-dimensional coordinate system; The displacement offset values ​​of all axes are vector synthesized to obtain the synthetic displacement vector of the first node, and the spatial ray pointed to by the synthetic displacement vector is defined as the main slip direction; Calculate the spatial direction vector of the line connecting the first node and the candidate nodes; Calculate the inner product of the spatial direction vector and the composite displacement vector, and select candidate nodes whose absolute values ​​of the corresponding inner product values ​​are within the zero tolerance range, and mark them as the reference nodes. 8.The data processing method for geological disaster warning according to claim 7, characterized in that, Following the designation as the reference node, the following is also included: Continuously collect the displacement data fluctuation variance of all the reference nodes, and calculate the group fluctuation average of the displacement data fluctuation variance; If the variance of displacement data fluctuation of a specific reference node exceeds the average value of the group fluctuation, it is determined that the associated reference node is affected by a local independent disturbance. Associated reference nodes affected by local independent disturbances are removed from the common-mode reference pool, and the remaining reference nodes are retained to provide the reference displacement sequence. 9.The data processing method for geological disaster warning according to claim 1, characterized in that, The process of acquiring the reference displacement sequence collected by the reference node, performing differential processing on the encrypted displacement sequence using the reference displacement sequence, and generating an updated risk state based on the encrypted displacement sequence after filtering out common-mode interference includes: Calculate the spatial linear intervals between the node to be tested and all the reference nodes; Using the reciprocal of the spatial linear interval as a weight, the reference displacement sequences output by all the reference nodes are weighted and fused to generate the estimated common-mode interference sequence of the location of the node to be examined; Subtract the estimated common-mode interference sequence from the encrypted displacement sequence to obtain the net deformation displacement sequence; Based on the cumulative change magnitude and acceleration mapping of the net deformation displacement sequence, the corresponding disaster alarm level is output as the updated risk status.

10. A data processing system for geological disaster warning, characterized by, include: The basic monitoring module is configured to compare the environmental state sequence collected by multiple nodes deployed in the monitoring area with the safety boundary, and output a basic early warning prompt when it is determined that the environmental state sequence exceeds the safety boundary; The spacing extraction module is configured to extract the first node where the environmental state sequence undergoes a sudden change, and to extract the physical spacing between the first node and the adjacent nodes to be examined by combining the spatial distribution topology. The time window calculation module is configured to combine the current geological stress transmission rate with the physical distance to calculate the stress wave lag time from the first node to the node to be tested; A high-frequency scheduling module is configured to control the node under test to start a high-frequency sampling process when it enters the stress wave and hysteresis period, and extract the encrypted displacement sequence. The benchmark screening module is configured to determine the main slip direction based on the displacement vector output by the first node, and select reference nodes that are perpendicular to the main slip direction from the plurality of nodes; The differential risk assessment module is configured to acquire the reference displacement sequence collected by the reference node, perform differential processing on the encrypted displacement sequence using the reference displacement sequence, and generate an updated risk status based on the encrypted displacement sequence after filtering out common-mode interference.

Images