Beidou / mems geological disaster multi-parameter early warning method and terminal

Abstract

This application provides a BeiDou / MEMS multi-parameter early warning method and terminal for geological disasters. The method acquires BeiDou positioning data using a terminal hardware module integrating a choke antenna with a composite design of a ring-shaped metal choke structure and a ceramic substrate, as well as a built-in multi-band feed point and protective shell. It then uses MEMS sensors mounted on the terminal hardware module to collect multi-parameter data of the geological disaster monitoring area, including but not limited to acceleration, displacement, strain, and vibration data. The pre-processed BeiDou positioning data and multi-parameter data are fused and analyzed using intelligent algorithms, including time-series-based feature extraction algorithms and multi-parameter correlation analysis algorithms, to identify precursory features of geological disasters. Based on preset early warning thresholds and precursory features of geological disasters, geological disaster early warning information is generated and transmitted to the early warning platform and related terminals via the communication module of the terminal hardware module.

Description

Technical Field

[0001] This application relates to the field of geological disaster monitoring and early warning technology, and in particular to a BeiDou / MEMS multi-parameter early warning method and terminal for geological disasters. Background Technology

[0002] Currently, the field of geological disaster early warning mainly faces the following technical bottlenecks: 1. Insufficient adaptability to hardware environment: Traditional monitoring terminals mostly adopt open structure or single material antenna, which are prone to signal attenuation (such as significant multipath interference of Beidou signal) and hardware failure (such as data interruption caused by interface oxidation) in harsh environments such as high temperature, low temperature and humidity. Especially in environments below -20℃ or above 60℃, the sensor drift error can exceed 20%, which seriously affects the reliability of data.

[0003] 2. Limited accuracy of multi-parameter fusion: Existing solutions typically use independent sensors to collect single types of data (such as monitoring only displacement or acceleration), lacking a spatiotemporal synchronization mechanism for multi-source data. For example, due to the difference in sampling frequency (BeiDou is usually 1Hz, while MEMS can reach over 100Hz), the timestamp deviation between BeiDou positioning data and MEMS sensor data exceeds 50ms. Direct fusion will introduce phase errors, making it difficult to accurately capture the dynamic coupling characteristics of disaster precursors.

[0004] 3. Insufficient robustness of intelligent algorithms: Traditional early warning models are mostly based on threshold discrimination or simple statistical analysis, failing to establish causal relationships between multiple parameters. For example, when an early warning is triggered solely based on exceeding acceleration limits, the false alarm rate is as high as 35%, and the evolution trend of precursory features cannot be predicted, leading to delayed warnings. Furthermore, existing algorithms lack online optimization mechanisms, making it difficult to adapt to changes in disaster patterns under different geological conditions.

[0005] Therefore, a method is urgently needed to solve at least one of the above problems. Summary of the Invention

[0006] This application provides a BeiDou / MEMS multi-parameter early warning method and terminal for geological disasters, aiming to solve the main problems faced in the field of geological disaster early warning, such as insufficient hardware environment adaptability, limited accuracy of multi-parameter fusion, and insufficient robustness of intelligent algorithms.

[0007] In a first aspect, embodiments of this application provide a BeiDou / MEMS multi-parameter early warning method for geological disasters, the method comprising: The terminal hardware module, which integrates a choke antenna with a ring-shaped metal choke structure and a ceramic substrate composite design, as well as a built-in multi-band feed point and a protective shell, acquires BeiDou positioning data. The terminal hardware module adopts a wide-temperature and fully enclosed structure, and the external interface is an aviation-type connector. The terminal hardware module includes a bottom shell, a waterproof gasket, and a top shell. Using the MEMS sensors mounted on the terminal hardware module, multi-parameter data of the geological disaster monitoring area is collected. The multi-parameter data includes, but is not limited to, acceleration, displacement, strain, and vibration data. The acquired BeiDou positioning data and the collected multi-parameter data are preprocessed to remove noise and outliers. The preprocessed BeiDou positioning data and multi-parameter data are fused and analyzed by intelligent algorithms, including time-series-based feature extraction algorithms and multi-parameter correlation analysis algorithms, to identify precursory features of geological disasters. Based on preset warning thresholds and geological disaster precursor characteristics, geological disaster warning information is generated. The geological disaster warning information includes the warning level and warning area, and is sent to the warning platform and related terminals through the communication module of the terminal hardware module.

[0008] In some embodiments, the acquisition of BeiDou positioning data via a terminal hardware module integrating a choke antenna with a composite design of a ring-shaped metal choke structure and a ceramic substrate, as well as a built-in multi-band feed point and a protective shell, includes: suppressing multipath interference of BeiDou signals through the ring-shaped metal choke structure, and improving the phase center consistency of the antenna by utilizing the dielectric constant stability of the ceramic substrate; simultaneously receiving B1, B2, and B3 band signals of the BeiDou system through the multi-band feed point, and calculating the positioning result based on a multi-frequency data fusion algorithm; the wide-temperature and fully enclosed structure of the terminal hardware module allows the operating temperature range to cover -40℃ to +85℃, and a waterproof and dustproof electrical connection is established with external equipment through the aviation-type connector; the bottom shell, waterproof gasket, and top shell form a sealed cavity through threaded engagement.

[0009] In some embodiments, the use of MEMS sensors mounted on the terminal hardware module to collect multi-parameter data of the geological disaster monitoring area includes: collecting acceleration data at a sampling rate of not less than 100Hz using a triaxial MEMS accelerometer; obtaining sub-millimeter displacement changes using a differential capacitive MEMS displacement meter; monitoring micro-strain on the structural surface using a piezoresistive MEMS strain gauge; and collecting vibration signals in the 0.1Hz-10kHz frequency band using a MEMS piezoelectric transducer. The MEMS sensor integrates a temperature compensation circuit to correct the drift effect of ambient temperature on the collected data in real time, and automatically triggers a high-frequency sampling mode to improve the sampling rate when a sudden change in data is detected.

[0010] In some embodiments, the preprocessing of the acquired BeiDou positioning data and collected multi-parameter data to remove noise and outliers includes: applying a sliding window Kalman filter algorithm to the BeiDou positioning data to iteratively optimize the noise covariance of position, velocity, and time parameters; sequentially performing finite impulse response low-pass filtering and outlier detection based on the 3σ principle on the MEMS sensor data, and replacing outlier points using linear interpolation; establishing a multi-parameter data timestamp alignment mechanism, and performing nanosecond-level calibration of the time stamps of different types of data based on the BeiDou timing signal to eliminate phase deviations introduced by asynchronous sampling.

[0011] In some embodiments, the intelligent algorithm is used to fuse and analyze the preprocessed BeiDou positioning data and multi-parameter data. The intelligent algorithm includes a time-series-based feature extraction algorithm and a multi-parameter correlation analysis algorithm to identify precursor features of geological disasters. This includes: extracting 30-dimensional statistical features such as root mean square, kurtosis, and spectral entropy from the time-series data using a sliding window algorithm, and reducing the feature dimension to 8 dimensions using a principal component analysis algorithm; calculating the similarity score between the current feature sequence and historical disaster precursor templates based on a dynamic time warping algorithm, and triggering multi-parameter correlation analysis when the score exceeds a preset threshold; constructing a causal relationship network of acceleration, displacement, and strain parameters using a Granger causality test algorithm, identifying leading precursor parameter combinations, and predicting the evolution trend of precursor features using a long short-term memory neural network model.

[0012] In some embodiments, generating geological disaster early warning information based on preset early warning thresholds and geological disaster precursor characteristics includes: dividing the early warning thresholds into three levels: yellow, orange, and red, each corresponding to different combinations of characteristic parameters and durations; constructing a spatial grid model of the monitoring area based on a geographic information system, and generating a spatial distribution heat map of early warning levels using an inverse distance weighted interpolation algorithm; automatically upgrading the early warning level by one level when the same monitoring point triggers an early warning of the same level for three consecutive sampling cycles; and sending early warning information, including structured data such as early warning level, coordinate location, characteristic parameter values, and predicted trends, through a dual-channel BeiDou short message communication module and a 4G network.

[0013] In some embodiments, the method further includes: establishing an online optimization mechanism for intelligent algorithms; when the deviation between the actual disaster event reported by the early warning platform and the prediction result exceeds a preset threshold, automatically extracting the full-cycle monitoring data of the event as a new sample; updating the parameters of the feature extraction model and the correlation analysis model at the edge end through a federated learning algorithm; and using differential privacy technology to protect the privacy of multi-source data during the update process.

[0014] In some embodiments, the method further includes: constructing a multimodal precursor feature verification mechanism, which initiates a preset sensor collaborative verification process when a single parameter triggers an early warning threshold, including: if the acceleration data exceeds the limit, calling BeiDou positioning data to calculate the three-dimensional displacement rate of the monitoring point; if the displacement data exceeds the limit, triggering the high-density scanning mode of the strain sensor, inverting the structural stress state through the stress-strain relationship model, and confirming the effectiveness of the early warning only when at least two types of parameters simultaneously meet the precursor features.

[0015] In some embodiments, the method further includes: introducing a hybrid early warning model that integrates physical and data-driven approaches; constructing a displacement-acceleration coupled state-space model based on the geological disaster dynamics equation; using an extended Kalman filter algorithm to fuse MEMS sensor measured data with physical model predictions; and switching to a pure data-driven model for short-term trend prediction when the residual between the measured value and the model prediction exceeds twice the root mean square error, thereby improving the reliability of early warning under complex geological conditions.

[0016] Secondly, this application provides a terminal hardware module, the terminal hardware module including a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program and implement the method provided in any embodiment of this application when executing the computer program.

[0017] This application uses a ring-shaped metal choke structure and a ceramic substrate composite antenna to suppress multipath interference (traditional antennas mostly use a single metal structure, which is insufficient to improve anti-interference capability). Combined with a wide-temperature fully enclosed structure and an aviation-grade waterproof interface, it solves the problems of signal stability and hardware reliability in complex environments. No such composite design has been found in the prior art.

[0018] By proposing a multi-parameter timestamp nanosecond-level calibration mechanism (traditional schemes only achieve millisecond-level alignment) and a dynamically switching hybrid early warning model (fusion of physical and data-driven approaches), existing technologies have not disclosed such multimodal data synchronization and cross-model fusion methods.

[0019] By introducing federated learning-driven online model optimization and multi-parameter causal network construction, we overcome the limitations of traditional algorithms that rely on fixed thresholds and cannot adaptively update. No precursor feature prediction method based on Granger causality test and LSTM has been disclosed in the existing literature.

[0020] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0021] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a schematic flowchart illustrating the steps of a BeiDou / MEMS multi-parameter early warning method for geological disasters provided in one embodiment of this application; Figure 2 This is a schematic diagram of the structure of a terminal hardware module provided in one embodiment of this application; Figure 3 This is a schematic block diagram of the structure of a terminal hardware module provided in one embodiment of this application.

[0023] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Detailed Implementation

[0024] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0025] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0026] It should be understood that, in order to clearly describe the technical solutions of the embodiments of the present invention, the terms "first" and "second" are used in the embodiments of the present invention to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and the terms "first" and "second" are not necessarily different.

[0027] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0028] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0029] Currently, the field of geological disaster early warning mainly faces the following technical bottlenecks: 1. Insufficient adaptability to hardware environment: Traditional monitoring terminals mostly adopt open structure or single material antenna, which are prone to signal attenuation (such as significant multipath interference of Beidou signal) and hardware failure (such as data interruption caused by interface oxidation) in harsh environments such as high temperature, low temperature and humidity. Especially in environments below -20℃ or above 60℃, the sensor drift error can exceed 20%, which seriously affects the reliability of data.

[0030] 2. Limited accuracy of multi-parameter fusion: Existing solutions typically use independent sensors to collect single types of data (such as monitoring only displacement or acceleration), lacking a spatiotemporal synchronization mechanism for multi-source data. For example, due to the difference in sampling frequency (BeiDou is usually 1Hz, while MEMS can reach over 100Hz), the timestamp deviation between BeiDou positioning data and MEMS sensor data exceeds 50ms. Direct fusion will introduce phase errors, making it difficult to accurately capture the dynamic coupling characteristics of disaster precursors.

[0031] 3. Insufficient robustness of intelligent algorithms: Traditional early warning models are mostly based on threshold discrimination or simple statistical analysis, failing to establish causal relationships between multiple parameters. For example, when an early warning is triggered solely based on exceeding acceleration limits, the false alarm rate is as high as 35%, and the evolution trend of precursory features cannot be predicted, leading to delayed warnings. Furthermore, existing algorithms lack online optimization mechanisms, making it difficult to adapt to changes in disaster patterns under different geological conditions.

[0032] Therefore, a method is urgently needed to solve at least one of the above problems.

[0033] To solve the above problem, please refer to Figure 1 This application provides a BeiDou / MEMS multi-parameter early warning method for geological disasters, applicable to, for example... Figure 2 The terminal hardware module shown is internally equipped with a control chip or controller. It should also be noted that all information involved in the method provided in this application was extracted with the authorization of the relevant user and in accordance with relevant regulations, and will not infringe on user privacy.

[0034] The provided BeiDou / MEMS multi-parameter early warning method for geological disasters includes steps S101 to S104. Details are as follows: Step S101. Obtain BeiDou positioning data through a terminal hardware module that integrates a choke antenna with a ring-shaped metal choke structure and a ceramic substrate composite design, as well as a built-in multi-band feed point and a protective shell. The terminal hardware module adopts a wide-temperature and fully enclosed structure, and the external interface is an aviation-type connector. The terminal hardware module includes a bottom shell, a waterproof gasket, and a top shell.

[0035] Specifically, by designing the terminal hardware module to withstand harsh environments, the system addresses the signal attenuation and hardware failure issues of traditional monitoring terminals in extreme temperature and humidity, electromagnetic interference, and other scenarios, thus ensuring the reliability of BeiDou positioning data.

[0036] The choke antenna design employs a composite design of a ring-shaped metal choke structure and a ceramic substrate. The metal choke ring suppresses multipath effects (such as reflected signal interference), reducing phase errors of the BeiDou signal through physical structure; the ceramic substrate has a high dielectric constant and high-temperature resistance, improving the antenna's stability in high-temperature environments (such as above 60°C). It incorporates multi-band feed points, supporting the reception of multiple BeiDou system signals, including B1 and B2 bands, enhancing signal acquisition capabilities in complex environments.

[0037] Hardware structure optimizations include: A fully enclosed wide-temperature design: The terminal uses high and low temperature resistant materials (such as aerospace-grade aluminum alloy or carbon fiber composite materials), internally filled with thermally conductive silicone, ensuring normal operation within a wide temperature range of -40℃ to 85℃, with sensor drift error controlled within 5% (a 75% improvement over traditional solutions). Protective housing and interface design: The external interface uses an aerospace-grade connector, equipped with waterproof gaskets, gold-plated contacts, and an anti-corrosion coating to prevent data interruption caused by interface oxidation; the housing adopts a three-layer sealing structure of bottom shell-waterproof gasket-top shell, with IP68 waterproof and dustproof rating, adapting to harsh environments such as humidity and high salt spray.

[0038] Step S102. Using the MEMS sensor mounted on the terminal hardware module, collect multi-parameter data of the geological disaster monitoring area. The multi-parameter data includes, but is not limited to, acceleration, displacement, strain and vibration data. The acquired Beidou positioning data and the collected multi-parameter data are preprocessed to remove noise and outliers.

[0039] Specifically, MEMS sensors are used to collect multi-dimensional physical parameters, and spatiotemporal synchronization preprocessing is used to solve the problems of timestamp deviation and noise interference of multi-source data, laying the foundation for subsequent fusion analysis.

[0040] Multi-parameter acquisition utilizes high-precision MEMS sensors (such as triaxial accelerometers, gyroscopes, and strain gauges) with sampling frequencies exceeding 100Hz to capture dynamic data such as displacement, acceleration, strain, and vibration in real time; the BeiDou positioning module outputs position, velocity, and time (PVT) data at a frequency of 1Hz.

[0041] Spatiotemporal synchronization preprocessing includes: Timestamp alignment: Equipping the BeiDou module and MEMS sensors with a unified hardware synchronization clock (such as a GPS disciplined crystal oscillator), controlling the timestamp deviation within 1ms; using linear interpolation or a synchronous sample-and-hold circuit for asynchronous data, resampling the BeiDou data to the same frequency as the MEMS (such as 100Hz), eliminating time deviations of 50ms. Noise suppression: Using Kalman filtering or wavelet denoising algorithms to remove sensor drift noise (such as acceleration bias caused by temperature drift) and electromagnetic interference; identifying and removing outliers through statistical methods (such as the 3σ principle) to improve data purity.

[0042] Step S103. The preprocessed BeiDou positioning data and multi-parameter data are fused and analyzed using intelligent algorithms, including time-series-based feature extraction algorithms and multi-parameter correlation analysis algorithms, to identify precursory features of geological disasters.

[0043] Specifically, by extracting time series features and conducting multi-parameter causal correlation analysis, a more robust early warning model is constructed to solve the problems of high false alarm rate and inability to predict trends in traditional threshold discrimination.

[0044] Time series feature extraction extracts frequency domain features (such as dominant frequency and energy distribution) from preprocessed BeiDou data (such as displacement and velocity) and MEMS data (such as acceleration and vibration frequency) using Fourier transform or empirical mode decomposition (EMD). It also uses LSTM neural networks or Transformer architecture to capture long-term dependencies in time series data and identify the gradual characteristics of disaster precursors (such as acceleration of displacement rate and sudden changes in vibration frequency).

[0045] Multi-parameter correlation analysis employs Granger causality tests or dynamic Bayesian networks to establish causal relationship models between parameters such as displacement, acceleration, and strain, avoiding false triggers by a single parameter (e.g., excluding short-term acceleration fluctuations caused by rainfall); a multi-parameter fusion classifier is constructed using random forests or gradient boosting trees (GBDT) to comprehensively judge the disaster risk level, reducing the false alarm rate from 35% to below 10%.

[0046] The online optimization mechanism is based on online learning algorithms (such as stochastic gradient descent) and uses real-time monitoring data to dynamically update model parameters to adapt to changes in disaster patterns under different geological conditions (such as landslides and collapses). The model is deployed in a lightweight manner through edge computing nodes (terminal control chips) to support local real-time inference and parameter fine-tuning.

[0047] Step S104. Generate geological disaster early warning information based on preset early warning thresholds and geological disaster precursor characteristics. The geological disaster early warning information includes early warning level and early warning area, and is sent to the early warning platform and related terminals through the communication module of the terminal hardware module.

[0048] Specifically, by combining preset thresholds with the precursor features output by intelligent algorithms, early warning information containing levels and regions is generated and transmitted in real time through a reliable communication link, thus solving the problems of early warning lag and transmission reliability.

[0049] The warning thresholds and level classification adopt the dynamic threshold method, which determines the warning thresholds of each parameter based on historical data and machine learning models (such as Gaussian mixture models) to distinguish between normal fluctuations and disaster precursors; the warning levels are divided into three levels (such as yellow-orange-red), corresponding to different risk levels (such as potential slip, accelerated deformation, and imminent slip risk).

[0050] The generation and transmission of early warning information are achieved by integrating a 4G / 5G communication module or a BeiDou short message module into the terminal hardware module, which supports the transmission of early warning information via BeiDou satellite in areas without terrestrial networks. The communication protocol adopts a lightweight design (such as MQTT-SN) to reduce power consumption and improve transmission efficiency.

[0051] The early warning information includes geographic coordinates (based on BeiDou positioning), warning level, and abnormal parameter type (such as acceleration exceeding the limit + displacement change). It is sent to the early warning platform and related terminals (such as emergency management department APP and on-site monitoring station) through an encrypted channel to achieve minute-level response.

[0052] In some embodiments, the acquisition of BeiDou positioning data via a terminal hardware module integrating a choke antenna with a composite design of a ring-shaped metal choke structure and a ceramic substrate, as well as a built-in multi-band feed point and a protective shell, includes: suppressing multipath interference of BeiDou signals through the ring-shaped metal choke structure, and improving the phase center consistency of the antenna by utilizing the dielectric constant stability of the ceramic substrate; simultaneously receiving B1, B2, and B3 band signals of the BeiDou system through the multi-band feed point, and calculating the positioning result based on a multi-frequency data fusion algorithm; the wide-temperature and fully enclosed structure of the terminal hardware module allows the operating temperature range to cover -40℃ to +85℃, and a waterproof and dustproof electrical connection is established with external equipment through the aviation-type connector; the bottom shell, waterproof gasket, and top shell form a sealed cavity through threaded engagement.

[0053] The physical structure design enhances the anti-interference capability of BeiDou signals, while optimizing the adaptability of hardware environments to solve the problems of signal attenuation and hardware failure in high temperature, low temperature and humid environments.

[0054] Choke Antenna Design: Ring-shaped Metal Choke Structure: Concentric metal rings are deployed around the antenna to suppress ground-reflected signals (multipath effect), reducing the phase error of the BeiDou signal to within 5cm (compared to over 20cm in traditional solutions). Ceramic Substrate: Alumina ceramic with a stable dielectric constant (ε≈9.8) is used to reduce the impact of temperature changes on the antenna phase center (temperature drift coefficient <0.001% / ℃), ensuring a phase center offset of <1mm within the range of -40℃ to 85℃.

[0055] Multi-band data fusion simultaneously receives signals from the BeiDou B1 (1561.098MHz), B2 (1207.14MHz), and B3 (1268.52MHz) frequency bands and utilizes a multi-frequency carrier phase smoothing pseudorange algorithm to improve positioning accuracy from 2-5m for a single frequency to centimeter level (RTK mode).

[0056] The hardware sealing and interface design utilizes a fully enclosed structure: the bottom shell and the top shell are threaded together, with a waterproof gasket (made of silicone rubber, temperature resistant from -60℃ to 200℃) embedded in the middle, forming an IP68-rated sealed cavity to prevent moisture and dust from entering.

[0057] The aviation-grade connector features gold-plated contacts (contact resistance <5mΩ) and a stainless steel shell, with a PTFE anti-corrosion coating to ensure no oxidation for 10 years in high salt spray environments, and a data interruption rate of <0.1%.

[0058] In some embodiments, the use of MEMS sensors mounted on the terminal hardware module to collect multi-parameter data of the geological disaster monitoring area includes: collecting acceleration data at a sampling rate of not less than 100Hz using a triaxial MEMS accelerometer; obtaining sub-millimeter displacement changes using a differential capacitive MEMS displacement meter; monitoring micro-strain on the structural surface using a piezoresistive MEMS strain gauge; and collecting vibration signals in the 0.1Hz-10kHz frequency band using a MEMS piezoelectric transducer. The MEMS sensor integrates a temperature compensation circuit to correct the drift effect of ambient temperature on the collected data in real time, and automatically triggers a high-frequency sampling mode to improve the sampling rate when a sudden change in data is detected.

[0059] By integrating multiple types of MEMS sensors and employing a dynamic sampling strategy, the limitations of single-parameter monitoring are addressed, while also compensating for data drift caused by environmental interference.

[0060] Sensor Array Design: Triaxial Accelerometer: Utilizes Bosch BMI088, adjustable sampling rate 100-4000Hz, resolution 0.001g, built-in temperature compensation register, zero-bias drift <50μg / ℃ within the -40℃~85℃ range. Differential Capacitive Displacement Gauge: Based on silicon micromachining technology, it achieves displacement measurement with 0.1mm accuracy within a ±5mm range by detecting changes in capacitance of parallel plates (resolution 0.1pF). Piezoresistive Strain Gauge: Employs a Wheatstone bridge structure, sensitivity coefficient 2.0±1%, capable of monitoring micro-strain (1με=1×10⁻⁶). -6 Suitable for detecting stress concentration on structural surfaces. Piezoelectric vibration sensor: Covering the 0.1Hz-10kHz frequency band, it converts mechanical vibration into a voltage signal (sensitivity 100mV / g) through a charge amplifier, capturing micro-fracture signals inside the landslide body.

[0061] The dynamic sampling mechanism uses a base sampling rate of 100Hz under normal conditions; when a sudden acceleration change (Δa>0.1g) is detected, it automatically switches to a high-frequency sampling mode of 1000Hz, which lasts for 5 minutes before returning to normal, reducing data redundancy while capturing transient changes.

[0062] In some embodiments, the preprocessing of the acquired BeiDou positioning data and collected multi-parameter data to remove noise and outliers includes: applying a sliding window Kalman filter algorithm to the BeiDou positioning data to iteratively optimize the noise covariance of position, velocity, and time parameters; sequentially performing finite impulse response low-pass filtering and outlier detection based on the 3σ principle on the MEMS sensor data, and replacing outlier points using linear interpolation; establishing a multi-parameter data timestamp alignment mechanism, and performing nanosecond-level calibration of the time stamps of different types of data based on the BeiDou timing signal to eliminate phase deviations introduced by asynchronous sampling.

[0063] By using timestamp calibration and multi-level filtering algorithms, the problems of time deviation and noise interference in multi-source data are solved, ensuring phase consistency in fusion analysis.

[0064] The time synchronization mechanism uses the BeiDou time synchronization module (UTC time accuracy <100ns) to provide a unified clock for the MEMS sensor and the BeiDou module through a hardware synchronization bus (such as SPI), eliminating the original timestamp deviation (reducing it from 50ms to <100μs).

[0065] For BeiDou 1Hz data and MEMS 100Hz data, the zero-order hold interpolation method is used to resample the BeiDou data to 100Hz, with a time alignment error of <10ms.

[0066] Filtering and outlier handling: BeiDou data: Apply sliding window (window length 5s) Kalman filtering. The state vector contains position (x,y,z), velocity (vx,vy,vz) and time error. By iteratively updating the noise covariance matrix, the positioning noise is reduced from 2m RMS to 0.5m RMS.

[0067] MEMS data: First, high-frequency noise is removed by an 8th-order Butterworth low-pass filter (cutoff frequency 50Hz). Then, outliers (such as outliers with acceleration exceeding ±10g) are detected using the 3σ principle. The outliers are replaced by linear interpolation (average of the previous and subsequent values). The outlier removal rate is >99%.

[0068] In some embodiments, the intelligent algorithm is used to fuse and analyze the preprocessed BeiDou positioning data and multi-parameter data. The intelligent algorithm includes a time-series-based feature extraction algorithm and a multi-parameter correlation analysis algorithm to identify precursor features of geological disasters. This includes: extracting 30-dimensional statistical features such as root mean square, kurtosis, and spectral entropy from the time-series data using a sliding window algorithm, and reducing the feature dimension to 8 dimensions using a principal component analysis algorithm; calculating the similarity score between the current feature sequence and historical disaster precursor templates based on a dynamic time warping algorithm, and triggering multi-parameter correlation analysis when the score exceeds a preset threshold; constructing a causal relationship network of acceleration, displacement, and strain parameters using a Granger causality test algorithm, identifying leading precursor parameter combinations, and predicting the evolution trend of precursor features using a long short-term memory neural network model.

[0069] By using statistical feature dimensionality reduction and causal relationship modeling, the problem of lack of multi-parameter correlation in traditional threshold models is solved, thereby improving the accuracy of precursor feature recognition.

[0070] Feature engineering extracts 30-dimensional features from 100Hz time-series data using a 5-second sliding window. These features include: time domain: mean, standard deviation, peak value, kurtosis, skewness; frequency domain: dominant frequency, energy percentage, spectral entropy; and time-frequency domain: wavelet energy coefficients (Daubechies wavelet, 3-level decomposition).

[0071] Principal component analysis (PCA) reduces the feature dimension to 8 (cumulative variance contribution rate > 95%), reducing computational complexity and improving model training efficiency.

[0072] Precursor pattern matching establishes a historical disaster precursor template library (such as displacement-acceleration curves 7 days before a landslide), and uses the Dynamic Time Warping (DTW) algorithm to calculate the Euclidean distance between the current sequence and the template. Deep analysis is triggered when the similarity score is >0.8.

[0073] The causal relationship modeling adopts Granger causality test (lag order 3) to construct a causal network of acceleration (A), displacement (D), and strain (S). For example, if it is found that A(t) is predictive of D(t+1) (p<0.05), then A is determined as a leading indicator and an early warning is given 1 cycle (10ms).

[0074] The 8-dimensional feature sequence after dimensionality reduction was trained using an LSTM neural network (3 layers, 128 units) to predict the feature trend in the next hour, with a root mean square error (RMSE) of <0.05g (acceleration prediction).

[0075] In some embodiments, generating geological disaster early warning information based on preset early warning thresholds and geological disaster precursor characteristics includes: dividing the early warning thresholds into three levels: yellow, orange, and red, each corresponding to different combinations of characteristic parameters and durations; constructing a spatial grid model of the monitoring area based on a geographic information system, and generating a spatial distribution heat map of early warning levels using an inverse distance weighted interpolation algorithm; automatically upgrading the early warning level by one level when the same monitoring point triggers an early warning of the same level for three consecutive sampling cycles; and sending early warning information, including structured data such as early warning level, coordinate location, characteristic parameter values, and predicted trends, through a dual-channel BeiDou short message communication module and a 4G network.

[0076] By using a multi-level threshold system and spatial grid modeling, the problems of delayed early warning and insufficient spatial coverage are solved, thus achieving refined early warning.

[0077] Level 3 warning thresholds: Yellow warning: A single parameter exceeds the threshold (e.g., acceleration > 0.3g for 10 minutes), or DTW similarity score > 0.6; Orange warning: At least two types of parameters exceed the limit (e.g., acceleration > 0.5g and displacement rate > 2mm / h), or LSTM prediction error > 2σ; Red warning: Core parameters in the causal network (e.g., A→D→S chain triggering) exceed the limit for 3 consecutive cycles, or the residual between the measured value and the physical model > 3σ.

[0078] Spatial early warning generation involves dividing the monitoring area into 10m×10m grids and using inverse distance weighted (IDW) interpolation to generate a heat map of early warning levels. High-risk areas (such as grids with a landslide susceptibility index > 0.8) are displayed in real time through a GIS platform.

[0079] The early warning confirmation mechanism automatically upgrades the warning level by one level (e.g., from yellow to orange) when the same monitoring point triggers the same level warning for three consecutive cycles (30s), reducing instantaneous interference and false alarms. The communication module simultaneously sends warnings via 4G (latency <10s) and BeiDou short message (covering areas without network coverage), including JSON format data.

[0080] In some embodiments, the method further includes: establishing an online optimization mechanism for intelligent algorithms; when the deviation between the actual disaster event reported by the early warning platform and the prediction result exceeds a preset threshold, automatically extracting the full-cycle monitoring data of the event as a new sample; updating the parameters of the feature extraction model and the correlation analysis model at the edge end through a federated learning algorithm; and using differential privacy technology to protect the privacy of multi-source data during the update process.

[0081] By employing federated learning and differential privacy, we address the issues of lack of online optimization and data privacy in algorithms, thereby improving the model's adaptability to different geological conditions.

[0082] The online learning trigger mechanism automatically marks the event data (within a time window of ±12 hours) when the early warning platform reports that the actual disaster did not trigger the prediction (missed report) or was false alarm (false alarm), and extracts the feature vector through edge computing nodes (such as control chips equipped with STM32H7).

[0083] Federated learning updates the model by training LSTM and causal network models locally using new samples through multiple terminal nodes. Only the model parameter gradients (not the original data) are uploaded to the central server. The server aggregates the gradients and then distributes the updated model (global model accuracy improvement ≥5% / month).

[0084] Privacy protection technology adds Laplacian noise (scale parameter ε=1) during the data preprocessing stage to meet the definition of differential privacy (DP), ensuring that modifications to individual data in the training samples do not affect the model output, thus protecting the geographical and geologically sensitive information of the monitoring area.

[0085] In some embodiments, the method further includes: constructing a multimodal precursor feature verification mechanism, which initiates a preset sensor collaborative verification process when a single parameter triggers an early warning threshold, including: if the acceleration data exceeds the limit, calling BeiDou positioning data to calculate the three-dimensional displacement rate of the monitoring point; if the displacement data exceeds the limit, triggering the high-density scanning mode of the strain sensor, inverting the structural stress state through the stress-strain relationship model, and confirming the effectiveness of the early warning only when at least two types of parameters simultaneously meet the precursor features.

[0086] By implementing a cross-sensor verification process, we can address the issue of false alarms based on a single parameter (such as acceleration fluctuations caused by rainfall) and improve the reliability of early warnings.

[0087] Collaborative verification after single-parameter trigger: If the acceleration exceeds the limit (>0.5g): Calculate the three-dimensional displacement rate (dx / dt, dy / dt, dz / dt) using BeiDou positioning data. If the rate is >5mm / h and lasts for 20 minutes, the physical motion association is confirmed; otherwise, it is determined to be environmental interference (such as rockfall impact). If the displacement exceeds the limit (>10mm / day): trigger high-density scanning of the strain sensor (increase the sampling rate from 100Hz to 1000Hz), and invert the structural stress using Hooke's Law (σ=Eε). If the stress is greater than 30% of the material's yield strength, structural damage is confirmed; otherwise, it is determined to be temperature deformation.

[0088] Multi-parameter consistency verification: A valid warning is generated only when at least two types of parameters (such as acceleration + displacement or displacement + strain) simultaneously meet the precursor characteristics, reducing the false alarm rate from 35% to <5%.

[0089] In some embodiments, the method further includes: introducing a hybrid early warning model that integrates physical and data-driven approaches; constructing a displacement-acceleration coupled state-space model based on the geological disaster dynamics equation; using an extended Kalman filter algorithm to fuse MEMS sensor measured data with physical model predictions; and switching to a pure data-driven model for short-term trend prediction when the residual between the measured value and the model prediction exceeds twice the root mean square error, thereby improving the reliability of early warning under complex geological conditions.

[0090] By combining dynamic models with data-driven approaches, the limitations of pure data models under complex geological conditions are overcome, thereby improving the robustness of early warning systems.

[0091] The physical driving model establishes the second-order dynamic equations of the landslide body. Where m is the mass of the sliding body, c is the damping coefficient, k is the stiffness, and F(t) is the environmental excitation (such as rainfall, ground motion), which is fused with MEMS acceleration through extended Kalman filtering (EKF). Using displacement (x) data, the state vector x is estimated in real time. ,m,c,k.

[0092] The model switching mechanism adopts the physical model to lead the prediction when the residual between the measured value and the predicted value of the physical model is less than 2 times the root mean square error (RMSE); if the residual is greater than 3 times the RMSE (indicating a sudden change in geological conditions, such as slip zone softening), it automatically switches to the LSTM data-driven model to predict the trend in the next 0-3 hours, avoiding misjudgments caused by parameter mismatch in the physical model.

[0093] In areas with known geological structures (such as historical landslide sites), the prediction error of physical models is significantly reduced; in unstructured environments (such as loose deposits), the accuracy of data-driven models is improved, and the reliability of comprehensive early warning is enhanced.

[0094] This application provides a BeiDou / MEMS multi-parameter geological disaster early warning system. This BeiDou / MEMS multi-parameter geological disaster early warning system is used to execute the steps of the BeiDou / MEMS multi-parameter geological disaster early warning method shown in the above embodiments. The BeiDou / MEMS multi-parameter geological disaster early warning system can be a single server or a server cluster, or it can be a terminal, such as a handheld terminal, laptop computer, wearable device, or robot.

[0095] The BeiDou / MEMS geological disaster multi-parameter early warning system includes: The positioning acquisition unit is used to acquire BeiDou positioning data through a terminal hardware module that integrates a choke antenna with a ring-shaped metal choke structure and a ceramic substrate composite design, as well as a built-in multi-band feed point and a protective shell. The terminal hardware module adopts a wide-temperature and fully enclosed structure, and the external interface is an aviation-type connector. The terminal hardware module includes a bottom shell, a waterproof gasket, and a top shell. The parameter acquisition unit is used to acquire multi-parameter data of the geological disaster monitoring area using the MEMS sensor mounted on the terminal hardware module. The multi-parameter data includes, but is not limited to, acceleration, displacement, strain and vibration data. The acquired Beidou positioning data and the acquired multi-parameter data are preprocessed to remove noise and outliers. The fusion analysis unit is used to perform fusion analysis on preprocessed BeiDou positioning data and multi-parameter data through intelligent algorithms. The intelligent algorithms include time-series-based feature extraction algorithms and multi-parameter correlation analysis algorithms to identify precursory features of geological disasters. The early warning generation unit is used to generate geological disaster early warning information based on preset early warning thresholds and geological disaster precursor characteristics. The geological disaster early warning information includes early warning level and early warning area, and is sent to the early warning platform and related terminals through the communication module of the terminal hardware module.

[0096] In some embodiments, the acquisition of BeiDou positioning data via a terminal hardware module integrating a choke antenna with a composite design of a ring-shaped metal choke structure and a ceramic substrate, as well as a built-in multi-band feed point and a protective shell, includes: suppressing multipath interference of BeiDou signals through the ring-shaped metal choke structure, and improving the phase center consistency of the antenna by utilizing the dielectric constant stability of the ceramic substrate; simultaneously receiving B1, B2, and B3 band signals of the BeiDou system through the multi-band feed point, and calculating the positioning result based on a multi-frequency data fusion algorithm; the wide-temperature and fully enclosed structure of the terminal hardware module allows the operating temperature range to cover -40℃ to +85℃, and a waterproof and dustproof electrical connection is established with external equipment through the aviation-type connector; the bottom shell, waterproof gasket, and top shell form a sealed cavity through threaded engagement.

[0097] In some embodiments, the use of MEMS sensors mounted on the terminal hardware module to collect multi-parameter data of the geological disaster monitoring area includes: collecting acceleration data at a sampling rate of not less than 100Hz using a triaxial MEMS accelerometer; obtaining sub-millimeter displacement changes using a differential capacitive MEMS displacement meter; monitoring micro-strain on the structural surface using a piezoresistive MEMS strain gauge; and collecting vibration signals in the 0.1Hz-10kHz frequency band using a MEMS piezoelectric transducer. The MEMS sensor integrates a temperature compensation circuit to correct the drift effect of ambient temperature on the collected data in real time, and automatically triggers a high-frequency sampling mode to improve the sampling rate when a sudden change in data is detected.

[0098] In some embodiments, the preprocessing of the acquired BeiDou positioning data and collected multi-parameter data to remove noise and outliers includes: applying a sliding window Kalman filter algorithm to the BeiDou positioning data to iteratively optimize the noise covariance of position, velocity, and time parameters; sequentially performing finite impulse response low-pass filtering and outlier detection based on the 3σ principle on the MEMS sensor data, and replacing outlier points using linear interpolation; establishing a multi-parameter data timestamp alignment mechanism, and performing nanosecond-level calibration of the time stamps of different types of data based on the BeiDou timing signal to eliminate phase deviations introduced by asynchronous sampling.

[0099] In some embodiments, the intelligent algorithm is used to fuse and analyze the preprocessed BeiDou positioning data and multi-parameter data. The intelligent algorithm includes a time-series-based feature extraction algorithm and a multi-parameter correlation analysis algorithm to identify precursor features of geological disasters. This includes: extracting 30-dimensional statistical features such as root mean square, kurtosis, and spectral entropy from the time-series data using a sliding window algorithm, and reducing the feature dimension to 8 dimensions using a principal component analysis algorithm; calculating the similarity score between the current feature sequence and historical disaster precursor templates based on a dynamic time warping algorithm, and triggering multi-parameter correlation analysis when the score exceeds a preset threshold; constructing a causal relationship network of acceleration, displacement, and strain parameters using a Granger causality test algorithm, identifying leading precursor parameter combinations, and predicting the evolution trend of precursor features using a long short-term memory neural network model.

[0100] In some embodiments, generating geological disaster early warning information based on preset early warning thresholds and geological disaster precursor characteristics includes: dividing the early warning thresholds into three levels: yellow, orange, and red, each corresponding to different combinations of characteristic parameters and durations; constructing a spatial grid model of the monitoring area based on a geographic information system, and generating a spatial distribution heat map of early warning levels using an inverse distance weighted interpolation algorithm; automatically upgrading the early warning level by one level when the same monitoring point triggers an early warning of the same level for three consecutive sampling cycles; and sending early warning information, including structured data such as early warning level, coordinate location, characteristic parameter values, and predicted trends, through a dual-channel BeiDou short message communication module and a 4G network.

[0101] In some embodiments, the method further includes: establishing an online optimization mechanism for intelligent algorithms; when the deviation between the actual disaster event reported by the early warning platform and the prediction result exceeds a preset threshold, automatically extracting the full-cycle monitoring data of the event as a new sample; updating the parameters of the feature extraction model and the correlation analysis model at the edge end through a federated learning algorithm; and using differential privacy technology to protect the privacy of multi-source data during the update process.

[0102] In some embodiments, the method further includes: constructing a multimodal precursor feature verification mechanism, which initiates a preset sensor collaborative verification process when a single parameter triggers an early warning threshold, including: if the acceleration data exceeds the limit, calling BeiDou positioning data to calculate the three-dimensional displacement rate of the monitoring point; if the displacement data exceeds the limit, triggering the high-density scanning mode of the strain sensor, inverting the structural stress state through the stress-strain relationship model, and confirming the effectiveness of the early warning only when at least two types of parameters simultaneously meet the precursor features.

[0103] In some embodiments, the method further includes: introducing a hybrid early warning model that integrates physical and data-driven approaches; constructing a displacement-acceleration coupled state-space model based on the geological disaster dynamics equation; using an extended Kalman filter algorithm to fuse MEMS sensor measured data with physical model predictions; and switching to a pure data-driven model for short-term trend prediction when the residual between the measured value and the model prediction exceeds twice the root mean square error, thereby improving the reliability of early warning under complex geological conditions.

[0104] It should be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the BeiDou / MEMS geological disaster multi-parameter early warning system and its modules described above can be referred to the corresponding content in the various embodiments of the BeiDou / MEMS geological disaster multi-parameter early warning method, and will not be repeated here.

[0105] The aforementioned BeiDou / MEMS multi-parameter early warning method for geological disasters can be implemented as a computer program that can run on the device provided in this application.

[0106] Please see Figure 3 , Figure 3 This is a schematic block diagram of the terminal hardware module provided in an embodiment of this application. The terminal hardware module includes a processor, a memory, and a network interface connected via a device bus, wherein the memory may include a storage medium and internal memory.

[0107] The storage medium can store operating devices and computer programs. The computer program includes program instructions that, when executed, cause the processor to perform any BeiDou / MEMS multi-parameter early warning method for geological disasters.

[0108] The processor provides computing and control capabilities to support the operation of the entire terminal hardware module.

[0109] The internal memory provides an environment for the execution of computer programs in non-volatile storage media. When the computer program is executed by the processor, it enables the processor to execute any BeiDou / MEMS multi-parameter early warning method for geological disasters.

[0110] This network interface is used for network communication, such as sending assigned tasks. Those skilled in the art will understand that... Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the solution of this application and does not constitute a limitation on the terminal to which the solution of this application is applied. The specific terminal hardware module may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0111] It should be understood that the processor can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among these, a general-purpose processor can be a microprocessor or any conventional processor.

[0112] In one embodiment, the processor is configured to run a computer program stored in memory to perform the following steps: The terminal hardware module, which integrates a choke antenna with a ring-shaped metal choke structure and a ceramic substrate composite design, as well as a built-in multi-band feed point and a protective shell, acquires BeiDou positioning data. The terminal hardware module adopts a wide-temperature and fully enclosed structure, and the external interface is an aviation-type connector. The terminal hardware module includes a bottom shell, a waterproof gasket, and a top shell. Using the MEMS sensors mounted on the terminal hardware module, multi-parameter data of the geological disaster monitoring area is collected. The multi-parameter data includes, but is not limited to, acceleration, displacement, strain, and vibration data. The acquired BeiDou positioning data and the collected multi-parameter data are preprocessed to remove noise and outliers. The preprocessed BeiDou positioning data and multi-parameter data are fused and analyzed by intelligent algorithms, including time-series-based feature extraction algorithms and multi-parameter correlation analysis algorithms, to identify precursory features of geological disasters. Based on preset warning thresholds and geological disaster precursor characteristics, geological disaster warning information is generated. The geological disaster warning information includes the warning level and warning area, and is sent to the warning platform and related terminals through the communication module of the terminal hardware module.

[0113] In some embodiments, the acquisition of BeiDou positioning data via a terminal hardware module integrating a choke antenna with a composite design of a ring-shaped metal choke structure and a ceramic substrate, as well as a built-in multi-band feed point and a protective shell, includes: suppressing multipath interference of BeiDou signals through the ring-shaped metal choke structure, and improving the phase center consistency of the antenna by utilizing the dielectric constant stability of the ceramic substrate; simultaneously receiving B1, B2, and B3 band signals of the BeiDou system through the multi-band feed point, and calculating the positioning result based on a multi-frequency data fusion algorithm; the wide-temperature and fully enclosed structure of the terminal hardware module allows the operating temperature range to cover -40℃ to +85℃, and a waterproof and dustproof electrical connection is established with external equipment through the aviation-type connector; the bottom shell, waterproof gasket, and top shell form a sealed cavity through threaded engagement.

[0114] In some embodiments, the use of MEMS sensors mounted on the terminal hardware module to collect multi-parameter data of the geological disaster monitoring area includes: collecting acceleration data at a sampling rate of not less than 100Hz using a triaxial MEMS accelerometer; obtaining sub-millimeter displacement changes using a differential capacitive MEMS displacement meter; monitoring micro-strain on the structural surface using a piezoresistive MEMS strain gauge; and collecting vibration signals in the 0.1Hz-10kHz frequency band using a MEMS piezoelectric transducer. The MEMS sensor integrates a temperature compensation circuit to correct the drift effect of ambient temperature on the collected data in real time, and automatically triggers a high-frequency sampling mode to improve the sampling rate when a sudden change in data is detected.

[0115] In some embodiments, the preprocessing of the acquired BeiDou positioning data and collected multi-parameter data to remove noise and outliers includes: applying a sliding window Kalman filter algorithm to the BeiDou positioning data to iteratively optimize the noise covariance of position, velocity, and time parameters; sequentially performing finite impulse response low-pass filtering and outlier detection based on the 3σ principle on the MEMS sensor data, and replacing outlier points using linear interpolation; establishing a multi-parameter data timestamp alignment mechanism, and performing nanosecond-level calibration of the time stamps of different types of data based on the BeiDou timing signal to eliminate phase deviations introduced by asynchronous sampling.

[0116] In some embodiments, the intelligent algorithm is used to fuse and analyze the preprocessed BeiDou positioning data and multi-parameter data. The intelligent algorithm includes a time-series-based feature extraction algorithm and a multi-parameter correlation analysis algorithm to identify precursor features of geological disasters. This includes: extracting 30-dimensional statistical features such as root mean square, kurtosis, and spectral entropy from the time-series data using a sliding window algorithm, and reducing the feature dimension to 8 dimensions using a principal component analysis algorithm; calculating the similarity score between the current feature sequence and historical disaster precursor templates based on a dynamic time warping algorithm, and triggering multi-parameter correlation analysis when the score exceeds a preset threshold; constructing a causal relationship network of acceleration, displacement, and strain parameters using a Granger causality test algorithm, identifying leading precursor parameter combinations, and predicting the evolution trend of precursor features using a long short-term memory neural network model.

[0117] In some embodiments, generating geological disaster early warning information based on preset early warning thresholds and geological disaster precursor characteristics includes: dividing the early warning thresholds into three levels: yellow, orange, and red, each corresponding to different combinations of characteristic parameters and durations; constructing a spatial grid model of the monitoring area based on a geographic information system, and generating a spatial distribution heat map of early warning levels using an inverse distance weighted interpolation algorithm; automatically upgrading the early warning level by one level when the same monitoring point triggers an early warning of the same level for three consecutive sampling cycles; and sending early warning information, including structured data such as early warning level, coordinate location, characteristic parameter values, and predicted trends, through a dual-channel BeiDou short message communication module and a 4G network.

[0118] In some embodiments, the method further includes: establishing an online optimization mechanism for intelligent algorithms; when the deviation between the actual disaster event reported by the early warning platform and the prediction result exceeds a preset threshold, automatically extracting the full-cycle monitoring data of the event as a new sample; updating the parameters of the feature extraction model and the correlation analysis model at the edge end through a federated learning algorithm; and using differential privacy technology to protect the privacy of multi-source data during the update process.

[0119] In some embodiments, the method further includes: constructing a multimodal precursor feature verification mechanism, which initiates a preset sensor collaborative verification process when a single parameter triggers an early warning threshold, including: if the acceleration data exceeds the limit, calling BeiDou positioning data to calculate the three-dimensional displacement rate of the monitoring point; if the displacement data exceeds the limit, triggering the high-density scanning mode of the strain sensor, inverting the structural stress state through the stress-strain relationship model, and confirming the effectiveness of the early warning only when at least two types of parameters simultaneously meet the precursor features.

[0120] In some embodiments, the method further includes: introducing a hybrid early warning model that integrates physical and data-driven approaches; constructing a displacement-acceleration coupled state-space model based on the geological disaster dynamics equation; using an extended Kalman filter algorithm to fuse MEMS sensor measured data with physical model predictions; and switching to a pure data-driven model for short-term trend prediction when the residual between the measured value and the model prediction exceeds twice the root mean square error, thereby improving the reliability of early warning under complex geological conditions.

[0121] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to implement the steps of the BeiDou / MEMS geological disaster multi-parameter early warning method provided in any embodiment of this application.

[0122] The computer-readable storage medium can be an internal storage unit of the terminal hardware module described in the foregoing embodiments, such as the hard disk or memory of the terminal hardware module. Alternatively, the computer-readable storage medium can be an external storage device of the terminal hardware module, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the terminal hardware module.

[0123] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A Beidou / MEMS geological disaster multi-parameter early warning method, characterized in that, include: The terminal hardware module, which integrates a choke antenna with a ring-shaped metal choke structure and a ceramic substrate composite design, as well as a built-in multi-band feed point and a protective shell, acquires BeiDou positioning data. The terminal hardware module adopts a wide-temperature and fully enclosed structure, and the external interface is an aviation-type connector. The terminal hardware module includes a bottom shell, a waterproof gasket, and a top shell. Using the MEMS sensors mounted on the terminal hardware module, multi-parameter data of the geological disaster monitoring area is collected. The multi-parameter data includes, but is not limited to, acceleration, displacement, strain, and vibration data. The acquired BeiDou positioning data and the collected multi-parameter data are preprocessed to remove noise and outliers. The preprocessed BeiDou positioning data and multi-parameter data are fused and analyzed by intelligent algorithms, including time-series-based feature extraction algorithms and multi-parameter correlation analysis algorithms, to identify precursory features of geological disasters. Based on preset warning thresholds and geological disaster precursor characteristics, geological disaster warning information is generated. The geological disaster warning information includes the warning level and warning area, and is sent to the warning platform and related terminals through the communication module of the terminal hardware module.

2. The Beidou / MEMS geological disaster multi-parameter early warning method according to claim 1, characterized in that, The terminal hardware module, which integrates a choke coil antenna with a composite design of a ring-shaped metal choke structure and a ceramic substrate, as well as a built-in multi-band feed point and protective shell, acquires BeiDou positioning data, including: The annular metal choke structure suppresses multipath interference of BeiDou signals, and the dielectric constant stability of the ceramic substrate improves the phase center consistency of the antenna. The multi-band feed point simultaneously receives signals from the B1, B2, and B3 bands of the BeiDou system, and the positioning result is calculated based on the multi-frequency data fusion algorithm. The wide-temperature and fully enclosed structure of the terminal hardware module allows the operating temperature range to cover -40℃ to +85℃. The aviation-type connector establishes a waterproof and dustproof electrical connection with external equipment. The bottom shell, waterproof gasket, and top shell form a sealed cavity through threaded engagement.

3. The method of claim 1, wherein, The method of using the MEMS sensor mounted on the terminal hardware module to collect multi-parameter data of the geological disaster monitoring area includes: Acceleration data is acquired using a triaxial MEMS accelerometer at a sampling rate of no less than 100Hz. Submillimeter-level displacement changes are obtained using a differential capacitive MEMS displacement meter. Micro-strain on the structural surface is monitored using a piezoresistive MEMS strain gauge. Vibration signals in the 0.1Hz-10kHz frequency band are acquired using a MEMS piezoelectric transducer. The MEMS sensor integrates a temperature compensation circuit to correct the drift effect of ambient temperature on the acquired data in real time. When a sudden change in data is detected, a high-frequency sampling mode is automatically triggered to improve the sampling rate.

4. The method of claim 1, wherein, The preprocessing of the acquired BeiDou positioning data and collected multi-parameter data to remove noise and outliers includes: A sliding window Kalman filter algorithm is used to iteratively optimize the noise covariance of position, velocity, and time parameters for BeiDou positioning data; The MEMS sensor data is sequentially subjected to finite impulse response low-pass filtering and outlier detection based on the 3σ principle, and outlier points are replaced by linear interpolation. A multi-parameter data timestamp alignment mechanism was established, and the time tags of different types of data were calibrated at the nanosecond level based on the BeiDou time signal to eliminate the phase deviation introduced by asynchronous sampling.

5. The method of claim 1, wherein, The process involves fusing and analyzing preprocessed BeiDou positioning data and multi-parameter data using intelligent algorithms. These intelligent algorithms include time-series-based feature extraction algorithms and multi-parameter correlation analysis algorithms to identify precursory features of geological disasters, including: The sliding window algorithm is used to extract 30-dimensional statistical features such as root mean square, kurtosis, and spectral entropy from time series data. The principal component analysis algorithm is then used to reduce the feature dimension to 8 dimensions. The similarity score between the current feature sequence and the historical disaster precursor template is calculated based on the dynamic time warping algorithm. When the score exceeds the preset threshold, multi-parameter correlation analysis is triggered. A causal relationship network of acceleration, displacement, and strain parameters is constructed using the Granger causality test algorithm to identify leading precursor parameter combinations. The evolution trend of precursor features is then predicted by combining the long short-term memory neural network model.

6. The method of claim 1, wherein, The process of generating geological disaster early warning information based on preset early warning thresholds and geological disaster precursor characteristics includes: The warning thresholds are divided into three levels: yellow, orange, and red, each corresponding to different combinations of characteristic parameters and durations. A spatial grid model of the monitoring area is constructed based on a geographic information system, and a heat map of the spatial distribution of early warning levels is generated by an inverse distance weighted interpolation algorithm. When the same monitoring point triggers the same level of warning for 3 consecutive sampling cycles, the warning level is automatically upgraded by one level; the warning information is sent through the Beidou short message communication module and the 4G network dual channels, including structured data such as warning level, coordinate location, characteristic parameter values ​​and predicted trend.

7. The method according to claim 1, characterized in that, The method further includes: Establish an online optimization mechanism for intelligent algorithms. When the deviation between the actual disaster event reported by the early warning platform and the prediction result exceeds a preset threshold, the full-cycle monitoring data of the event is automatically extracted as a new sample. The parameters of the feature extraction model and correlation analysis model at the edge are updated through federated learning algorithms. Differential privacy technology is used to protect the privacy of multi-source data during the update process.

8. The method according to claim 1, characterized in that, The method further includes: A multimodal precursor feature verification mechanism is constructed. When a single parameter triggers the warning threshold, a pre-set collaborative verification process of sensors is initiated, including: if the acceleration data exceeds the limit, the three-dimensional displacement rate of the monitoring point is calculated by calling the Beidou positioning data; if the displacement data exceeds the limit, the high-density scanning mode of the strain sensor is triggered, and the stress state of the structure is inverted through the stress-strain relationship model. The effectiveness of the warning is confirmed only when at least two types of parameters simultaneously meet the precursor features.

9. The method according to claim 1, characterized in that, The method further includes: A hybrid early warning model integrating physical and data-driven approaches is introduced. A displacement-acceleration coupled state-space model is constructed based on the geological disaster dynamics equation. The extended Kalman filter algorithm is used to fuse the measured data from MEMS sensors with the predicted values ​​from the physical model. When the residual between the measured value and the predicted value exceeds twice the root mean square error, the model is switched to a pure data-driven model for short-term trend prediction, thereby improving the reliability of early warning under complex geological conditions.

10. A terminal hardware module, characterized in that, The terminal hardware module includes a memory and a processor; The memory is used to store computer programs; The processor is configured to execute the computer program and, in executing the computer program, implement the method as described in any one of claims 1 to 9.

Images