Geological disaster real-time monitoring system

Abstract

The invention relates to the technical field of geological disaster real-time monitoring, and discloses a geological disaster real-time monitoring system comprising a sensor module used for collecting geological deformation data; the data processing module is used for carrying out filtering processing and feature extraction on the collected geological deformation data; the multi-sensor data fusion module is used for fusing geological deformation data acquired by different sensors through a weighted least square method; the state estimation module is used for performing state estimation and dynamic updating on the fused geological deformation data based on a recursive optimization algorithm; and the communication module is used for sending the geological deformation data after state estimation and dynamic updating to a remote server in a wireless mode. According to the invention, the real-time acquisition of the multi-dimensional deformation data of the geologic body is realized through the combined sensor module integrating the accelerometer, the tilt angle sensor and the strain sensor. The technical effect of stably outputting the monitoring signal under the multi-scene and multi-medium conditions is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of real-time monitoring of geological disasters, and in particular to a real-time monitoring system for geological disasters. Background Art

[0002] Geological disasters, especially sudden geological events such as landslides, ground fissures, and foundation subsidence, are widely distributed in key areas such as mountainous areas, along highways, around water conservancy facilities, and in mining areas, posing a direct threat to people's lives and property. With the advancement of urbanization and the expansion of engineering construction, the need for identifying and dynamically monitoring early signs of disasters is becoming increasingly urgent. Geological disasters generally develop through a combination of slow evolution and sudden, sudden events, placing high demands on the real-time, accuracy, and stability of monitoring systems.

[0003] Most existing geological hazard monitoring solutions rely on the deployment of a single sensor type, such as strain gauges, inclinometers, or single-axis accelerometers, supplemented by data acquisition terminals to collect physical quantities. With this technology, sensors are typically deployed point-to-point, with collected data transmitted to on-site terminals via cables or local wireless transmission for processing. Some systems utilize solar power and intermittent GPRS transmission to enable remote data upload.

[0004] However, the existing geological disaster monitoring technology, with its single physical quantity perception means, is unable to fully capture the internal correlation when facing geological disasters caused by the coupling of multiple factors, and its ability to judge the overall complex geological response is limited. In addition, the filtering and processing algorithms used by traditional systems are relatively static, and it is difficult to flexibly respond to changes in on-site noise levels or signal fluctuations. As the monitoring time goes by, errors are prone to accumulation and sudden events may be missed. In terms of multi-source information processing, common data fusion methods mainly rely on linear superposition or artificially set fixed weights, which are difficult to effectively reflect the quality differences and correlations between sensors, and the accuracy and adaptability of the fusion results are affected to a certain extent. Therefore, the present invention provides a real-time monitoring system for geological disasters to address the shortcomings of the existing technology. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention provides a real-time monitoring system for geological disasters, which solves the problem that the existing geological disaster monitoring technology, whose single physical quantity perception method, is difficult to fully capture the internal correlation when facing geological disasters caused by the coupling of multiple factors, and has limited ability to comprehensively judge complex geological responses.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: A real-time geological disaster monitoring system, comprising: A sensor module, used for collecting geological deformation data, wherein the sensor module includes an accelerometer, an inclination sensor and a strain sensor; Data processing module, used for filtering and feature extraction of collected geological deformation data; The multi-sensor data fusion module fuses the geological deformation data collected by different sensors through the weighted least squares method; the state estimation module is used to perform state estimation and dynamic update of the fused geological deformation data based on the recursive optimization algorithm; the communication module is used to send the geological deformation data after state estimation and dynamic update to the remote server via wireless.

[0007] Preferably, the accelerometer is used to collect acceleration change data of the geological body, the inclination sensor is used to measure the inclination angle of the geological body, and the strain sensor is used to measure the strain of the geological body.

[0008] Preferably, the filtering process includes: The Kalman filter algorithm is used to estimate the true value of the data through recursive optimization, eliminate noise interference, and improve monitoring accuracy; the feature extraction module is used to extract features from sensor data that are helpful for geological disaster judgment, such as the dynamic change trends of acceleration, inclination changes, and strain changes.

[0009] Preferably, the multi-sensor data fusion module includes: A weighted least squares processing unit for fusing data from different sensors by minimizing weighted errors; Weight calculation unit, used to dynamically calculate the weight of each sensor data; The data fusion unit is used to output the fused data.

[0010] Preferably, the state estimation module includes: A recursive optimization unit is used to recursively update the state of the current data and the previous data to perform dynamic state estimation; a Kalman gain calculation unit is used to calculate the Kalman gain and optimize the estimation process; The state update unit is used to update the system state based on the recursive algorithm of the Kalman filter.

[0011] Preferably, the communication module is a low-power wide area network communication module selected from NB-IoT, LoRa or 4G modules, which is suitable for realizing remote transmission of data in different communication environments.

[0012] Preferably, the Kalman filter algorithm: K(t)=P(t|t-1)H T [HP(t|t-1)H T +R] -1 ; x(t∣t)=x(t∣t-1)+K(t)(z(t)-Hx(t∣t-1)); Where K(t) is the Kalman gain; P(t|t-1) is the prediction covariance matrix; H is the measurement matrix; z(t) is the observation value; R is the measurement noise covariance; x(t|t) is the state estimate at the current moment; x(t|t-1) is the state estimate at the previous moment; and T is the matrix transpose operation.

[0013] Preferably, the objective function of the weighted least squares method is: Where J(W) represents the loss function of the weighted least squares method; x(t) represents the state of the system at time t; is the predicted value based on the physical model; W represents the sum of the expected values ​​of the squared errors; i is the weight coefficient of the i-th sensor, λ is the regularization parameter; n is the number of sensors involved in the fusion.

[0014] Preferably, the formula of the recursive optimization algorithm is: x(t+1)=Ax(t)+Bu(t)+w(t); Among them, x(t+1) represents the state of the system at time t+1; x(t) represents the state of the system at time t; A is the state transfer matrix; B is the control matrix; u(t) is the control input; and w(t) is the process noise.

[0015] A method for real-time monitoring of geological disasters is also provided, comprising the following steps: Collect geological deformation data, including acceleration, inclination and strain data; Filter and extract features from the collected geological deformation data to remove noise and extract key features; The collected geological deformation data are fused and the weighted least square method is used to combine the weights of each sensor to generate optimized data; The state of the fused data is estimated and dynamically updated based on the Kalman filter recursive algorithm; The state estimation and dynamically updated data are transmitted wirelessly to a remote server for further analysis and early warning.

[0016] The present invention provides a real-time geological disaster monitoring system. It has the following beneficial effects: 1. This invention utilizes a composite sensor module integrating an accelerometer, an inclination sensor, and a strain sensor to achieve real-time acquisition of multi-dimensional deformation data of geological bodies. This achieves the technical effect of stably outputting monitoring signals in multiple scenarios and under multiple media conditions. Compared to existing monitoring methods that rely on a single sensor structure, this overcomes the problem of slow response to specific deformation types and improves adaptability to complex geological processes.

[0017] 2. This invention integrates a fusion weighted least squares algorithm and a dynamic weight allocation mechanism to uniformly process multi-source heterogeneous sensor data. This significantly improves the stability and robustness of geological state estimation after data fusion. Compared with existing fusion strategies based on static averaging or fixed weighting, this approach addresses the issue of low fusion accuracy when sensor data reliability fluctuates.

[0018] 3. This invention uses a recursive state estimation model based on Kalman filtering to dynamically track geological conditions. This enables continuous monitoring and prediction of time-varying geological events. Existing solutions often rely on one-time data judgment or static evaluation models, which suffer from slow response to sudden events and large estimation jumps.

[0019] 4. This invention utilizes low-power, wide-area wireless transmission technology, primarily based on NB-IoT modules, combined with a modular replacement strategy. This enables flexible data reporting in diverse geographic communication environments. Compared to traditional, high-power or short-range communication methods based on GPRS or Wi-Fi, this approach avoids the weaknesses of high energy consumption and susceptibility to interruption, while also supporting long-term field deployments. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is a system architecture diagram of the present invention; Figure 2 The figure is a flow chart of the method steps of the present invention. DETAILED DESCRIPTION

[0021] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the present specification. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0022] Please see the attached Figure 1 The embodiment of the present invention provides a real-time geological disaster monitoring system, including: A sensor module, used for collecting geological deformation data, wherein the sensor module includes an accelerometer, an inclination sensor and a strain sensor; Data processing module, used for filtering and feature extraction of collected geological deformation data; The multi-sensor data fusion module fuses the geological deformation data collected by different sensors through the weighted least squares method; the state estimation module is used to perform state estimation and dynamic update of the fused geological deformation data based on the recursive optimization algorithm; the communication module is used to send the geological deformation data after state estimation and dynamic update to the remote server via wireless.

[0023] As for the sensor module, in this embodiment, the sensor module integrates three types of sensor units, namely, an accelerometer, an inclination sensor, and a strain sensor, to form a three-dimensional perception capability of the motion state of the geological body.

[0024] In some embodiments, the accelerometer is a three-axis MEMS-based acceleration sensor, designed to collect acceleration information along the X, Y, and Z axes of the geological body. Specifically, the sensor's range is set to ±16g, meeting the dual requirements of detecting both gradual geological changes and sudden events. Its output resolution can reach 0.001g, effectively capturing minute vibrations and subtle movements.

[0025] Alternatively, the inclination sensor utilizes a digital dual-axis tilt unit, capable of real-time measurement of the horizontal and vertical tilt angles of geological bodies with an accuracy of up to 0.1°. This makes it suitable for discriminant analysis of slowly deforming structures such as landslides, slopes, and piles. In one possible implementation, the sensor uses a high-sampling-rate digital output, facilitating subsequent fitting analysis of angle change trends.

[0026] In other embodiments, the strain sensor uses a metal foil strain gauge, which is adhesively attached to the surface of a geological body or at key structural nodes to respond to changes in stress in real time. This sensor has a sensitivity coefficient of 2.0 and can convert tiny strains into stable electrical signals, which can then be used to derive the stress distribution and evolution trends of the geological body.

[0027] Specifically, the sensors within the sensor module are connected to the microcontroller's data acquisition port via an analog front-end circuit. Sensor signals are digitized via an analog-to-digital converter (ADC). Acceleration and inclination sensors typically communicate with the microcontroller using the SPI bus. Strain gauges are connected to the microcontroller's differential input port via a Wheatstone bridge and amplifier circuit.

[0028] In one implementation, data collected by various sensors undergoes preliminary time synchronization calibration and basic outlier removal before being sent to the data processing module. Taking strain signals as an example, the system sets a baseline zero point and dynamic thresholds, and uses a sliding window range method for initial judgment, thereby reducing the risk of false triggering.

[0029] In some specific scenarios, the sensor module can customize the sensor type and installation method based on the material properties of different geological bodies, burial environment conditions, etc. For example, in rock crack monitoring, strain gauges are preferably placed in areas prone to cracking, while accelerometers are used in areas affected by falling rocks to form a complementary response mechanism.

[0030] To further improve the consistency and fusion accuracy of multi-sensor data, in some embodiments, the raw data output by the sensor module will be given a timestamp and spatial identifier as an important basis for subsequent weight adjustment and fusion calculation.

[0031] Furthermore, the sensor module's overall packaging design prioritizes the use of pressure-resistant, waterproof, and corrosion-resistant materials to meet the demands of long-term outdoor operation. A low-power wake-up pin is reserved for the power supply interface, allowing it to coordinate with the main control chip to enter sleep / activate mode, thereby extending the system's overall battery life.

[0032] In terms of expansion, the sensor module can be pre-integrated with an abnormal event trigger mechanism. For example, when the acceleration change suddenly increases or the inclination angle suddenly changes and reaches the set threshold, data reporting acceleration or auxiliary event logging is triggered.

[0033] Regarding the data processing module in this embodiment, the sensory data from geological environments is often accompanied by high noise, nonlinearity, and irregular fluctuations. Without effective processing, this can easily lead to failures in subsequent fusion, judgment, and early warning logic. Therefore, the configuration of this module needs to have a certain degree of dynamic response capability and robustness to adapt to changes in complex field environments.

[0034] In this embodiment, the data processing module is mainly integrated on a microcontroller unit (MCU), preferably an STM32 series microcontroller with a main frequency of 72MHz, multi-channel ADC, DMA support and rich serial communication interfaces, which is suitable for implementing embedded algorithm computing tasks and multi-sensor signal access.

[0035] In some embodiments, the data processing module first performs a pre-filtering operation on the sampled data to reduce the impact of high-frequency noise and occasional sudden signal mutations. Alternatively, the pre-processing module introduces an algorithm based on a sliding mean or median filter, which is suitable for initial noise suppression and data smoothing.

[0036] Specifically, in more complex or real-time scenarios, the module uses the Kalman filter algorithm to recursively estimate and dynamically correct sensor data. The filtering process is based on the following standard Kalman update formula: K(t) = P(t|t-1)H T [HP(t|t-1)H T +R] -1 ; x(t)=x(t-1)+K(t)(z(t)-Hx(t∣t-1)); Where K(t) is the Kalman gain; P(t|t-1) is the prediction covariance matrix; H is the measurement matrix; z(t) is the observation; R is the measurement noise covariance; x(t|t) is the current state estimate; x(t|t-1) is the previous state estimate; and T is the matrix transpose. This Kalman filter formula uses the sensor observation z(t) and the system's prediction Hx(t|t-1) to perform a recursive estimation using a weighted update method, thereby optimizing the system's state prediction. This process can effectively remove noise in complex environments, improving system accuracy and stability.

[0037] In one possible implementation, the system sets corresponding measurement models and covariance matrix parameters for different sensor signals, which are obtained through experimental calibration and training with historical data to ensure the dynamic adaptability of state estimation.

[0038] In other embodiments, in addition to performing filtering operations, the data processing module also includes a feature extraction unit for extracting structured information from time series data that is useful for identifying geological hazards. For example, for acceleration signals, the mean, extreme values, fluctuation amplitude, and frequency components can be extracted; for inclination signals, the first-order derivative and slope change rate can be calculated; and for strain signals, the strain rate change breakpoints can be identified.

[0039] Generally speaking, the feature extraction process adopts the sliding window segmentation method to calculate statistics and dynamic change indicators by time period to provide quantitative input for subsequent fusion and modeling.

[0040] As an option, to improve response efficiency, some feature calculations can be completed on the edge side, and only the results are uploaded to the upper-level analysis system to reduce wireless bandwidth pressure.

[0041] In a possible expansion method, the data processing module can also set a trigger threshold logic, that is, when the change amplitude of the data in a certain dimension exceeds the set threshold, the high-frequency sampling mode is started and the relevant information is reported first.

[0042] In addition, to meet the requirements of low-power design, the microcontroller supports an interrupt wake-up mechanism, enters a low-power state during idle time, and resumes normal operation when the sensor module changes beyond the limit or periodic sampling is triggered.

[0043] Regarding the multi-sensor data fusion module, in this embodiment, after completing the filtering processing and preliminary feature extraction of the geological deformation data, the system needs to further unify the modeling and fusion of the data collected by the multi-source sensors. Generally speaking, the data of a single sensor is greatly affected by its range limitation, noise disturbance or environmental drift, and cannot accurately and comprehensively reflect the overall state of the geological body. Therefore, setting up a stable and efficient data fusion module has become a key component for achieving high-reliability judgment in the real-time geological disaster monitoring system. This module aims to achieve a weighted unified estimation of geological deformation parameters by fusing the output results of multiple sensors, and provide consistent input for subsequent status updates.

[0044] In this embodiment, the wireless data fusion module uses weighted least squares as the core fusion algorithm to achieve error suppression and weighted integration of data from different sources. Specifically, the fusion process aims to minimize the weighted residual sum of squares between the observed data and the target state, which is expressed as follows: Where J(W) represents the loss function of the weighted least squares method; x(t) represents the state of the system at time t; is the predicted value based on the physical model, W represents the sum of the expected values ​​of the squared errors; i is the weight coefficient of the i-th sensor, λ is the regularization parameter; n is the number of sensors involved in the fusion.

[0045] In general, the weight w i The calculation of needs to be combined with historical error variance, current signal-to-noise ratio, and environmental factors for comprehensive assignment. As an option, in actual implementation, the weight can be dynamically determined by the following expression: Among them, w i is the weight of the i-th sensor; is the measurement variance of the i-th sensor; represents the sum of the inverse variances of all sensors; j is the summation index. This formula ensures that sensors with high measurement accuracy occupy a larger proportion in the fusion.

[0046] In some embodiments, the system models the data of each sensor individually and calculates its estimated error variance, which is then fed into a weighted calculation unit. This error can be obtained by measuring the variance within a sliding window or by averaging historical backtest errors.

[0047] Specifically, the module includes a weight estimation unit, a fusion calculation unit, and a fusion result verification unit. The weight estimation unit outputs dynamic weight values ​​based on real-time input and historical reliability indicators; the fusion calculation unit performs the optimization solution of the least squares function described above; and the fusion result verification unit performs robustness verification on the estimated values, such as checking whether they exceed physical boundaries or produce discontinuous jumps.

[0048] In one possible implementation, the fusion module also pre-configures a multi-channel fusion pathway, with separate fusion sub-channels for acceleration, inclination, and strain signals. Before the fusion results are output, the cross-modal consistency of each channel is checked. If the difference exceeds a threshold, the system triggers a fusion anomaly flag.

[0049] In some embodiments, the module has an abnormal rejection function. When a sensor outputs an abnormal mutation value or a static offset exceeding a preset range within a specific period of time, its corresponding weight will be forced to zero or reduced to the lower limit of the weight to avoid affecting the overall fusion.

[0050] As an extended technical solution, in special application scenarios such as slope engineering or high-fill sections, the fusion module can incorporate physical or empirical models of geological bodies to participate in weight adjustment. For example, sensor response priorities can be adjusted based on stratum type or vegetation cover to enhance the structural adaptability of geological perception.

[0051] In addition, to ensure data processing efficiency in low-power operation mode, the fusion module can be set to execute the fusion logic in periodic triggering or event-driven mode, activating the operation process only when the sensor output changes trend or reaches the trigger threshold.

[0052] Regarding the state estimation module, in this embodiment, after completing the fusion processing of multi-source data, the system enters the dynamic state estimation and recursive update phase to achieve time-series tracking and trend identification of the geological body's motion state. Generally speaking, the formation of geological hazards is characterized by the coexistence of continuous evolution and sudden uncertainty. Relying solely on static or single-point estimates makes it difficult to form an effective early warning mechanism. Therefore, introducing a state estimation module based on the fusion results and recursively predicting and updating the target state in real time based on a time series model becomes a key step in ensuring the continuity of system monitoring and the effectiveness of response.

[0053] In this embodiment, the state estimation module uses a linear state-space model as its foundation and incorporates a recursive optimization algorithm to construct a dynamic update mechanism. This module performs state modeling and predictive correction on the fused geological deformation data. By incorporating state transition equations and observation models, this module performs a filtered estimate of the actual state of the geological body, suppressing the influence of observation noise.

[0054] In some embodiments, state estimation is modeled using the following state transition equation: x(t+1)=Ax(t)+Bu(t)+w(t); Where x(t+1) represents the state of the system at time t+1; x(t) represents the state variable of the system at time t; A is the state transition matrix, which represents the evolution of the system from the previous moment to the current state; B is the control matrix; u(t) is the control input (which can be regarded as an external disturbance or a known intervention in this application); w(t) is the process noise, which satisfies the Gaussian distribution Q is the process noise covariance matrix, which is used to characterize the system modeling error or unobserved disturbance.

[0055] As an option, combined with the Kalman recursion algorithm, the system further updates the state using the observation equation expressed as follows: z(t)=Hx(t)+v(t); Among them, x(t) represents the state variable of the system at time t; z(t) is the observation value at time t; H is the observation matrix, which represents the mapping relationship between the observation quantity and the state quantity; is the observation noise; R is the observation noise covariance matrix, which is used to represent measurement error or perception uncertainty.

[0056] In other embodiments, to enhance the system's ability to respond to sudden events, the state estimation module introduces a residual determination mechanism to monitor the current observation and prediction residuals. It is considered as a potential abnormal event and the high-frequency update mode is activated, in which is the predicted value of the state estimation at the previous moment; ∈ is the preset threshold, which is set according to the actual monitoring needs.

[0057] Typically, the state estimation module runs as part of a microcontroller, sharing the fusion results as input with the data fusion module and implementing continuous updates using a loop structure. If power resources permit, a double buffering structure can be used to store historical estimated states for backtracking analysis or trend extrapolation.

[0058] As an extension solution, the state estimation module can introduce a stratigraphic type weight adjustment mechanism in combination with geological expertise. For example, in soft soil areas, red layer areas, or under pile carriers, the state prediction sensitivity and error amplification factor can be adjusted respectively to better fit the geological evolution characteristics.

[0059] In addition, to meet the needs of long-term field operation of the system, the module is designed to support a hierarchical power consumption control strategy. When the state change rate is lower than the threshold, the sampling rate and update frequency are automatically reduced to reduce computing load and energy consumption.

[0060] Regarding the communication module, in this embodiment, after the system completes the estimation and dynamic update of the geological deformation state, the relevant information needs to be quickly and stably transmitted to the remote server to support the subsequent data analysis and disaster warning process. Generally speaking, since geological monitoring points are mostly distributed in areas with complex communication conditions such as mountainous areas, forested areas, or construction slopes, traditional communication methods often have problems such as high power consumption, weak signals, and difficult deployment. Therefore, setting up a communication module in the system to achieve low-power, high-reliability remote data transmission is the key to ensuring the effective linkage of the entire monitoring system.

[0061] In this embodiment, the communication module uses a wireless communication unit based on the Low Power Wide Area Network (LPWAN) architecture, with NB-IoT (Narrowband Internet of Things) as the primary implementation. Specifically, the module connects to the microcontroller unit via a UART serial interface, receives the structured data packets output by the state estimation module, and reports them to a remote server via a cellular network protocol.

[0062] In some embodiments, the communication module has a built-in data packaging mechanism to compress and label the original transmission data structure so that it has a timestamp, node number, status code and exception identification field to support server-side decoding and traceability management.

[0063] As an option, the NB-IoT module supports multi-band adaptive communication and has a built-in TCP / UDP protocol stack. In terms of transmission mode, the system can switch between two modes: event-triggered reporting and scheduled reporting. The specific reporting interval is T r Can be preset based on on-site deployment strategies or modified remotely by central commands: T r =T0+ΔT·f(x); Among them, T0 is the basic reporting period; ΔT is the unit of change; f(x) is the state factor function, which reflects the stability score of the current geological state.

[0064] In general, in order to achieve power consumption control and reasonable use of channel resources, the communication module introduces an average power consumption control model, and its single transmission power consumption E tx It can be expressed as: E tx =V·I tx ·t tx ; Where, V is the supply voltage; I tx is the emission current; t tx The duration of data packet transmission.

[0065] In one possible implementation, the module adopts a double buffer structure. When communication fails or the signal is weak, the data is temporarily stored in the local buffer area and automatically resent after the channel is restored to ensure data integrity.

[0066] In some embodiments, to enhance communication adaptability, the system also supports replacing the communication module with a LoRa or 4G module. In areas with complex communication environments or insufficient signal coverage, the LoRa module is preferred for multi-hop relay and extended signal coverage. In towns, built-up areas, or areas with a high density of base stations, the 4G module can be used to improve bandwidth and timeliness.

[0067] Specifically, when replacing a communication module, the microcontroller has a built-in communication protocol switching program that automatically identifies the communication module type and loads the corresponding drive logic to ensure system compatibility.

[0068] In expanded applications, the communication module also supports short-range communication with local gateway devices (such as edge servers or mobile relays). For example, in drone-assisted inspection scenarios, the communication module can briefly transmit data to inspection drones via Wi-Fi or Bluetooth interfaces, enabling periodic upload of data from offline areas.

[0069] In addition, to deal with risks such as electromagnetic interference and signal blocking in field operating environments, the module antenna structure adopts an omnidirectional ceramic antenna or FPC flexible antenna design, and uses a metal shielding cover to isolate the MCU's high-frequency interference to improve link stability.

[0070] The real-time geological disaster monitoring method described below and the real-time geological disaster monitoring system described above can be referenced to each other.

[0071] Please see the attached Figure 2 The present invention also provides a method for real-time monitoring of geological disasters, comprising the following steps: S1. Collect geological deformation data, including acceleration, inclination and strain data; S2. Filtering and feature extraction are performed on the collected geological deformation data to remove noise and extract key features; S3, fuse the collected geological deformation data, use the weighted least squares method to combine the weights of each sensor, and generate optimized data; S4, perform state estimation and dynamic update on the fused data based on the Kalman filter recursive algorithm; S5. Transmit the state estimation and dynamically updated data to a remote server via wireless means for further analysis and early warning.

[0072] For step S1, real-time data collection is performed on the target area by deploying multiple types of sensors (including accelerometers, inclination sensors, and strain sensors) on the surface of the geological body. The collected data covers key physical parameters such as three-dimensional acceleration, angle changes, and stress and strain of the geological body.

[0073] Generally, each sensor is connected to the main control chip through a corresponding signal interface (such as SPI, analog channel) and continuously obtains geological information at a set sampling frequency.

[0074] In step S2, the raw data is initially filtered to suppress environmental interference and measurement noise. Alternatively, a Kalman filter can be introduced in this step to dynamically adjust the estimation accuracy using the difference between the current observation and the previous state estimate.

[0075] The system extracts statistical or dynamic features such as maximum value, variance, slope, and rate of change from each channel data to support fusion link modeling.

[0076] In step S3, a weighted least squares method is used to construct an objective function for minimizing the fusion error. The mathematical meaning of this objective function is: the deviations between the observed values ​​measured by each sensor and the target estimate are squared and summed according to a preset weight factor. A regularization term for the weight coefficient is introduced into the objective function to prevent excessive weight bias towards a single sensor and avoid model overfitting.

[0077] In step S4, the system's prediction model is used to predict the current system state based on the state estimate and state transition relationship at the previous moment. Then, by introducing the Kalman gain, the predicted value is corrected based on the residual between the current observation and the predicted value, thereby obtaining the optimal state estimate at the current moment.

[0078] In step S5, the state estimation results are uploaded to a cloud server in a standardized data packet format via a wireless communication module (preferably NB-IoT). This server can further perform data aggregation, risk assessment, and disaster warning processing. The communication process supports breakpoint resuming and power consumption optimization mechanisms to ensure data transmission stability in complex field environments.

[0079] The method of this embodiment can be used to execute the above system embodiment. Its principles and technical effects are similar and will not be described in detail here.

[0080] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A real-time geological disaster monitoring system, characterized in that: include: A sensor module, used for collecting geological deformation data, wherein the sensor module includes an accelerometer, an inclination sensor and a strain sensor; Data processing module, used for filtering and feature extraction of collected geological deformation data; Multi-sensor data fusion module, which fuses geological deformation data collected by different sensors through weighted least squares method; The state estimation module is used to perform state estimation and dynamic update on the fused geological deformation data based on a recursive optimization algorithm; The communication module is used to send the state estimation and dynamically updated geological deformation data to the remote server via wireless means.

2. A real-time geological disaster monitoring system according to claim 1, characterized in that: The accelerometer is used to collect acceleration change data of the geological body, the inclination sensor is used to measure the inclination angle of the geological body, and the strain sensor is used to measure the strain of the geological body.

3. A real-time geological disaster monitoring system according to claim 1, characterized in that: The filtering process includes: Kalman filter algorithm, which is used to estimate the true value of data through recursive optimization, eliminate noise interference, and improve monitoring accuracy; The feature extraction module is used to extract features from sensor data that are helpful for geological disaster judgment, such as the dynamic change trends of acceleration, inclination change and strain change.

4. A real-time geological disaster monitoring system according to claim 1, characterized in that: The multi-sensor data fusion module includes: A weighted least squares processing unit for fusing data from different sensors by minimizing weighted errors; Weight calculation unit, used to dynamically calculate the weight of each sensor data; The data fusion unit is used to output the fused data.

5. A real-time geological disaster monitoring system according to claim 1, characterized in that: The state estimation module includes: A recursive optimization unit for recursively updating the state of the current data and the previous data to perform dynamic state estimation; A Kalman gain calculation unit, used to calculate the Kalman gain and optimize the estimation process; The state update unit is used to update the system state based on the recursive algorithm of the Kalman filter.

6. A real-time geological disaster monitoring system according to claim 1, characterized in that: The communication module is a low-power wide area network communication module selected from NB-IoT, LoRa or 4G modules, which is suitable for realizing remote data transmission in different communication environments.

7. A real-time geological disaster monitoring system according to claim 3, characterized in that: The Kalman filter algorithm: K(t) = P(t|t-1)H T [HP(t|t-1)H T +R] -1 ; x(t∣t)=x(t∣t-1)+K(t)(z(t)-Hx(t∣t-1)); Where K(t) is the Kalman gain; P(t|t-1) is the prediction covariance matrix; H is the measurement matrix; z(t) is the observation value; R is the measurement noise covariance; x(t|t) is the state estimate at the current moment; x(t|t-1) is the state estimate at the previous moment; and T is the matrix transpose operation.

8. A real-time geological disaster monitoring system according to claim 4, characterized in that: The objective function of the weighted least squares method is: Where J(W) represents the loss function of the weighted least squares method; x(t) represents the state of the system at time t; is the predicted value based on the physical model; W represents the sum of the expected values ​​of the squared errors; i is the weight coefficient of the i-th sensor, λ is the regularization parameter; n is the number of sensors involved in fusion.

9. A real-time geological disaster monitoring system according to claim 5, characterized in that: The formula of the recursive optimization algorithm is: x(t+1)=Ax(t)+Bu(t)+w(t); Among them, x(t+1) represents the state of the system at time t+1; x(t) represents the state of the system at time t; A is the state transfer matrix; B is the control matrix; u(t) is the control input; and w(t) is the process noise.

10. A method for real-time monitoring of geological disasters, applied to a real-time monitoring system for geological disasters according to any one of claims 1 to 9, characterized in that: The following steps are involved: Collect geological deformation data, including acceleration, inclination and strain data; Filter and extract features from the collected geological deformation data to remove noise and extract key features; The collected geological deformation data are fused and the weighted least square method is used to combine the weights of each sensor to generate optimized data; The state of the fused data is estimated and dynamically updated based on the Kalman filter recursive algorithm; The state estimation and dynamically updated data are transmitted wirelessly to a remote server for further analysis and early warning.

Images