Tower footing geological landslide monitoring and early warning system and method

Abstract

The invention relates to the technical field of geological landslide monitoring, in particular to a tower footing geological landslide monitoring and early warning system and method. The method comprises the following steps: arranging an inclination angle sensor, a Beidou-GNSS module, a soil moisture content probe and a micro-seismic accelerometer on a tower footing and the periphery of the tower footing, and collecting the inclination amount, three-dimensional displacement, soil moisture content and vibration response information of the tower footing to form a monitoring original data set; timestamp unification, WGS-84 projection coordinate system conversion and multi-frequency vibration elimination are carried out on the monitored original data set, and an air-ground integrated displacement profile is constructed; the whole process realizes high integration and automation in the aspects of data acquisition, processing, fusion and transmission, so that the timeliness, accuracy and data integrity of landslide risk monitoring are ensured, and scientific decision support is provided for disaster prevention and reduction.

Description

Technical Field

[0001] The present invention relates to the technical field of geological landslide monitoring, and in particular to a geological landslide monitoring and early warning system and method for a tower base. Background Art

[0002] Existing methods for monitoring and early warning of geological landslides at tower bases suffer from numerous shortcomings. First, traditional monitoring systems often rely on a single sensing technology and lack effective fusion of multi-source data, resulting in limited sensing of landslide activity and difficulty accurately capturing the complex dynamics of landslides. Second, many methods fail to achieve high-precision, high-frequency, synchronous sampling during data acquisition. This leads to inconsistent timestamps and inaccurate spatial coordinate conversion, compromising the spatiotemporal consistency of displacement and deformation data, and thus reducing the reliability and effectiveness of monitoring results. Furthermore, existing early warning models often employ fixed thresholds or simple statistical analysis, lacking dynamic risk assessment mechanisms based on multidimensional comprehensive features. This makes it difficult to scientifically categorize and predict landslide risks. Regarding data processing, traditional technologies inadequately handle noise and abnormal data, lacking systematic multi-frequency vibration rejection and signal noise reduction methods. This results in uneven monitoring data quality, impacting the accuracy of subsequent analysis. Furthermore, the transmission mechanism for early warning information is incomplete. Particularly in weak signal environments such as mountainous areas, data transmission is subject to the risk of loss or delay, making it difficult to ensure timely delivery of warning information and effective response. Finally, most existing systems lack the ability to deeply mine and retrospectively analyze historical monitoring data, making it difficult to extract patterns of potential risk changes from long-term data, thus limiting the scientific decision-making support capabilities of landslide warnings. Summary of the Invention

[0003] Based on this, it is necessary to provide a tower base geological landslide monitoring and early warning system and method to solve at least one of the above technical problems.

[0004] To achieve the above object, a method for monitoring and early warning of geological landslides at tower bases is provided, the method comprising the following steps: Step S1: Deploy inclination sensors, BeiDou-GNSS modules, soil moisture probes, and microseismic accelerometers at and around the tower base to collect tower base inclination, three-dimensional displacement, soil moisture content, and vibration response information to form a raw monitoring data set. Step S2: The original monitoring data set is timestamp unified, WGS-84 projection coordinate system converted, and multi-frequency vibration is eliminated to construct an integrated air-ground displacement profile. At the same time, the displacement rate-acceleration joint feature is calculated with a 1-hour window and Pearson correlation analysis is performed with the water content series to generate a comprehensive characteristic tensor of the tower foundation deformation. Step S3: Call the landslide stability assessment rule library, perform fuzzy reasoning on the tower base deformation comprehensive characteristic tensor, output the landslide probability, hazard level and predicted displacement in the next 6 hours, and obtain the risk assessment result; Step S4: Trigger the four-level warning strategy based on the risk assessment results. The warning information is transmitted back via the NB-IoT-4G dual link. The platform simultaneously pushes SMS, voice and pop-up windows, and generates a warning log with a timestamp for tracing.

[0005] The beneficial effect of the present invention is that, through the systematic acquisition and deep integration of multi-source sensor data, high-precision dynamic monitoring and early warning of tower base landslide risks are achieved. First, step S1 collects multi-dimensional raw data, including angular changes of the tilt sensor, three-dimensional displacement information from the Beidou-GNSS module, humidity data from the soil moisture probe, and vibration response from the microseismic accelerometer, forming a rich raw monitoring data set to ensure multi-angle capture of geological environmental changes. Subsequently, step S2 eliminates data asynchrony through unified timestamps, achieves a unified expression of spatial data using coordinate system transformation, and employs multi-frequency vibration rejection technology to reduce the impact of environmental noise, thereby constructing an integrated air-ground displacement profile and enhancing the spatial and temporal consistency of the data. In addition, the joint characteristics of displacement rate and acceleration are calculated through a sliding time window, and statistical correlation analysis is performed in combination with the moisture content series to form a high-dimensional comprehensive feature tensor, providing a quantitative representation of complex deformation patterns. Step S3 then inputs this comprehensive feature into the landslide stability assessment rule library, using fuzzy inference methods to integrate multivariate uncertainty, accurately determine landslide probability and hazard level, and perform short-term displacement trend prediction, improving the foresight and reliability of risk warnings. Finally, step S4 automatically triggers a four-level warning strategy based on the assessment results. Redundant data backhaul is achieved via NB-IoT and 4G dual links, ensuring stable and real-time information transmission. Multi-channel push notifications via SMS, voice, and pop-up windows ensure rapid response and widespread coverage of warning information. A structured, time-stamped warning log is generated to facilitate subsequent data tracing and event review. The overall process is highly integrated and automated across data collection, processing, fusion, and transmission, ensuring the timeliness, accuracy, and data integrity of landslide risk monitoring and providing scientific decision-making support for disaster prevention and mitigation.

[0006] Preferably, step S1 includes the following steps: Step S11: Fix at least three ±0.01° MEMS tilt sensors at the tower base, the middle and the upper part of the landslide body, respectively, with a sampling frequency of 5 Hz; Step S12: Beidou-GNSS antennas are deployed at the tower base and the trailing edge of the mountain to continuously track the carrier wave, achieving relative positioning accuracy at the millimeter level. Step S13: Lower the steel cable and anti-twist rod in sections; bury soil moisture probes in a vertical direction along the potential slip surface, with a measurement range of 0-100% Vol, wherein the moisture probes are buried in layers at depths of 0.5m, 1m, and 2m; Step S14: Install triaxial accelerometers at the middle of the tower and at the base, with a range of ±16g, to capture microseismic signals. The microseismic accelerometers are installed with bidirectional damping to suppress errors. Step S15: After performing wavelet noise reduction and outlier removal on the multi-source sampling sequence, an original monitoring data set is formed.

[0007] This invention installs multiple high-precision ±0.01° MEMS tilt sensors at the tower base and in the middle and upper parts of the landslide mass, continuously capturing tilt angle changes at a sampling frequency of 5Hz. This enables simultaneous multi-point spatial monitoring and enhances sensitive response to small deformations. In step S12, Beidou-GNSS antennas are deployed at the tower base and the trailing edge of the mountain, utilizing continuous carrier tracking technology to obtain three-dimensional displacement data with millimeter-level relative positioning accuracy, ensuring high precision and stability in ground displacement measurements. In step S13, a segmented lowering method using steel cables combined with anti-twist rods ensures that soil moisture probes can be accurately buried vertically along the potential sliding surface at different depths (0.5m, 1m, and 2m). This enables continuous monitoring of stratified moisture levels and helps detect hydrological changes that could be precursors to landslides. In step S14, triaxial accelerometers installed in the middle of the tower and at the base have a ±16g range and can capture microseismic signals with high sensitivity. Bidirectional damping effectively suppresses installation errors and environmental noise interference, ensuring the authenticity and stability of vibration data. Finally, step S15 applies wavelet transform noise reduction and outlier removal to the multi-source sampling sequences from different sensors, effectively removing high-frequency noise and sudden abnormal data, ensuring a high signal-to-noise ratio and data consistency in the resulting raw monitoring data set. Overall, this process, through multi-point high-frequency sampling, multi-dimensional sensor fusion, and advanced signal processing techniques, establishes an accurate, stable, and comprehensive raw monitoring data foundation in terms of temporal and spatial coverage, providing reliable data support for subsequent landslide dynamic analysis and risk assessment.

[0008] Preferably, constructing the air-ground combined displacement profile in step S2 includes: Calculate the tower base horizontal displacement curve based on the BeiDou-GNSS displacement point sequence and perform least squares smoothing; The accumulated angle of the tilt sensor is converted into vertical displacement through the rod length, and then fused with the GNSS curve to generate a unified ground-underground displacement profile; The Pearson correlation coefficient between water content and displacement rate is used to mark the potential sliding surface in layers, and a comprehensive characteristic tensor of tower foundation deformation is formed.

[0009] The present invention uses the least squares method to curve fit and smooth the horizontal displacement data of the tower base based on the continuous displacement point series collected by the Beidou / GNSS system, significantly reducing the influence of data noise and measurement errors, ensuring the continuity and physical rationality of the displacement curve, and accurately reflecting the dynamic change trend of the horizontal displacement of the tower base. Secondly, the cumulative angle data collected by the inclination sensor is converted into the corresponding vertical displacement through rod length conversion, and is spatially and temporally integrated with the horizontal displacement data obtained by the GNSS system to achieve a unified profile construction of ground and underground displacements, compensate for the limitations of a single measurement method in spatial dimensions, and enhance the stereoscopic observation capability of the overall deformation of the landslide body. Furthermore, combined with the layered buried soil moisture sensor data, the Pearson correlation coefficient is used to perform statistical correlation analysis on the moisture content sequence and the displacement rate sequence to reveal the linear relationship between moisture change and deformation rate, and the potential sliding surface is layered and marked based on the correlation coefficient, thereby achieving fine-grained identification of landslide potential areas at the data level. Ultimately, the multidimensional characteristics of moisture content and displacement rate were organized into a tensor structure, forming a comprehensive representation of the tower base deformation, facilitating subsequent landslide stability analysis and risk assessment. Overall, this method, through data cleaning, conversion, fusion, and multivariate correlation analysis, improved the spatial integrity and temporal consistency of monitoring data, providing a solid data foundation for refined landslide dynamics research and early warning.

[0010] Preferably, the landslide stability assessment rule base in step S3 includes: The daily displacement increment threshold is: if the horizontal displacement growth rate > 2mm / d, it is considered as a Level III warning; Tilt sudden change threshold: a single change of >0.05° for any sensor triggers a Level II warning; Moisture content saturation threshold: When the sensor detects that the moisture content is greater than 80% Vol. and the displacement acceleration increases, a Level I warning is directly triggered; The microseismic frequency threshold is 1-10Hz, and the energy peak lasting for more than 3 minutes will increase to level 1 warning.

[0011] The present invention calculates the daily incremental horizontal displacement rate based on monitored displacement data. When the threshold of 2mm / d is exceeded, the system automatically identifies it as a Level III warning. This indicator quantifies the displacement velocity change of the landslide body, reflects the dynamic trend of deformation development, and helps to timely capture potential accelerated sliding stages. Secondly, for the angle change data collected in real time by the tilt sensor, a single change exceeding 0.05° is set to trigger a Level II warning, reflecting the ability to keenly capture local sudden deformations. High-frequency sampling data is used to dynamically monitor tilt anomalies to ensure immediate response to small fluctuations in landslide structural deformation. Thirdly, combined with soil moisture sensor data, when the moisture content exceeds 80% volumetric water content and the displacement acceleration shows an upward trend, the system directly triggers the highest level Level I warning. This threshold combination reflects the direct impact of water saturation on landslide stability. The conditional judgment based on the simultaneous activation of multiple variables improves the accuracy and timeliness of the warning. Finally, in terms of microseismic monitoring, the system analyzes energy peaks within the 1-10 Hz frequency range. When this energy peak persists for more than three minutes, the system automatically raises the warning level by one level. By continuously detecting vibration signals, it captures microseismic activity during the loosening and rupture of the landslide's internal structure, providing early warning signs of changes in landslide dynamics. In summary, this threshold system, through quantitative time series data threshold setting and a multivariate triggering mechanism, forms a multidimensional, time-series dynamic landslide risk assessment system. This improves the scientific nature of warnings and the feasibility of operations, effectively supporting landslide risk management and emergency response.

[0012] Preferably, step S4 uses dual-channel alarm: The wireless NB-IoT-4G channel is used for regular data backhaul and platform pop-up windows; The SMS voice gateway sends alarms to the operation and maintenance manager and the power supply command center at levels I and II, ensuring that the success rate of alarms in weak signal environments in mountainous areas is ≥99.5%.

[0013] This invention utilizes wireless NB-IoT and 4G dual-channel communication to enable real-time upload of multi-source time-series data collected by monitoring equipment (including key indicators such as tilt, displacement, water content, and microseismicity), ensuring the platform receives a continuous and high-frequency stream of raw monitoring data. NB-IoT, with its wide coverage and low power consumption, adapts to the complex geographical environment of mountainous areas, while 4G channels provide backup high-bandwidth transmission. The two complement each other to reduce the risk of data loss. Secondly, when the warning trigger conditions reach Level I or II, the system simultaneously sends an alarm message to the operations and maintenance manager and the power supply command center via SMS and voice gateway. This process relies on multi-path data channels and a distributed alarm platform, utilizing synchronous management of time-stamped log data to achieve dual confirmation and tracking of alarms. This mechanism ensures that even in mountainous environments with weak signals and unstable communication conditions, alarm messages can still reach their designated recipients with a success rate exceeding 99.5%. Multi-point verification and multi-protocol redundant transmission strategies at the data level significantly enhance the system's ability to safeguard critical alarm information, effectively reducing missed responses due to communication delays or packet loss. In summary, the technical implementation of wireless multi-channel data transmission and multi-mode alarm push has formed a closed-loop management from data collection, transmission, processing to notification, which improves the real-time, reliability and anti-interference ability of the landslide monitoring system, ensures the rapid and accurate distribution of key safety information, and supports subsequent emergency response and maintenance decisions.

[0014] Preferably, it also includes: Step S5: Utilize the wind-solar-lithium hybrid power supply system to continuously supply energy to each monitoring node for ≥5 years, and maintain normal operation under the extreme condition of continuous light loss for 15 days; Step S6: When the battery state of charge (SOC) is less than 20%, the platform automatically issues a low sampling rate instruction and switches to narrowband standby mode; Step S7: Obtain warning information from the Meteorological Bureau; when the cloud platform detects that the continuous rainfall for 3 consecutive days is greater than or equal to 100 mm or the Meteorological Bureau's warning information is an upstream debris flow orange warning, the inclination and acceleration sampling frequency is automatically increased from 5 Hz to 20 Hz, and the high-density observation mode is enabled to obtain the full-process geological monitoring data of the tower base; Step S8: After the disaster is lifted, perform multi-strategy backtracking on the full-process geological monitoring data of the tower base to generate a tower base landslide monitoring and reinforcement recommendation report, and archive and manage it according to the project.

[0015] This invention uses a wind-solar-lithium hybrid power supply system to integrate wind, solar, and lithium battery energy storage data to establish a multi-dimensional energy input and output data model. This model supports real-time monitoring and predictive analysis of the power supply of each monitoring node, ensuring energy supply stability for more than five consecutive years. Specifically targeting extreme weather conditions, the system uses multi-day no-sun data (15 consecutive days of no sun) for risk assessment. It uses backup batteries and power consumption control data to ensure the normal operation of nodes and prevent monitoring data interruptions. Secondly, when the battery state of charge (SOC) falls below 20%, the platform automatically adjusts the sampling frequency and device operating mode based on real-time power data, reducing the data collection frequency and communication load, switching to narrowband standby mode to maximize energy utilization, and achieving dynamic energy closed-loop management. This process monitors the SOC change curve and historical energy consumption data to adjust the sampling task scheduling and ensure a balance between data collection and the power supply system. Furthermore, if rainfall exceeds 100 mm for three consecutive days or an orange mudslide warning is triggered upstream, the platform automatically increases the sampling frequency of inclination and acceleration sensors from 5 Hz to 20 Hz using real-time meteorological and warning data. Simultaneously, it activates a high-density monitoring mode based on "landslide body zoning," enabling high-precision and timely data collection during the critical period of disaster development, enhancing data granularity and sensitivity. Finally, after the disaster is resolved, the system uses full-cycle, multi-strategy data backtracking technology to conduct multi-dimensional analysis of inclination, acceleration, meteorological, and hydrological data, integrating spatiotemporal features to achieve deep mining and anomaly pattern identification in the monitoring data. It also automatically generates a report document containing monitoring analysis results and reinforcement recommendations. Through structured and time-series data management, it supports project-level archiving, ensuring data integrity and traceability. Overall, this technology forms a closed loop across data collection, transmission, storage, and processing, optimizing energy consumption management at monitoring nodes, improving data quality during critical periods, strengthening disaster response capabilities and supporting post-disaster decision-making, and meeting the needs for long-term, continuous, and accurate landslide monitoring.

[0016] Preferably, it also includes: The monitoring installation benchmark point is located in a stable area 5-10m outside the extension line of the tower line direction, and the bottom of the hole below the benchmark point remains relatively still with the landslide body.

[0017] The present invention significantly improves the accuracy and reliability of the overall displacement data of the tower base by rationally arranging monitoring benchmark points and static conditions at the bottom of the hole. From the data level analysis, first, the benchmark point is set in a stable area 5 to 10 meters outside the extension line of the tower line direction, which effectively isolates the impact of the potential activity of the landslide body on the benchmark point displacement data and ensures the stability of the spatial coordinates of the benchmark point in the time series. This stable spatial reference point serves as the absolute coordinate benchmark for multi-phase displacement measurement, providing a unified benchmark framework for all subsequent displacement data and eliminating the systematic errors introduced by the movement of the benchmark point itself. Secondly, the bottom of the hole below the benchmark point maintains a relatively static state with the landslide body, which means that the geological layer at the bottom of the hole does not move relative to each other, thereby ensuring the stability and continuity of the deep benchmark data. This condition is verified by the time series change curve of the long-term monitoring of the micro-displacement data of the bottom of the hole, eliminating the interference of the sliding of the bottom hole layer, so that when the overall displacement of the tower base is calculated by the inclination sensor and GNSS data, the motion components of the surface and deep layers of the landslide body can be effectively separated, thereby improving the accuracy of the overall displacement vector estimation. Furthermore, the relative static state of the benchmarks reduces potential noise and abnormal fluctuations in the data, facilitating the stable operation of trend analysis and anomaly detection algorithms in displacement time series, ensuring that the monitoring system can output continuous, consistent, and high-quality displacement data across multiple time periods and operating conditions. Therefore, by establishing a stable spatial reference frame and deep static baseline, this benchmark placement scheme significantly improves the accuracy of overall displacement monitoring data for the landslide mass and tower base, providing a solid data foundation for subsequent landslide risk assessment and safety warnings.

[0018] Preferably, it also includes: Before installing the inclinometer, the verticality of the inclinometer tube must be checked. The vertical deviation must not exceed 30% of the full scale of the sensor. All sensors should be aligned with the guide groove in the same direction and connected and straightened through universal joints before being lowered in sequence.

[0019] The present invention verifies the verticality of the inclinometer tube to ensure that its vertical deviation does not exceed 30% of the sensor's full-scale range. This measure limits the impact of installation errors on angle measurement results and avoids systematic deviations caused by tube tilt. Verticality control ensures the stability of the sensor's orientation reference in the spatial reference system, ensuring that the collected inclination data accurately reflects actual deformation rather than equipment installation errors. Secondly, all sensors are aligned with the guide slot in a unified direction, ensuring consistency in measurement direction between sensors. This facilitates spatial comparison and fusion of multi-sensor data and reduces data heterogeneity caused by inconsistent installation orientations. The alignment process, achieved through a universal joint connection, further eliminates angular errors between the sensor and the inclinometer tube, achieving high-precision alignment of the sensor and the guide slot. This step reduces the degree of freedom error in the mechanical connection and stabilizes the sensor's measurement posture. The sequential lowering of sensors ensures the continuity of the entire measurement chain and the integrity of the spatial sequence, resulting in an ordered and physically continuous sequence of angle changes in the inclination data. Based on this installation verification and structural fixation process, the signal-to-noise ratio of the collected inclination time series data is significantly improved, reducing the presence of noise and outliers, and enhancing the data's usability and interpretability. This not only supports the accurate execution of subsequent data processing steps, such as deformation trend analysis and critical threshold determination, but also provides high-quality foundational data for the multi-sensor fusion algorithms within the landslide monitoring system, ensuring accurate responses for risk assessment and early warning systems. Therefore, by strictly controlling verticality and directional consistency, this installation technology establishes a solid and accurate measurement foundation for data acquisition, fundamentally improving the reliability and data quality of the entire monitoring system.

[0020] Preferably, it also includes: The warning level adopts four levels: Level IV prompt, Level III attention, Level II alert, and Level I emergency. The platform supports retrospective analysis of monitoring data of any period and automatically generates reinforcement treatment recommendations.

[0021] This invention categorizes continuously collected monitoring data into four warning levels based on preset risk thresholds, classifying them according to different risk intensities. This ensures hierarchical identification of landslide or structural deformation risks and enables layered analysis and processing of monitoring data. After data is collected and stored in real time on the platform, the system compares and analyzes real-time and historical data based on thresholds at each level, automatically determining whether key indicators such as current deformation, displacement rate, and vibration frequency meet a certain level of warning criteria, thereby achieving a quantitative expression of risk status. Secondly, the platform supports backtracking of monitoring data for any period, meaning it is not limited to real-time data analysis but also enables systematic review and trend mining of data within historical time periods. By analyzing and comparing historical data in a time series, the evolution of potential risks and critical change points can be revealed, enriching the spatiotemporal dimensions of risk identification and making warnings more dynamic and accurate. During the backtracking analysis process, the system combines multidimensional features such as cumulative deformation, rate of change, and environmental factors to generate reinforcement treatment recommendations through data fusion technology and statistical models, achieving a transition from data-driven to engineering decision-making. Based on data-derived landslide risk levels and deformation trends, the proposal provides targeted engineering reinforcement solutions, facilitating the scientific development of maintenance plans. Overall, this four-tiered early warning system, through refined threshold setting, multi-period time series backtracking, and intelligent decision-making support at the data level, ensures hierarchical presentation and scientific response of early warning information. This provides a data-driven theoretical basis and application support for risk management, and promotes the effective translation of monitoring data into practical engineering applications.

[0022] In this specification, a tower base geological landslide monitoring and early warning system is provided, which is used to implement the above-mentioned tower base geological landslide monitoring and early warning method. The tower base geological landslide monitoring and early warning system includes: The multi-source monitoring and acquisition module is used to deploy inclination sensors, Beidou-GNSS modules, soil moisture probes, and microseismic accelerometers at and around the tower base to collect tower base tilt, three-dimensional displacement, soil moisture content, and vibration response information to form a raw monitoring data set. The data spatiotemporal fusion and feature construction module is used to unify the timestamps of the original monitoring data set, convert the WGS-84 projection coordinate system, and eliminate multi-frequency vibrations to construct an integrated air-ground displacement profile. Simultaneously, the displacement rate-acceleration joint feature is calculated using a 1-hour window and Pearson correlation analysis is performed with the moisture content series to generate a comprehensive feature tensor of the tower foundation deformation. The landslide stability assessment module is used to call the landslide stability assessment rule library, perform fuzzy reasoning on the comprehensive characteristic tensor of the tower base deformation, output the landslide probability, hazard level and the predicted displacement in the next 6 hours, and obtain the risk assessment result; The intelligent hierarchical warning and information push module is used to trigger a four-level warning strategy based on risk assessment results. Warning information is transmitted back via the NB-IoT-4G dual link. The platform simultaneously pushes SMS, voice, and pop-up notifications, and generates a time-stamped warning log for traceability. Wind-solar-lithium hybrid power supply module, which uses a wind-solar-lithium hybrid power supply system to provide continuous power to each monitoring node for ≥5 years and maintain normal operation under extreme conditions of 15 days of no sunlight; The energy closed-loop management module is used to automatically issue low-power sampling instructions through the platform when the battery state of charge is less than 20%, thus achieving a safe closed-loop power supply. Disaster enhancement observation trigger module, used to obtain warning information from the Meteorological Bureau; when the cloud platform detects that the continuous rainfall for 3 days is greater than or equal to 100mm or the Meteorological Bureau's warning information is an orange warning for upstream debris flow, the inclination and acceleration sampling frequency will be automatically increased from 5Hz to 20Hz, and the high-density observation mode will be enabled to obtain the full-process geological monitoring data of the tower base; data backtracking and reinforcement decision-making module: after the disaster is lifted, multi-strategy backtracking is performed on the full-process geological monitoring data of the tower base to generate a tower base landslide monitoring and reinforcement recommendation report, which is archived and managed by project.

[0023] The beneficial effect of the present invention is that, through the systematic acquisition and deep integration of multi-source sensor data, high-precision dynamic monitoring and early warning of tower base landslide risks are achieved. First, step S1 collects multi-dimensional raw data, including angular changes of the tilt sensor, three-dimensional displacement information from the Beidou-GNSS module, humidity data from the soil moisture probe, and vibration response from the microseismic accelerometer, forming a rich raw monitoring data set to ensure multi-angle capture of geological environmental changes. Subsequently, step S2 eliminates data asynchrony through unified timestamps, achieves a unified expression of spatial data using coordinate system transformation, and employs multi-frequency vibration rejection technology to reduce the impact of environmental noise, thereby constructing an integrated air-ground displacement profile and enhancing the spatial and temporal consistency of the data. In addition, the joint characteristics of displacement rate and acceleration are calculated through a sliding time window, and statistical correlation analysis is performed in combination with the moisture content series to form a high-dimensional comprehensive feature tensor, providing a quantitative representation of complex deformation patterns. Step S3 then inputs this comprehensive feature into the landslide stability assessment rule library, using fuzzy inference methods to integrate multivariate uncertainty, accurately determine landslide probability and hazard level, and perform short-term displacement trend prediction, improving the foresight and reliability of risk warnings. Finally, step S4 automatically triggers a four-level warning strategy based on the assessment results. Redundant data backhaul is achieved via NB-IoT and 4G dual links, ensuring stable and real-time information transmission. Multi-channel push notifications via SMS, voice, and pop-up windows ensure rapid response and widespread coverage of warning information. A structured, time-stamped warning log is generated to facilitate subsequent data tracing and event review. The overall process is highly integrated and automated across data collection, processing, fusion, and transmission, ensuring the timeliness, accuracy, and data integrity of landslide risk monitoring and providing scientific decision-making support for disaster prevention and mitigation. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A schematic diagram of the steps of a tower base geological landslide monitoring and early warning method; Figure 2 for Figure 1 Detailed implementation steps of step S1 in FIG. The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0025] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. Obviously, the embodiments described are 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 those skilled in the art without making any creative work are within the scope of protection of the present invention.

[0026] In addition, the accompanying drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. Identical reference numerals in the figures denote identical or similar parts, and thus repetitive descriptions thereof will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor and / or microcontroller approaches.

[0027] It should be understood that although the terms "first," "second," and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used solely to distinguish one element from another. For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element, without departing from the scope of the exemplary embodiments. The term "and / or" as used herein includes any and all combinations of one or more of the listed associated items.

[0028] To achieve this, please refer to Figures 1 to 2 A method for monitoring and early warning of geological landslides at a tower base, comprising the following steps: Step S1: Deploy inclination sensors, BeiDou-GNSS modules, soil moisture probes, and microseismic accelerometers at and around the tower base to collect tower base inclination, three-dimensional displacement, soil moisture content, and vibration response information to form a raw monitoring data set. Step S2: The original monitoring data set is timestamp unified, WGS-84 projection coordinate system converted, and multi-frequency vibration is eliminated to construct an integrated air-ground displacement profile. At the same time, the displacement rate-acceleration joint feature is calculated with a 1-hour window and Pearson correlation analysis is performed with the water content series to generate a comprehensive characteristic tensor of the tower foundation deformation. Step S3: Call the landslide stability assessment rule library, perform fuzzy reasoning on the tower base deformation comprehensive characteristic tensor, output the landslide probability, hazard level and predicted displacement in the next 6 hours, and obtain the risk assessment result; Step S4: Trigger the four-level warning strategy based on the risk assessment results. The warning information is transmitted back via the NB-IoT-4G dual link. The platform simultaneously pushes SMS, voice and pop-up windows, and generates a warning log with a timestamp for tracing.

[0029] In the embodiment of the present invention, reference Figure 1 FIG. 1 is a schematic flow chart of a method for monitoring and warning landslides at a tower base according to the present invention. In this example, the method for monitoring and warning landslides at a tower base includes the following steps: Step S1: Deploy inclination sensors, BeiDou-GNSS modules, soil moisture probes, and microseismic accelerometers at and around the tower base to collect tower base inclination, three-dimensional displacement, soil moisture content, and vibration response information to form a raw monitoring data set. In an embodiment of the present invention, multiple types of sensors are deployed at the tower base and its surroundings, including high-precision MEMS inclination sensors, Beidou / GNSS dual-frequency positioning modules, time-domain reflectometry soil moisture probes, and triaxial microseismic accelerometers. The inclination sensor is typically installed on the tower base concrete component or shallow surface. By recording its X- and Y-axis posture changes, it forms a time-evolution sequence θ(t) of the tower base inclination angle, which is used to monitor the inclination trend of the tower body. The Beidou-GNSS module uses a relative positioning method, with a reference point located in a remote stable area as a reference, to record the three-dimensional coordinates P(t) = (x, y, z) of the tower base in real time, generate a millimeter-level displacement sequence ΔP(t), and realize the joint monitoring of the horizontal and vertical displacements of the tower base. Soil moisture probes are buried in layers along the vertical profile of the potential slip surface, typically at depths of 0.5 m, 1 m, and 2 m. This generates moisture profile curves W(z,t). These data are used to characterize the moisture response mechanisms of different soil layers under rainfall, evaporation, or seepage. Microseismic accelerometers are deployed at the junction of the tower body and the base, collecting high-frequency acceleration signals a(t). A Fourier transform is used to generate a frequency domain characteristic spectrum F(f), which is used to identify microseismic responses caused by localized slip, structural impact, or deep deformation. All sensor data are aligned along a unified sampling time axis T, forming a multivariate joint monitoring raw dataset consisting of θ(t), ΔP(t), W(z,t), and F(f). This dataset is highly timely and spatially multidimensional, and can be subsequently used for landslide trend assessment, warning threshold trigger analysis, and dynamic risk modeling, providing fundamental support for the data processing and reasoning logic of the entire monitoring system.

[0030] Step S2: The original monitoring data set is timestamp unified, WGS-84 projection coordinate system converted, and multi-frequency vibration is eliminated to construct an integrated air-ground displacement profile. At the same time, the displacement rate-acceleration joint feature is calculated with a 1-hour window and Pearson correlation analysis is performed with the water content series to generate a comprehensive characteristic tensor of the tower foundation deformation. In this embodiment of the present invention, the raw data generated by various sensors is time-stamped and aligned using a unified UTC or local standard time format to eliminate time drift caused by sampling delays and communication synchronization errors between different devices, ensuring that multi-source time series data has a unified reference axis during analysis. Regarding spatial processing, the displacement point series recorded by the Beidou-GNSS module in the WGS-84 geographic coordinate system is converted to a plane rectangular coordinate system using the Gauss-Krüger projection (or UTM projection) to facilitate a unified scale representation with the vertical displacement values ​​converted from the tilt sensor. For the raw vibration signals collected by the microseismic accelerometer, a multi-component wavelet packet transform or bandpass filtering method is used to frequency-domain remove non-geological disturbance noise in the frequency range above 10 Hz and below 0.1 Hz, extracting valid vibration data within the target frequency band. Based on this, the sensor layout coordinates are spatially aligned with the projected displacement, tilt, and acceleration data to construct an integrated air-ground three-dimensional displacement profile containing both surface and subsurface information, displaying deformation responses at different depths and horizontal positions. Subsequently, using a one-hour sliding window, a first-order difference process was performed on the continuous displacement sequence ΔP(t) to obtain the displacement rate v(t) = ΔP(t) / Δt. Combined with the accelerometer-generated a(t) data, a joint eigenvector [v(t), a(t)] was constructed. A Pearson correlation analysis was performed on this eigenvector with the multi-depth soil moisture time series w(z, t), yielding the coupling coefficient matrix R(z, t) between moisture content at different depths and the tower foundation motion characteristics. Ultimately, the displacement, acceleration, moisture content, and their statistical correlations were mapped to a multidimensional tensor structure T(i, j, k), where i represents spatial position, j represents time step, and k represents feature dimension. This allows for a structured representation of the deformation mechanism and standardized input for subsequent modeling and calculations.

[0031] Step S3: Call the landslide stability assessment rule library, perform fuzzy reasoning on the tower base deformation comprehensive characteristic tensor, output the landslide probability, hazard level and predicted displacement in the next 6 hours, and obtain the risk assessment result; In this embodiment of the present invention, key monitoring indicators are extracted from the tower foundation deformation comprehensive feature tensor T(i, j, k) generated in the previous step. These include displacement rate, acceleration, moisture content, and their statistical correlation characteristics calculated at different spatial locations i and time series j. This tensor serves as the input data structure for the fuzzy inference engine and is converted into fuzzy variable inputs through mapping predefined fuzzy sets and membership functions in the rule base. For example, displacement rate is divided into multiple membership intervals such as "low," "medium," and "high," while moisture content is similarly mapped to fuzzy levels such as "dry," "moderately wet," and "saturated." The hazard level L is defined as a fuzzy set ranging from Level IV (warning) to Level I (emergency). Based on this, the landslide stability assessment rule base incorporates multiple expert rules and empirical fuzzy inference rules. Fuzzy inference mechanisms (such as the Mamdani or Sugeno methods) are used to derive fuzzy relationships and perform comprehensive judgments on the input fuzzy variables, generating a fuzzy output of the landslide occurrence probability P. This fuzzy probability is converted into a clear numerical output using a defuzzification algorithm (such as the centroid method) for subsequent processing. The hazard level L is output in a graded manner based on the fuzzy output probability and its membership function, combined with a weighted decision-making process based on multiple indicators. Simultaneously, based on the landslide deformation time series and prediction models (such as recursive fuzzy time series models or state-space models based on fuzzy logic), and in conjunction with current deformation trends, the system predicts a fuzzy estimate of the tower base displacement Δd within the next six hours. By integrating the fuzzy prediction with actual observation data, prediction accuracy is improved. Ultimately, the risk assessment results, including the landslide probability P, the graded hazard level L, and the predicted future displacement Δd, are output in a structured data format, providing a scientific basis for subsequent risk response strategies and early warning mechanisms.

[0032] Step S4: Trigger the four-level warning strategy based on the risk assessment results. The warning information is transmitted back via the NB-IoT-4G dual link. The platform simultaneously pushes SMS, voice and pop-up windows, and generates a warning log with a timestamp for tracing.

[0033] In this embodiment of the present invention, the system receives structured risk output data from the risk assessment module, including landslide probability P, hazard level L, and associated time tags. This data serves as the input for triggering warnings. Based on a pre-set four-level classification rule, the warning triggering logic analyzes the hazard level L in real time, maps it to the corresponding warning level (Level IV Warning to Level I Emergency), and generates a warning data packet containing the level identifier, timestamp, and associated deformation parameters. To ensure the reliability and coverage of information transmission, the system utilizes a dual-link NB-IoT and 4G communication architecture, establishing two independent data return channels. Warning data is transmitted simultaneously over both links, achieving channel redundancy and backup. Data packets undergo encoding and compression before transmission to reduce transmission latency and bandwidth usage, and a checksum is added to ensure data integrity. After transmission to the central platform, the data is decoded, verified, and stored in a real-time warning database. Based on the warning level configuration, the platform triggers a multi-channel information distribution system, which simultaneously pushes warning content to operations and maintenance personnel and relevant management terminals via text, voice, and pop-up notifications via the Short Message Service Center (SMSC), voice gateway, and web push service. Each channel is assigned a unified timestamp and level identifier, facilitating cross-channel information consistency verification. All warning events, along with detailed data and timestamps, are archived in a distributed log management system, supporting subsequent query and traceability analysis. This system utilizes log indexing technology to enable rapid retrieval, ensuring comprehensive tracking and review of historical warning events.

[0034] As an example of the present invention, refer to Figure 2 As shown, in this example, step S1 includes: Step S11: Fix at least three ±0.01° MEMS tilt sensors at the tower base, the middle and the upper part of the landslide body, respectively, with a sampling frequency of 5 Hz; Step S12: Beidou-GNSS antennas are deployed at the tower base and the trailing edge of the mountain to continuously track the carrier wave, achieving relative positioning accuracy at the millimeter level. Step S13: Lowering the steel cable and anti-twist rod in sections; burying soil moisture probes in a vertical direction along the potential slip surface, with a measurement range of 0-100% Vol., wherein the moisture probes are buried in layers at depths of 0.5m, 1m, and 2m; Step S14: Install triaxial accelerometers at the middle of the tower and at the base, with a range of ±16g, to capture microseismic signals. The microseismic accelerometers are installed with bidirectional damping to suppress errors. Step S15: After performing wavelet noise reduction and outlier removal on the multi-source sampling sequence, an original monitoring data set is formed.

[0035] In an embodiment of the present invention, at least three MEMS inclination sensors with an accuracy of ±0.01° are installed at the tower base, the middle part, and the upper part of the landslide. The sampling frequency is set to 5 Hz to continuously acquire data on the attitude changes of the sliding body in different partitions. In its original state, the inclination data contains a certain amplitude of environmental noise and attitude disturbances. Therefore, its subsequent processing needs to rely on a highly stable wavelet denoising algorithm to improve data readability. Secondly, a Beidou-GNSS antenna system is deployed at the tower base and the rear edge of the landslide. The system supports continuous carrier phase tracking and achieves relative displacement data acquisition with sub-centimeter to millimeter accuracy through real-time kinematic differential (RTK) or network RTK, forming a high-frequency continuous observation record of the absolute position of the sliding body and micro-displacement changes. Third, given the importance of subsurface moisture factors in their impact on landslides, a soil profile was vertically deployed using steel cables and anti-twist rods. Volumetric moisture probes were buried at 0.5m, 1m, and 2m. These probes have a measurement range of 0-100% Vol., enabling precise monitoring of temporal changes in soil moisture in multiple layers above and below the potential sliding surface. All moisture content data was structured along a time axis to form a two-dimensional spatial-depth moisture matrix. Fourth, triaxial accelerometers with a range of ±16g were installed in the middle of the tower and at the base to capture local microseismic events. The accelerometers were secured with a bidirectional damping structure to reduce measurement errors caused by structural resonance or installation angle offset. The acquired acceleration signals were further filtered and normalized in separate channels. Finally, after collection, the time series data generated by the above-mentioned multi-source sensors are connected to the edge computing unit through a unified data interface, and the wavelet multi-scale decomposition method is executed to filter out high-frequency noise. At the same time, the median sliding window and isolation forest algorithms are used to identify and eliminate outliers that significantly deviate from the background trend, thereby constructing an original data set that is consistent in time and space, has good noise suppression effect, and has engineering monitoring significance, providing a unified data foundation for subsequent landslide risk identification and dynamic trend modeling.

[0036] Preferably, constructing the air-ground combined displacement profile in step S2 includes: Calculate the tower base horizontal displacement curve based on the BeiDou-GNSS displacement point sequence and perform least squares smoothing; The accumulated angle of the tilt sensor is converted into vertical displacement through the rod length, and then fused with the GNSS curve to generate a unified ground-underground displacement profile; The Pearson correlation coefficient between water content and displacement rate is used to mark the potential sliding surface in layers, and a comprehensive characteristic tensor of tower foundation deformation is formed.

[0037] In this embodiment of the present invention, the results of the multi-point carrier phase difference decomposition generated by the Beidou / GNSS receiver at the same epoch are organized into a three-dimensional coordinate discrete point sequence P(t) sorted by time series, and a relative coordinate system is constructed with the tower base calibration point as a reference. Subsequently, B-spline curve fitting is performed on the east-west and north-south components of P(t) using weighted least squares polynomial regression. While simultaneously constraining curvature continuity, a smooth curve H(t) of the tower base horizontal displacement is obtained, filtering out random measurement noise and jump errors. Secondly, the cumulative tilt angle θ(t) recorded by the tilt sensor within the same time window is cosine-corrected and then multiplied by the known pole length L to obtain the vertical displacement increment ΔV(t) = L·(1-cosθ(t)). ΔV(t) is then bivariately time-aligned with H(t) synchronized to the same timestamp via a Kalman filter. The sampling frequency difference is corrected by linear interpolation to form a unified surface-underground displacement profile dataset D(z,t), where z represents the depth layer index. Next, the time-order first-order difference of the displacement profile is taken to obtain the rate matrix V(z,t)=∂D / ∂t. The soil moisture matrix W(z,t) for the same layer is also extracted. The Pearson correlation coefficient R(z)=corrτ[V(z,t),W(z,t)] between V and W is calculated at each depth within a fixed sliding window τ. A confidence interval test is performed on R(z). When |R(z)| exceeds the threshold R0, the depth segment is marked as a potential slip zone. Finally, the horizontal displacement curve H(t), vertical displacement profile D(z,t), displacement rate V(z,t), moisture content W(z,t), and layer-by-layer correlation coefficient R(z) are combined according to tensor dimensional mapping rules into a four-dimensional feature tensor T(x,z,t,f), where x is the horizontal sequence index and f is the feature channel number. This provides the data foundation for subsequent spatiotemporal coupling analysis.

[0038] Preferably, the landslide stability assessment rule base in step S3 includes: The daily displacement increment threshold is: if the horizontal displacement growth rate > 2mm / d, it is considered as a Level III warning; Tilt sudden change threshold: a single change of >0.05° for any sensor triggers a Level II warning; Moisture content saturation threshold: When the sensor detects that the moisture content is greater than 80% Vol. and the displacement acceleration increases, a Level I warning is directly triggered; The microseismic frequency threshold is 1-10Hz, and the energy peak lasting for more than 3 minutes will increase to level 1 warning.

[0039] In this embodiment of the present invention, for the daily increment threshold of horizontal displacement, the system constructs a discrete time series H(t) using the relative displacement results at the same time each day in the Beidou / GNSS positioning sequence. The daily growth rate ΔH(t) = H(t) − H(t−1) is calculated through first-order forward differencing. When ΔH(t) exceeds 2mm, it is recorded as a potential acceleration event and enters the Level III warning buffer state. This process is implemented by a threshold comparator with a fixed time interval. Secondly, for the inclination mutation threshold, the system uses the single-sample angle difference |θ_i(t) − θ_i(t−1)| of each inclination sensor θ_i as the detection indicator. All inclination channels are processed in parallel through edge computing nodes. When the mutation amplitude of any θ_i exceeds 0.05°, a Level II warning event is triggered. This threshold judgment is based on single-point mutation recognition and relies primarily on sliding window calculation and absolute value comparison algorithms. Third, for water content saturation threshold warnings, the system constructs a joint judgment mechanism based on W(z,t) and its first-order derivative, ∂²D / ∂t². ​​First, it determines whether any deep probe value in the water content data exceeds 80% Vol. Secondly, it calculates the second-order derivative of the displacement sequence, i.e., the landslide acceleration signal. If both conditions hold simultaneously, a Level I warning is directly triggered. This joint judgment implements cross-state judgment through a Boolean logic controller, with synchronization constraints. Finally, for microseismic frequency threshold judgment, the system performs a short-time Fourier transform (STFT) or continuous wavelet transform on the time series a(t) acquired by the triaxial accelerometer, extracting the spectral power P(f,t) concentrated in the 1-10 Hz frequency band and integrating it within a sliding time window. If the power peak value remains above the set threshold for more than 3 minutes, the system automatically increases the warning level by one level based on the current level. This process is based on a joint detection algorithm based on frequency domain judgment and time persistence. Overall, the above four types of criteria are all based on the extraction of statistical fluctuation characteristics of quantitative time series data, and combined with the threshold trigger mechanism to complete the dynamic judgment and update of the warning level.

[0040] Preferably, step S4 uses dual-channel alarm: The wireless NB-IoT-4G channel is used for regular data backhaul and platform pop-up windows; The SMS voice gateway sends alarms to the operation and maintenance manager and the power supply command center at levels I and II, ensuring that the success rate of alarms in weak signal environments in mountainous areas is ≥99.5%.

[0041] An embodiment of the present invention involves a multi-channel redundant design for data communication and alarm information transmission in a landslide monitoring system. Its core technical approaches include a data backhaul mechanism for wireless narrowband Internet of Things (NB-IoT) and 4G cellular communications, a multi-target short-term and wide-spread messaging strategy for SMS and voice gateways, and a communication quality assurance model. First, conventional monitoring data, such as GNSS displacement, tilt angle changes, moisture content, and acceleration signals, are uniformly formatted into time-stamped data packets by edge computing nodes, encapsulated according to a preset communication cycle, and then uploaded to a cloud platform database via the NB-IoT or 4G module using the UDP or MQTT protocols. The NB-IoT channel is primarily suitable for narrowband, low-power scenarios, ensuring wide-area accessibility for data uploads. The 4G channel is used for concurrent data uploads in areas with strong signal coverage. Priority or automatic switching strategies can be set within the edge node for both, enabling adaptive multi-link transmission. To ensure timely delivery of alerts, the system invokes the SMS and Voice Gateway module when Level I and Level II alerts are triggered. This module sends structured short messages and voice templates to the operator platform via the SMPP protocol or HTTP API. This module automatically translates the alert content into both text and voice formats and distributes it to the mobile phones of designated operations and maintenance personnel and the power grid command center. To ensure alert delivery in mountainous areas with unstable signal conditions, the SMS and Voice Channels feature redundant retransmission logic. If the initial receipt fails to return a success flag within a set time window, the system automatically initiates multi-base station path expansion and performs up to three network path switches and retransmissions, each with an interval of no more than 30 seconds. Dynamic channel power adjustment is also employed to enhance signal penetration. The alert success rate of the communication link is calculated through receipt comparison, two-way handshake log analysis, and a packet loss rate assessment model. The platform maintains daily updates to the communication log to ensure that the high-confidence alert delivery rate remains above 99.5%. This mechanism establishes monitoring nodes at the data link layer, session control layer, and application service layer to ensure the timeliness and robustness of alert communication.

[0042] Preferably, it also includes: Step S5: Utilize the wind-solar-lithium hybrid power supply system to continuously supply energy to each monitoring node for ≥5 years, and maintain normal operation under the extreme condition of continuous light loss for 15 days; Step S6: When the battery state of charge (SOC) is less than 20%, the platform automatically issues a low sampling rate instruction and switches to narrowband standby mode; Step S7: Obtain warning information from the Meteorological Bureau; when the cloud platform detects that the continuous rainfall for 3 consecutive days is greater than or equal to 100 mm or the Meteorological Bureau's warning information is an upstream debris flow orange warning, the inclination and acceleration sampling frequency is automatically increased from 5 Hz to 20 Hz, and the high-density observation mode is enabled to obtain the full-process geological monitoring data of the tower base; Step S8: After the disaster is lifted, perform multi-strategy backtracking on the full-process geological monitoring data of the tower base to generate a tower base landslide monitoring and reinforcement recommendation report, and archive and manage it according to the project.

[0043] In this embodiment of the present invention, the wind-solar-lithium hybrid energy supply system consists of a small wind turbine, high-efficiency photovoltaic panels, and a lithium battery pack. The operating status of each component is centrally managed by an energy harvesting controller. The system records daily power generation, battery state of charge (SOC), and load energy consumption curves to construct a time-series energy flow model E(t), where SOC(t) is a key state parameter. This model can be used to predict the remaining energy supply duration under extreme weather conditions and establish a threshold judgment mechanism. When the SOC drops below 20%, the platform issues a control command to automatically reduce the sampling frequency of all sensors, such as the inclination and acceleration sensors, to below 1Hz, and switches the communication module to NB-IoT narrowband standby mode to minimize energy consumption. This process is triggered by the downlink command control module, ensuring closed-loop execution of the low-power strategy. Secondly, the system's integrated meteorological data module obtains real-time rainfall data R(t) through an API or a local rain gauge and compares it with the orange warning data from the upstream debris flow monitoring center. When cumulative rainfall reaches 100 mm or more for three consecutive days or an orange warning signal is received, the control system automatically increases the core sensor sampling frequency from 5 Hz to 20 Hz and activates a high-density landslide monitoring mode. This mode involves simultaneous sampling at nodes at different elevations to improve spatial resolution and response speed. This switching behavior is controlled by an event-driven mechanism, with edge nodes performing real-time command parsing and device parameter adjustment. Furthermore, the disaster relief criterion is that rainfall has ceased for more than 48 hours and all monitoring indicators have returned to the background stable range, at which point the system enters the data backtracking phase. Complete time series data from all sensors is imported into the platform's data analysis module for a multi-strategy backtracking process, including microseismic source identification based on acceleration anomalies, inclination angle trend regression, and reconstruction of the correlation between water content changes and slip acceleration. The analysis results are output as a structured report covering anomaly identification records, regional risk assessment, power supply and energy consumption assessment, and emergency response process review. Finally, the platform's archiving system links tower base codes for project-level documentation, forming a traceable data chain for landslide monitoring and reinforcement recommendations.

[0044] Preferably, it also includes: The monitoring installation benchmark point is located in a stable area 5-10m outside the extension line of the tower line direction, and the bottom of the hole below the benchmark point remains relatively still with the landslide body.

[0045] In this embodiment of the present invention, the spatial selection principle for the benchmark point is based on the mechanical characteristics of the line structure and geological stability analysis. It is located 5-10 meters outside the tower along the extended line of the line, ensuring that it is not affected by the tower base structural load and is far away from the boundary of the potential sliding body, thereby achieving geometric independence of the displacement observation baseline. The stability of this location is generally confirmed through initial geological surveys, surface deformation history analysis, and small-scale geological drilling. Specifically, static parameters such as the lithology, groundwater activity, and signs of slope slip in the area are evaluated to ensure that it can be used as a static reference point for displacement calculations. Secondly, the drilling depth below the benchmark point should penetrate the surface loose layer until it reaches stable bedrock or hard formations. Engineering measures such as steel casing or grouting are often used to isolate the benchmark point from surrounding weak structures, forming a fixed support point that does not move relative to the landslide body. At the data processing level, this benchmark point participates in continuous carrier phase differential measurement by deploying GNSS antennas. All tower base GNSS data are relative to this point for position calculation, resulting in a high-precision displacement sequence P_rel(t) of the tower base relative to the benchmark point. To verify the stationary nature of the benchmark, its historical displacement sequence, P_base(t), must be continuously monitored. When the change in horizontal or vertical displacement is within a set threshold (typically less than 0.3 mm), its stability is confirmed. This mechanism eliminates false displacement calculation errors caused by unstable benchmarks at the source, ensuring the uniqueness and high accuracy of the spatial reference frame in tower base deformation analysis. This provides fundamental data support for subsequent landslide trend monitoring, early warning threshold determination, and surface-deep displacement profile calculation.

[0046] Preferably, it also includes: Before installing the inclinometer, the verticality of the inclinometer tube must be checked. The vertical deviation must not exceed 30% of the full scale of the sensor. All sensors should be aligned with the guide groove in the same direction and connected and straightened through universal joints before being lowered in sequence.

[0047] In this embodiment of the present invention, quantitative verification of the inclinometer casing's verticality, unified sensor orientation, and a multi-section, adaptive connection structure ensure the consistency and geometric rationality of measurement data during lowering. Before sensor installation, the casing's vertical reference state must be precisely verified. This is typically done using a high-precision digital plumb line or a combined laser rangefinder and laser plumb line. A vertical direction vector, V_ref, is established relative to a fixed surface reference. The vertical deviation, θ_max = max(θ_i), is calculated by measuring the angle θ_i between the inclinometer casing's centerline unit vector, V_i, at each depth segment. This deviation is then compared to the sensor's maximum range, α_full. To ensure data reliability, θ_max must not exceed 30% of α_full. In other words, for an inclinometer with a full-scale range of ±0.5°, the allowable vertical deviation should be less than ±0.15°. After confirming that the inclinometer casing meets installation requirements, all inclinometer sensors are oriented in a unified direction. This direction is aligned with the mechanical positioning surfaces pre-set in the guide grooves to prevent data deviations caused by directional mismatches between sensors. In addition, the inclination sensor adopts a universal joint connection structure, and each section has a 2-degree-of-freedom rotation compensation capability. During the lowering process, gravity automatically completes the posture leveling to keep each section aligned and stable on the vertical link. At the data level, this structure ensures that the posture data collected by each sensor can be uniformly solved in a three-dimensional coordinate system to avoid angular drift caused by mechanical errors or non-centered installation. All sensors synchronously record the initial value of the installation posture during the lowering process, and construct a deformation reference curve through the lowering interval distance Δz and the initial posture sequence θ(z) to facilitate correction and restoration in subsequent analysis, ensuring the physical consistency of the monitoring profile and data interpretability. This process realizes a complete closed loop from geometric calibration, posture control to data consistency, laying a technical foundation for high-precision, multi-node integration of inclination monitoring.

[0048] Preferably, it also includes: The warning level adopts four levels: Level IV prompt, Level III attention, Level II alert, and Level I emergency. The platform supports retrospective analysis of monitoring data of any period and automatically generates reinforcement treatment recommendations.

[0049] In this embodiment of the present invention, a four-level warning system is established: Level IV Warning, Level III Caution, Level II Alert, and Level I Emergency. Each level corresponds to a set of joint judgment rules for multiple monitoring indicators, such as displacement increment thresholds, inclination mutation thresholds, water content saturation thresholds, and microseismic signal energy thresholds. These rules are embedded in the platform's event triggering engine as logical expressions. At the data level, all sensor data is archived and stored using a time series structure. Each data record includes a timestamp, spatial location code, physical quantity category, numerical value, and sampling status identifier, supporting flexible indexing on a scale from minutes to days. By establishing a unified "data-threshold-level" mapping matrix, the system performs sliding window analysis on the data stream within each sampling period, automatically determining whether the trigger conditions for a specific warning level are met. To enable data backtracking for any period, the platform provides a time period selection interface, allowing users to select any time interval T1 to T2. The platform then automatically retrieves all archived raw data and event records within that interval in the background, constructing a cross-time indicator evolution trajectory. The retrospective analysis module re-executes the early warning logic reasoning chain based on a dynamic rule reconstruction algorithm to determine the change process of the warning level of each indicator under the conditions of historical data, and compares it with the actual early warning records to identify whether there are omissions or misjudgments. Subsequently, the platform's built-in reinforcement suggestion generation module combines high-risk indicators in the retrospective data, such as high-frequency inclination mutation sections, high water content continuous saturation sections, and continuous microseismic intervals, and matches them with the preset treatment strategy templates in the database to automatically generate a structured reinforcement suggestion report, including recommended construction methods (such as anchoring, drainage, slope cutting, etc.), the coordinate range of the reinforcement site, the treatment priority and the recommended implementation period. Finally, it is presented in a visual combination of graphics and text, and archived in the engineering management database to form a complete data-driven decision-making support process.

[0050] In this specification, a wireless communication system of a fingerprint machine is provided, which is used to execute the wireless communication method of the fingerprint machine mentioned above. The wireless communication system of the fingerprint machine includes: The multi-source monitoring and acquisition module is used to deploy inclination sensors, Beidou-GNSS modules, soil moisture probes, and microseismic accelerometers at and around the tower base to collect tower base tilt, three-dimensional displacement, soil moisture content, and vibration response information to form a raw monitoring data set. The data spatiotemporal fusion and feature construction module is used to unify the timestamps of the original monitoring data set, convert the WGS-84 projection coordinate system, and eliminate multi-frequency vibrations to construct an integrated air-ground displacement profile. Simultaneously, the displacement rate-acceleration joint feature is calculated using a 1-hour window and Pearson correlation analysis is performed with the moisture content series to generate a comprehensive feature tensor of the tower foundation deformation. The landslide stability assessment module is used to call the landslide stability assessment rule library, perform fuzzy reasoning on the comprehensive characteristic tensor of the tower base deformation, output the landslide probability, hazard level and the predicted displacement in the next 6 hours, and obtain the risk assessment result; The intelligent hierarchical warning and information push module is used to trigger a four-level warning strategy based on risk assessment results. Warning information is transmitted back via the NB-IoT-4G dual link. The platform simultaneously pushes SMS, voice, and pop-up notifications, and generates a time-stamped warning log for traceability. Wind-solar-lithium hybrid power supply module, which uses a wind-solar-lithium hybrid power supply system to provide continuous power to each monitoring node for ≥5 years and maintain normal operation under extreme conditions of 15 days of no sunlight; The energy closed-loop management module is used to automatically issue low-power sampling instructions through the platform when the battery state of charge is less than 20%, thus achieving a safe closed-loop power supply. Disaster enhancement observation trigger module, used to obtain warning information from the Meteorological Bureau; when the cloud platform detects that the continuous rainfall for 3 days is greater than or equal to 100mm or the Meteorological Bureau's warning information is an orange warning for upstream debris flow, the inclination and acceleration sampling frequency will be automatically increased from 5Hz to 20Hz, and the high-density observation mode will be enabled to obtain the full-process geological monitoring data of the tower base; data backtracking and reinforcement decision-making module: after the disaster is lifted, multi-strategy backtracking is performed on the full-process geological monitoring data of the tower base to generate a tower base landslide monitoring and reinforcement recommendation report, which is archived and managed by project.

[0051] The beneficial effect of the present invention is that, through the systematic acquisition and deep integration of multi-source sensor data, high-precision dynamic monitoring and early warning of tower base landslide risks are achieved. First, step S1 collects multi-dimensional raw data, including angular changes of the tilt sensor, three-dimensional displacement information from the Beidou-GNSS module, humidity data from the soil moisture probe, and vibration response from the microseismic accelerometer, forming a rich raw monitoring data set to ensure multi-angle capture of geological environmental changes. Subsequently, step S2 eliminates data asynchrony through unified timestamps, achieves a unified expression of spatial data using coordinate system transformation, and employs multi-frequency vibration rejection technology to reduce the impact of environmental noise, thereby constructing an integrated air-ground displacement profile and enhancing the spatial and temporal consistency of the data. In addition, the joint characteristics of displacement rate and acceleration are calculated through a sliding time window, and statistical correlation analysis is performed in combination with the moisture content series to form a high-dimensional comprehensive feature tensor, providing a quantitative representation of complex deformation patterns. Step S3 then inputs this comprehensive feature into the landslide stability assessment rule library, using fuzzy inference methods to integrate multivariate uncertainty, accurately determine landslide probability and hazard level, and perform short-term displacement trend prediction, improving the foresight and reliability of risk warnings. Finally, step S4 automatically triggers a four-level warning strategy based on the assessment results. Redundant data backhaul is achieved via NB-IoT and 4G dual links, ensuring stable and real-time information transmission. Multi-channel push notifications via SMS, voice, and pop-up windows ensure rapid response and widespread coverage of warning information. A structured, time-stamped warning log is generated to facilitate subsequent data tracing and event review. The overall process is highly integrated and automated across data collection, processing, fusion, and transmission, ensuring the timeliness, accuracy, and data integrity of landslide risk monitoring and providing scientific decision-making support for disaster prevention and mitigation.

[0052] The present invention is therefore intended to be illustrative and non-restrictive in all respects, with the scope of the invention being defined by the appended claims rather than the foregoing description, and all changes that come within the meaning and range of equivalents of the application documents are intended to be embraced therein.

[0053] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. A method for monitoring and early warning of geological landslides at tower bases, characterized in that: The following steps are involved: Step S1: Deploy inclination sensors, BeiDou-GNSS modules, soil moisture probes, and microseismic accelerometers at and around the tower base to collect tower base inclination, three-dimensional displacement, soil moisture content, and vibration response information to form a raw monitoring data set. Step S2: The original monitoring data set is timestamp unified, WGS-84 projection coordinate system converted, and multi-frequency vibration is eliminated to construct an integrated air-ground displacement profile. Using a 1-hour window, the displacement rate-acceleration joint feature is calculated and Pearson correlation analysis is performed with the water content series to generate a comprehensive characteristic tensor of the tower foundation deformation. Step S3: Call the landslide stability assessment rule library, perform fuzzy reasoning on the tower base deformation comprehensive characteristic tensor, output the landslide probability, hazard level and predicted displacement in the next 6 hours, and obtain the risk assessment result; Step S4: Based on the risk assessment results, the four-level warning strategy is triggered. The warning information is transmitted back via the NB-IoT-4G dual link. The platform simultaneously pushes SMS, voice and pop-up windows, and generates a warning log with a timestamp.

2. The tower base landslide monitoring and early warning method according to claim 1 is characterized in that: Step S1 includes: Step S11: Fix at least three ±0.01° MEMS tilt sensors at the tower base, the middle and the upper part of the landslide body, respectively, with a sampling frequency of 5 Hz; Step S12: Beidou-GNSS antennas are deployed at the tower base and the trailing edge of the mountain to continuously track the carrier wave, achieving relative positioning accuracy at the millimeter level. Step S13: Lower the steel cable and anti-twist rod in sections; bury soil moisture probes in a vertical direction along the potential slip surface, with a measurement range of 0-100% Vol, wherein the moisture probes are buried in layers at depths of 0.5m, 1m, and 2m; Step S14: Install triaxial accelerometers at the middle of the tower and at the base, with a range of ±16g, to capture microseismic signals. The microseismic accelerometers are installed with bidirectional damping to suppress errors. Step S15: After performing wavelet noise reduction and outlier removal on the multi-source sampling sequence, an original monitoring data set is formed.

3. The tower base landslide monitoring and early warning method according to claim 1 is characterized in that: The construction of the air-ground combined displacement profile in step S2 includes: Calculate the tower base horizontal displacement curve based on the BeiDou-GNSS displacement point sequence and perform least squares smoothing; The accumulated angle of the tilt sensor is converted into vertical displacement through the rod length, and then fused with the GNSS curve to generate a unified ground-underground displacement profile; The Pearson correlation coefficient between water content and displacement rate is used to mark the potential sliding surface in layers, and a comprehensive characteristic tensor of tower foundation deformation is formed.

4. The tower base landslide monitoring and early warning method according to claim 1 is characterized in that: The landslide stability assessment rule base in step S3 includes: The daily displacement increment threshold is: if the horizontal displacement growth rate > 2mm / d, it is considered as a Level III warning; Tilt sudden change threshold: a single change of >0.05° for any sensor triggers a Level II warning; Moisture content saturation threshold: if the sensor detects that the moisture content is greater than 80% Vol. and the displacement acceleration increases, a Level I warning will be directly triggered; The microseismic frequency threshold is 1-10Hz, and the energy peak lasting for more than 3 minutes will increase to level 1 warning.

5. The tower base geological landslide monitoring and early warning method according to claim 1 is characterized in that: Step S4 uses dual-channel alarm: The wireless NB-IoT-4G channel is used for regular data backhaul and platform pop-up windows; The SMS voice gateway sends alarms to the operation and maintenance manager and the power supply command center at the same time when the alarm is at level I or II.

6. The tower base landslide monitoring and early warning method according to claim 1 is characterized in that: Also includes: Step S5: Utilize the wind-solar-lithium hybrid power supply system to continuously power each monitoring node for ≥5 years, and maintain normal operation under the extreme condition of continuous light absence for 15 days, thereby obtaining the battery state of charge; Step S6: When the battery state of charge (SOC) is less than 20%, the platform automatically issues a low sampling rate instruction and switches to narrowband standby mode; Step S7: Obtaining warning information from the Meteorological Bureau; when the cloud platform detects that the rainfall for three consecutive days is greater than or equal to 100 mm or the Meteorological Bureau's warning information is an orange warning for upstream debris flow, the inclination and acceleration sampling frequency is automatically increased from 5 Hz to 20 Hz, and the high-density observation mode is enabled to obtain full-process geological monitoring data of the tower base; Step S8: After the disaster is resolved, perform multi-strategy backtracking on the full-process geological monitoring data of the tower base to generate a tower base landslide monitoring and reinforcement recommendation report, and archive and manage it according to the project.

7. The tower base landslide monitoring and early warning method according to claim 1 is characterized in that: The monitoring installation benchmark point is located in a stable area 5-10m outside the extension line of the tower line direction, and the bottom of the hole below the benchmark point remains relatively still with the landslide body.

8. The tower base landslide monitoring and early warning method according to claim 1 is characterized in that: Before installing the inclinometer, the verticality of the inclinometer tube must be checked. The vertical deviation must not exceed 30% of the full scale of the sensor. All sensors should be aligned with the guide groove in the same direction and connected and straightened through universal joints before being lowered in sequence.

9. The tower base landslide monitoring and early warning method according to claim 1, characterized in that: The warning level adopts four levels: Level IV prompt, Level III attention, Level II alert, and Level I emergency. The platform supports retrospective analysis of monitoring data of any period and automatically generates reinforcement treatment recommendations.

10. A tower base geological landslide monitoring and early warning system, characterized in that: The system is used to implement the tower base geological landslide monitoring and early warning method according to any one of claims 1 to 9, comprising: The multi-source monitoring and acquisition module is used to deploy inclination sensors, Beidou-GNSS modules, soil moisture probes, and microseismic accelerometers at and around the tower base to collect tower base tilt, three-dimensional displacement, soil moisture content, and vibration response information to form a raw monitoring data set. The data spatiotemporal fusion and feature construction module is used to unify the timestamps of the original monitoring data set, convert the WGS-84 projection coordinate system, and eliminate multi-frequency vibrations to construct an integrated air-ground displacement profile. Simultaneously, the displacement rate-acceleration joint feature is calculated using a 1-hour window and Pearson correlation analysis is performed with the moisture content series to generate a comprehensive feature tensor of the tower foundation deformation. The landslide stability assessment module is used to call the landslide stability assessment rule library, perform fuzzy reasoning on the comprehensive characteristic tensor of the tower base deformation, output the landslide probability, hazard level and the predicted displacement in the next 6 hours, and obtain the risk assessment result; The intelligent hierarchical warning and information push module is used to trigger a four-level warning strategy based on risk assessment results. Warning information is transmitted back via the NB-IoT-4G dual link. The platform simultaneously pushes SMS, voice, and pop-up notifications, and generates a time-stamped warning log for traceability. Wind-solar-lithium hybrid power supply module, which uses a wind-solar-lithium hybrid power supply system to provide continuous power to each monitoring node for ≥5 years and maintain normal operation under extreme conditions of 15 days of no sunlight; Energy closed-loop management module, used to automatically issue low sampling rate instructions and switch to narrowband standby mode when the battery state of charge (SOC) is less than 20%; Disaster enhancement observation trigger module, used to obtain warning information from the Meteorological Bureau; when the cloud platform detects that the continuous rainfall for 3 days is greater than or equal to 100mm or the Meteorological Bureau's warning information is an orange warning for upstream debris flow, the inclination and acceleration sampling frequency will be automatically increased from 5Hz to 20Hz, and the high-density observation mode will be enabled to obtain the full-process geological monitoring data of the tower base; data backtracking and reinforcement decision-making generation module: after the disaster is lifted, multi-strategy backtracking is performed on the full-process geological monitoring data of the tower base to generate a tower base landslide monitoring and reinforcement recommendation report, which is archived and managed by project.

Images