Internet of things driven intelligent termite monitoring system for water conservancy projects

Abstract

The application discloses an Internet of Things driven intelligent termite monitoring system for hydraulic engineering and relates to the technical field of water conservancy monitoring.The system comprises the following steps: collecting real-time termite activity sound wave signals, soil vibration data and environmental parameters, performing pretreatment such as sliding window filtering and cubic spline interpolation, and constructing a three-dimensional correlation entropy matrix of temperature and humidity-soil disturbance-wood damage.A fuzzy Bayesian neural network model based on random forest feature weight distribution can output low / medium / high risk probability distribution, and regional risk quantification can be performed in combination with water level fluctuation, soil permeability coefficient and facility importance coefficient.The system fuses facility abnormal use coefficient and diffusion enhanced structure risk coefficient, generates a priority index through a dynamic early warning adjustment model, and triggers a hierarchical response strategy.The system plans a drug spraying path through a genetic algorithm, realizes 72-hour diffusion thermal force prediction and precise prevention and control, improves termite monitoring recognition rate, reduces energy consumption, and improves the intelligent management and protection level of hydraulic engineering facilities.

Description

Technical Field

[0001] This invention belongs to the field of water conservancy monitoring technology, specifically relating to an intelligent termite monitoring system for water conservancy projects driven by the Internet of Things. Background Technology

[0002] Termite damage in water conservancy projects is a key risk threatening the safety of facilities such as dikes and sluices. Traditional monitoring relies on manual patrols and fixed-point observations, which suffer from problems such as high lag and limited coverage, making it difficult to detect the hidden damage caused by termite activity (such as internal wood decay) in a timely manner.

[0003] Existing technologies mostly use single sensors (such as sound wave detection or humidity monitoring), but lack collaborative analysis of multi-dimensional environmental parameters (temperature, humidity, soil vibration) and have not established dynamic risk assessment models, resulting in insufficient accuracy of early warning. While existing technologies attempt to use single sensors for monitoring, they suffer from significant shortcomings in physical signal detection: First, acoustic detection devices are mostly deployed at single points, capturing only signals in a limited frequency band. They lack frequency band optimization for characteristic vibrations caused by termites gnawing on wood and low-frequency acoustic emission pulses caused by worker ant movement (pulse repetition rates vary significantly due to species specificity). This results in insufficient sensitivity to capture weak physical signals, making it difficult to distinguish termite activity from environmental noise. Second, soil vibration monitoring often uses general-purpose sensors, lacking the ability to synchronously acquire three-dimensional vibration vectors. This prevents nest location from being determined by phase differences and propagation time differences in vibration signals, leading to a lack of correlation analysis between physical signals and biological activities. Third, existing sensing systems lack a distributed network architecture, enabling only single-point data acquisition. They cannot achieve spatial positioning and propagation path tracking of termite activity through multi-node collaboration, and lack physical signal correction mechanisms for complex environments in water conservancy projects (such as changes in soil moisture and water level fluctuations), making the detection results susceptible to environmental interference.

[0004] Furthermore, the spread of termite damage exhibits nonlinear spatiotemporal characteristics, making it difficult for existing systems to quantify its propagation paths and risk levels. Moreover, they cannot prioritize warnings based on the importance of water conservancy facilities (such as the differentiated protection needs of main dams and secondary channels). Simultaneously, traditional control measures (such as pesticide spraying) lack intelligent path planning, easily leading to pesticide waste or overlooking high-risk areas.

[0005] To address the aforementioned issues, this system proposes an IoT-driven, multi-dimensional intelligent termite monitoring method to provide intelligent solutions. Summary of the Invention

[0006] To overcome the shortcomings and deficiencies of the existing technology, the present invention adopts the following technical solution:

[0007] The Internet of Things-driven intelligent termite monitoring system for water conservancy projects includes: Data acquisition and preprocessing module: Real-time acquisition of termite activity acoustic signals, soil vibration data and environmental parameters; data preprocessing includes noise filtering, missing value interpolation; calculation of correlation entropy matrix to quantify the regularity of parameter fluctuations; Hazard assessment and regional feature extraction module: Extracts key indicators including humidity mutation rate, vibration anomaly frequency, and stress change gradient, and generates termite damage dominant factors by combining a weight model trained with historical data; outputs low / medium / high risk probability distributions by using a fuzzy Bayesian neural network model combined with wood damage index and temperature and humidity data; identifies environmental stability parameters and facility importance coefficients to assist in risk assessment. Multi-dimensional data fusion and risk quantification module: Analyzes facility usage frequency and combines it with structural risk coefficients to quantify the impact of termite spread on water conservancy projects; identifies the geometric and dynamic characteristics of termite activity hotspots; and integrates facility usage frequency and structural risk weights to generate comprehensive early warning priorities. Dynamic early warning decision and execution module: Based on hazard score and early warning priority, calculate early warning level and classify it into high / medium / low risk; trigger response strategies include level 1 early warning, level 2 early warning and routine monitoring; adjust the volume or frequency of early warning equipment according to changes in hazard score to optimize energy consumption; generate a 72-hour diffusion heat map and optimize the drug spraying path through genetic algorithm to prioritize coverage of high-risk areas.

[0008] Preferably, the system workflow is as follows: Data acquisition and preprocessing: Real-time acquisition of termite activity acoustic signals, soil vibration data, and environmental temperature and humidity data is performed through an acoustic emission sensor array, a miniature soil pressure sensor, and a temperature and humidity sensor network. Noise filtering, missing value interpolation, and correlation entropy calculation are also performed. Hazard assessment and regional feature extraction: Based on the preprocessed data, the dominant factors, the probability distribution of termite hazard levels, the hazard score, and regional environmental characteristic parameters are calculated; Multi-dimensional data fusion and risk quantification: combining facility usage frequency analysis, diffusion-enhanced structural risk calculation, diffusion feature extraction, and early warning priority quantification model to generate risk assessment results; Dynamic early warning decision-making and execution: Trigger a graded early warning mechanism based on risk assessment results, dynamically adjust monitoring strategies, and generate diffusion heat maps and optimize drug spraying paths through GIS maps.

[0009] Preferably, the data acquisition and preprocessing steps include: The acoustic emission sensor array is used to collect the acoustic wave signals of termite activity and generate the wood damage index MD (0~1). Soil vibration frequency Fs and amplitude Sa were collected using a miniature soil pressure sensor and denoted as Sn(t). Monitor the ambient temperature T and relative humidity RH through a temperature and humidity sensor network; Perform sliding window filtering on soil pressure data with a window size of 5s to eliminate outliers; Complement the missing temperature and humidity data using the cubic spline interpolation method, and construct a three-dimensional correlation entropy matrix E(RHT,Fs,MD).

[0010] Preferably, the steps of hazard assessment and regional feature extraction include: Calculate the humidity mutation rate, vibration anomaly frequency and stress change gradient as key indicators; Determine the weight w=[0.4, 0.3, 0.3] through random forest feature importance analysis, and calculate the dominant factor; Input the preprocessed parameters into the fuzzy Bayesian neural network model, and output the hazard level probability distribution, where the low-risk yd probability is <30%, the medium-risk yz probability is in the range of 30%-70%, and the high-risk yg probability is >70%; Calculate the final hazard score in combination with the hazard scoring formula, where the weight coefficient is determined by training on historical cases.

[0011] Preferably, the steps of multi-dimensional data fusion and risk quantification include: Record the sluice operation frequency and inspection path through IoT devices, and calculate the abnormal usage coefficient YCuf; Calculate the structural risk coefficient based on the distance from the monitoring point to the key structure, the norm of diffusion intensity and the shape complexity; Extract the geometric features and dynamic features of the diffusion features; the geometric features include the hot spot convex hull area Area and the shape complexity C; the dynamic features include the area change rate and the diffusion direction angle θ; Fuse the abnormal usage coefficient and the structural risk coefficient to construct an early warning regulation assessment model.

[0012] Preferably, the steps of dynamic early warning decision-making and execution include:

[0013] Trigger hierarchical responses according to the early warning level WLevel: Level 1 early warning (red, WLevel>0.7): UAV inspection, encrypted sensor sampling, and dam structure acoustic scanning; Level 2 early warning (yellow, 0.4<WLevel≤0.7): GIS heat map push and manual verification task dispatch; Conventional monitoring (green, WLevel≤0.4): Data synchronization to the database and model iterative storage; Dynamic adjustment strategy: If the current hazard score increases by more than 20% compared with the previous cycle, increase the volume of the early warning device by 3dB; if the score is <0.3 for 3 consecutive cycles, reduce the early warning frequency to once per hour.

[0014] Preferably, the decision support system includes: Generate a termite spread heat map for the next 72 hours based on GIS maps, and mark high-risk areas; The genetic algorithm is used to optimize the drug spraying path, minimize drug consumption, and prioritize coverage of hotspot areas.

[0015] Preferably, the hazard score calculation formula is as follows: ; in, The dominant factor is yd; yz and yg are the low / medium / high risk probabilities output by the fuzzy Bayesian network; weight coefficients are used to determine the weighting factors. , , Determined through training with historical cases; The dominant factor weights are determined using the AHP method as correction terms.

[0016] Preferably, the formula for calculating the risk coefficient of the diffusion-enhanced structure is as follows: ; in, The distance from the monitoring point to the nearest critical structure; Average distance; This represents the distance from the standard deviation. (where the diffusion intensity modulus is dimensionless) ; The average area change rate over 3 hours; The diffusion direction angle; Let represent the shape complexity.

[0017] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. This invention utilizes a three-dimensional sensing system comprised of an acoustic emission sensor array, a miniature soil pressure sensor, and a temperature and humidity sensor network to simultaneously acquire acoustic signals of termite activity (gnawing frequency, movement intensity), soil vibration parameters (frequency, amplitude), and environmental microclimate (temperature and humidity). Combined with preprocessing techniques such as sliding window filtering and cubic spline interpolation, noise interference is effectively eliminated and missing data is supplemented. A correlation entropy matrix of temperature / humidity, soil disturbance, and wood damage is constructed to quantify the coupling patterns of these parameters. This multi-source data fusion mechanism enhances the sensitivity of termite activity identification, reduces the false negative rate of traditional single-parameter monitoring, and enables a matched intelligent sensing monitoring system to achieve full-domain coverage and accurate capture of physical signals. 2. This invention utilizes a fuzzy Bayesian neural network model based on random forest feature weight allocation, trained with historical termite nest data, to output low / medium / high risk probability distributions in real time. The system further incorporates environmental stability indicators such as water level fluctuations and soil permeability coefficients, as well as facility importance coefficients, and corrects the weights of dominant factors using the AHP method to construct a diffusion-enhanced structural risk quantification model. This model can dynamically calculate the early warning priority index and trigger a three-level response strategy. 3. This invention generates a termite spread heatmap for the next 72 hours using a genetic algorithm and plans pesticide spraying paths, reducing pesticide consumption while covering high-risk areas. Dynamic adjustment strategies (such as adjusting warning volume or frequency based on hazard scores) further reduce ineffective energy consumption. Furthermore, the fusion assessment of facility anomaly usage coefficients and structural risk coefficients allows maintenance resources to be allocated towards key facilities such as main dams, while secondary channels are flexibly allocated according to risk levels, forming a "prediction-decision-execution" closed loop, significantly improving the intelligent management and maintenance level of water conservancy facilities. It not only significantly improves the intelligent management and maintenance level of water conservancy facilities but also constructs a system architecture suitable for intelligent termite monitoring through innovation in core physical detection technologies. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This invention illustrates a block diagram of an IoT-driven intelligent termite monitoring system for water conservancy projects.

[0020] Figure 2 A flowchart of the IoT-driven intelligent termite monitoring system for water conservancy projects of the present invention is shown;

[0021] Figure 3 The flowchart of the hazard assessment and regional feature extraction of the present invention is shown. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more exemplary embodiments. Numerous specific details are provided in the following description to give a full understanding of exemplary embodiments of this disclosure. However, those skilled in the art will recognize that the technical solutions of this disclosure can be practiced with one or more of the specific details omitted, or other methods, components, steps, etc., can be employed. In other instances, well-known structures, methods, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of this disclosure.

[0024] Example 1: See Figure 1 As shown, the IoT-driven intelligent termite monitoring system for water conservancy projects in this embodiment includes: Data acquisition and preprocessing module: Real-time acquisition of termite activity acoustic signals, soil vibration data, and environmental parameters (temperature and humidity) through a network of acoustic emission sensors, soil pressure sensors, and temperature and humidity sensors; data preprocessing includes noise filtering (sliding window filtering), missing value interpolation (cubic spline interpolation); calculation of the correlation entropy matrix to quantify the regularity of parameter fluctuations.

[0025] The hazard assessment and regional feature extraction module extracts key indicators such as humidity mutation rate, vibration anomaly frequency, and stress change gradient, and generates the dominant factor of termite damage by combining a weight model trained with historical data; it outputs low / medium / high risk probability distributions by using a fuzzy Bayesian neural network model combined with wood damage index, temperature and humidity data, etc.; and it identifies environmental stability parameters (water level fluctuation, soil permeability coefficient) and facility importance coefficients to assist in risk assessment.

[0026] Multi-dimensional data fusion and risk quantification module: Analyzes facility usage frequency (calculates abnormal usage coefficient), combines structural risk coefficient (diffusion intensity and direction angle of distance from key structures, etc.) to quantify the impact of termite spread on water conservancy projects; identifies the geometric characteristics (convex hull area, shape complexity) and dynamic characteristics (area change rate, diffusion direction) of termite activity hotspots; integrates facility usage frequency and structural risk weights to generate comprehensive early warning priorities.

[0027] Dynamic early warning decision and execution module: Based on hazard score and early warning priority, calculate early warning level and classify it into red (high risk), yellow (medium risk), and green (low risk); trigger response strategies include first-level early warning, second-level early warning, and routine monitoring; adjust the volume or frequency of early warning equipment according to changes in hazard score to optimize energy consumption; generate a 72-hour diffusion heat map and optimize the drug spraying path through genetic algorithm to prioritize coverage of high-risk areas.

[0028] The beneficial effects of this embodiment are as follows: the system achieves real-time monitoring, accurate risk assessment and dynamic response to termite disasters through multi-source data fusion and intelligent algorithms, which significantly improves the efficiency of identifying hidden dangers in water conservancy projects, reduces the false alarm rate, optimizes resource allocation, extends facility life, and supports intelligent decision-making to minimize prevention and control costs.

[0029] Example 2: See Figure 2 As shown in this embodiment, the IoT-driven intelligent termite monitoring system for water conservancy projects operates as follows: Step 1: Data Acquisition and Preprocessing.

[0030] S11, Multidimensional Data Acquisition: Acoustic emission sensor array: captures the acoustic signals of termite activity in real time (gnawing frequency Fa, movement intensity Am), and generates the wood damage index MD (0~1, the higher the value, the more severe the internal decay) through spectrum analysis. Miniature soil pressure sensor: Collects soil vibration frequency Fs and amplitude Sa, labeled as Sn(t) (n = sensor number); Temperature and humidity sensor network: monitors ambient temperature (T) and relative humidity (RH) to capture changes in the microenvironment of termite activity.

[0031] S12, Data Preprocessing: Noise filtering: for raw soil pressure data Perform sliding window filtering (window size 5s) to remove outliers (threshold set to the mean). , (Standard deviation), retain valid data. ; Missing value interpolation: Missing temperature and humidity data are filled using cubic spline interpolation. The formula is as follows: Where a, b, and d are interpolation coefficients, obtained by fitting adjacent known points; The time is for adjacent known points.

[0032] Correlation entropy calculation: Extract 10-minute time-series data for each parameter (e.g., temperature and humidity RHT, soil disturbance frequency Fs, wood damage index MD), and calculate the self-correlation entropy: ; Where S is the parameter sequence; The probability of the value occurring reflects the regularity of data fluctuations, and a three-dimensional correlation entropy matrix E(RHT,Fs,MD) is constructed.

[0033] Step 2: Hazard assessment and regional feature extraction.

[0034] See Figure 3 As shown, the specific process for hazard assessment and regional feature extraction is as follows: S20, Calculation of dominant factors; Key metrics extraction: Humidity mutation rate: ( ); Abnormal vibration frequency: (times / minute); Stress variation gradient: (Reflects stress relaxation in wood); Weight allocation: The weights w=[0.4,0.3,0.3] were determined by training with historical termite activity data (random forest feature importance analysis).

[0035] Dominant factor formula: .

[0036] S21. Prediction of termite damage level.

[0037] Preprocessed soil disturbance parameters Damage characteristics of timber structures (MD), temperature and humidity data (RHT), and correlation entropy matrix (E) are input into a fuzzy Bayesian neural network model, which outputs a probability distribution of hazard levels.

[0038] Low-risk yd (<30% probability)

[0039] Medium risk (yz) (30%-70% probability)

[0040] High-risk YG (>70% probability)

[0041] S22, Hazard score calculation.

[0042] ; in, The dominant factor is yd; yz and yg are the low / medium / high risk probabilities output by the fuzzy Bayesian network; weight coefficients are used to determine the weighting factors. , , Determined through training with historical cases; It is the dominant factor weight, used as a correction term (determined by AHP method).

[0043] S23. Regional Feature Collection: Obtain the water level fluctuation frequency (WF) and soil permeability coefficient (SP) of the monitoring area to reflect the stability of the surrounding environment of the water conservancy project. Facility importance: Define the facility importance coefficient CI (with a value of 0 or 1; the higher the value, the more critical the facility, such as main dam CI=1, secondary canal CI=0.5, sluice gate CI=0.8).

[0044] Step 3: Multi-dimensional data fusion and risk quantification.

[0045] S31. Facility Usage Frequency Analysis: Data such as sluice gate operation frequency and inspection routes are recorded through IoT devices, and an abnormal usage coefficient YCuf is defined. Set the frequency threshold to use. The frequency of use exceeding the threshold within the statistical period ; The calculation formula is as follows: This reflects the impact of abnormal facility operation on termite activity monitoring.

[0046] S32, Diffusion-enhanced structural risk.

[0047] Structural risk coefficient ;in, The distance from the monitoring point to the nearest critical structure; Average distance; This represents the distance from the standard deviation. Diffusion intensity modulus (dimensionless): ;in, The average area change rate over 3 hours; The diffusion direction angle; Let represent the shape complexity.

[0048] S33, Diffusion Feature Extraction.

[0049] Multi-parameter hotspot identification: ; Geometric features: Area (convex hull of hotspot), shape complexity C.

[0050] Dynamic characteristics: Area change rate 2. Diffusion direction angle θ (angle with the direction of water flow, 0° in the same direction, 90° perpendicular).

[0051] S34. Quantitatively construct an early warning adjustment and evaluation model by prioritizing early warnings, integrating abnormal usage coefficients and structural risk coefficients: ; Among them, the weighting coefficient (Facility usage frequency weight) (Structural risk weights).

[0052] Step 4: Dynamic early warning decision-making and execution.

[0053] S41, Dissemination Perception Early Warning Level.

[0054] Warning level: ; .

[0055] S42, Warning Level Trigger Logic.

[0056] Level 1 Warning (Red, WLevel>0.7): 1. Drones along Inspection of the fan-shaped area (focusing on scanning in the direction of water diffusion); 2. (Maximum 2 times).

[0057] 3. Initiate acoustic scanning of the dam structure (once per hour).

[0058] Level II Warning (Yellow, 0.4) <WLevel≤0.7):

[0059] 1. The operation and maintenance platform pushes GIS heatmaps containing diffusion vector arrows (arrow length = area change rate, direction = θ).

[0060] 2. Encrypt sensor sampling (every 10 minutes), manual verification of task assignment.

[0061] Routine monitoring (green, WLevel≤0.4): 1. Data is synchronized to the database every hour; 2. Edge nodes locally store the association entropy matrix for model iteration.

[0062] S43. Dynamic adjustment strategy: If the current PGTJ increases by more than 20% compared to the previous cycle, increase the volume of the warning device by 3dB to enhance the warning effect; if PGTJ < 0.3 for 3 consecutive cycles, reduce the warning frequency to once per hour to reduce energy consumption.

[0063] S44. The decision support system generates a termite spread heat map for the next 72 hours based on a GIS map, marking high-risk areas; it optimizes the pesticide spraying path using a genetic algorithm to minimize pesticide consumption and prioritize coverage. Hotspot areas.

[0064] The beneficial effects of this embodiment are as follows: by using multi-source sensor fusion and intelligent algorithms to achieve real-time monitoring, accurate prediction and dynamic response to termite disasters, it significantly improves the efficiency of identifying hidden dangers in water conservancy projects, reduces the false alarm rate, optimizes resource allocation, extends facility life, and provides scientific decision support for the intelligent management and maintenance of water conservancy facilities.

[0065] The weighting coefficients of this invention are used to measure the degree of influence of different factors or variables on a certain outcome or decision. The weighting coefficient is defined as the numerical value assigned to each factor when comparing and evaluating multiple factors, reflecting their importance or priority. These weighting coefficients can be determined according to specific circumstances and needs, and are usually jointly formulated and confirmed by professionals or relevant stakeholders. By reasonably setting the weighting coefficients, programs or systems can be helped to make decisions or predictions more accurately.

[0066] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0067] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0068] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0069] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0070] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0071] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An IoT-driven intelligent termite monitoring system for water conservancy projects, characterized in that: The system includes: Data Acquisition and Preprocessing Module: This module collects termite activity acoustic signals, soil vibration data, and environmental temperature and humidity data in real time using an acoustic emission sensor array, a miniature soil pressure sensor, and a temperature and humidity sensor network. It then performs noise filtering, missing value interpolation, and correlation entropy calculation. The acoustic emission sensor array collects termite activity acoustic signals to generate a wood damage index (MD), with values ​​ranging from 0 to 1. The miniature soil pressure sensor collects soil vibration frequency and amplitude, labeled as Sn(t). The temperature and humidity sensor network monitors environmental temperature (T) and relative humidity (RH). Sliding window filtering is applied to the soil pressure data, with a set window size to remove outliers. Missing temperature and humidity data are completed using cubic spline interpolation, and a three-dimensional correlation entropy matrix E(RHT,Fs,MD) is constructed. The correlation entropy matrix is ​​used to analyze the regularity of parameter fluctuations. The hazard assessment and regional feature extraction module calculates dominant factors, termite hazard level probability distribution, hazard score, and regional environmental characteristic parameters based on preprocessed data; it calculates humidity mutation rate, vibration anomaly frequency, and stress change gradient as key indicators; it determines weights w through random forest feature importance analysis and calculates dominant factors; it inputs the preprocessed parameters into a fuzzy Bayesian neural network model to output hazard level probability distributions: low risk yd probability <30%, medium risk yz probability ∈ 30%-70%, and high risk yg probability >70%; it calculates the final hazard score using the hazard scoring formula, where weight coefficients are determined through training with historical cases; and it identifies environmental stability parameters and facility importance coefficients to assist in risk assessment. Multi-dimensional data fusion and risk quantification module: Analyzes facility usage frequency and combines it with structural risk coefficients to quantify the impact of termite spread on water conservancy projects; identifies the geometric and dynamic characteristics of termite activity hotspots; integrates facility usage frequency and structural risk weights to generate comprehensive early warning priorities; combines facility usage frequency analysis, diffusion-enhanced structural risk calculation, diffusion feature extraction, and early warning priority quantification models to generate risk assessment results; records sluice gate operation frequency and inspection paths through IoT devices to calculate the abnormal usage coefficient YCuf; calculates the structural risk coefficient based on the distance from monitoring points to key structures, diffusion intensity modulus, and shape complexity; extracts the geometric and dynamic characteristics of diffusion features; geometric features include the hotspot convex hull area (Area) and shape complexity (C); dynamic features include the area change rate and diffusion direction angle (θ); integrates the abnormal usage coefficient and structural risk coefficient to construct an early warning adjustment and assessment model. Dynamic early warning decision-making and execution module: calculating early warning levels, which are divided into high / medium / low risks; triggering response strategies; adjusting the volume or frequency of early warning devices according to changes in hazard scores; generating a 72-hour spread heat map, optimizing the pesticide spraying path, and giving priority to covering high-risk areas; triggering a hierarchical early warning mechanism according to the risk assessment result, dynamically adjusting the monitoring strategy, and generating a spread heat map and optimizing the pesticide spraying path through a GIS map; triggering hierarchical responses according to the early warning level WLevel: Red alert, WLevel>0.7 is level-I alert: UAV inspection, encrypted sensor sampling, and sonic scanning of dam structures; Yellow alert, 0.4<WLevel≤0.7 is level-II alert: GIS heat map push and manual verification task dispatching; Green alert, WLevel≤0.4 is routine monitoring: synchronizing data to the database and storing it for model iteration; Dynamic adjustment strategy: if the current hazard score rises by more than the set score threshold compared with the previous cycle, increase the volume of the early warning device to the set volume value; if the score is less than the set score threshold for 3 consecutive cycles, reduce the early warning frequency to once per hour.

2. The IoT-driven intelligent termite monitoring system for water conservancy projects according to claim 1, characterized in that, The dynamic early warning decision-making and execution module further comprises: generating a 72-hour termite spread heat map in the future based on a GIS map, and marking high-risk areas; optimizing the pesticide spraying path through a genetic algorithm, so as to minimize the pesticide consumption and give priority to covering hot spot areas.

3. The IoT-driven intelligent termite monitoring system for water conservancy projects according to claim 1, characterized in that, The calculation formula for the hazard score is: ； in, The dominant factor is yd; yz and yg are the low / medium / high risk probabilities output by the fuzzy Bayesian network; weight coefficients are used to determine the weighting factors. , , Determined through training with historical cases; The dominant factor weights are determined using the AHP method as correction terms.

4. The IoT-driven intelligent termite monitoring system for water conservancy projects according to claim 1, characterized in that, The calculation formula for the diffusion-enhanced structural risk coefficient is: ； Wherein, CI is the facility importance coefficient, with a value ranging from 0 to 1. The distance from the monitoring point to the nearest critical structure; Average distance; This represents the distance from the standard deviation. The diffusion intensity modulus is dimensionless. ;in, The average area change rate over 3 hours; The diffusion direction angle; Let represent the shape complexity.

Images