A multi-factor fusion method and system for real-time risk assessment of ant infestations

Abstract

This application discloses a multi-factor fusion method and system for real-time risk assessment of ant infestations, relating to the field of water conservancy engineering safety monitoring technology. The method includes: acquiring multiple ant infestation data sequences, multiple humidity sequences, and multiple temperature sequences; performing soil excavability analysis to obtain multiple excavability sequences; performing ant infestation evolution rationality analysis to obtain multiple evolution rationality coefficients; performing distribution rationality analysis on the multiple humidity sequences and multiple temperature sequences to obtain distribution rationality coefficients; and obtaining multiple real-time risk parameters based on the excavability sequences and ant infestation data sequences, performing risk compensation, and obtaining multiple compensated real-time risk parameters as the assessment result. This solves the technical problem that existing technologies fail to fully consider the dynamic correlation between soil excavability and ant infestation evolution, and the impact of the spatiotemporal distribution rationality of environmental factors on the assessment results, leading to insufficient accuracy in risk assessment.

Description

Technical Field

[0001] This application relates to the field of water conservancy project safety monitoring technology, specifically to a multi-factor fusion method and system for real-time risk assessment of ant infestations. Background Technology

[0002] With the continuous expansion of the scale of water conservancy projects and the sustained increase in their service life, termite damage to water conservancy facilities such as dikes and reservoirs is becoming increasingly prominent. As a social insect, termites' nest-building, feeding, and digging activities can cause gradual damage to the soil structure, which may lead to major safety accidents such as piping, leakage, or even dam failure in severe cases.

[0003] However, traditional termite monitoring methods mainly rely on regular manual inspections and simple baiting devices, which suffer from low monitoring frequency, limited coverage, and poor data continuity. In existing technologies, with the development of IoT and sensor technologies, intelligent monitoring methods based on multi-source data fusion have gradually become a research hotspot. Some solutions have achieved automatic collection of environmental parameters and termite data by deploying temperature and humidity sensors and termite activity monitoring equipment, but they still have shortcomings in risk assessment.

[0004] On the one hand, existing methods often simply correlate environmental factors such as temperature and humidity with termite data, failing to fully consider the constraints of soil physical properties on termite digging behavior. Soil diggability directly affects the expansion speed and direction of termite nests, and soil diggability is affected by the coupling effect of temperature and humidity.

[0005] On the other hand, existing risk assessment models are mostly based on static threshold judgments or data comparisons at a single time point, ignoring the temporal characteristics and spatial distribution patterns of ant infestation evolution. This makes it difficult for the assessment results to reflect the true development trend of ant infestation, and it is easy to have false alarms or omissions. Summary of the Invention

[0006] This application provides a multi-factor fusion method and system for real-time risk assessment of ant infestations, which solves the technical problem that existing technologies fail to fully consider the dynamic correlation between soil excavability and ant infestation evolution, as well as the impact of the spatiotemporal distribution of environmental factors on the assessment results, resulting in insufficient accuracy of risk assessment.

[0007] The technical solution to the above-mentioned technical problems in this application is as follows:

[0008] Firstly, this application provides a multi-factor fusion method for real-time risk assessment of ant infestations, the method comprising:

[0009] Acquire multiple ant infestation data sequences, multiple humidity sequences, and multiple temperature sequences from multiple monitoring points within the target monitoring area;

[0010] Based on the multiple humidity sequences and multiple temperature sequences, soil excavability analysis is performed to obtain multiple excavability sequences. Combined with the multiple ant infestation data sequences, ant infestation evolution rationality analysis is performed to obtain multiple evolution rationality coefficients. Among them, the ant infestation evolution rationality analysis includes trend rationality analysis and time series rationality analysis.

[0011] Based on historical monitoring data of the target monitoring area, a distribution rationality analysis was conducted on multiple humidity sequences and multiple temperature sequences to obtain the distribution rationality coefficient;

[0012] Based on multiple mineability sequences and multiple ant population data sequences, multiple real-time risk parameters are obtained. Risk compensation is performed based on the multiple evolution rationality coefficients and distribution rationality coefficients to obtain multiple compensated real-time risk parameters, which serve as the evaluation results.

[0013] Secondly, this application provides a multi-factor fusion-based real-time risk assessment system for ant infestations, including:

[0014] The information acquisition module is used to acquire multiple ant infestation data sequences, multiple humidity sequences, and multiple temperature sequences from multiple monitoring points within the target monitoring area;

[0015] The sequence analysis module is used to perform soil excavability analysis based on the multiple humidity sequences and multiple temperature sequences to obtain multiple excavability sequences. Combined with the multiple ant infestation data sequences, it performs ant infestation evolution rationality analysis to obtain multiple evolution rationality coefficients. The ant infestation evolution rationality analysis includes trend rationality analysis and time series rationality analysis.

[0016] The coefficient calculation module is used to perform distribution rationality analysis on multiple humidity sequences and multiple temperature sequences based on historical monitoring data of the target monitoring area, and obtain the distribution rationality coefficient.

[0017] The result acquisition module is used to process multiple real-time risk parameters based on multiple mineability sequences and multiple ant population data sequences, and to perform risk compensation based on the multiple evolution rationality coefficients and distribution rationality coefficients to obtain multiple compensated real-time risk parameters as evaluation results.

[0018] This application provides one or more technical solutions, which have at least the following technical effects or advantages:

[0019] This application provides a multi-factor fusion method and system for real-time risk assessment of ant infestations. First, a soil analysis agent is constructed to transform the coupling effect of temperature and humidity into soil excavability. Second, a dual verification mechanism combining trend rationality analysis and time-series rationality analysis is employed to assess the rationality of ant infestation evolution from two dimensions: response trend similarity and time matching accuracy. Third, a distribution rationality coefficient is introduced to quantitatively assess the spatial consistency of the monitoring network. By comparing the similarity between real-time temperature and humidity gradients and historical gradients, it is determined whether the current environmental field distribution conforms to regional climate characteristics and topographic patterns, thereby identifying possible monitoring point anomalies or data transmission errors. Finally, a risk compensation mechanism is constructed based on the evolution rationality coefficient and the distribution rationality coefficient, transforming the multi-dimensional rationality assessment into quantifiable error coefficients, thereby dynamically correcting real-time risk parameters. This approach not only reflects the true ant infestation threat in a timely manner but also possesses robustness against data quality issues.

[0020] In summary, this application achieves full-chain optimization from environmental parameter collection to risk assessment output by organically integrating soil excavability modeling, dual verification of evolutionary rationality, spatial distribution rationality testing, and multi-factor risk compensation, thereby improving the accuracy, reliability, and engineering practicality of real-time ant infestation risk assessment. Attached Figure Description

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

[0022] Figure 1 This is a flowchart illustrating a multi-factor fusion method for real-time risk assessment of ant populations provided in an embodiment of this application.

[0023] Figure 2 This is a schematic diagram of the structure of a multi-factor fusion real-time risk assessment system for ant populations provided in an embodiment of this application.

[0024] The components represented by each number in the attached diagram are explained below:

[0025] Information acquisition module 11, sequence analysis module 12, coefficient calculation module 13, and result acquisition module 14. Detailed Implementation

[0026] This application provides a multi-factor fusion method and system for real-time risk assessment of ant infestations, which addresses the technical problem that existing technologies fail to adequately consider the dynamic correlation between soil excavability and ant infestation evolution, as well as the impact of the spatiotemporal distribution of environmental factors on the assessment results, leading to insufficient accuracy in risk assessment.

[0027] Example 1, as Figure 1 As shown in the embodiments of this application, a multi-factor fusion method for real-time risk assessment of ant infestations is provided, including:

[0028] S10: Acquire multiple ant infestation data sequences, multiple humidity sequences, and multiple temperature sequences from multiple monitoring points within the target monitoring area;

[0029] In this embodiment of the application, the target monitoring area refers to a specific geographical area around water conservancy project dams and reservoirs where there is a risk of termite damage. The layout of monitoring points needs to take into account the topography, historical termite distribution, soil type and structural characteristics of water conservancy facilities. For example, a monitoring section is set up every 50 to 100 meters along the dam axis, and monitoring points are arranged on the top of the dam, the upstream slope and the downstream slope of each section to form a three-dimensional monitoring network.

[0030] The termite infestation data sequence was obtained through a termite trapping and monitoring device. This device is equipped with termite-preferred bait and a counting sensor. When termites enter to feed, a counting signal is triggered, recording the frequency of termite activity per unit time, forming a continuous data sequence with a time resolution of 1 hour.

[0031] Meanwhile, humidity and temperature sequences are collected by temperature and humidity sensors buried in the shallow soil layer, which is the main activity area of ​​termites. The sensor sampling frequency is synchronized with the termite data to ensure that the three types of data sequences are strictly aligned in the time dimension.

[0032] Specifically, step S10 in the method includes:

[0033] The average values ​​of termite infestation data, average temperature, and average humidity at multiple monitoring points within the target monitoring area over the most recent time period are collected to obtain multiple real-time termite infestation data, multiple real-time temperature data, and multiple real-time humidity data. Among them, the termite infestation data includes the monitoring of termite activity characteristics data.

[0034] Based on historical monitoring data from multiple monitoring points within the target monitoring area, obtain multiple historical ant infestation data sequences, multiple historical humidity sequences, and multiple historical temperature sequences from multiple monitoring points over multiple past time periods.

[0035] Based on the multiple historical ant infestation data sequences, multiple historical humidity sequences, multiple historical temperature sequences, multiple real-time ant infestation data, multiple real-time temperature and multiple real-time humidity, multiple ant infestation data sequences, multiple humidity sequences and multiple temperature sequences are constructed.

[0036] In this embodiment of the application, firstly, the raw data of multiple monitoring points within the most recent time period are averaged to eliminate the interference of short-term random fluctuations on data quality. The most recent time period can be set to 1 to 24 hours according to monitoring needs. The arithmetic mean of the termite activity frequency within this period is calculated as real-time termite data. Simultaneously, the arithmetic mean of temperature and humidity within this period is calculated as real-time temperature and real-time humidity.

[0037] Secondly, extract continuous records from each monitoring point over multiple time periods from the historical monitoring database. The number of historical time periods should cover at least one complete annual cycle to include the seasonal variation patterns of termite activity. For example, select hourly data from the past 12 or 24 months to form historical termite data sequences, historical humidity sequences, and historical temperature sequences.

[0038] Finally, the real-time data and historical data are spliced ​​and integrated, with the real-time data as the end of the sequence and the historical data as the beginning of the sequence, to construct a complete data sequence with a unified time base.

[0039] S20: Based on the multiple humidity sequences and multiple temperature sequences, soil excavability analysis is performed to obtain multiple excavability sequences. Combined with the multiple ant infestation data sequences, ant infestation evolution rationality analysis is performed to obtain multiple evolution rationality coefficients. Among them, the ant infestation evolution rationality analysis includes trend rationality analysis and time series rationality analysis.

[0040] In this embodiment, the core of soil excavability analysis lies in establishing a quantitative relationship between the coupling effect of temperature and humidity and soil mechanical properties. Termites' excavation activities require overcoming the cohesive and frictional forces between soil particles, and soil moisture content and temperature directly determine excavation energy consumption by affecting soil strength parameters. Specifically, when the soil moisture content is within a suitable range, the soil exhibits a plastic state, neither becoming too hard and difficult to excavate due to excessive dryness, nor becoming too sticky and clumpy due to excessive moisture; temperature indirectly regulates excavation efficiency by affecting the termite metabolic rate and the intensity of soil moisture evaporation.

[0041] Specifically, the rationality analysis of ant infestation evolution is verified from two complementary dimensions. The trend rationality analysis focuses on the degree of matching of macroscopic change patterns, using dynamic time warping algorithms or similarity calculation methods based on sliding windows to quantify the consistency between the ant infestation data sequence and the soil excavability sequence in rising, falling, and stable trend segments. If the ant infestation data at a certain monitoring point shows a significant upward trend, while the soil excavability continues to decline during the same period, it indicates that the ant infestation change lacks an environmental driving basis, and the trend rationality coefficient will be reduced accordingly.

[0042] The temporal rationality analysis delves into the microscopic time structure, examining the lag characteristics of termite infestation changes relative to environmental changes. In this embodiment, termites, as poikilothermic animals, have their activity intensity directly regulated by ambient temperature. Soil temperature changes lag behind air temperature changes, and there is a depth gradient in soil heat conduction, resulting in a characteristic time delay in termite infestation response relative to changes in mineability. The temporal rationality analysis determines the optimal time lag by calculating the cross-correlation function between the termite infestation data sequence and the mineability sequence, and assesses whether this lag conforms to the ecological laws of termites. If the measured optimal lag is negative or far exceeds the reasonable range, it indicates that there is a temporal misalignment or abnormal interference in the data, and the temporal rationality coefficient will be corrected.

[0043] Specifically, step S20 in the method includes:

[0044] Input each set of temperature and humidity within multiple temperature and humidity sequences into the soil analysis agent, and output multiple exploitability sequences.

[0045] Based on multiple mineability sequences and multiple ant population data sequences, a rationality analysis of the ant population evolution trend is conducted to obtain the rationality coefficient of the trend evolution.

[0046] Based on multiple mineability sequences and multiple ant population data sequences, an analysis of the temporal rationality of ant populations was conducted to obtain the temporal evolution rationality coefficient.

[0047] The evolution rationality coefficient is calculated based on the trend evolution rationality coefficient and the time-series evolution rationality coefficient.

[0048] In this embodiment, firstly, the soil analysis agent is used as a mapping model of temperature-humidity coupling to soil excavability. This agent is trained based on soil mechanics experimental data and termite behavioral observation data. The input layer receives normalized temperature and humidity values, the hidden layer captures the interaction effect between the two through a nonlinear activation function, and the output layer generates an excavability index between 0 and 1, where 1 indicates that the soil is in the optimal excavation state for termites, and 0 indicates that the soil is completely unexcavable.

[0049] Secondly, for the analysis of the rationality of trend evolution, this embodiment adopts the Pearson correlation coefficient calculation method based on a sliding window. The sliding window length is set to cover a typical environmental change cycle of 3 to 7 days. Within each window, the correlation coefficient between the mineability sequence and the ant population data sequence is calculated separately. Then, a weighted average is calculated for the correlation coefficients of all windows, with the weights decaying exponentially according to the time distance between the window and the current moment, to obtain the rationality coefficient of trend evolution. This coefficient ranges from -1 to 1, with positive values ​​indicating that the two trends are in the same direction, and negative values ​​indicating that the trends are divergent. The absolute value reflects the strength of the correlation. When the coefficient is lower than a preset threshold, a data quality warning is triggered, indicating that there may be equipment failure or abnormal interference at the monitoring point.

[0050] Furthermore, regarding the temporal rationality analysis, this application employs a combined phase matching and cross-correlation optimization method. First, wavelet decomposition is performed on the mineability sequence and termite infestation data sequence to extract the dominant periodic components. The phase difference between the two in the characteristic frequency band is calculated to preliminarily determine the time-series lead-lag relationship. Subsequently, the extreme value of the cross-correlation function is searched within a reasonable lag range to determine the optimal lag amount. According to termite ecology research, the typical lag time of soil temperature change relative to air temperature change is 2 to 6 hours, and the reasonable lag time of termite response relative to soil mineability change is 6 to 24 hours. If the measured optimal lag amount exceeds this range, the temporal rationality is determined to be abnormal, and the temporal evolution rationality coefficient is calculated with a discount based on the degree of deviation.

[0051] Ultimately, the trend evolution rationality coefficient and the time series evolution rationality coefficient are weighted and fused to form a comprehensive evolution rationality coefficient. The weight allocation takes into account the climate type and seasonal characteristics of the monitoring area. The weight of trend analysis is increased in seasons with drastic temperature and humidity fluctuations, and the weight of time series analysis is increased in seasons with stable environments, so as to adapt to the assessment needs under different working conditions.

[0052] For example, suppose a dam monitoring area contains 5 monitoring points, and data were continuously collected over the past 72 hours. The temperature sequence at monitoring point A shows the daily average temperature gradually rising from 28℃ to 32℃, while the humidity sequence shows soil moisture content remaining between 18% and 22%. The soil exploitability sequence output by the soil analysis agent shows an initial increase followed by stabilization, with the peak occurring at hour 36. Simultaneously, the termite activity data sequence shows termite activity frequency increasing from 15 times / hour to 45 times / hour, with the inflection point occurring at hour 42.

[0053] Through sliding window correlation analysis, the trend correlation coefficient between the mineability and ant population data reached 0.78, the trend evolution rationality coefficient was rated at 0.85, and the weighting coefficient was set at 0.6. Cross-correlation analysis determined the optimal lag to be 6 hours, which is within a reasonable range. The time-series evolution rationality coefficient was rated at 0.92, and the weighting coefficient was set at 0.4. At this point, the comprehensive evolution rationality coefficient is 0.78×0.6+0.92×0.4=0.836, indicating that the ant population evolution at this monitoring point has high environmental rationality.

[0054] The training steps for the soil analysis agent include:

[0055] Based on historical soil termite monitoring data, a set of sample temperature and humidity groups was collected, and the rate of soil excavation and erosion by termites under different sample temperature and humidity groups was collected. The ratio of the rate to the maximum rate was calculated and labeled as the sample excavability, thus obtaining the sample excavability set.

[0056] Based on machine learning, construct an intelligent agent for soil analysis;

[0057] The soil analysis agent is trained under supervision using the sample temperature and humidity set and the sample exploitability set until the test converges, and then the training is completed and the agent is configured for use.

[0058] In this embodiment, firstly, termite excavation activity data under different temperature and humidity combinations are extracted from historical monitoring records of water conservancy project dams. For example, the temperature sampling range covers the active temperature range of termites from 15°C to 35°C, with temperature gradient nodes set at 2°C intervals; the humidity sampling range covers soil moisture content from 10% to 30%, with humidity gradient nodes set at 2% intervals, forming a sample grid covering the two-dimensional parameter space of temperature and humidity. At each grid node, the soil excavation volume of termites per unit time is obtained through controlled experiments or field observations, and the maximum observation rate at that node is recorded as a normalization benchmark.

[0059] Secondly, the relative digging rate of each sample node is calculated, which is the ratio of the actual digging rate to the maximum rate. The maximum rate is the highest value monitored over a historical period. This ratio quantifies the ease with which the soil is excavated by termites under specific temperature and humidity conditions, and is labeled as the sample's digability, ranging from 0 to 1. The construction of the sample digability set must ensure the balance of data distribution. For medium-high temperature and medium humidity ranges where termite activity is dense, the sampling density is appropriately increased. For sparse data under extreme temperature and humidity conditions, physical model interpolation is used to supplement the data, ensuring the agent's generalization ability in the edge regions of the parameter space.

[0060] Furthermore, based on machine learning algorithms, a network architecture for the soil analysis agent is constructed. This embodiment employs a three-layer fully connected neural network structure. The input layer contains two neurons that receive normalized temperature and humidity data respectively. The hidden layer has 16 to 32 neurons and uses the ReLU activation function to capture the nonlinear interaction effect of temperature and humidity. The output layer contains one neuron that uses the Sigmoid activation function to output the mineability index. The network training uses the Adam optimization algorithm, and the loss function is the mean squared error to minimize the deviation between the predicted mineability and the sample mineability.

[0061] Finally, the sample temperature and humidity groups are divided into training, validation, and test sets, with a typical ratio of 7:1:2. An early stopping mechanism is used during training to prevent overfitting. Training is terminated when the validation set loss no longer decreases after 10 consecutive training rounds. The coefficient of determination R² and root mean square error RMSE on the test set are used as convergence criteria, requiring R² to be no less than 0.85 and RMSE to be no more than 0.1. After meeting the accuracy requirements, the agent is configured and deployed to the risk assessment system.

[0062] Furthermore, based on multiple mineability sequences and multiple ant population data sequences, a rationality analysis of the ant population evolution trend was conducted, resulting in multiple trend evolution rationality coefficients, including:

[0063] Randomly select a pre-set length of mineability within multiple mineability sequences to obtain multiple first selection time periods and multiple first mineability subsequences, and calculate multiple first mineability change rates;

[0064] Obtain the response time period of soil change and ant infestation change;

[0065] Within multiple ant infestation data sequences, ant infestation data of a preset length after multiple selected time periods and response time cycles are selected to obtain multiple first ant infestation data subsequences, and multiple first ant infestation change rates are calculated.

[0066] The similarity between multiple rates of change in mineability and multiple rates of change in ant population is calculated to obtain multiple first response trend similarities, which are used as multiple trend evolution rationality coefficients.

[0067] In this embodiment, firstly, from multiple excavability sequences, a continuous data segment of length L is randomly selected as the first selected time period by tracing back from the current time. This length L corresponds to a typical environmental change cycle of 3 to 7 days. For each excavability subsequence within the first selected time period, the first excavability change rate is calculated to quantify the overall evolution trend of soil excavability within that time period. Positive values ​​indicate improved excavation conditions, while negative values ​​indicate deteriorated excavation conditions.

[0068] For example, the first rate of change of mineability is calculated by fitting a time trend line of the mineability subsequence using the least squares method; the slope of this line is the first rate of change of mineability index. Assuming the subsequence contains n sampling points, and time ti corresponds to the mineability value di, the rate of change k1 is calculated using the formula: k1 = [n...] (ti×di)- ti× di]÷[n (ti²)-( ti)²).

[0069] Secondly, the response time period is determined based on termite ecological experimental data and field observation statistics, characterizing the typical time span required for soil environmental changes to be transmitted to termite behavioral responses. In this embodiment, the response time period is set as an adjustable parameter of 6 to 24 hours, which can be calibrated according to the specific termite species, soil type, and seasonal characteristics of the monitoring area. For example, a shorter response period is used in areas with good thermal conductivity of sandy soil, and a longer response period is used in areas with high thermal inertia of clay soil.

[0070] Furthermore, for each selected first time period, after superimposing the response time period at its end, a continuous data segment of equal length L is selected from the termite infestation data sequence as the first termite infestation data subsequence, ensuring that the two sequences strictly correspond in causal time sequence. Using the same calculation method as the rate of change of mineability, the first termite infestation change rate is obtained, quantifying the evolution trend of termite activity intensity driven by the environment.

[0071] Finally, the similarity between the first rate of change in mineability and the first rate of change in ant population is calculated as the first response trend similarity, i.e., the trend evolution rationality coefficient. In this embodiment, the normalized distance method is used as the measurement index. When the two rates of change have the same sign and are close in value, the similarity approaches 1, indicating that the ant population change is highly consistent with the environmental driving force. When the signs are opposite or the values ​​are significantly different, the similarity decreases or even becomes negative, suggesting that there may be abnormal interference or non-environmental factors.

[0072] For example, the similarity between the rate of change in mineability and the rate of change in ant population can be calculated. Let the first rate of change in mineability be k1 and the first rate of change in ant population be k2. Then, the formula for calculating the first response trend similarity s is: s = 1 - |k1 - k2| ÷ (max(|k1|,|k2|) + ε), where ε is a very small positive number to prevent division by zero anomalies. When k1 and k2 have the same sign and similar absolute values, s approaches 1; when k1 and k2 have opposite signs, s is negative, indicating a trend divergence.

[0073] Furthermore, based on multiple mineability sequences and multiple ant population data sequences, a temporal rationality analysis of ant populations is conducted to obtain a temporal evolution rationality coefficient, including:

[0074] Select the earliest mineability of a preset length from multiple mineability sequences to obtain a second selected time period and multiple second mineability subsequences, and calculate the multiple second mineability change rates.

[0075] Obtain the response time period of soil change and ant infestation change;

[0076] Within multiple ant infestation data sequences, ant infestation data of a preset length are randomly selected to obtain multiple randomly selected time periods and multiple second ant infestation data subsequences, and multiple second ant infestation change rates are calculated.

[0077] Calculate the similarity between multiple second-minableness change rates and multiple second-ant population change rates to obtain multiple second-response trend similarities;

[0078] Continue to randomly select ant population data subsequences and calculate response trend similarity until the convergence count is reached. Filter the randomly selected time period corresponding to the largest response trend similarity, calculate the time span with the second selected time period, and obtain multiple matching time periods.

[0079] Calculate the similarity between multiple matching time periods and response time periods to obtain multiple temporal evolution rationality coefficients.

[0080] In this embodiment, the temporal rationality analysis first verifies whether the time delay of the ant infestation response relative to environmental driving forces conforms to ecological principles. Unlike trend analysis, temporal analysis employs a reverse matching strategy, starting from a fixed starting point in the mineability sequence and searching for the optimal matching time period in the ant infestation data sequence to reveal the true response lag characteristics.

[0081] Within multiple exploitability sequences, using the start time of the time series as the anchor point, a continuous data segment of a preset length L is selected as the second selected time period, which also covers a typical cycle of 3 to 7 days. For each second exploitability subsequence, the rate of change of the second exploitability is calculated, and the same least squares method as in trend analysis is used for fitting to obtain the slope of soil exploitability evolution within that time period.

[0082] Secondly, the acquisition of the response time period is consistent with the trend analysis, still set as an ecologically reasonable range of 6 to 24 hours. However, this parameter in the time series analysis serves both as a priori constraint to limit the search range and as a reference benchmark to evaluate the rationality of the measured matching results.

[0083] Furthermore, a random selection mechanism is established within the ant population data sequence. Unlike the fixed time sequence correspondence in trend analysis, time series analysis implements Monte Carlo random sampling within the ant population data sequence to generate a large number of candidate second ant population data subsequences. Each random selection must meet two constraints: first, the subsequence length is constant at L to ensure scale consistency with the mineability subsequence; second, the starting time of the subsequence must be within a reasonable interval after the starting time of the second selection period, defined by the minimum response time of 6 hours and the maximum response time of 24 hours.

[0084] Next, for each randomly generated second ant population data subsequence, the second ant population change rate is calculated, and the similarity is calculated with the second mineability change rate to obtain the second response trend similarity. The calculation of this similarity follows the normalized distance formula in trend analysis to quantify the degree of matching between the evolution patterns of the two sequences.

[0085] Specifically, random selection and similarity calculation constitute an iterative loop until a preset number of convergences is reached. The convergence criterion is that the coverage rate of the candidate subsequence to the reasonable interval exceeds 95% in multiple consecutive iterations. After convergence, the global maximum value is selected from all second response trend similarities, and the randomly selected time period corresponding to the maximum value is the optimal matching time period.

[0086] Then, the time difference between the start time of the optimal matching time period and the start time of the second selected time period is calculated to obtain the matching time period, which is the data-driven measured response lag. The similarity between the matching time period and the prior response time period is calculated to obtain the temporal evolution rationality coefficient. In this embodiment, the similarity between the matching time period and the response time period is calculated using "1-|AB| / [(A+B) / 2]". When the matching time period is exactly equal to the response time period, r=1, and the temporal sequence is reasonable; when the matching time period deviates from the response time period, r decays exponentially, and the greater the deviation, the lower the rationality.

[0087] For example, continuing the aforementioned dam monitoring scenario, a temporal rationality analysis was performed on monitoring point A. The second selected time period was set as the exploitability sequence from hour 0 to hour 72. Least square fitting yielded a second exploitability change rate k12 = 0.012 / hour, showing a slow upward trend. Random selection was performed within the ant infestation data sequence from hour 6 to hour 96, generating 200 candidate second ant infestation data subsequences. Through iterative calculation, the optimal match occurred in the period from hour 6 to hour 78, with a second ant infestation change rate k22 = 0.011 / hour and a second response trend similarity of 0.91, indicating extremely high morphological matching. The matching time period was 6 hours, precisely at the lower limit of the response time period. The temporal evolution rationality coefficient was calculated as r = 1 - |6 - 15| / [(6 + 15) / 2] = 1 - 9 / 10.5 = 0.143, where the median response time period was 15 hours. Specifically, the closer the calculated matching time period and response time period are, the closer the time sequence evolution rationality coefficient is to 1, indicating a higher level of monitoring credibility.

[0088] S30: Based on historical monitoring data of the target monitoring area, conduct a distribution rationality analysis on multiple humidity sequences and multiple temperature sequences to obtain the distribution rationality coefficient;

[0089] In this embodiment, the distribution rationality can assess the consistency between the spatial distribution characteristics of temperature and humidity data at monitoring points and historical norms, and identify abnormal distribution patterns caused by sensor malfunctions, sudden environmental changes, or human interference. Since the temperature and humidity at different locations of the dam exhibit certain patterns, a distribution rationality coefficient is obtained by determining whether the currently collected temperature and humidity sequences conform to this pattern. This coefficient reflects the reliability of the data collection.

[0090] Specifically, step S30 in the method includes:

[0091] Based on historical monitoring data from multiple monitoring points within the target monitoring area, multiple historical humidity gradients and multiple historical temperature gradients for these monitoring points are calculated.

[0092] Filter multiple real-time temperatures and multiple real-time humidity within the multiple humidity sequences and multiple temperature sequences, and calculate multiple real-time humidity gradients and multiple real-time temperature gradients;

[0093] The similarity of the multiple historical humidity gradients, multiple historical temperature gradients, multiple real-time humidity gradients, and multiple real-time temperature gradients is calculated to obtain the distribution rationality coefficient.

[0094] In this embodiment, firstly, based on historical monitoring data of the target monitoring area, monitoring records from the past complete hydrological year or at least 12 consecutive months are selected. According to the spatial coordinates of the monitoring points, the humidity difference and temperature difference between each monitoring point and its spatial neighboring monitoring points are calculated, and then divided by the corresponding spatial distance to obtain the historical humidity gradient and historical temperature gradient. This gradient calculation considers the typical spatial structural characteristics of the dam, including the gradient directional differences of different functional zones such as the water-facing side, the backwater side, the dam crest, and the dam toe. For example, the water-facing side to the backwater side typically exhibits a decreasing humidity gradient, while the dam toe to the dam crest typically exhibits a increasing temperature gradient.

[0095] During the real-time monitoring phase, temperature and humidity data from each monitoring point are collected synchronously. Following the same spatial neighborhood definition and calculation method as historical data, real-time humidity and temperature gradients are obtained. To eliminate the influence of absolute value differences on gradient comparison, both historical and real-time gradients are normalized in direction, retaining the gradient direction vector while ignoring the absolute value, thus focusing the comparison on the similarity of spatial distribution patterns.

[0096] Furthermore, the gradient similarity is calculated using a weighted combination of the cosine of the vector angle and the relative amplitude ratio. Let the historical humidity gradient vector of a monitoring point and its neighboring points be Gh, and the real-time humidity gradient vector be Gr. Then, the formula for calculating the humidity gradient similarity sh is: sh = α × (Gh·Gr) ÷ (|Gh| × |Gr|) + (1-α) × [1-||Gh|-|Gr|| ÷ (max(|Gh|,|Gr|) + ε)], where α is the directional weight coefficient, which is taken as 0.6 in this embodiment to emphasize the dominant role of spatial distribution pattern, and ε is a very small positive number. The temperature gradient similarity st is calculated using the same formula.

[0097] Finally, the distribution rationality coefficient d is obtained by combining the humidity gradient similarity and temperature gradient similarity. In this embodiment, the geometric mean method is used: d = The coefficient ranges from 0 to 1. The closer it is to 1, the better the real-time spatial distribution matches the historical norm, and the higher the reliability of the data collection. If the coefficient is lower than the preset threshold of 0.7, a spatial distribution anomaly alarm is triggered, indicating that there may be abnormalities such as sensor drift, local leakage or human damage.

[0098] For example, a certain earth-rock dam has 12 temperature and humidity monitoring points arranged in a 3x4 grid. Based on historical data, the historical humidity gradients of monitoring point B2 and its four neighboring points B1, B3, A2, and C2 are calculated to be -0.8% / m, +0.6% / m, -1.2% / m, and +0.4% / m, respectively, with negative values ​​indicating an increase in humidity towards the water-facing side. During real-time monitoring, the humidity gradient of point B2 is calculated to be -0.5% / m, +0.9% / m, -0.9% / m, and +0.7% / m, showing good directional consistency but with some amplitude deviation. After directional normalization, the humidity gradient similarity sh=0.82, the temperature gradient similarity st=0.91, and the distribution rationality coefficient d=0.86, all within the normal range. Therefore, the data collection of this monitoring point is deemed reliable, and no abnormalities are observed in its spatial distribution.

[0099] S40: Based on multiple mineability sequences and multiple ant population data sequences, multiple real-time risk parameters are obtained. Risk compensation is performed based on the multiple evolution rationality coefficients and distribution rationality coefficients to obtain multiple compensated real-time risk parameters as the evaluation result.

[0100] In this embodiment of the application, the processing of real-time risk parameters is based on the coupled analysis of the mineability sequence and the termite infestation data sequence. The purpose is to quantify the immediate threat level of current termite activity to the safety of the dam structure, and then perform risk compensation correction to obtain the evaluation result, thereby avoiding errors caused by inaccurate analysis and inaccurate data collection.

[0101] Among them, based on multiple mineability sequences and multiple ant population data sequences, several real-time risk parameters are obtained, including:

[0102] Filter multiple real-time mineability sequences and multiple real-time ant infestation data within multiple mineability sequences and multiple ant infestation data sequences;

[0103] Multiple real-time mineability data and multiple real-time ant population data are normalized to calculate multiple real-time risk parameters.

[0104] In this embodiment, firstly, the latest data points within the current monitoring time and one preceding response time period are extracted from multiple exploitability sequences and multiple termite infestation data sequences as real-time exploitability and real-time termite infestation data. Considering the differences in soil texture and termite species in different monitoring areas, the normalization process adopts a dynamic scaling method based on historical extreme values. Specifically, for each monitoring point, the historical maximum and minimum values ​​of exploitability over the past two complete hydrological years, as well as the historical peak and baseline values ​​of termite infestation data during the same period, are retrieved to establish an adaptive normalization interval.

[0105] Specifically, let w be the real-time exploitability of a certain monitoring point, and let w be its historical maximum value. max The historical minimum value is w minThe normalized mineability index W is calculated using the formula: W = (ww min )÷(w max -w min The index is denoted by ε, where ε is a very small positive number. The closer this index is to 1, the closer the current soil excavability is to historical extreme conditions, and the higher the vulnerability of the dam structure.

[0106] Similarly, let the real-time ant population data be 'a', and the historical peak value be 'a'. max The baseline value is a base The formula for calculating the normalized ant activity index A is: A = (aa base )÷(a max -a base The index (+ε) reflects the degree to which the current intensity of termite activity deviates from historical norms.

[0107] Finally, by multiplying the normalized real-time mineability with the real-time ant infestation data, multiple real-time risk parameters were calculated, namely the real-time risk parameter R=W×A. This parameter comprehensively quantifies the coupling effect between structural vulnerability and biological threat, and its value ranges from 0 to 1. The closer it is to 1, the higher the current ant infestation risk, and the more necessary it is to take immediate prevention and control measures; the closer it is to 0, the more controllable the risk is, and the more normal monitoring frequency can be maintained.

[0108] Furthermore, risk compensation is performed based on the aforementioned multiple evolutionary rationality coefficients and distribution rationality coefficients to obtain multiple real-time risk parameters for compensation, which serve as the evaluation results, including:

[0109] Based on the aforementioned multiple evolution rationality coefficients and distribution rationality coefficients, calculate multiple evolution error coefficients and distribution error coefficients;

[0110] Calculate multiple risk compensation coefficients based on multiple evolution error coefficients and distribution error coefficients;

[0111] Multiple risk compensation coefficients are used to calculate and compensate multiple real-time risk parameters, resulting in multiple compensated real-time risk parameters, which serve as the evaluation results.

[0112] In this embodiment, the evolution error coefficient and distribution error coefficient first characterize the degree of deviation of data reliability in temporal evolution analysis and spatial distribution analysis, respectively. Specifically, the evolution error coefficient is obtained by converting the temporal evolution rationality coefficient r, i.e., evolution error coefficient = 1 - r. The closer this coefficient is to 0, the higher the temporal matching degree, the more the ant infestation response lag characteristics conform to ecological laws, and the better the temporal reliability of the data. The distribution error coefficient is obtained by converting the distribution rationality coefficient d, i.e., distribution error coefficient = 1 - d. The closer this coefficient is to 0, the more the spatial distribution matches the historical normal state, the more stable the sensor network working state, and the better the spatial reliability of the data.

[0113] Furthermore, the risk compensation coefficient is calculated using a weighted fusion strategy, comprehensively considering the error contributions from both temporal and spatial dimensions. In this embodiment, the risk compensation coefficient = 1 - β × evolution error coefficient - (1 - β) × distribution error coefficient, where β is the temporal weight coefficient. In this embodiment, it is set to 0.55, slightly biased towards the temporal dimension, because the dynamic evolution of ant populations has strong time sensitivity, while spatial distribution anomalies can often be partially mitigated through cross-validation of redundant monitoring points. The compensation coefficient ranges from 0 to 1. The closer it is to 1, the higher the overall data quality and the better the reliability of the real-time risk parameters; the closer it is to 0, the more significant the data anomalies, and the risk assessment results need to be interpreted with caution.

[0114] Then, the risk compensation coefficient is used to calculate and compensate the real-time risk parameters, and the compensated real-time risk parameters are used as the final evaluation result.

[0115] Specifically, the real-time risk parameter compensation = (1 + risk compensation coefficient) × real-time risk parameter. When the risk compensation coefficient is close to 1, the real-time risk parameter is significantly amplified, reflecting the risk confirmation supported by low-reliability data; when the risk compensation coefficient is close to 0, the real-time risk parameter remains at its original value.

[0116] In summary, compared with existing technologies, this application adopts a risk compensation mechanism that integrates multiple factors, using the rationality of temporal evolution and spatial distribution as dual verification dimensions of data quality, and dynamically corrects real-time risk parameters. This effectively solves the limitations of single data source assessment, which is susceptible to collection errors and environmental disturbances, and improves the robustness and engineering practicality of real-time ant population risk assessment.

[0117] In summary, the embodiments of this application have at least the following technical effects:

[0118] This application provides a multi-factor fusion method for real-time risk assessment of ant infestations. First, a soil analysis agent is constructed to transform the coupling effect of temperature and humidity into soil excavability. Second, a dual verification mechanism combining trend rationality analysis and time-series rationality analysis is employed to assess the rationality of ant infestation evolution from two dimensions: response trend similarity and time matching accuracy. Third, a distribution rationality coefficient is introduced to quantitatively assess the spatial consistency of the monitoring network. By comparing the similarity between real-time temperature and humidity gradients and historical gradients, it is determined whether the current environmental field distribution conforms to regional climate characteristics and topographic patterns, thereby identifying possible monitoring point anomalies or data transmission errors. Finally, a risk compensation mechanism is constructed based on the evolution rationality coefficient and the distribution rationality coefficient, transforming the multi-dimensional rationality assessment into quantifiable error coefficients, thereby dynamically correcting real-time risk parameters. This method not only reflects the true ant infestation threat in a timely manner but also possesses robustness against data quality issues.

[0119] In summary, this application achieves full-chain optimization from environmental parameter collection to risk assessment output by organically integrating soil excavability modeling, dual verification of evolutionary rationality, spatial distribution rationality testing, and multi-factor risk compensation, thereby improving the accuracy, reliability, and engineering practicality of real-time ant infestation risk assessment.

[0120] Example 2, as Figure 2 As shown, based on the same inventive concept as the multi-factor fusion real-time risk assessment method for ant infestations provided in Embodiment 1, this application also provides a multi-factor fusion real-time risk assessment system for ant infestations, including:

[0121] Information acquisition module 11 is used to acquire multiple ant infestation data sequences, multiple humidity sequences, and multiple temperature sequences from multiple monitoring points within the target monitoring area;

[0122] The sequence analysis module 12 is used to perform soil excavability analysis based on the multiple humidity sequences and multiple temperature sequences to obtain multiple excavability sequences. Combined with the multiple ant infestation data sequences, it performs ant infestation evolution rationality analysis to obtain multiple evolution rationality coefficients. The ant infestation evolution rationality analysis includes trend rationality analysis and time series rationality analysis.

[0123] The coefficient calculation module 13 is used to perform distribution rationality analysis on multiple humidity sequences and multiple temperature sequences based on historical monitoring data of the target monitoring area, and obtain the distribution rationality coefficient.

[0124] The result acquisition module 14 is used to process multiple real-time risk parameters based on multiple mineability sequences and multiple ant population data sequences, perform risk compensation based on the multiple evolution rationality coefficients and distribution rationality coefficients, and obtain multiple compensated real-time risk parameters as evaluation results.

[0125] In one embodiment, the information acquisition module 11 is specifically used for:

[0126] The average values ​​of termite infestation data, average temperature, and average humidity at multiple monitoring points within the target monitoring area over the most recent time period are collected to obtain multiple real-time termite infestation data, multiple real-time temperature data, and multiple real-time humidity data. Among them, the termite infestation data includes the monitoring of termite activity characteristics data.

[0127] Based on historical monitoring data from multiple monitoring points within the target monitoring area, obtain multiple historical ant infestation data sequences, multiple historical humidity sequences, and multiple historical temperature sequences from multiple monitoring points over multiple past time periods.

[0128] Based on the multiple historical ant infestation data sequences, multiple historical humidity sequences, multiple historical temperature sequences, multiple real-time ant infestation data, multiple real-time temperature and multiple real-time humidity, multiple ant infestation data sequences, multiple humidity sequences and multiple temperature sequences are constructed.

[0129] In one embodiment, the sequence analysis module 12 is specifically used for:

[0130] Input each set of temperature and humidity within multiple temperature and humidity sequences into the soil analysis agent, and output multiple exploitability sequences.

[0131] Based on multiple mineability sequences and multiple ant population data sequences, a rationality analysis of the ant population evolution trend is conducted to obtain the rationality coefficient of the trend evolution.

[0132] Based on multiple mineability sequences and multiple ant population data sequences, an analysis of the temporal rationality of ant populations was conducted to obtain the temporal evolution rationality coefficient.

[0133] The evolution rationality coefficient is calculated based on the trend evolution rationality coefficient and the time-series evolution rationality coefficient.

[0134] Furthermore, in one embodiment of the application, the training steps of the soil analysis agent include:

[0135] Based on historical soil termite monitoring data, a set of sample temperature and humidity groups was collected, and the rate of soil excavation and erosion by termites under different sample temperature and humidity groups was collected. The ratio of the rate to the maximum rate was calculated and labeled as the sample excavability, thus obtaining the sample excavability set.

[0136] Based on machine learning, construct an intelligent agent for soil analysis;

[0137] The soil analysis agent is trained under supervision using the sample temperature and humidity set and the sample exploitability set until the test converges, and then the training is completed and the agent is configured for use.

[0138] Furthermore, in one embodiment of the application, based on multiple mineability sequences and multiple ant population data sequences, a rationality analysis of the ant population evolution trend is performed to obtain multiple trend evolution rationality coefficients, including:

[0139] Randomly select a pre-set length of mineability within multiple mineability sequences to obtain multiple first selection time periods and multiple first mineability subsequences, and calculate multiple first mineability change rates;

[0140] Obtain the response time period of soil change and ant infestation change;

[0141] Within multiple ant infestation data sequences, ant infestation data of a preset length after multiple selected time periods and response time cycles are selected to obtain multiple first ant infestation data subsequences, and multiple first ant infestation change rates are calculated.

[0142] The similarity between multiple rates of change in mineability and multiple rates of change in ant population is calculated to obtain multiple first response trend similarities, which are used as multiple trend evolution rationality coefficients.

[0143] Furthermore, in one embodiment, based on multiple mineability sequences and multiple ant population data sequences, an ant population time-series rationality analysis is performed to obtain a time-series evolution rationality coefficient, including:

[0144] Select the earliest mineability of a preset length from multiple mineability sequences to obtain a second selected time period and multiple second mineability subsequences, and calculate the multiple second mineability change rates.

[0145] Obtain the response time period of soil change and ant infestation change;

[0146] Within multiple ant infestation data sequences, ant infestation data of a preset length are randomly selected to obtain multiple randomly selected time periods and multiple second ant infestation data subsequences, and multiple second ant infestation change rates are calculated.

[0147] Calculate the similarity between multiple second-minableness change rates and multiple second-ant population change rates to obtain multiple second-response trend similarities;

[0148] Continue to randomly select ant population data subsequences and calculate response trend similarity until the convergence count is reached. Filter the randomly selected time period corresponding to the largest response trend similarity, calculate the time span with the second selected time period, and obtain multiple matching time periods.

[0149] Calculate the similarity between multiple matching time periods and response time periods to obtain multiple temporal evolution rationality coefficients.

[0150] Furthermore, the coefficient calculation module 13 is specifically used for:

[0151] Based on historical monitoring data from multiple monitoring points within the target monitoring area, multiple historical humidity gradients and multiple historical temperature gradients for these monitoring points are calculated.

[0152] Filter multiple real-time temperatures and multiple real-time humidity within the multiple humidity sequences and multiple temperature sequences, and calculate multiple real-time humidity gradients and multiple real-time temperature gradients;

[0153] The similarity of the multiple historical humidity gradients, multiple historical temperature gradients, multiple real-time humidity gradients, and multiple real-time temperature gradients is calculated to obtain the distribution rationality coefficient.

[0154] Furthermore, in one embodiment, multiple real-time risk parameters are obtained based on multiple mineability sequences and multiple ant population data sequences, including:

[0155] Filter multiple real-time mineability sequences and multiple real-time ant infestation data within multiple mineability sequences and multiple ant infestation data sequences;

[0156] Multiple real-time mineability data and multiple real-time ant population data are normalized to calculate multiple real-time risk parameters.

[0157] Furthermore, risk compensation is performed based on the aforementioned multiple evolutionary rationality coefficients and distribution rationality coefficients to obtain multiple real-time risk parameters for compensation, which serve as the evaluation results, including:

[0158] Based on the aforementioned multiple evolution rationality coefficients and distribution rationality coefficients, calculate multiple evolution error coefficients and distribution error coefficients;

[0159] Calculate multiple risk compensation coefficients based on multiple evolution error coefficients and distribution error coefficients;

[0160] Multiple risk compensation coefficients are used to calculate and compensate multiple real-time risk parameters, resulting in multiple compensated real-time risk parameters, which serve as the evaluation results.

Claims

1. A multi-factor fusion method for real-time risk assessment of ant infestations, characterized in that, The method includes: Acquire multiple ant infestation data sequences, multiple humidity sequences, and multiple temperature sequences from multiple monitoring points within the target monitoring area; Based on the multiple humidity and temperature sequences, soil excavability analysis is performed to obtain multiple excavability sequences. Combined with the multiple ant infestation data sequences, an ant infestation evolution rationality analysis is conducted to obtain multiple evolution rationality coefficients. The ant infestation evolution rationality analysis includes trend rationality analysis and time-series rationality analysis, including: Input each set of temperature and humidity within multiple temperature and humidity sequences into the soil analysis agent, and output multiple exploitability sequences. Based on multiple mineability sequences and multiple ant population data sequences, a rationality analysis of the ant population evolution trend is conducted to obtain the rationality coefficient of the trend evolution. Based on multiple mineability sequences and multiple ant population data sequences, an analysis of the temporal rationality of ant populations was conducted to obtain the temporal evolution rationality coefficient. The evolution rationality coefficient is calculated based on the trend evolution rationality coefficient and the time-series evolution rationality coefficient. Based on historical monitoring data of the target monitoring area, a distribution rationality analysis was conducted on multiple humidity sequences and multiple temperature sequences to obtain distribution rationality coefficients, including: Based on historical monitoring data from multiple monitoring points within the target monitoring area, multiple historical humidity gradients and multiple historical temperature gradients for these monitoring points are calculated. Filter multiple real-time temperatures and multiple real-time humidity within the multiple humidity sequences and multiple temperature sequences, and calculate multiple real-time humidity gradients and multiple real-time temperature gradients; Calculate the similarity of the multiple historical humidity gradients, multiple historical temperature gradients, multiple real-time humidity gradients, and multiple real-time temperature gradients to obtain the distribution rationality coefficient; Based on multiple mineability sequences and multiple ant population data sequences, multiple real-time risk parameters are obtained. Risk compensation is performed based on the multiple evolution rationality coefficients and distribution rationality coefficients to obtain multiple compensated real-time risk parameters, which serve as the evaluation results.

2. The multi-factor fusion method for real-time risk assessment of ant infestations according to claim 1, characterized in that, Acquire multiple ant infestation data sequences, multiple humidity sequences, and multiple temperature sequences from multiple monitoring points within the target monitoring area, including: The average values ​​of termite infestation data, average temperature, and average humidity at multiple monitoring points within the target monitoring area over the most recent time period are collected to obtain multiple real-time termite infestation data, multiple real-time temperature data, and multiple real-time humidity data. Among them, the termite infestation data includes the monitoring of termite activity characteristics data. Based on historical monitoring data from multiple monitoring points within the target monitoring area, obtain multiple historical ant infestation data sequences, multiple historical humidity sequences, and multiple historical temperature sequences from multiple monitoring points over multiple past time periods. Based on the multiple historical ant infestation data sequences, multiple historical humidity sequences, multiple historical temperature sequences, multiple real-time ant infestation data, multiple real-time temperature and multiple real-time humidity, multiple ant infestation data sequences, multiple humidity sequences and multiple temperature sequences are constructed.

3. The multi-factor fusion method for real-time risk assessment of ant infestations according to claim 1, characterized in that, The training steps for the soil analysis agent include: Based on historical soil termite monitoring data, a set of sample temperature and humidity groups was collected, and the rate of soil excavation and erosion by termites under different sample temperature and humidity groups was collected. The ratio of the rate to the maximum rate was calculated and labeled as the sample excavability, thus obtaining the sample excavability set. Based on machine learning, construct an intelligent agent for soil analysis; The soil analysis agent is trained under supervision using the sample temperature and humidity set and the sample exploitability set until the test converges, and then the training is completed and the agent is configured for use.

4. The multi-factor fusion method for real-time risk assessment of ant infestations according to claim 1, characterized in that, Based on multiple mineability sequences and multiple ant population data sequences, a rationality analysis of the ant population evolution trend is conducted to obtain the rationality coefficient of the trend evolution, including: Randomly select a pre-set length of mineability within multiple mineability sequences to obtain multiple first selection time periods and multiple first mineability subsequences, and calculate multiple first mineability change rates; Obtain the response time period of soil change and ant infestation change; Within multiple ant infestation data sequences, ant infestation data of a preset length after multiple selected time periods and response time cycles are selected to obtain multiple first ant infestation data subsequences, and multiple first ant infestation change rates are calculated. The similarity between multiple rates of change in mineability and multiple rates of change in ant population is calculated to obtain multiple first response trend similarities, which are used as multiple trend evolution rationality coefficients.

5. The multi-factor fusion method for real-time risk assessment of ant infestations according to claim 1, characterized in that, Based on multiple mineability sequences and multiple ant population data sequences, a temporal rationality analysis of ant populations is conducted to obtain a temporal evolution rationality coefficient, including: Select the earliest mineability of a preset length from multiple mineability sequences to obtain a second selected time period and multiple second mineability subsequences, and calculate the multiple second mineability change rates. Obtain the response time period of soil change and ant infestation change; Within multiple ant infestation data sequences, ant infestation data of a preset length are randomly selected to obtain multiple randomly selected time periods and multiple second ant infestation data subsequences, and multiple second ant infestation change rates are calculated. Calculate the similarity between multiple second-minableness change rates and multiple second-ant population change rates to obtain multiple second-response trend similarities; Continue to randomly select ant population data subsequences and calculate response trend similarity until the convergence count is reached. Filter the randomly selected time period corresponding to the largest response trend similarity, calculate the time span with the second selected time period, and obtain multiple matching time periods. Calculate the similarity between multiple matching time periods and response time periods to obtain multiple temporal evolution rationality coefficients.

6. The multi-factor fusion method for real-time risk assessment of ant infestations according to claim 1, characterized in that, Based on multiple mineability sequences and multiple ant population data sequences, several real-time risk parameters are obtained, including: Filter multiple real-time mineability sequences and multiple real-time ant infestation data within multiple mineability sequences and multiple ant infestation data sequences; Multiple real-time mineability data and multiple real-time ant population data are normalized to calculate multiple real-time risk parameters.

7. The multi-factor fusion method for real-time risk assessment of ant infestations according to claim 1, characterized in that, Risk compensation is performed based on the aforementioned multiple evolutionary rationality coefficients and distribution rationality coefficients to obtain multiple real-time risk parameters for compensation, which serve as the evaluation results, including: Based on the aforementioned multiple evolution rationality coefficients and distribution rationality coefficients, calculate multiple evolution error coefficients and distribution error coefficients; Calculate multiple risk compensation coefficients based on multiple evolution error coefficients and distribution error coefficients; Multiple risk compensation coefficients are used to calculate and compensate multiple real-time risk parameters, resulting in multiple compensated real-time risk parameters, which serve as the evaluation results.

8. A multi-factor fusion real-time risk assessment system for ant infestations, characterized in that, A method for performing a multi-factor fusion-based real-time risk assessment of ant populations as described in any one of claims 1-7 includes: The information acquisition module is used to acquire multiple ant infestation data sequences, multiple humidity sequences, and multiple temperature sequences from multiple monitoring points within the target monitoring area; The sequence analysis module is used to perform soil excavability analysis based on the multiple humidity sequences and multiple temperature sequences, obtaining multiple excavability sequences. Combined with the multiple ant infestation data sequences, it performs ant infestation evolution rationality analysis, obtaining multiple evolution rationality coefficients. The ant infestation evolution rationality analysis includes trend rationality analysis and time-series rationality analysis, including: Input each set of temperature and humidity within multiple temperature and humidity sequences into the soil analysis agent, and output multiple exploitability sequences. Based on multiple mineability sequences and multiple ant population data sequences, a rationality analysis of the ant population evolution trend is conducted to obtain the rationality coefficient of the trend evolution. Based on multiple mineability sequences and multiple ant population data sequences, an analysis of the temporal rationality of ant populations was conducted to obtain the temporal evolution rationality coefficient. The evolution rationality coefficient is calculated based on the trend evolution rationality coefficient and the time-series evolution rationality coefficient. The coefficient calculation module is used to perform distribution rationality analysis on multiple humidity sequences and multiple temperature sequences based on historical monitoring data of the target monitoring area, and obtain distribution rationality coefficients, including: Based on historical monitoring data from multiple monitoring points within the target monitoring area, multiple historical humidity gradients and multiple historical temperature gradients for these monitoring points are calculated. Filter multiple real-time temperatures and multiple real-time humidity within the multiple humidity sequences and multiple temperature sequences, and calculate multiple real-time humidity gradients and multiple real-time temperature gradients; Calculate the similarity of the multiple historical humidity gradients, multiple historical temperature gradients, multiple real-time humidity gradients, and multiple real-time temperature gradients to obtain the distribution rationality coefficient; The result acquisition module is used to process multiple real-time risk parameters based on multiple mineability sequences and multiple ant population data sequences, and to perform risk compensation based on the multiple evolution rationality coefficients and distribution rationality coefficients to obtain multiple compensated real-time risk parameters as evaluation results.

Images