Tower inclination evaluation method and system based on soil water content and resistivity correlation mechanism

Abstract

The invention discloses a tower inclination evaluation method and system based on a soil water content and resistivity correlation mechanism, and relates to the technical field of tower safety monitoring. According to the method, multi-source parameters of soil around a tower are collected, a prediction model of soil moisture content / temperature and resistivity and a correlation model of a tower inclination angle and a soil resistivity change rate are established, and then inclination risk assessment is realized based on resistivity change. According to the method, the defects that a traditional method depends on a single tilt angle sensor and neglects the soil parameter coupling effect are overcome, conduction and risk assessment from soil parameter abnormity to structure tilt risks are achieved by associating the soil physical characteristics with the tower mechanical state, and the timeliness of early warning is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of tower safety monitoring, and particularly relates to a tower tilt evaluation method and system based on a correlation mechanism of soil water content and resistivity. BACKGROUND

[0002] The stability of a transmission line tower is the core of ensuring the safe operation of a power grid, and tower tilt or collapse is often caused by unbalanced stress on the soil foundation. The coupling effect of soil water content and resistivity is a key factor affecting the stability of the foundation.

[0003] Traditional methods only rely on inclinometer sensors to directly measure the tilt angle of the tower, ignoring the coupling effect of soil water content on the performance of the grounding system and the mechanical properties of the soil by changing the resistivity, resulting in evaluation results lacking support from underlying physical mechanisms and a high false positive rate. Although existing grounding resistance calculations take into account the spatial distribution of soil moisture, they do not correlate the dynamic relationship between water content and resistivity, and cannot real-time correct the impact of changes in soil resistivity on the lightning current dispersion capacity, thereby underestimating the tilt risk caused by grounding abnormalities and making it difficult to achieve the conduction and prediction of tilt risk from soil parameter abnormalities.

[0004] In view of this, there is a need for a tower tilt evaluation method and system based on a correlation mechanism of soil water content and resistivity. SUMMARY

[0005] In view of the problem that existing grounding resistance calculations do not correlate the dynamic relationship between water content and resistivity, cannot real-time correct the impact of changes in soil resistivity on the lightning current dispersion capacity, and make it difficult to achieve the conduction and prediction of tilt risk from soil parameter abnormalities, the present application provides a tower tilt evaluation method and system based on a correlation mechanism of soil water content and resistivity, which establishes a risk conduction path from soil parameter abnormalities to resistivity changes to tower tilt, and achieves early prediction of tilt risk from soil parameter abnormalities. The specific technical solutions are as follows: The present application provides a tower tilt evaluation method based on a correlation mechanism of soil water content and resistivity, comprising the following steps: Obtain the multi-source parameters of the soil around the target tower and the parameters of the tower itself; Denoise, normalize and encode the multi-source parameters, and based on correlation analysis, screen out key data features associated with the tilt state of the tower to form an input feature set; Based on the input feature set, a prediction model is constructed and trained with soil water content and soil temperature as input and soil resistivity as output, for outputting predicted resistivity according to real-time multi-source parameters; According to the multi-source parameters and the state parameters, a tower-pile-foundation-soil coupling mechanical model is established, and an association model between a tower inclination angle and a resistivity change rate is constructed by combining a preset soil mechanical state and resistivity association rule. Based on the predicted resistivity, a resistivity change rate of a monitoring point is calculated, a tower inclination evaluation value is obtained by combining the association model, and a tower inclination risk level is evaluated.

[0006] Preferably, the multi-source parameters are synchronously collected through a plurality of monitoring points arranged in a monitoring area centered on the tower; and a spatial layout of the monitoring area is configured based on a foundation structure geometric feature of the tower.

[0007] Preferably, the prediction model is a BP neural network model, a number of nodes in a hidden layer of the BP neural network model is configured according to a soil type of the multi-source parameters; and the BP neural network model adopts an Adam optimization algorithm and is trained by taking a mean square error as a loss function.

[0008] Preferably, when the prediction model fuses the soil water content and soil temperature data, a seasonal time weight factor is introduced, the weight factor is determined according to an analysis of influences of historical same-period data on resistivity, so as to optimize input of the prediction model.

[0009] Preferably, the preset soil mechanical state and resistivity association rule is obtained by the following way: A sample with the same soil property as a surrounding soil of a target tower is prepared; Different mechanical loads are applied to the sample to make it generate different volume strains, and resistivity of the sample under each volume strain state is synchronously measured; Based on the measurement data, a continuous function relationship or a relationship curve between the volume strain and the resistivity of the soil is fitted.

[0010] Preferably, the association model between the tower inclination angle and the resistivity change rate is constructed by the following way: Geometric boundaries and material properties of the coupling mechanical model are determined according to a foundation size of a target tower and soil geological survey parameters; In simulation analysis, a beam element is adopted to simulate a tower structure, a solid element is adopted to simulate a tower foundation, an elastoplastic constitutive model is adopted to simulate a surrounding soil, and the coupling mechanical model is assembled; A horizontal displacement load is applied to a top of the coupling mechanical model to simulate states of the tower reaching a plurality of predetermined inclination angles; Volume strain data of the soil at the monitoring points under the states of the predetermined inclination angles are calculated and extracted. Based on the soil volume strain and resistivity relationship curve calibrated in advance through the test, the volume strain data is converted into the corresponding soil resistivity change rate.

[0011] Preferably, the criterion for evaluating the tilt risk level of the tower includes: The resistivity change rate of a single monitoring point exceeds an independent threshold value, and the resistivity change rate gradient between a plurality of monitoring points with a preset geometric correlation exceeds a gradient threshold value; The monitoring points with the preset geometric correlation are point pairs symmetric about the central axis or center of the tower foundation.

[0012] Preferably, the tilt risk level includes three levels, including: When the absolute value of the resistivity change rate of any monitoring point reaches a first threshold value, or the difference value of the resistivity change rate between any two spatially symmetric monitoring points reaches a second threshold value, a first-level early warning is triggered; When the absolute value of the resistivity change rate of any monitoring point reaches a third threshold value, or the difference value of the resistivity change rate between any two spatially symmetric monitoring points reaches a fourth threshold value, a second-level early warning is triggered; When the absolute value of the resistivity change rate of any monitoring point reaches a fifth threshold value, or the difference value of the resistivity change rate between any two spatially symmetric monitoring points reaches a sixth threshold value, a third-level early warning is triggered; The third threshold value is greater than the first threshold value, and the fifth threshold value is greater than the third threshold value; the fourth threshold value is greater than the second threshold value, and the sixth threshold value is greater than the fourth threshold value.

[0013] Preferably, a tower tilt evaluation method based on the correlation mechanism of soil water content and resistivity further includes: Obtaining a real-time tower tilt angle and comparing it with the tilt evaluation value; According to the comparison difference, the correlation relationship between the tower tilt angle and the resistivity change rate in the correlation model is feedback calibrated.

[0014] The application also provides a tower tilt evaluation system based on the correlation mechanism of soil water content and resistivity, which is applied to the aforementioned method and includes: A data acquisition unit is configured to acquire multi-source parameters of soil around a target tower and parameters of the tower itself; A data processing unit is configured to denoise, normalize and encode the multi-source parameters, and filter out key data features associated with the tilt state of the tower based on correlation analysis to form an input feature set; The resistivity prediction unit is configured to construct and train a prediction model with soil water content and soil temperature as input and soil resistivity as output based on the input feature set, and output predicted resistivity according to real-time multi-source parameters. The correlation model construction unit is configured to establish a tower-pile-foundation-soil coupling mechanical model based on the multi-source parameters and the state parameters, and construct a correlation model between the tower inclination angle and the resistivity change rate by combining a preset soil mechanical state and resistivity correlation rule. The inclination risk assessment unit is configured to calculate the resistivity change rate of the monitoring point based on the predicted resistivity, obtain a tower inclination assessment value by combining the correlation model, and assess the inclination risk level of the tower.

[0015] Compared with the prior art, the method has the following advantages: The tower inclination assessment method based on the soil water content and resistivity correlation mechanism establishes a risk transmission path from soil parameter anomaly to resistivity change to tower inclination, calculates the resistivity change rate by monitoring the soil multi-source parameters, obtains the inclination assessment value by combining the correlation model, and divides the risk level, thereby realizing the early prediction of the risk from the soil parameter anomaly to the tower inclination, and providing a basis for the preventive maintenance of the power transmission line tower. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the specific embodiments of the present application or the technical solutions in the prior art, the drawings needed in the specific embodiments or the prior art description will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, each element or part is not necessarily drawn according to the actual scale.

[0017] Figure 1 The flow chart of the tower inclination assessment method based on the soil water content and resistivity correlation mechanism.

[0018] Figure 2 The tower inclination assessment system principle diagram based on the soil water content and resistivity correlation mechanism. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0020] It should be understood that the terms "comprise" and "comprising" when used in this specification, indicate the presence of the described features, integers, steps, operations, elements, and / or components but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0021] It should also be understood that the terms used in the present application specification are only for the purpose of describing particular embodiments and are not intended to limit the present application. As used in the present application specification and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0022] It should be further understood that the term "and / or" used in the present application specification means any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0023] The following examples are based on Figure 1 and Figure 2 .

[0024] The embodiments of the present application provide a tower tilt evaluation method based on the correlation mechanism of soil moisture content and resistivity, comprising the following steps: Step S1, acquiring multi-source parameters of soil around the target tower and tower self parameters; Wherein, the multi-source parameters also include soil basic characteristic parameters and soil physical and mechanical parameters in the region; the soil basic characteristic parameters include soil temperature, soil moisture content and soil type, etc.; the soil physical and mechanical parameters include soil cohesion, internal friction angle, initial resistivity, etc. The tower self parameters include the structure parameters and state parameters of the tower itself, wherein the structure parameters include tower model, foundation type and size, material strength, etc.; the state parameters include tilt angle time series data.

[0025] Specifically, the multi-source parameters are synchronously collected by multiple monitoring points arranged in a monitoring area centered on the tower; the spatial layout of the monitoring area is configured based on the geometric characteristics of the foundation structure of the tower. For example, a 30m range around the tower is delimited as the monitoring area, 9 monitoring points are arranged in a 3x3 grid, including the four corners of the grounding grid, the upper and lower sides, the left and right sides, and the center, and the following equipment is used to collect data: Time domain reflectometer (TDR) or ground penetrating radar (GPR) is used to collect soil volume moisture content ( ) of each monitoring point at a depth of 0~60cm. Wherein, the data sampling frequency is set based on meteorological factors. For example, set once a week in rainy season and once every two weeks in dry season.

[0026] Multi-electrode array system is used to collect apparent resistivity data of each monitoring point, and soil temperature and soil type are recorded synchronously; Collecting the time series data of the inclination angle of the top and bottom of the tower by using the carrier positioning sensor; and collecting the parameters of the tower grounding net: the cross-sectional area of the grounding net , the rectangular area of the grounding net , the burial depth H, the longest side length , and the lightning impulse current amplitude in the historical lightning stroke record .

[0027] Step S2, denoising, normalizing and encoding the multi-source parameters, and screening out key data features associated with the inclination state of the tower based on correlation analysis to form an input feature set; The apparent resistivity data are denoised by using the moving average filtering method, and the filtering formula is: wherein, is the i-th apparent resistivity data after filtering, is the original data, and n is the size of the sliding window; The numerical data such as soil moisture, apparent resistivity and soil temperature are mapped to the interval [0, 1] by normalization, and are expressed as: wherein, is the original value, are the maximum value and the minimum value of the parameter, respectively; The non-numerical data such as soil type are processed by one-hot encoding, such as clay [1, 0, 0], sandy land [0, 1, 0], grassland [0, 0, 1]; The mean vector of the soil-related information is calculated by using correlation analysis to screen key data features: wherein, P is the total number of soil-related information, is the p-th information; The feature mean of each type of data is calculated: wherein, is the k-th data in the p-th information. The correlation coefficient is calculated by the following formula , and the features with ≥0.6 are selected as the input feature set: Step S3, based on the input feature set, a prediction model is constructed and trained with soil moisture and soil temperature as input and soil resistivity as output, which is used to output the predicted resistivity according to real-time multi-source parameters; Specifically, the prediction model is a BP neural network model, the number of nodes in the hidden layer of which is configured according to the soil type of the multi-source parameters; the BP neural network model is trained by using an Adam optimization algorithm and taking a mean square error as a loss function.

[0028] In a specific implementation, the key data feature set extracted in step S2 is divided into a training set (70%) and a test set (30%) according to a ratio of 7:3, and a BP neural network model is trained. In this embodiment, the network architecture of the BP neural network model is as follows: Input layer: 2 nodes, normalized soil water content , soil temperature t; Hidden layer: the number of nodes is optimized according to the soil type, for example, 8 nodes for farmland, 4 nodes for sandy land, and 5 nodes for grassland, and the activation function uses ReLU; Output layer: 1 node, normalized resistivity ; The Adam optimization algorithm is used, and the loss function is a mean square error: wherein, is a real value of resistivity, is a predicted value, and Q is the total number of samples.

[0029] The training is performed for 100-500 epochs, and the model is verified once every 5 epochs. When the loss function of the test set no longer decreases, the training is stopped, and the model is saved.

[0030] In the saved trained model, the soil water content and soil temperature data collected in real time in the area to be evaluated are input, and the predicted value of the resistivity at the corresponding depth is output .

[0031] The existing grounding resistance calculation does not associate the dynamic relationship between the water content and the resistivity, and cannot correct the influence of the resistivity change on the lightning current dispersion ability in real time. Step S3 of this embodiment can output the predicted resistivity according to the real-time multi-source parameters by constructing a resistivity prediction model with soil water content and temperature as inputs, can reflect the dynamic change of the soil resistivity in real time, and can accurately evaluate the influence of the resistivity change on the lightning current dispersion ability, thereby avoiding the tower tilt hidden danger caused by underestimating the grounding abnormal risk.

[0032] Step S4, according to the multi-source parameters and the state parameters, a tower-base-soil coupling mechanical model is established, and a correlation model between the tower tilt angle and the resistivity change rate is constructed by combining a preset soil mechanical state and a resistivity correlation rule; In a specific implementation, the construction process of the correlation model is as follows: Step one, determine the model base parameters; The finite element analysis software with high efficient structure-soil coupling analysis ability and meeting the complex mechanical response calculation requirement of tower-foundation-soil system is used as the numerical simulation tool. In the finite element analysis software simulation, the spatial range of the model is determined based on the actual power transmission tower engineering geological survey report and tower design drawings. For example, the horizontal direction takes the center of the tower foundation as the origin and extends outward to 5-8 times the width of the foundation, and the vertical direction extends downward from the ground surface to 3-5 times the depth of the foundation, ensuring that the boundary effect does not significantly affect the calculation results. Through field investigation and indoor soil test, the soil physical and mechanical parameters of the target tower area are obtained, including soil cohesion c, internal friction angle φ, initial resistivity p0, as well as structural parameters such as tower model, foundation type and size, material strength, etc.

[0033] Step two, build a tower-foundation-soil coupling mechanical model; Beam element BEAM188 is used for numerical simulation of the tower, which is suitable for slender structures bearing combined loads of tension, compression, bending and shear, and can reflect the bending deformation characteristics of the tower. According to the tower design drawings, the tower segment and node model is established in a 1:1 scale, and the mechanical parameters of the tower material such as elastic modulus, Poisson's ratio and density are given. Solid element SOLID65 is used to simulate the tower foundation, which supports cracking and crushing analysis of concrete materials and meets the stress and deformation characteristics of the foundation. According to the foundation construction drawings, the elastic modulus, Poisson's ratio and compressive strength of the foundation concrete material are given. The Drucker-Prager model in the elastic-plastic constitutive model is used to describe the mechanical behavior of soil, which can effectively reflect the yield and flow characteristics of soil under complex stress state and is suitable for mixed stratum of cohesive soil and sandy soil. According to the obtained soil parameters, the cohesion c and internal friction angle φ of the model are input, and the basic mechanical parameters of the soil such as elastic modulus, Poisson's ratio and density are given. The tower model, foundation model and soil model are geometrically assembled to ensure that the connection nodes of the tower and foundation are completely coincident, and the contact interface between the foundation and soil is set as frictional contact, completing the construction of the entire coupled finite element model.

[0034] Step three, set the inclined displacement working condition and load application; Monitoring points are arranged along the foundation perimeter and depth in the soil area, and the monitoring point distribution should cover the foundation stress affected area. Each monitoring point is marked as a synchronous collection point of resistivity and volumetric strain. By setting 3 groups of typical inclined displacement working conditions, corresponding to 0.5°, 1.0° and 1.5° inclination angles of the tower, the angle range includes the critical interval of normal operation and dangerous state of the tower.

[0035] A concentrated load is applied in the horizontal direction on the top of the tower, and the tower is adjusted to a preset inclination angle by gradually adjusting the load size, and the loading process is in displacement control mode to avoid stress mutation.

[0036] Step four, calculate the soil volume strain and extract data; In the finite element software, set the solver to the static general solver, and use the force and displacement double control as the convergence criterion, with the force convergence accuracy being 1x10 -5 and the displacement convergence accuracy being 1x10 -4。 When the model reaches the preset inclination angle and stabilizes, the soil volume strain data at each monitoring point is extracted, which directly reflects the change of soil density under the action of the tower inclination. The increase of the volume strain indicates the decrease of the soil density, and the decrease of the volume strain indicates the increase of the soil density; The monitoring point coordinates and volume strain values under each inclination angle condition are recorded one by one to form a correlation data set.

[0037] Step five, convert the resistivity change rate and output the results; The soil volume strain-resistivity correlation curve established by the indoor soil test in the early stage is called, and the method for obtaining the curve includes: Prepare a sample with the same soil properties as the target tower surrounding soil; Different mechanical loads are applied to the sample to produce different volume strains, and the resistivity of the sample under each volume strain state is measured synchronously; Based on the measured data, a continuous function relationship or a relationship curve between the volume strain and the resistivity of the soil is fitted.

[0038] According to the volume strain values of each monitoring point extracted in step four, the corresponding resistivity p is queried on the correlation curve, and the resistivity change rate k is calculated using the following formula: In the formula, k is the resistivity change rate, p is the soil resistivity of the monitoring point under a certain inclination angle, and p0 is the initial resistivity of the soil; Correlate the resistivity change rate k of each monitoring point corresponding to different inclination angles to form a correlation data set of inclination angle and resistivity change rate.

[0039] The step S4 of the embodiment combines the tower-foundation-soil coupled mechanical model to establish a correlation model of inclination angle and resistivity change rate, so that the tower inclination evaluation is based on the bottom mechanism driving.

[0040] Step S5, based on the predicted resistivity, calculate the resistivity change rate of the monitoring point, combine the correlation model to obtain the tower inclination evaluation value, and evaluate the inclination risk level of the tower.

[0041] wherein, based on the predicted resistivity , the resistivity change rate of the monitoring point is calculated , expressed as: Specifically, the criterion for evaluating the tilt risk level of the tower includes: The resistivity change rate of a single monitoring point exceeds an independent threshold value, and the resistivity change rate gradient between multiple monitoring points with a preset geometric correlation exceeds a gradient threshold value. Wherein, the monitoring point with a preset geometric correlation is a point pair about the axis or center of the tower foundation.

[0042] The tilt risk level includes three levels, including: When the absolute value of the resistivity change rate of any monitoring point reaches the first threshold value, or the difference in resistivity change rate between any two spatially symmetric monitoring points reaches the second threshold value, a first-level early warning is triggered; for example, when the resistivity change rate of a single monitoring point is ≥ ± 10%, or the difference between any two symmetric monitoring points is ≥ 8%, corresponding to a tilt angle θ ≈ 0.5°, the tower is in a slight tilt state, and there is no immediate safety risk. Push prompt information through the operation and maintenance APP, including abnormal monitoring point position, value, real-time weather data; automatically start data encryption collection mode, and strengthen monitoring intensity.

[0043] When the absolute value of the resistivity change rate of any monitoring point reaches the third threshold value, or the difference in resistivity change rate between any two spatially symmetric monitoring points reaches the fourth threshold value, a second-level early warning is triggered; for example, when the resistivity change rate of a single monitoring point k is ≥ ± 20%, or the difference between symmetric monitoring points k is ≥ 15%, corresponding to a tilt angle θ ≈ 1.0°, the system triggers a short message and platform double early warning to notify the operation and maintenance personnel to check on site.

[0044] When the absolute value of the resistivity change rate of any monitoring point reaches the fifth threshold value, or the difference in resistivity change rate between any two spatially symmetric monitoring points reaches the sixth threshold value, a third-level early warning is triggered; for example, when the resistivity change rate of a single monitoring point k is ≥ ± 30%, or the difference between symmetric monitoring points k is ≥ 25%, corresponding to a tilt angle θ ≈ 1.5°, the system immediately sends an emergency alert to the power transmission dispatching center, suggesting to shut down the relevant line and implement emergency disposal.

[0045] Wherein, the third threshold value is greater than the first threshold value, and the fifth threshold value is greater than the third threshold value; the fourth threshold value is greater than the second threshold value, and the sixth threshold value is greater than the fourth threshold value.

[0046] The tower tilt evaluation method based on the correlation mechanism of soil water content and resistivity of the application establishes a risk conduction path from soil parameter anomaly to resistivity change to tower tilt, calculates the resistivity change rate by monitoring the soil multi-source parameters, obtains the tilt evaluation value combined with the correlation model and divides the risk level, realizes the early prediction of the tower tilt risk from the soil parameter anomaly, and provides a basis for the preventive maintenance of the transmission line tower.

[0047] For the prediction of resistivity, in other embodiments, a support vector machine (SVM) based prediction model can also be used to predict resistivity, or an integrated learning prediction of resistivity based on random forest or gradient boosting tree (such as XGBoost). When using a support vector machine (SVM) based prediction model to predict resistivity, the soil resistivity is predicted according to the input features such as soil water content and temperature. SVM realizes regression prediction by finding the optimal hyperplane, which is suitable for small sample, nonlinear, high-dimensional data scenarios, and has good anti-overfitting characteristics. The specific implementation process is as follows: For the nonlinear relationship between soil water content, temperature and resistivity, radial basis function (RadialBasisFunction, RBF) is selected as the kernel function, and its expression is as follows: Among them, : represents the inner product of two input sample vectors and in the high-dimensional feature space; represents the input feature vector of the th and the th sample; is the kernel parameter, which controls the distribution of samples after mapping to high-dimensional space.

[0048] Support vector regression (SVR) is used, the target is to find a function , so that the deviation between the predicted value and the true value of all samples does not exceed , and at the same time minimize the model complexity (i.e. ).

[0049] Among them, the key hyperparameters of the model include the penalty coefficient , which is used to control the penalty degree of samples exceeding ; the kernel parameter , which affects the measurement of sample similarity; and the insensitive loss parameter ​, which is used to control the width of the regression band. By employing Grid Search or Bayesian Optimization combined with cross-validation, the optimal combination of hyperparameters is found. Using the training set data, the support vectors and model weights are obtained by solving a convex quadratic programming problem, and the training of the SVM regression model is completed.

[0050] The real-time collected soil water content and temperature data (after the same preprocessing) are input into the trained SVM model, and the corresponding soil resistivity prediction value is output .

[0051] SVM is based on the principle of structural risk minimization, which can maintain good generalization performance under limited samples, and is especially suitable for the actual situation of limited monitoring points of transmission towers and small amount of data samples. Through the RBF kernel function, the complex nonlinear relationship between water content, temperature and resistivity can be effectively fitted, and the prediction accuracy is usually better than that of traditional linear regression methods.

[0052] Specifically, in a preferred embodiment of the present application, when the prediction model fuses the soil water content and soil temperature data, a seasonal time weight factor is introduced, which is determined according to the analysis of the influence of historical synchronous data on resistivity, to optimize the input of the prediction model.

[0053] The preferred embodiment introduces a seasonal time weight factor to optimize the soil water content and temperature data input into the prediction model, in order to improve the accuracy of resistivity prediction. The specific implementation process is as follows: Let the collected historical synchronous monitoring data set be , where is the timestamp of the th data point; are the measured values of soil volume water content, soil temperature and soil apparent resistivity at the corresponding time, respectively. is the total number of historical data points.

[0054] Define a season division function , which maps the time to the season it belongs to (e.g. ). According to this, the historical data set is divided into four seasonal subsets: For each season , analyze the water content and temperature on its data set , respectively.the independent influence on resistivity .

[0055] the partial correlation coefficient with , and , controlling for , and , controlling for . The seasonal weight factor is defined as: This definition normalizes the weights so that features with greater influence get higher weights and features with less influence get lower weights in any season.

[0056] As an alternative, a simple linear influence model can be fitted for each season : where and are the fitted regression coefficients, and is the error term. The seasonal weight factor is then determined by the relative size of the coefficients: Let the current time be , and its belonging season be . The real-time collected soil moisture and temperature data are and , respectively. Before making the resistivity prediction, the input features are preprocessed by weighting: Subsequently, the preprocessed feature vector is input into the trained BP neural network prediction model , and the resistivity prediction value at this time is obtained: To adapt to long-term climate change, the weight factor can be established with a periodic update mechanism. After obtaining the new data of the th complete year, the above steps are re-executed using the entire historical data set containing the rolling time window of the new data to calculate the updated seasonal weight factor set , which is used for the prediction of the next period. At the end of each season, the weight factors are fine-tuned by re-analyzing the newly accumulated data of the season, so that the model can adapt to long-term climate change trends.

[0057] In the preferred embodiment, by embedding the seasonal physical laws in the form of weight factors into the prediction process, the importance of different features in the input is dynamically adjusted, so that the prediction output of the neural network model is more in line with the change law of the soil electrical properties in different seasons. A single neural network model can better adapt to the climate change throughout the year, avoiding the complexity of establishing different models for different seasons, thereby effectively reducing seasonal system errors and improving overall prediction accuracy.

[0058] Specifically, in one preferred embodiment of the present application, a tower tilt evaluation method based on the correlation mechanism between soil moisture content and resistivity further comprises: Obtaining the real-time tower tilt angle and comparing it with the tilt evaluation value; According to the comparison difference, the correlation between the tower tilt angle and the resistivity change rate in the correlation model is feedback calibrated.

[0059] The preferred embodiment is used to solve the prediction deviation problem of the tilt angle and resistivity change rate correlation model caused by long-term changes in soil parameters, accumulation of basic micro-damage, etc. By establishing a closed-loop feedback calibration mechanism, the directly measurable tower tilt angle is used as the true data to intermittently or trigger the correction of the correlation model. The specific refinement steps are as follows: Set the calibration trigger condition and data window, and set it to automatically start the calibration process once every quarter or every half year. When the system is in a first-level warning state and lasts for more than a preset time, or after extreme events such as lightning and gale, the calibration is automatically triggered.

[0060] Draw a stable data sampling window, during which the weather is stable and there is no external interference. At the same time, collect the measured average tilt angle of the high-precision tilt sensor and the multi-source physical parameters in the window .

[0061] Input the physical parameters in the data window into the prediction model to obtain the resistivity prediction value, and then calculate the measured resistivity change rate of each monitoring point .

[0062] According to the spatial distribution of , the current correlation model is used to deduce an evaluation tilt angle .

[0063] The absolute difference between the evaluation value and the measured value .

[0064] Set a difference threshold . If , it is considered that the model is still accurate at present and calibration is not needed; if , calibration is started. The calibration does not change the model structure, but adjusts the core mapping parameters. For example: If the correlation model is a lookup table derived from finite element simulation, according to the direction and size of the difference , the resistivity variation rate corresponding to each inclination angle in the table is globally scaled or offset adjusted in proportion.

[0065] If the model is an empirically fitted mathematical function, the measured points are used as new anchor points, and an online learning algorithm such as the promotion least square method is used to update the parameter function.

[0066] The embodiment of the application also provides a tower tilt evaluation system based on the correlation mechanism of soil moisture content and resistivity, which is applied to the method described above and comprises: A data acquisition unit is configured to acquire multi-source parameters of soil around a target tower and parameters of the tower itself. A data processing unit is configured to denoise, normalize and encode the multi-source parameters, and filter out key data features associated with the tower tilt state based on correlation analysis to form an input feature set. A resistivity prediction unit is configured to construct and train a prediction model based on the input feature set, with soil moisture content and soil temperature as inputs and soil resistivity as output, for outputting a predicted resistivity according to real-time multi-source parameters. A correlation model construction unit is configured to establish a tower-foundation-soil coupled mechanical model based on multi-source parameters and state parameters, and construct a correlation model between tower tilt angle and resistivity variation rate in combination with a pre-set soil mechanical state and resistivity correlation rule. A tilt risk evaluation unit is configured to calculate the resistivity variation rate of a monitoring point based on the predicted resistivity, obtain a tower tilt evaluation value in combination with the correlation model, and evaluate the tilt risk level of the tower.

[0067] The functions of each unit in the embodiment are explained in the same way as the tower tilt evaluation method based on the correlation mechanism of soil moisture content and resistivity, and the technical effects are the same, and thus are not repeated here.

[0068] Those skilled in the art can appreciate that the units of the examples described in combination with the embodiments disclosed herein can be realized in electronic hardware, computer software or a combination of both. In order to clearly illustrate the interchangeability of hardware and software, the components of the examples have been described in general terms in the above description. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. The skilled person can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0069] In the embodiments provided by the present application, it should be understood that the division of units is only a logical functional division, and there can be another division manner in actual implementation, for example, multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored, etc.

[0070] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0071] When the integrated unit is realized in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: a U disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a mobile hard disk, a magnetic disk or an optical disk, and various media that can store program codes.

[0072] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application, and they should be covered in the scope of the description of the present application.

Claims

1. A tower tilt evaluation method based on the correlation mechanism of soil water content and resistivity, characterized in that, The method comprises the following steps: obtaining multi-source parameters of the soil around the target tower and parameters of the tower itself; de-noising, normalizing and encoding the multi-source parameters, and screening out key data features associated with the tower tilt state based on correlation analysis to form an input feature set; based on the input feature set, constructing and training a prediction model with soil moisture content and soil temperature as inputs and soil resistivity as output, for outputting predicted resistivity according to real-time multi-source parameters; based on the multi-source parameters and state parameters, establishing a tower-foundation-soil coupling mechanical model, and combining a pre-set soil mechanical state and resistivity correlation rule to construct a correlation model between the tower tilt angle and the resistivity change rate; based on the predicted resistivity, calculating the resistivity change rate of the monitoring point, combining the correlation model to obtain a tower tilt evaluation value, and evaluating the tower tilt risk level.

2. The tower tilt evaluation method based on the correlation mechanism of soil water content and resistivity according to claim 1, characterized in that, The multi-source parameters are synchronously collected by multiple monitoring points arranged in a monitoring area centered on the tower; the spatial layout of the monitoring area is configured based on the geometric features of the foundation structure of the tower.

3. The tower tilt evaluation method based on the correlation mechanism of soil water content and resistivity according to claim 1, characterized in that, The prediction model is a BP neural network model, the number of nodes in the hidden layer of which is configured according to the soil type of the multi-source parameters; the BP neural network model is trained using the Adam optimization algorithm and the mean square error as the loss function.

4. The tower tilt evaluation method based on the correlation mechanism of soil water content and resistivity according to claim 3, characterized in that, When the prediction model fuses the soil moisture content and soil temperature data, a seasonal time weight factor is introduced, which is determined according to the analysis of the influence of historical same-period data on the resistivity to optimize the input of the prediction model.

5. The tower tilt evaluation method based on the correlation mechanism of soil water content and resistivity according to claim 1, characterized in that, The pre-set soil mechanical state and resistivity correlation rule is obtained by the following method: preparing a sample with the same soil properties as the soil around the target tower; applying different mechanical loads to the sample to produce different volume strains, and synchronously measuring the resistivity of the sample under each volume strain state; based on the measurement data, fitting to obtain a continuous function relationship or a relationship curve between the volume strain and the resistivity of the soil.

6. The tower tilt evaluation method based on the correlation mechanism of soil water content and resistivity according to claim 5, characterized in that, The method of establishing a tower-foundation-soil coupling mechanical model based on multi-source parameters and state parameters, and combining a pre-set soil mechanical state and resistivity correlation rule to construct a correlation model between the tower tilt angle and the resistivity change rate comprises: determining the geometric boundary and material properties of the coupling mechanical model according to the foundation size of the target tower and the soil geological survey parameters; in the simulation analysis, beam elements are used to simulate the tower structure, solid elements are used to simulate the tower foundation, and elastoplastic constitutive models are used to simulate the surrounding soil, and the coupling mechanical model is assembled; applying a horizontal displacement load to the top of the tower in the coupling mechanical model to simulate the state of the tower reaching a plurality of predetermined tilt angles; calculating and extracting the volume strain data of the soil at the monitoring points under each predetermined tilt angle state; based on the soil volume strain and resistivity relationship curve calibrated in advance by experiment, the volume strain data is converted into the corresponding soil resistivity change rate.

7. The tower tilt evaluation method based on the correlation mechanism of soil water content and resistivity according to claim 1, characterized in that, The criteria for evaluating the tilt risk level of the tower include: The resistivity change rate of a single monitoring point exceeds an independent threshold value, and the resistivity change rate gradient between multiple monitoring points with a preset geometric correlation exceeds a gradient threshold value; The monitoring points with the preset geometric correlation are point pairs that are symmetric about the axis or center of the tower foundation.

8. The tower tilt evaluation method based on the correlation mechanism of soil water content and resistivity according to claim 7, characterized in that, The tilt risk level includes three levels, including: When the absolute value of the resistivity change rate of any monitoring point reaches a first threshold value, or the difference in the resistivity change rate between any two spatially symmetric monitoring points reaches a second threshold value, a first-level early warning is triggered; When the absolute value of the resistivity change rate of any monitoring point reaches a third threshold value, or the difference in the resistivity change rate between any two spatially symmetric monitoring points reaches a fourth threshold value, a second-level early warning is triggered; When the absolute value of the resistivity change rate of any monitoring point reaches a fifth threshold value, or the difference in the resistivity change rate between any two spatially symmetric monitoring points reaches a sixth threshold value, a third-level early warning is triggered; The third threshold value is greater than the first threshold value, and the fifth threshold value is greater than the third threshold value; the fourth threshold value is greater than the second threshold value, and the sixth threshold value is greater than the fourth threshold value.

9. The tower tilt assessment method based on the mechanism of correlation between soil water content and resistivity according to any one of claims 1-7, characterized in that, Further comprising: Obtaining a real-time tower tilt angle and comparing it with the tilt evaluation value; According to the difference in comparison, the correlation between the tower tilt angle and the resistivity change rate in the correlation model is fed back and calibrated.

10. A tower tilt evaluation system based on the mechanism of correlation between soil water content and resistivity, characterized in that, Applied to the method of any one of claims 1-9, comprising: A data acquisition unit for acquiring multi-source parameters of the soil around the target tower and parameters of the tower itself; A data processing unit for denoising, normalizing and encoding the multi-source parameters, and filtering out key data features associated with the tilt state of the tower based on correlation analysis to form an input feature set; A resistivity prediction unit for constructing and training a prediction model based on the input feature set, with soil moisture content and soil temperature as inputs and soil resistivity as output, for outputting a predicted resistivity according to real-time multi-source parameters; An association model construction unit for establishing a tower-foundation-soil coupled mechanical model based on multi-source parameters and state parameters, and constructing an association model between the tower tilt angle and the resistivity change rate in combination with a preset soil mechanical state and resistivity correlation rule; A tilt risk assessment unit for calculating the resistivity change rate of the monitoring point based on the predicted resistivity, obtaining a tower tilt evaluation value in combination with the association model, and assessing the tilt risk level of the tower.

Images