Communication engineering-oriented digital twin basin infrastructure construction control system

Abstract

The invention belongs to the technical field of digital twinning communication of drainage basin infrastructure, and discloses a digital twinning drainage basin infrastructure control system for communication engineering, when a communication detector monitors signal attenuation, the sampling interval of a hydrological sensor can be automatically shortened, the frequency of an infrastructure collector is improved, and the communication efficiency is improved. Microcosmic signal fluctuation is associated with macroscopic hydrology and mesoscopic infrastructure data in real time; meanwhile, data of different scales are subjected to layered denoising, the cross-scale correlation degree is calibrated through wavelet transform and an attention mechanism in combination with a Pearson coefficient, the space error of the fused data is controlled within + / -1.5 m in cooperation with a space deviation compensation equation, and deep correlation fusion of the multi-scale data is achieved; the output calibration data can effectively reduce communication control errors, and the communication control problem caused by inaccurate data is avoided; a cross-domain physical rule base is established based on calibration data, the cross-domain physical rule base comprises dam stress-strain and base station tilt-link attenuation relations, and water conservancy-communication multi-physics field coupling equation quantitative correlation is further established.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of digital twin communication of river basin infrastructure, and specifically relates to a digital twin river basin infrastructure control system for communication engineering. BACKGROUND

[0002] With the deep integration of river basin infrastructure and communication engineering, digital twin technology becomes an important technical means for supporting the whole life cycle management and control of complex river basin infrastructure because it can realize real-time mapping of physical river basin and virtual river basin. However, when digital twin river basin and communication engineering are applied, there are many technical limitations, and the following technical problems exist: The hydrological data, communication signals and infrastructure construction in the river basin have large differences in time and space scales. The existing technology only aligns the data by time stamp to splice them together, and cannot establish the internal relationship between micro communication signal fluctuations and macro hydrological changes, resulting in large errors when the fused data is used for communication control, and cannot support precise operations such as base station power adjustment and link optimization.

[0003] The existing model generally models the correlation of historical data, such as the rule that water level rise leads to communication interruption. However, the complete cause-and-effect chain among hydrology, infrastructure and communication has not been sorted out, for example, water level rise will increase the water content of embankment soil, and then cause embankment deformation and base station tower foundation tilt, and finally cause communication link attenuation and interruption. Such a model that only looks at the correlation can only cope with regular working conditions, and when encountering unexpected situations such as pipe gushing and extreme rainstorms, the prediction accuracy will be greatly reduced, and effective communication control plans cannot be prepared in advance.

[0004] River basin infrastructure involves water conservancy, communication and construction departments. Water conservancy needs to ensure river basin safety, and may need to close nearshore base stations during flood season. Communication needs to ensure network coverage and keep base stations running smoothly. Construction needs to ensure high-frequency signal transmission of equipment such as shield machines. The existing scheme does not have a mechanism for dynamically adjusting target weights, and when conflicts occur, they can only be coordinated through manual meetings, which is slow and easy to miss the best control time of emergency situations such as flood season and pipe gushing, resulting in long communication interruption time, construction rework, or increased risk of river basin safety. SUMMARY

[0005] The purpose of the present application is to provide a digital twin river basin infrastructure control system for communication engineering to solve the problems raised in the background.

[0006] In order to achieve the above purpose, the present application provides the following technical scheme: a digital twin river basin infrastructure control system for communication engineering, wherein the data preprocessing module is specifically as follows: Firstly, the terminal collects data, when the communication detector monitors signal attenuation, the hydrological sensor will automatically trigger to shorten the sampling interval, and the sampling frequency of the infrastructure collector will be increased, so that the micro signal fluctuation energy can be associated with hydrological data and infrastructure data in real time; the wavelet threshold denoising method is used for the second-level high-frequency component of the communication signal, and the Kalman filter is used for the hour-level low-frequency component of the hydrological data; After decomposing the data characteristics of different scales by wavelet transform, the correlation area of hydrological changes and communication intermediate frequency components is focused on by combining the attention mechanism, the correlation degree between communication minute-level intermediate frequency components and hydrological water level changes is calculated by using the Pearson coefficient, and if the correlation degree is lower than 0.6, the attention weight of the Transformer network is adjusted; For the spatial scale difference between infrastructure data and communication data, a scale deviation compensation equation is established, and a correction coefficient based on distance weight is added, so that the spatial error of the fused data is controlled within ±1.5 meters, and finally the calibrated multi-scale data is output.

[0007] Further, the coupling modeling module is specifically as follows: Based on the multi-scale data calibrated by the data preprocessing module, a cross-domain physical rule library is first established, including the stress and strain relationship of dam materials, the tilt of base station tower and the link attenuation formula, and a water conservancy and communication cross-domain multi-physical field coupling equation is also established; when the water conservancy rules and communication rules conflict, the rule conflict resolver is combined with the multi-objective optimization function to dynamically adjust the rule threshold; In terms of causal modeling, a causal graph of the core node is established, the influence weight of each node on the communication fault is calculated by using the entropy weight method, and the causal strength parameters of the high-influence-degree nodes are preferentially iterated; at the same time, a real-time data and rule double-feedback closed loop is established, if the deviation between the monitored base station tilt value and the graph prediction value exceeds 15%, not only the causal strength parameters are corrected, but also the corresponding mechanical transmission equation coefficient in the physical rule library is adjusted.

[0008] Further, the intelligent adaptation module is specifically as follows: Based on the causal graph of the coupling modeling module, the root cause tree of the fault is formed by tracing back the link attenuation source in reverse, the influence weight of each root cause is calculated based on the fault propagation path by using the analytic hierarchy process, and a root cause influence degree heat map is generated, and implicit correlations are supplemented to the root cause tree by using association rule analysis; When different adjustment strategies are formulated for different root causes, the root cause influence degree and the strategy execution cost are combined to sort the strategy priority by using the TOPSIS method; at the same time, the model is introduced to predict the strategy effect, the scheme with the effect meeting the standard and the lowest cost is selected to realize root cause repair type link adjustment, and finally the fault root cause and adjustment effect data are output.

[0009] Further, the multi-objective decision-making module is specifically as follows: In combination with the fault root cause and adjustment effect data of the intelligent adaptation module, a working condition and weight mapping matrix is first established, so as to clearly define the target priority under different working conditions; when multiple targets conflict, the conflict root cause is traced back through the causal graph of the coupling modeling module, the target weight corresponding to the conflict root cause is temporarily increased, and the weight adjustment boundary is referred to; If an extreme conflict is encountered, a combined compensation scheme is automatically generated; after decision execution, the weight model for the next similar working condition is corrected through reinforcement learning by monitoring the communication coverage recovery rate and the basin safety risk, and finally the decision result is output.

[0010] Further, the early warning distributed module is as follows: Based on the upstream node monitoring demand of the causal chain and the working condition weight data of the multi-target decision module, a three-level system of root cause early warning, derivative early warning and fault early warning is established, the root cause early warning is pushed to the water conservancy department, the derivative early warning is pushed to the construction department, and the fault early warning is pushed to the communication department; based on fault records, different early warning thresholds are calculated by survival analysis method; at the same time, the early warning level dynamic upgrade rule is set, when the embankment displacement is 3.8mm and the rainfall in the next 24 hours is >50mm, the root cause early warning is automatically upgraded from yellow to orange, and the preparation is started in advance; The edge computing node tracks the upstream node state of the causal chain in real time, first removes the false data caused by sensor failure through the 3σ principle, and then uploads the effective data to the cloud; if a node triggers the root cause early warning, it will be automatically synchronized to the surrounding base station monitoring and construction monitoring nodes, starting collaborative monitoring, forming a global early warning network, and finally outputting early warning information.

[0011] The data statistical window length of the 3σ principle is set to 10 minutes, that is, the sensor sampling data in the last 10 minutes is taken to calculate the mean μ and the standard deviation σ, and the data beyond the range of [μ-3σ, μ+3σ] is determined as false.

[0012] Further, the causal correlation operation and maintenance cooperation module is as follows: In combination with the early warning information of the early warning distributed module and the decision guidance of the multi-target decision module, the influence of construction action on the causal chain is simulated through the discrete element method in the construction stage, and a construction parameter and risk correlation table is generated; for key processes including shield machine crossing the embankment, real-time upstream nodes of the causal chain monitoring embankment stress, if the stress exceeds the threshold, the construction is automatically suspended and the parameters are adjusted, and a construction taboo list is generated; After entering the operation and maintenance stage, historical operation and maintenance data are included in the causal graph, and when the water level exceeds the threshold, the drainage port cleaning instruction is automatically triggered; by analyzing historical data to mine implicit causal correlations, and feeding back the operation and maintenance effect to the causal graph, the correlation strength is corrected, and the construction taboo list is updated, and finally the operation and maintenance data is output.

[0013] Further, the security sharing module is as follows: Firstly, the sharing scene is divided into daily monitoring, emergency decision and fault tracing three categories, the risk level of each scene is evaluated by fuzzy comprehensive evaluation method;The sensitive field of data is automatically labeled, and the sharing content is screened according to the risk level of the scene, so that the sharing range is matched with the actual demand; Based on the differential privacy technology, the data is graded and desensitized, and the desensitization intensity is adjusted according to the scene risk;At the same time, the usability of the desensitized data is verified, if the usability is lower than 80%, the desensitization algorithm is adjusted, so that the security and efficiency are balanced.

[0014] The beneficial effects of the application are as follows: 1、The communication detector of the application can automatically shorten the sampling interval of the hydrological sensor and improve the frequency of the infrastructure collector when monitoring signal attenuation, so as to associate micro-signal fluctuation with macro-hydrological and meso-infrastructure data in real time;At the same time, the data of different scales is de-noised layer by layer, and then the cross-scale correlation is calibrated by wavelet transform, attention mechanism and Pearson coefficient, and the spatial error of the fused data is controlled within ±1.5 meters by using the spatial deviation compensation equation;Instead of simply splicing data, the application realizes deep correlation fusion of multi-scale data, and the calibrated data output can effectively reduce communication control error, provide reliable basis for precise operation such as base station power adjustment and link optimization, and avoid communication control problems caused by inaccurate data.

[0015] 2、The application establishes a cross-domain physical rule library based on the calibrated data, which includes the relationship between dam stress and strain, base station inclination and link attenuation, and also constructs a multi-physical field coupling equation to quantify the correlation;Around 12 core nodes, a causal map is built, the influence weight of the nodes on communication failure is calculated by entropy weight method, and the causal chain is checked by real-time data and rules every 3 minutes;When the deviation exceeds the threshold, the causal parameters and physical rule coefficients are corrected synchronously;This dual-driven system of physical rules and causal map can sort out the complete causal chain of hydrology, infrastructure and communication, even in the face of unexpected abnormal events such as sudden piping and extreme rainstorms, the influence can be deduced through causal logic, which greatly improves the prediction accuracy and creates conditions for formulating communication control plans in advance, so that passive response to failure is no longer needed.

[0016] 3、The application establishes a dynamic weight matrix by combining fault root cause and working condition through a multi-target decision module, adjusts target weight by tracing the root cause when there is a conflict, and automatically generates a combination compensation scheme of suspension construction + encryption monitoring + deployment of unmanned base station in extreme cases, and the weight model of similar working conditions can be corrected next time through reinforcement learning after decision-making; The causal correlation operation and maintenance cooperation module runs through the whole cycle with early warning information and decision direction, simulates the influence of action on the causal chain during the construction period, controls the key process, and excavates the implicit causal correlation and feedback optimization during the operation and maintenance period; Eliminate the long delay of manual coordination, quickly determine the decision under emergency working conditions such as flood season and pipe heaving, realize the linkage of construction and operation and maintenance, reduce the communication interruption time and construction rework, and effectively improve the efficiency of the coordinated management of basin infrastructure and communication. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The application is a digital twin basin infrastructure control system for communication engineering. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of the application.

[0019] As shown in Figure 1 The application provides a digital twin basin infrastructure control system for communication engineering. The data preprocessing module first realizes dynamic linkage of data acquisition through scale adaptation of the acquisition terminal. When the communication detector detects signal attenuation, for example, >5dB / 10ms, the hydrological sensor will automatically trigger the sampling interval to be shortened from 1 hour to 10 minutes, and the sampling frequency of the infrastructure collector is increased to 5 times / hour, so that the microscopic signal fluctuation can be associated with hydrological data and infrastructure data in real time. In order to reduce the noise interference of different scale data, wavelet threshold denoising is used for the second-level high-frequency component of the communication signal, and the wavelet threshold type is selected as a soft threshold, and the threshold calculation formula is: threshold = σ x n , wherein σ is the noise standard deviation, and n is the number of sampling points; Kalman filtering is used for the hour-level low-frequency component of the hydrological data to ensure the quality of the original data. The process noise covariance Q of Kalman filtering is set to 0.01 (corresponding to the noise intensity of the hour-level change of the hydrological data), the observation noise covariance R is set to 0.05 (corresponding to the measurement error of the hydrological sensor), and the initial state vector is set to the average of the first 3 times of sampling of the hydrological water level.

[0020] In the data fusion link, after decomposing the data characteristics of different scales by wavelet transform, the attention mechanism is combined to focus on the correlation area of hydrological changes and communication intermediate frequency components, and a cross-scale correlation degree calibration is introduced: the Pearson coefficient is used to calculate the correlation degree between the communication minute-level intermediate frequency components and the hydrological water level changes. If the correlation degree is less than 0.6, the attention weight of the correlation area of hydrological changes and communication intermediate frequency components in the Transformer network is increased from the initial value 0.3 to 0.5, and after adjustment, the correlation degree should be ≥0.6. The gradient descent method is used for iterative optimization during adjustment, the learning rate is set to 0.001, and the number of iterations is not more than 50 times, to ensure that the correlation degree is quickly met and overfitting is avoided.

[0021] In view of the spatial scale difference between infrastructure data (hundred-meter spatial scale) and communication data (meter spatial scale), a scale deviation compensation equation is established and a correction coefficient based on distance weight is added to control the spatial error of the fused data within ±1.5 meters, and finally the calibrated multi-scale data is output; Spatial deviation compensation equation: ; In the formula: represents the spatial deviation compensation value, which is used to correct the spatial position deviation of infrastructure data and communication data to ensure that the spatial error after data fusion is controllable, with a unit of meters and a range of [-1.5m, 1.5m]; represents the distance weight coefficient, which quantifies the influence of the distance between the infrastructure collection point and the base station on the spatial deviation, and dynamically adjusts the compensation strength with distance. The near distance (D≤100m) takes 0.002, and the far distance (D>100m) takes 0.005; represents the Euclidean distance, which refers to the straight-line distance between the infrastructure collection point and the communication base station, with a unit of meters, calculated by the coordinates of the two through the Euclidean formula; represents the system error correction term, which compensates for fixed errors such as sensor installation deviation and instrument inherent error, and is independent of distance, with a unit of meters and a range of [-0.2m, 0.2m], which needs to be calibrated and updated regularly; the calibration period is once a month, and the average value is updated by measuring the infrastructure-base station coordinate deviation of 3 groups of different distances

[0022] Multi-scale data cross-domain correlation degree calculation formula: ; In the formula: represents the cross-scale correlation degree between communication intermediate frequency components and hydrological water level changes, i.e. the Pearson correlation coefficient, with a value range of [-1, 1]. When ≥0.6 is determined as strong correlation, ​Adjust the attention weight of the Transformer network when < 0.6; represents the number of sampling data sets, which is determined by the synchronous sampling period of the scale adaptation acquisition terminal, such as the communication intermediate frequency component sampled at the minute level, the hydrological water level sampled at the 10-minute level, is the number of synchronous sampling times in the same time window; represents the first group of communication intermediate frequency component sampling values, with units of dB, corresponding to the effective components of the minute-level fluctuations of the communication signal, which are extracted after wavelet transform decomposition; represents the first group of hydrological water level change sampling values, with units of m, corresponding to the increment of the hydrological water level in the same time window, such as the water level rising from 12.5 m to 12.7 m in the first 10 minutes, = 0.2; represents the sum of all communication intermediate frequency component sampling values; represents the sum of all hydrological water level change sampling values; represents the sum of squares of all communication intermediate frequency component sampling values; represents the sum of squares of all hydrological water level change sampling values.

[0023] The coupling modeling module first establishes a cross-domain physical rule library based on the multi-scale data calibrated by the data preprocessing module, including the stress and strain relationship of dam materials, the tilt and link attenuation formula of base station towers, and other core laws. In addition, a water conservancy and communication cross-domain multi-physical field coupling equation is established, such as the mechanical transmission relationship between dam vertical displacement Δh and base station tower tilt angle θ. When water conservancy rules (such as dam allowable displacement ≤10 mm) and communication rules (such as base station tilt allowable angle ≤0.5°) conflict, a rule conflict resolver (a software algorithm module integrating multi-objective optimization functions, with the target variable being the basin safety risk value and the communication interruption probability value) is used in combination with a multi-objective optimization function. The core logic is to construct a weighted summation objective function with the basin safety risk value (weight 0.6) and the communication interruption probability value (weight 0.4), and solve the optimal rule threshold value through a particle swarm algorithm. At the same time, the dam safety risk and the communication interruption probability are minimized, and the rule threshold value is dynamically adjusted, such as reducing the dam allowable displacement to 8 mm during the flood season, and synchronously reducing the base station tilt allowable angle to 0.3°; The specific form of the water conservancy-communication multi-physical field coupling equation is: ; This represents the communication link attenuation value, measured in dB. The tilt effect coefficient is set to 0.8 dB / °. The displacement influence coefficient is set to 1.2 dB / m. The stress influence coefficient of the dam is taken as 0.5 dB / MPa; This represents the stress value of the dam, in units of... .

[0024] In terms of causal modeling, a causal graph is established, including 12 core nodes such as water level, dam deformation, base station displacement, and signal attenuation. The entropy weight method is used to calculate the influence weight of each node on communication failure. Data standardization adopts min-max standardization, mapping the original data to the [0,1] interval. The formula is: Standardized value = (original value - minimum value) / (maximum value - minimum value). For example, if the influence of water level exceeding the warning level is 70% and the influence of dam deformation is 85%, the causal intensity parameters of high-influence nodes are prioritized for iteration. At the same time, a real-time data and rule dual feedback closed loop is established, and the causal chain is checked every 3 minutes. If the deviation between the monitored base station tilt value and the graph prediction value exceeds 15%, not only are the causal intensity parameters corrected, but the corresponding mechanical transfer equation coefficients in the physical rule base are also adjusted to ensure that the model conforms to both data regularity and physical logic.

[0025] Formula for multi-physics coupling transfer between dam and base station: ; In the formula: This indicates the base station tilt angle, in degrees. The value must meet the communication rule threshold, such as ≤3° during the flood season and ≤5° during the normal season. It is monitored in real time by the base station displacement sensor. This represents the tower base stiffness coefficient, measured in ° / m. It is a dynamic variable that changes with the moisture content of the dam material; for example, it varies with the soil moisture content of the dam, increasing by 10%. The value increased from 0.05° / m to 0.06° / m; This represents the vertical displacement of the dam, measured in meters (m). The value must meet the threshold requirements stipulated by water conservancy regulations, such as ≤8mm during the flood season and ≤10mm during the normal season. It is monitored in real time by dam settlement sensors.

[0026] This represents a correction term, in degrees, with a value range of [-0.02°, 0.02°]. It is used to compensate for system deviations such as tower base installation errors and sensor measurement errors. For example, when... When the deviation between the measured value and the predicted value exceeds 15%, adjust. To deviation 5%.

[0027] Among them, the intelligent adaptation module forms a fault root cause tree by tracing back the link attenuation source based on the causal graph of the coupling modeling module, for example, weak signal → base station tilt → dam deformation → water level alarm; In order to accurately identify the key root cause, the influence weight of each root cause is calculated based on the fault propagation path using the analytic hierarchy process (AHP), such as water level alarm influence degree 60%, dam deformation influence degree 90%, and a root cause influence degree heat map is generated, and the implicit association is mined through association rule analysis (set minimum support ≥5%, minimum confidence ≥80%), the data statistical period is calculated according to the natural week, and the association rule library is updated once a week, and the invalid association with a support degree lower than 5% in the period is excluded, for example, when the water level alarm + soil moisture content > 25%, the dam deformation risk will increase by 3 times, which is supplemented to the root cause tree to ensure comprehensive coverage; When different root causes are formulated for differential adjustment strategies, the root cause influence degree and the strategy execution cost are combined, such as reinforcing the base station cost of 50,000 yuan and the emergency flood discharge cost of 200,000 yuan, and the strategy priority is sorted using the TOPSIS method, and the base station is preferentially reinforced, and then the necessity of flood discharge is evaluated; At the same time, the model predicts the strategy effect, for example, predicting the probability that the link attenuation rate decreases from 30% to 10% within 1 hour after reinforcing the base station, selecting the scheme with the lowest cost and effect meeting the standard to realize root cause repair type link adjustment, and finally outputting the fault root cause and adjustment effect data.

[0028] In the TOPSIS method, the index standardization adopts linear normalization (positive index: standardized value = original value / maximum value; negative index: standardized value = minimum value / original value), and the index weight is determined by the entropy weight method (consistent with the weight calculation logic of the coupling modeling module).

[0029] Among them, the multi-objective decision-making module combines the fault root cause and adjustment effect data of the intelligent adaptation module, first establishes a working condition and weight mapping matrix, so as to clearly define the target priority under different working conditions, such as flood season: basin safety weight 0.6, communication coverage weight 0.2, construction progress weight 0.2; Normal construction period: construction progress weight 0.5, communication coverage weight 0.3, basin safety weight 0.2; When multiple objectives conflict, such as construction progress and basin safety conflict, the conflict root cause is traced back through the causal graph of the coupling modeling module, for example, the additional stress of the dam caused by the construction shield machine load exceeds the alarm, and the target weight corresponding to the conflict root cause is temporarily increased, such as the basin safety weight from 0.2 to 0.8, and the weight adjustment boundary is set by referring to the “Basin Flood Control Standard” and “Communication Base Station Operation and Maintenance Specification”, for example, the communication coverage weight is 0.1-0.6, the basin safety weight is 0.2-0.8, and the construction progress weight is 0.1-0.5, to avoid decision imbalance caused by extreme weight; In the event of extreme conflicts, such as the necessity to shut down near-shore base stations, a combined compensation scheme is automatically generated. This could include suspending tunnel boring machine construction, reducing the monitoring interval for dam displacement to 10 minutes, and deploying drone base stations to ensure communication. After the decision is executed, by monitoring the communication coverage recovery rate (e.g., increasing from 60% to 95%) and watershed safety risks (e.g., dam stress decreasing from 1.2 MPa to 0.8 MPa), reinforcement learning is used to correct the weight model for the next similar working condition (iterations ≥ 100 times / working condition). The reward function is defined as: reward value = 0.6 × communication coverage recovery rate improvement + 0.3 × watershed safety risk decrease - 0.1 × construction cost increase rate, with the learning rate set to 0.01. Finally, the decision result is output. For example, in the next construction condition, the initial weight of watershed safety is adjusted from 0.2 to 0.3.

[0030] Cross-domain target dynamic weight calculation formula: ; In the formula: Indicates the first The dynamic weight values ​​of each target are in the range of [0, 1], and the sum of the weights of all targets is 1. Represents the judgment matrix of the first... Line number The elements of the column represent the first element. The evaluation index is for the first The importance score for each objective is an integer ranging from 1 to 9, where 1 indicates equal importance and 9 indicates extreme importance. This indicates the number of evaluation indicators, including four indicators: scope of impact of the fault, response timeliness, economic cost, and compliance. This indicates the number of decision-making objectives, which is fixed at 3, including watershed safety, communication coverage, and construction progress; Represents the judgment matrix of the first... The sum of row elements, used for Normalization is performed to eliminate the influence of dimensions.

[0031] The early warning distributed module establishes a three-level system of root cause early warning, derived early warning and fault early warning based on the causal chain upstream node monitoring demand of the coupling modeling module and the working condition weight data of the multi-objective decision module, the root cause early warning pushes the water conservancy department, the derived early warning pushes the construction department, and the fault early warning pushes the communication department, the trigger condition of the root cause early warning is embankment displacement > 4mm, the trigger condition of the derived early warning is construction induced embankment stress > 1.0MPa, and the trigger condition of the fault early warning is link attenuation rate > 30%; in order to adapt to the early warning demand of different working conditions, based on the fault record in the past 5 years in flood season or non-flood season, the survival analysis method is used to calculate different early warning thresholds, the Kaplan-Meier method is used to calculate the early warning threshold, the fault occurrence time is taken as the survival time variable, and the working condition type is taken as the grouping variable, so as to ensure that the threshold adapts to different scenes; the threshold value of embankment displacement in flood season is 4mm, and that in normal period is 5mm; the threshold value of base station inclination in flood season is 0.3°, and that in normal period is 0.5°; at the same time, the dynamic upgrade rule of early warning level is set, when the edge node monitors the embankment displacement of 3.8mm (close to the threshold value of 4mm in flood season) and the rainfall in the next 24 hours is greater than 50mm, the root cause early warning is automatically upgraded from yellow to orange, and the preparation is started in advance; The edge computing node monitors the state of the upstream node of the causal chain in real time, such as embankment displacement, first removes the false data caused by sensor failure, such as instantaneous 10mm displacement, and then uploads the effective data to the cloud; if a node triggers root cause early warning, such as embankment displacement exceeding the threshold, it will automatically synchronize to the base station monitoring and construction monitoring nodes within 5km, start common monitoring, such as encrypted inclination monitoring of base station node, form a global early warning network, and finally output early warning information.

[0032] The causal correlation operation and maintenance cooperation module combines the early warning information of the early warning distributed module and the decision guidance of the multi-objective decision module, simulates the influence of construction action on the causal chain by discrete element method in the construction stage, the time step is set to 0.01 second, the soil particle cohesion is 15kPa, and the internal friction angle is 28°, which matches the mechanical properties of common clay in the basin; when the excavation depth of foundation pit is 5m, soil settlement may be caused, which leads to the sinking of base station foundation and the increase of inclination risk by 20%, and the construction parameter and risk correlation table is generated, for example, the risk is less than 5% when the excavation depth is ≤3m; for key processes such as shield machine crossing embankment, the upstream nodes of causal chain such as embankment stress are monitored in real time, if the stress is greater than 1.2MPa, the construction is automatically suspended and the advancing speed of shield machine is adjusted to ≤5mm / min, such as reducing the advancing speed of shield machine, and a construction taboo list is generated to clearly indicate the operations that need to be avoided; After entering the operation and maintenance stage, historical operation and maintenance data, such as base station failure caused by rainwater immersion, root cause of poor drainage and associated water level change, are included in the causal graph, and when the water level exceeds the threshold, the cleaning instruction of the drainage outlet is automatically triggered; by analyzing historical data to mine implicit causal correlation, such as frequent base station failure in a certain area, which is related to poor drainage caused by soil compaction less than 90% during construction three years ago, and the operation and maintenance effect (such as base station failure reduced from 5 times per month to 1 time after cleaning the drainage outlet) is fed back to the causal graph, and the correlation strength is corrected, such as the causal strength of poor drainage leading to base station failure is increased from 0.7 to 0.9, and the construction taboo list is updated, such as the soil compaction degree after construction needs to be ≥92%, and various data generated during operation and maintenance is transferred across departments through a safe sharing mechanism to ensure the smoothness of the whole cycle collaboration; the soil compaction degree is detected by the ring knife method, with one detection point every 50 meters, and the sampling depth is 0.8 meters. The test results must meet the requirements of "Earthwork and Blasting Engineering Construction and Acceptance Specification" GB 50201.

[0033] Among them, the safe sharing module first divides the sharing scene into three categories: daily monitoring, emergency decision-making and fault tracing. The risk level of each scene is evaluated by fuzzy comprehensive evaluation method, such as fault tracing involving base station location and sampling log, the risk level is high; daily monitoring only shares statistical data, the risk level is low; the evaluation index includes data sensitivity level (weight 0.4), sharing range (weight 0.3), and leakage consequences (weight 0.3). The risk level is calculated by linear weighting; the sensitive fields of data are automatically labeled, such as base station latitude and longitude is extremely sensitive, signal average strength is low sensitive, and the sharing content is selected according to the risk level of the scene, such as emergency decision-making only shares base station tilt value, without leaking latitude and longitude, so that the sharing range matches the actual demand; Comment set: high risk = extremely high sensitivity + wide sharing range + serious leakage consequences, medium risk = medium sensitivity + medium sharing range + general leakage consequences, low risk = low sensitivity + narrow sharing range + slight leakage consequences; In desensitization processing, the data is graded desensitized based on differential privacy technology, and the desensitization strength is adjusted according to the scene risk. The high-risk scene uses ε =0.3 differential privacy, and the low-risk scene uses ε =1.0; At the same time, the usability of desensitized data is verified, such as whether the base station location processed by k-anonymity can meet the demand of regional signal coverage analysis. If the usability is less than 80%, the desensitization algorithm is adjusted, such as adjusting the k value of k-anonymity algorithm from 10 to 5, so as to balance the safety and efficiency; The safe sharing mechanism realized provides a basic guarantee for the data flow and cross-department collaboration among all modules of the system.

[0034] Differential privacy ε value is mapped according to the risk level of the scene: high risk (fault tracing) →ε = 0.3, medium risk (emergency decision) -> ε = 0.6, low risk (routine monitoring) -> ε = 1.0, mapping relationship is calibrated according to the Information Security Technology Data Desensitization Guide GB / T 37932.

[0035] It is to be noted that, in the present document, relational terms such as first and second and the like can be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0036] While embodiments of the present application have been shown and described with reference to particular embodiments thereof, it will be understood by those skilled in the art that various changes in form and details can be made therein without departing from the spirit and scope of the application. The scope of the application is defined by the appended claims and their equivalents.

Claims

1. A digital twin watershed infrastructure control system for communication engineering, characterized in that: The system includes: Data preprocessing module: Wavelet threshold denoising is applied to communication signals, Kalman filtering is applied to hydrological data, and data is fused by combining wavelet transform, attention mechanism and correlation calibration to compensate for spatial scale bias and output calibrated data; Coupled modeling module: Based on the calibrated data, a cross-domain physical rule base is established. When there is a rule conflict, the rule threshold is adjusted, a causal graph is constructed, the influence weight of nodes on communication failures is calculated, and when the deviation between the base station tilt value and the predicted value exceeds the threshold, the causal strength parameter is corrected and the rule base is adjusted, and the causal graph is output. Intelligent Adaptation Module: Based on the causal graph, establish a root cause tree for the fault, calculate the influence weight of the root cause, analyze and mine hidden correlations, prioritize the root cause sorting strategy, predict the effect of the strategy, and output the root cause and adjustment effect data. Multi-objective decision-making module: Combining the root causes of the faults and the adjustment effect data, it clarifies the objective priorities of different working conditions. When there is a conflict, it increases the objective weight of the root cause. In case of extreme conflict, it generates a combined compensation scheme, corrects the weight model for the next similar working condition, and outputs the decision results. Early warning distributed module: Establishes a three-level early warning system, calculates differentiated early warning thresholds, initiates response preparations in advance, uploads valid data after eliminating false alarm data, and outputs early warning information; Causal correlation operation and maintenance collaboration module: During the construction period, it simulates the impact of construction actions on the causal chain and controls key processes; during the operation and maintenance period, it mines hidden causal relationships by analyzing historical data. Secure sharing module: classifies scenarios to assess risk levels, adjusts the desensitization intensity according to scenario risks, and verifies the availability of desensitized data.

2. The digital twin watershed infrastructure control system for communication engineering according to claim 1, characterized in that: The data preprocessing module is as follows: First, data is collected through the acquisition terminal. When the communication detector detects signal attenuation, it will automatically trigger the hydrological sensor to shorten the sampling interval and increase the sampling frequency of the infrastructure acquisition device, so that the micro signal fluctuations can be correlated with the hydrological data and infrastructure data in real time. Wavelet threshold denoising is used for the second-level high-frequency components of the communication signal, and Kalman filtering is used for the hour-level low-frequency components of the hydrological data. After decomposing data features at different scales using wavelet transform, the attention mechanism is combined to focus on the correlation region between hydrological changes and communication mid-frequency components. The Pearson coefficient is used to calculate the correlation between the minute-level mid-frequency components of communication and hydrological water level changes. If the correlation is less than 0.6, the attention weights of the Transformer network are adjusted. To address the spatial scale differences between infrastructure data and communication data, a scale deviation compensation equation is established and a correction coefficient based on distance weight is added to control the spatial error of the fused data within ±1.5 meters, ultimately outputting calibrated multi-scale data.

3. The digital twin watershed infrastructure control system for communication engineering according to claim 2, characterized in that: The coupling modeling module is specifically as follows: Based on the multi-scale data calibrated by the data preprocessing module, a cross-domain physical rule base is first established, including the stress-strain relationship of dam materials, the tilt and link attenuation formula of base station towers, and the cross-domain multi-physics coupling equation of water conservancy and communication. When water conservancy rules and communication rules conflict, the rule threshold is dynamically adjusted by combining a rule conflict collaborative resolver with a multi-objective optimization function. In terms of causal modeling, a causal graph of core nodes is established, and the entropy weight method is used to calculate the influence weight of each node on communication failures. The causal strength parameters of high-influence nodes are iterated first. At the same time, a dual feedback loop of real-time data and rules is established. If the deviation between the monitored base station tilt value and the graph prediction value exceeds 15%, not only are the causal strength parameters corrected, but the corresponding mechanical transfer equation coefficients in the physical rule base are also adjusted.

4. The digital twin watershed infrastructure control system for communication engineering according to claim 3, characterized in that: The intelligent adaptation module is as follows: Based on the causal graph of the coupled modeling module, a root cause tree of the fault is formed by tracing the root cause of the link attenuation in reverse. The influence weight of each root cause is calculated by the hierarchical analysis method based on the fault propagation path, and a root cause influence heat map is generated. At the same time, implicit associations are mined by association rule analysis and added to the root cause tree. When formulating differentiated adjustment strategies for different root causes, the TOPSIS method is used to prioritize strategies based on the impact of the root cause and the cost of strategy implementation. At the same time, a model is introduced to predict the effect of the strategy, and the solution with the best effect and the lowest cost is selected to implement root cause repair-type link adjustment. Finally, the root cause of the failure and the adjustment effect data are output.

5. The digital twin watershed infrastructure control system for communication engineering according to claim 4, characterized in that: The multi-objective decision-making module is as follows: Combining the fault root cause and adjustment effect data of the intelligent adaptation module, a working condition and weight mapping matrix is ​​first established to clarify the target priority under different working conditions. When multiple targets conflict, the cause-effect graph of the coupled modeling module is used to trace the root cause of the conflict and automatically increase the target weight corresponding to the root cause of the conflict, while referring to the set weight adjustment boundary. In the event of extreme conflict, a combined compensation scheme is automatically generated. After the decision is implemented, the weight model for the next similar working condition is corrected by monitoring the communication coverage restoration rate and watershed safety risks, and finally the decision result is output.

6. The digital twin watershed infrastructure control system for communication engineering according to claim 5, characterized in that: The specific details of the early warning distributed module are as follows: Based on the monitoring requirements of upstream nodes in the causal chain of the coupled modeling module and the working condition weight data of the multi-objective decision-making module, a three-level system of root cause early warning, derivative early warning and fault early warning is established, and the root cause early warning is pushed to the water conservancy department, the derivative early warning is pushed to the construction department, and the fault early warning is pushed to the communication department. Based on fault records, survival analysis is used to calculate different early warning thresholds; at the same time, a dynamic upgrade rule for early warning levels is set. When an edge node detects a dam displacement of 3.8 mm and the rainfall in the next 24 hours is greater than 50 mm, the root cause warning is automatically upgraded from yellow to orange, and preparations for response are initiated in advance. Edge computing nodes track the status of upstream nodes in the causal chain in real time. They first eliminate false alarms caused by sensor failures using the 3σ principle, and then upload valid data to the cloud. If a node triggers a root cause warning, it will automatically synchronize with surrounding base station monitoring and construction monitoring nodes to start collaborative monitoring, form a global warning network, and finally output warning information.

7. The digital twin watershed infrastructure control system for communication engineering according to claim 6, characterized in that: The specific details of the causal relationship-based operation and maintenance collaboration module are as follows: Combining the early warning information from the early warning distributed module with the decision guidance from the multi-objective decision-making module, the impact of construction actions on the causal chain is simulated using the discrete element method during the construction phase, generating a construction parameter and risk correlation table; for key procedures including shield tunneling machine crossing the dam, the upstream nodes of the causal chain for real-time monitoring of dam stress are included, and if the stress exceeds the threshold, construction is automatically suspended and parameters are adjusted, while a construction prohibition list is generated. Once the operation and maintenance phase begins, historical operation and maintenance data is incorporated into the causal graph. When the water level exceeds the threshold, a drainage outlet cleaning command is automatically triggered. By analyzing historical data, hidden causal relationships are uncovered, and the operation and maintenance effects are fed back to the causal graph to correct the correlation strength. At the same time, the construction prohibition list is updated, and finally, the operation and maintenance data is output.

8. The digital twin watershed infrastructure control system for communication engineering according to claim 7, characterized in that: The secure sharing module is specifically as follows: First, the shared scenarios are divided into three categories: daily monitoring, emergency decision-making, and fault tracing. The risk level of each scenario is assessed by fuzzy comprehensive evaluation method. Sensitive fields of data are automatically labeled, and shared content is filtered according to the risk level of the scenario, so that the scope of sharing matches the actual needs. Data is anonymized in a tiered manner based on differential privacy technology, and the anonymization intensity is adjusted according to the risk of the scenario. At the same time, the usability of the anonymized data is verified. If the usability is lower than 80%, the anonymization algorithm is adjusted to achieve a balance between security and efficiency.

Images