A power grid fault prediction system and method integrating digital twins

Abstract

The present invention relates to the field of digital twin technology, specifically a power grid fault prediction system and method that integrates digital twins. The system includes: a state perception module, a path deduction module, a twin mapping module, and a prediction execution module. In the present invention, a structural state tensor is constructed by fusing multi-source heterogeneous sensor data and executing a spatiotemporal alignment algorithm. Laplace operations are combined with stress wave propagation models to analyze non-rigid deformation characteristics, and load transfer path maps are generated to reveal implicit mechanical correlations. The finite element grid dynamically corrects the Young's modulus to form a structural degradation coefficient matrix. Time series difference and long short-term memory networks are combined to capture equipment degradation trends. The Kolmogorov test quantifies the spatiotemporal distribution probability of faults, enabling power grid equipment to leap from a single threshold warning to a multi-physical field coupling prediction, improving the spatiotemporal resolution and dynamic adaptability of fault prediction, reducing dependence on manual experience, and enhancing the robustness of complex working condition systems.

Description

Technical Field

[0001] The present invention relates to the field of digital twin technology, and in particular to a power grid fault prediction system and method integrating digital twins. Background Art

[0002] The field of digital twin technology involves dynamic mapping of physical entities and virtual models throughout their entire lifecycle. This involves collecting physical data in real time through sensor networks to construct three-dimensional visualization models, and combining simulation algorithms to enable system status monitoring and trend analysis. Traditional power grid fault prediction systems, among other solutions, rely on statistical analysis of historical operating data and offline model calculations. These systems utilize SCADA systems to collect voltage and current parameters, manually set thresholds for warning of equipment anomalies, and rely on regular on-site maintenance personnel to monitor equipment aging and estimate the remaining lifecycle using empirical formulas.

[0003] Traditional methods rely on manually set fixed threshold intervals for anomaly judgment, which cannot adapt to the nonlinear attenuation characteristics of material properties during equipment operation, resulting in a contradiction between warning sensitivity and false alarm rate. The voltage and current parameters collected by the SCADA system are difficult to reflect the internal stress distribution of the equipment's mechanical structure. There are monitoring blind spots in the regular maintenance mode. The empirical formula ignores the dynamic changes of the load transfer path when estimating the remaining life, resulting in life prediction deviations. Manual on-site inspections are inefficient and cannot capture sudden deformations. When the equipment is under alternating loads or extreme environments, missed detections are likely to occur, affecting the safe and stable operation of the power grid. Summary of the Invention

[0004] The purpose of the present invention is to solve the shortcomings of the existing technology and propose a power grid fault prediction system and method integrating digital twins.

[0005] To achieve the above objectives, the present invention adopts the following technical solutions: A power grid fault prediction system integrating digital twins includes:

[0006] The state perception module is used to obtain the strain gradient of the transmission tower steel structure through distributed fiber optic sensors, collect the displacement trajectory of the transmission tower base through the Beidou positioning unit, calculate the angular velocity based on the six-axis inertial unit, perform a spatiotemporal alignment algorithm on the multi-source data, combine them into a structural state tensor, and pass the structural state tensor to the path deduction module;

[0007] The path deduction module is used to receive the structural state tensor, perform Laplace operation on the strain gradient, input the displacement trajectory coordinates into the three-dimensional velocity field correction model, and construct the stress wave propagation path in combination with the angular velocity parameters, and solve it by the finite difference method. The wave equation generates a load transfer path map;

[0008] in, Indicates the spatial position of the transmission tower steel structure Department The displacement vector at time and in meters, Represents a time variable and its unit is seconds. It represents the corrected longitudinal wave propagation velocity and the unit is m / s, represents the Laplace operator and has dimension m -2 , Represents three-dimensional space coordinates and the unit is meter, Represents the external excitation force function and the unit is N / m 3 ;

[0009] Extracting non-rigid deformation components, generating a load transfer path map, and transferring the load transfer path map to a twin mapping module;

[0010] a twin mapping module, configured to receive the load transfer path map, establish a finite element mesh model, input stress wave parameters into a damage accumulation model, modify the mesh Young's modulus, generate a structural degradation coefficient matrix, and transmit the structural degradation coefficient matrix to a prediction execution module;

[0011] The prediction execution module is used to receive the structural degradation coefficient matrix, perform time series difference on the degradation coefficient, input the sequence into the long short-term memory network, perform the Kolmogorov test, generate the fault time and space coordinate code, and enter the fault time and space coordinate code into the power grid dispatch decision.

[0012] As a further solution of the present invention, the structural state tensor includes strain gradient distribution, displacement trajectory coordinates, and angular velocity parameters. The load transfer path map is specifically the stress wave attenuation path corrected by the displacement trajectory, the non-rigid deformation area identified based on the angular velocity parameters, and the main load conduction direction obtained by strain gradient analysis. The structural degradation coefficient matrix specifically refers to material fatigue, modulus correction value, and grid unit degradation index. The fault spatiotemporal coordinate encoding includes a prediction period sequence, a confidence interval threshold, and a fault geographic coordinate.

[0013] As a further solution of the present invention, when the path deduction module executes the stress wave propagation model, the formula is used: ;

[0014] in, It represents the longitudinal wave propagation speed of stress wave in the steel structure of transmission tower and its unit is m / s. It represents the benchmark Young's modulus of the steel structure measured by the distributed optical fiber sensor and the unit is GPa. The dimensionless parameter representing the Poisson's ratio of the transmission tower steel, Indicates the density of steel and the unit is kg / m 3 ;

[0015] When calculating the stress wave propagation velocity, the strain gradient distribution is used as an initial input parameter, and the velocity field is corrected in combination with the displacement trajectory coordinates.

[0016] As a further solution of the present invention, the twin mapping module adopts a damage accumulation model when correcting the mesh Young's modulus: ;

[0017] in, represents the modified mesh element Young's modulus and is in GPa. Indicates the initial Young's modulus of the steel material as specified by the factory and the unit is GPa. represents the dimensionless damage factor generated by the qth load cycle, is the load cycle number, represents the total number of load cycles, pass Calculated, It represents the amplitude attenuation rate of the stress wave on the propagation path in the qth cycle and is calculated as (peak amplitude - trough amplitude) / peak amplitude. It represents the energy dissipation rate in the main load conduction direction during the qth cycle and is calculated as 1-(output energy / input energy).

[0018] As a further solution of the present invention, the long short-term memory network in the predictive execution module includes = 128 hidden units, whose gating mechanism uses sigmoid activation function;

[0019] The significance level of the Kolmogorov test is set at 0.05 to verify the distribution consistency between the predicted period sequence in the fault spatiotemporal coordinate encoding and the historical fault data;

[0020] in, represents the number of neurons in the hidden layer and is optimized on the training set by the gradient descent method. 0.05 represents the upper limit of the probability of the first type of error allowed in the statistical test.

[0021] As a further solution of the present invention, the spatiotemporal alignment algorithm uses a dynamic time warping method to unify the sampling frequency of the strain gradient, displacement trajectory, and angular velocity parameters to 100 Hz;

[0022] When extracting the non-rigid deformation component, the curvature change threshold is set to 0.15 rad / m 2 ;

[0023] Among them, 100 Hz represents the sampling frequency of collecting 100 data points per second and is set according to the maximum sampling capacity of the distributed optical fiber sensor, 0.15 rad / m 2It represents the critical judgment value of the surface curvature change of the transmission tower structure and is obtained through the Q235 steel yield test calibration.

[0024] As a further solution of the present invention, the unit size of the finite element mesh model is adaptively adjusted according to the main load transmission direction;

[0025] A 5mm×5mm grid is used in the load conduction dense area, and a 10mm×10mm grid is used in the conduction sparse area;

[0026] Among them, 5mm represents the side length of the square grid unit and is applicable to areas where the stress value in the main load transmission direction exceeds 50MPa; 10mm represents the side length of the square grid unit and is applicable to areas where the stress value in the main load transmission direction is less than 50MPa.

[0027] As a further solution of the present invention, the update period of the structural degradation coefficient matrix is ​​set to 30 minutes;

[0028] When the grid unit degradation index exceeds 0.7, the real-time warning mechanism of the prediction execution module is triggered;

[0029] Among them, 30 minutes represents the minimum time interval for the system to update the structural degradation state, 0.7 represents the critical threshold of the grid unit safety factor and is determined according to the failure critical value of the 85% confidence interval in the tensile test of metal materials in GB / T228.1-2021.

[0030] A method for predicting power grid faults by integrating digital twins is provided. The method is based on the above-mentioned power grid fault prediction system integrating digital twins and includes the following steps: S1: obtaining the strain gradient of the steel structure of a transmission tower through a distributed optical fiber sensor, collecting the displacement trajectory of the transmission tower base through a Beidou positioning unit, calculating the angular velocity based on a six-axis inertial unit, performing a spatiotemporal alignment algorithm on the multi-source data, and combining the data into a structural state tensor;

[0031] S2: performing a Laplace operation on the strain gradient in the structural state tensor, inputting the result into a stress wave propagation model, extracting non-rigid deformation components, and generating a load transfer path map;

[0032] S3: establishing a finite element mesh model based on the load transfer path map, inputting stress wave parameters into a damage accumulation model, modifying the mesh Young's modulus, and generating a structural degradation coefficient matrix;

[0033] S4: Performing time series difference on the degradation coefficients in the structural degradation coefficient matrix, inputting the sequence into a long short-term memory network, performing a Kolmogorov test, generating a fault spatiotemporal coordinate code and entering it into a power grid dispatch decision.

[0034] As a further solution of the present invention, the establishment of the stress wave propagation model in step S2 includes: S201, calculating the reference propagation velocity according to the material property parameters ;

[0035] S202, constructing a three-dimensional velocity field distribution based on the displacement trajectory coordinates;

[0036] S203, using the finite difference method to solve the stress wave propagation partial differential equation;

[0037] The Young's modulus correction in step S3 includes: S301, calculating the cumulative damage factor of each grid unit ;

[0038] S302, performing modulus reduction according to the damage factor;

[0039] S303. Update the finite element model stiffness matrix.

[0040] Compared with the prior art, the advantages and positive effects of the present invention are:

[0041] In the present invention, by fusing multi-source heterogeneous sensor data and executing the spatiotemporal alignment algorithm, a structural state tensor is constructed to realize the full-dimensional feature coupling of physical entities, the Laplace operation is combined with the stress wave propagation model to analyze the non-rigid deformation characteristics, and a load transfer path map is generated to reveal the implicit mechanical correlation. The Young's modulus is dynamically corrected based on the finite element grid to form a structural degradation coefficient matrix, the equipment degradation trend is captured by combining time series difference and long short-term memory network, and the spatiotemporal distribution probability of faults is quantified through the Kolmogorov test, so as to realize the leap from single parameter threshold warning to multi-physical field coupling prediction of power grid equipment, improve the spatiotemporal resolution and dynamic adaptability of fault prediction, reduce the dependence on manual experience, and enhance the system robustness under complex working conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is a flow chart of the power grid fault prediction system integrated with digital twins of the present invention;

[0043] Figure 2 This is a flow chart of the state perception module of the present invention;

[0044] Figure 3 This is the path deduction module process of the present invention;

[0045] Figure 4 This is a flow chart of the twin mapping module of the present invention;

[0046] Figure 5 This is a flow chart of the predictive execution module of the present invention. DETAILED DESCRIPTION

[0047] To make the purpose, technical solutions and advantages of the present invention clearer, the following is a detailed description of the technical solutions based on software implementation in conjunction with the system architecture diagram and embodiments. It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present invention and do not constitute a limitation on the scope of protection.

[0048] In the description of this invention, the system architecture relationships or data processing flows indicated by terms such as "layer," "module," "interface," "data flow," "client," and "server" are defined based on the architecture diagrams or flow charts corresponding to the embodiments. This expression is intended solely to clarify the logical relationships between the various elements of the technical solution and does not limit the physical deployment form. The term "plurality" encompasses two or more technical units, including but not limited to scalable elements such as multiple data nodes, processing threads, service instances, or functional components. The specific number will be determined based on the actual business scenario and requires special explanation.

[0049] See also Figure 1 and Figure 2 The present invention provides a technical solution: a power grid fault prediction system integrating digital twins includes:

[0050] The state perception module is used to obtain the strain gradient of the transmission tower steel structure through distributed fiber optic sensors, collect the displacement trajectory of the transmission tower base through the Beidou positioning unit, calculate the angular velocity based on the six-axis inertial unit, perform a spatiotemporal alignment algorithm on the multi-source data, combine them into a structural state tensor, and pass the structural state tensor to the path deduction module;

[0051] The structural state tensor includes strain gradient distribution, displacement trajectory coordinates, and angular velocity parameters;

[0052] The spatiotemporal alignment algorithm uses a dynamic time warping method to unify the sampling frequency of strain gradients, displacement trajectories, and angular velocity parameters to 100 Hz;

[0053] 100 Hz represents a sampling frequency of collecting 100 data points per second and is set according to the maximum sampling capacity of the distributed fiber optic sensor.

[0054] First, on a 220kV "ZBC-35" straight tower in Zhangjiakou, Hebei Province, distributed fiber optic sensors were placed vertically every 10 meters along the windward and leeward sides of its four main legs, starting from 5 meters at the tower foot, all the way to 55 meters at the top of the tower. At each placement point, four sensor units were evenly installed along the circumference of the pole, for a total of 100 units installed on the tower body. sensor units. When the timestamp is When the sensor at point A (located on the windward side) at a height of 15 meters above the tower collects a strain of 152.3 microstrain ( ), while the strain collected by the sensor at point B, which is 0.2 meters away from the axial direction of the rod, is 153.5 At this time, the strain gradient value at this position is The calculation results in 6.0 This calculation is applied simultaneously to all adjacent pairs of sensing units, forming a real-time strain gradient distribution matrix covering the entire tower. The raw data acquisition frequency of this distributed fiber optic sensor is 50Hz.

[0055] At the same time, a monitoring terminal integrating a BeiDou high-precision positioning module and a six-axis inertial measurement unit (IMU) is installed at the center of the tower top. The BeiDou positioning unit continuously collects the three-dimensional geographic coordinates of the tower top at a frequency of 10Hz. For example, At , the coordinates collected were (114.88251° east longitude, 40.75162° north latitude, 1535.62 meters above sea level). At 11:00 PM, the coordinates were (114.88252°E, 40.75161°N, 1535.61 meters above sea level). This series of consecutive coordinate points constitutes the displacement trajectory of the tower top.

[0056] The six-axis IMU collects the three-axis acceleration and three-axis angular velocity of the tower top at a frequency of 200Hz. The gyroscope inside the IMU directly outputs the rotation rate around its internal x, y, and z axes. When the wind load is applied, the output angular velocity vector is (0.03rad / s, -0.01rad / s, 0.08rad / s). This data reflects the real-time torsion and swing rate of the transmission tower under wind load.

[0057] After acquiring the three heterogeneous data sets of strain gradient, displacement trajectory, and angular velocity, a spatiotemporal alignment process was performed to unify the sampling frequency of all data streams to 100 Hz. This frequency was set based on the following experiment: a Q235 steel component, made of the same material as a transmission tower, was mounted on a vibration table with a distributed fiber optic sensor attached to it. A sinusoidal sweep excitation from 1 Hz to 150 Hz was applied to the vibration table, while a laser Doppler vibrometer was used as a true reference. The response signals from the fiber optic sensor were recorded and compared with the vibrometer signals.

[0058] Table 1: Fiber optic sensor sampling frequency verification experiment data table

[0059]

[0060] As shown in Table 1, experimental data shows that when the excitation frequency exceeds 100Hz, the signal-to-noise ratio of the sensor signal is less than 40dB, and the signal distortion increases sharply. Therefore, 100Hz is set as the maximum effective sampling frequency. The alignment process is as follows: establish a reference time axis with an interval of 0.01 seconds (i.e., 100Hz). For strain data with an original frequency of 50Hz, a linear interpolation method is used between two original data points to calculate the new data at the reference time point. For example, The strain is 152.3 , The strain is 152.9 , then interpolation is calculated The strain is For the 10Hz displacement data, the same method is used to interpolate to generate 100Hz data points. For the 200Hz angular velocity data, at each 100Hz reference time point, take the two original data points before and after it (for example, take and ), downsample to 100Hz by taking the arithmetic mean. For example, The angular velocity is 0.08rad / s, The angular velocity is 0.09 rad / s, then The angular velocity is .

[0061] Finally, at each 0.01 second timestamp, the aligned strain gradient distribution matrix, a three-dimensional displacement trajectory coordinate vector, and a three-dimensional angular velocity parameter vector are combined into a structural state tensor. An example of a data slice at this time is: {strain gradient matrix [96x1], displacement coordinates (114.882515°, 40.751615°, 1535.615m), angular velocity (0.035rad / s, -0.012rad / s, 0.085rad / s)}. This structural state tensor is passed as a whole to the path deduction module.

[0062] See also Figure 1 and Figure 3 , the path deduction module is used to receive the structural state tensor, perform Laplace operation on the strain gradient, input the displacement trajectory coordinates into the three-dimensional velocity field correction model, and construct the stress wave propagation path in combination with the angular velocity parameters, and solve it by the finite difference method The wave equation generates a load transfer path map;

[0063] in, Indicates the spatial position of the transmission tower steel structure Department The displacement vector at time and in meters, Represents a time variable and its unit is seconds. It represents the corrected longitudinal wave propagation velocity and the unit is m / s, represents the Laplace operator and has dimension m -2 , Represents three-dimensional space coordinates and the unit is meter, Represents the external excitation force function and the unit is N / m 3 ;

[0064] Extract non-rigid deformation components, generate load transfer path maps, and transfer the load transfer path maps to the twin mapping module;

[0065] The load transfer path map specifically includes the stress wave attenuation path corrected by the displacement trajectory, the non-rigid deformation area identified based on the angular velocity parameters, and the main load conduction direction obtained by strain gradient analysis;

[0066] The stress wave propagation model uses the formula: ;

[0067] in, It represents the longitudinal wave propagation speed of stress wave in the steel structure of transmission tower and its unit is m / s. It represents the benchmark Young's modulus of the steel structure measured by the distributed optical fiber sensor and the unit is GPa. The dimensionless parameter representing the Poisson's ratio of the transmission tower steel, Indicates the density of steel and the unit is kg / m 3 ;

[0068] The stress wave propagation model innovatively introduces a displacement trajectory feedback mechanism to correct velocity field parameters in real time using Beidou positioning data. The specific implementation is as follows: in is the displacement influence factor, is the base displacement, is the maximum allowable displacement threshold;

[0069] When calculating the stress wave propagation velocity, the strain gradient distribution is used as the initial input parameter, and the velocity field is corrected in combination with the displacement trajectory coordinates;

[0070] When extracting non-rigid deformation components, set the curvature change threshold to 0.15rad / m 2 ;

[0071] 0.15rad / m 2 It represents the critical judgment value of the surface curvature change of the transmission tower structure and is obtained through the Q235 steel yield test calibration.

[0072] After receiving the structural state tensor from the state perception module, the path deduction module first performs a Laplace operation on the strain gradient distribution data. The specific execution method of this operation on the discrete data field is as follows: for a certain sensor unit on the tower structure, , the strain gradient is , the strain gradient values ​​of the four adjacent sensing units in the structure are , , , , the physical distance between adjacent units is .but The Laplace value of a point is Taking a node at 15 meters of the tower as an example, the strain gradient value is 6.0 The gradient values ​​of its four adjacent nodes above, below, left, and right are 5.8, 6.4, 6.1, and 5.9 respectively. , the adjacent physical distance is 0.2m, then the Laplace value of the node is This operation traverses all nodes and generates a Laplace value distribution graph, the magnitude of which reflects the local curvature of the strain field, and areas with higher values ​​are marked as stress concentration areas.

[0073] Subsequently, the above calculation results are input into the stress wave propagation model to calculate the propagation speed of the stress wave in the tower structure. The stress wave propagation model uses the formula: This formula is used to calculate the propagation speed of longitudinal waves (P waves) in an ideal elastic medium. The logic of the formula is that the wave speed is determined by the elasticity (ability to resist deformation) and inertia (density) of the material. It represents the longitudinal wave propagation speed of stress wave in the steel structure of transmission tower, in m / s; is the benchmark Young's modulus of the steel structure measured by the distributed optical fiber sensor, in GPa; is the Poisson's ratio of the transmission tower steel, which is a dimensionless parameter; is the density of steel, in kg / m 3 . Numerator Represents the longitudinal stiffness of the material under constraint conditions, the denominator The square root operation converts the ratio of stiffness to density into velocity dimension. The formula is useful in that it uses the benchmark Young's modulus obtained by field measurement. The nominal Young's modulus of the replacement material at the factory makes the calculation results closer to the actual service status of the transmission tower.

[0074] The parameter assignment process in the formula is as follows: (Steel density): Directly adopt the national standard density value of Q235 steel used in transmission towers, =7850kg / m 3 . (Poisson's ratio): Obtained from the Q235 steel manual, dimensionless parameter =0.28. (Benchmark Young's modulus): Calibrate during the initial deployment of the system. Use a hammer to apply a transient impact load of 10kN for 1ms on a main material at the tower foot, 5.0 meters away from the fiber optic sensor point A. The fiber optic sensors at point A and point B, 10.0 meters away from point A, record the exact time when the stress wave peak arrives. If the time recorded at point A is , the time recorded at point B is , then the measured wave velocity is .Will Substitute into the formula and calculate back The Young's modulus is 198.5 GPa, which is used as the benchmark for subsequent calculations.

[0075] Substitute the above parameters into the formula for calculation: The results show that the theoretical propagation speed of stress waves in the current structural state is 5685m / s.

[0076] Next, the velocity field is corrected based on the displacement trajectory coordinates. arrive During the time period, the vertical displacement of the tower top is -0.01m, so the vertical vibration speed of the tower top during this time period is (The negative sign indicates downward motion.) When the stress wave propagates downward from the top of the tower, its corrected velocity is .

[0077] When extracting non-rigid deformation components, it is necessary to set a curvature change threshold. The threshold is 0.15rad / m 2 The following experimental calibration was performed: 20 Q235 steel standard specimens were selected and subjected to uniaxial tension using an MTS universal material testing machine according to the GB / T228.1-2021 standard. During the test, three-dimensional digital image correlation (3D-DIC) technology was used to monitor and calculate the curvature changes of the specimen surface in real time. The critical curvature value of each specimen at the moment it entered the plastic yield stage from the elastic stage was recorded. The average of the 20 sets of data after removing the highest and lowest values ​​was obtained to obtain 0.151 rad / m 2 , with a standard deviation of 0.009 rad / m 2 Therefore, the engineering threshold is set to 0.15rad / m 2In the system, the stress concentration area value obtained by the Laplace operation is converted into the equivalent surface curvature. 2 The areas are judged to have undergone non-rigid deformation.

[0078] Finally, the above analysis results are combined: the difference between the corrected wave velocity and the measured wave velocity is used to map the attenuation path of the stress wave in the tower structure; all non-rigid deformation areas with excessive curvature are marked in the 3D model; and the transmission direction of the main load is determined based on the path with the highest strain gradient and the lowest attenuation. These three together form a load transfer path map, which is passed to the twin mapping module.

[0079] See also Figure 1 and Figure 4 ,Twin mapping module is used to receive the load transfer path map, establish a finite element mesh model, input stress wave parameters into the damage accumulation model, modify the mesh Young's modulus, generate a structural degradation coefficient matrix, and transfer the structural degradation coefficient matrix to the prediction execution module;

[0080] The structural degradation coefficient matrix specifically refers to material fatigue, modulus correction value, and grid unit degradation index;

[0081] The damage accumulation model uses: ;

[0082] in, represents the modified mesh element Young's modulus and is in GPa. Indicates the initial Young's modulus of the steel material as specified by the factory and the unit is GPa. represents the dimensionless damage factor generated by the qth load cycle, is the load cycle number, represents the total number of load cycles, pass Calculated, It represents the amplitude attenuation rate of the stress wave on the propagation path in the qth cycle and is calculated as (peak amplitude - trough amplitude) / peak amplitude. It represents the energy dissipation rate in the main load conduction direction during the qth cycle and is calculated as 1-(output energy / input energy);

[0083] The unit size of the finite element mesh model is adaptively adjusted according to the main load transmission direction;

[0084] A 5mm×5mm grid is used in the load conduction dense area, and a 10mm×10mm grid is used in the conduction sparse area;

[0085] 5mm represents the side length of the square grid unit and is applicable to areas where the stress value in the main load transmission direction exceeds 50MPa; 10mm represents the side length of the square grid unit and is applicable to areas where the stress value in the main load transmission direction is less than 50MPa;

[0086] The update period of the structural degradation coefficient matrix is ​​set to 30 minutes;

[0087] When the grid cell degradation index exceeds 0.7, the real-time warning mechanism of the prediction execution module is triggered;

[0088] 30 minutes represents the minimum time interval for the system to update the structural degradation status, and 0.7 represents the critical threshold of the grid unit safety factor and is determined based on the failure critical value of the 85% confidence interval in the tensile test of metal materials in GB / T228.1-2021.

[0089] After receiving the load transfer path map, the twin mapping module immediately starts building the finite element mesh model. This process is adaptive: first, the base 3D CAD model of the transmission tower is retrieved, and then the mesh is divided according to the load transfer path map. For the areas marked as the main load transmission direction in the map, such as the windward side of the tower leg and the key cross-arm connection, the stress value converted from the strain monitoring value (for example, through stress = strain) is calculated. , If the stress is below 50 MPa, a 10 mm × 10 mm mesh size is used. If the stress at another location is 65 MPa, exceeding the 50 MPa threshold, the mesh in that area is automatically refined, using a 5 mm × 5 mm element size. The 50 MPa threshold is based on the fatigue SN curve for Q235 steel. This value lies in the transition zone between high-cycle fatigue and low-cycle fatigue. Stress cycles below this value generally do not result in significant damage accumulation in the short term, while stresses above this value require more sophisticated calculations to capture their damage effects.

[0090] Next, the stress wave parameters are input into the damage accumulation model to modify the Young's modulus of the mesh element. The damage accumulation model is: The logic of this formula is to dynamically adjust the mechanical properties of materials by simulating the fatigue damage process of materials under cyclic loads. It represents the modified Young's modulus of the mesh element, in GPa; It is the initial Young's modulus of the steel as specified by the factory, in GPa; is the dimensionless damage factor generated by the qth load cycle, q is the load cycle number, and n is the total number of load cycles. Represents the total damage caused by all load cycles within a period of time (30 minutes in this example). This factor directly reflects the degree of degradation of material properties. By multiplying this degradation factor with the initial modulus, we can get the corrected Young's modulus that is closer to the actual state. Damage Factor pass Calculated, where represents the amplitude attenuation rate of the stress wave on the propagation path in the qth cycle, It represents the energy dissipation rate in the direction of the main load conduction in the qth cycle. The benefit of the formula is that it will damage the factor Two physical quantities that can be directly monitored: stress wave amplitude attenuation rate and energy dissipation rate Correlation allows damage assessment to be driven directly by real-time sensor data.

[0091] The parameter assignment and calculation process in the formula are as follows: (Initial Young's modulus): The nominal value of Q235 steel on the factory certificate is set to 206GPa. In a 30-minute update cycle, the system uses the rain flow counting method to identify the strain data of a key node. A complete load-unload cycle. Take a loop as an example to calculate: (Amplitude decay rate): In this cycle, the stress wave propagates from upstream point A to downstream point B. The peak value of the stress wave monitored at point A is 70 MPa and the trough value is -60 MPa; the peak value monitored at point B is 65 MPa and the trough value is -55 MPa. The calculation calls the peak and trough amplitudes of point A, =(70MPa-65MPa) / 70MPa=0.0714. (Energy dissipation rate): The stress wave energy is proportional to the square of the stress amplitude. The input energy is proportional to , the output energy is proportional to . = . (Damage factor for a single load cycle): Assuming that in 150 cycles, the calculated damage factor The average value of is 0.00009, so the total damage accumulation . (Corrected Young's modulus): The results show that after 30 minutes of loading cycle, the Young's modulus of the mesh element has dropped from 206 GPa to 203.22 GPa.

[0092] The above calculations are applied to each mesh element in the finite element model. Subsequently, the system generates a structural degradation coefficient matrix, which is updated every 30 minutes. For the mesh element calculated above, the corresponding degradation coefficient entry is: {Material fatigue: 0.0135, Modulus correction value: 203.22GPa, Mesh element degradation index: 0.0135}. The mesh element degradation index is given by Calculated.

[0093] The system will continuously monitor the degradation index of all grid cells. When the degradation index of any cell exceeds 0.7, an alert will be triggered. The 0.7 threshold is determined based on the following experimental analysis:

[0094] Table 2: Correlation between degradation index and failure state of Q235 steel tensile test

[0095]

[0096] As shown in Table 2, Q235 steel was subjected to a tensile test in accordance with the GB / T228.1-2021 standard, and the ratio of its tangent modulus to the initial modulus was calculated in real time to obtain the degradation index. The experimental results show that when the degradation index reaches 0.7, the material has entered the necking stage, and its bearing capacity is about to or has begun to decline. At this time, the lower limit of the 85% confidence interval of the structural failure probability has exceeded the critical value. Therefore, 0.7 is set as the critical threshold for early warning. In this case, the calculated degradation index of 0.0135 is much lower than 0.7. The structural degradation coefficient matrix is ​​passed as a whole to the prediction execution module.

[0097] See also Figure 1 and Figure 5 ,The prediction execution module is used to receive the structural degradation coefficient matrix, perform time series difference on the degradation coefficient, input the sequence into the long short-term memory network, perform Kolmogorov test, generate fault time and space coordinate coding, and enter the fault time and space coordinate coding into the power grid dispatch decision;

[0098] The fault spatiotemporal coordinate encoding includes the prediction period sequence, confidence interval threshold, and fault geographic coordinates;

[0099] Long short-term memory networks include = 128 hidden units, whose gating mechanism uses sigmoid activation function;

[0100] The significance level of the Kolmogorov test was set at 0.05 to verify the distribution consistency between the predicted period sequence and the historical fault data in the fault spatiotemporal coordinate encoding;

[0101] represents the number of neurons in the hidden layer and is optimized on the training set by the gradient descent method. 0.05 represents the upper limit of the probability of the first type of error allowed in the statistical test.

[0102] After receiving the structural degradation coefficient matrix, which is updated every 30 minutes, the prediction execution module performs a time-series difference operation on the key indicator, the "grid cell degradation index." This operation aims to shift the focus from the absolute value of the degradation state to the rate and acceleration of its change. For example, for a grid cell in a high-stress area, its degradation index for three consecutive cycles is recorded as: , , The time series difference calculation process is to calculate the difference between adjacent cycles: Difference 1 = 0.0135 - 0.0130 = 0.0005; Difference 2 = 0.0142 - 0.0135 = 0.0007. This results in a time series representing the degradation rate [0.0005, 0.0007].

[0103] This differenced sequence is input into a pre-trained long short-term memory network (LSTM). The network structure contains 128 hidden units ( ). Number of hidden units The determination process was as follows: During the model development phase, historical monitoring data was used as the training set to construct LSTM models with 32, 64, 128, and 256 hidden units, respectively. These models were evaluated on an independent validation set, using the root mean square error (RMSE) of the prediction error. Testing found that the 128-hidden-unit model achieved an RMSE of 0.00008 on the validation set, while the 64-unit model achieved an RMSE of 0.00011 and the 256-unit model achieved an RMSE of 0.000085. However, this significantly increased training time. Therefore, 128 units was chosen as a balance between accuracy and efficiency. The network's gating mechanisms, such as the input, forget, and output gates, all utilize the sigmoid activation function, which maps internally computed values ​​to a range between 0 and 1.

[0104] Currently, the sequence [0.0005, 0.0007] is fed into the LSTM network. Based on the historical fault evolution patterns it has learned, the network predicts the future degradation rate. The network outputs the predicted rates for the next three cycles (90 minutes) as [0.0010, 0.0015, 0.0022]. By summing these predicted rates, the future degradation index can be calculated: ; ; The sequence [0.0152, 0.0167, 0.0189] is the “forecast period sequence”.

[0105] Subsequently, the Kolmogorov-Smirnov test is performed on the predicted results to verify their distribution consistency with the actual failure mode. Set to 0.05. This value is a common standard in statistics and engineering practice, meaning that there is a 5% probability of misjudging a prediction that is actually consistent with the historical failure mode as inconsistent (Type I error). When performing the test, the prediction cycle sequence output by LSTM is used as sample one, and the degradation evolution data of 10 cases similar to the current working conditions (such as wind speed and temperature) that eventually led to failure are retrieved from the historical database as sample two. The test calculates a p-value. If the p-value is 0.42, due to , then the null hypothesis is accepted, that is, it is believed that there is no significant difference in statistical distribution between the current predicted degradation trend and the actual fault evolution trend in history.

[0106] Finally, the system integrates this information to generate a fault spatiotemporal coordinate code. This code is a structured data object containing the following: {Prediction period sequence: [0.0152, 0.0167, 0.0189], Confidence interval threshold: 0.95, Fault geographic coordinates: {Tower number: 'ZBC-35', Structural location: 'South main leg, height 25m', Geographic coordinates: (114.88251° East, 40.75162° North)}}. This code is immediately generated and entered into the decision support system of the power grid dispatch center using a standard data interface format.

[0107] The power grid fault prediction method integrating digital twins includes the following steps:

[0108] S1: The strain gradient of the transmission tower steel structure is acquired through distributed fiber optic sensors. The displacement trajectory of the transmission tower base is collected through the Beidou positioning unit. The angular velocity is calculated based on the six-axis inertial unit. The spatiotemporal alignment algorithm is performed on the multi-source data to combine them into a structural state tensor.

[0109] S2: Perform Laplace operation on the strain gradient in the structural state tensor and input the result into the stress wave propagation model to extract the non-rigid deformation components and generate the load transfer path map;

[0110] Establishing a stress wave propagation model includes: S201, calculating the reference propagation velocity based on material property parameters ;

[0111] S202, constructing a three-dimensional velocity field distribution based on the displacement trajectory coordinates;

[0112] S203, using the finite difference method to solve the stress wave propagation partial differential equation;

[0113] S3: Establish a finite element mesh model based on the load transfer path map, input the stress wave parameters into the damage accumulation model, modify the mesh Young's modulus, and generate the structural degradation coefficient matrix;

[0114] Young's modulus correction includes: S301, calculation of the cumulative damage factor of each grid element ;

[0115] S302, performing modulus reduction according to the damage factor;

[0116] S303, updating the finite element model stiffness matrix;

[0117] S4: Perform temporal difference on the degradation coefficients in the structural degradation coefficient matrix, input the sequence into the long short-term memory network, perform the Kolmogorov test, generate the fault spatiotemporal coordinate encoding and enter it into the grid dispatch decision.

[0118] The above examples illustrate preferred implementations of the present invention. Any equivalent adjustments to the technical solutions based on software engineering methods are within the scope of protection, including but not limited to: implementing algorithmic logic using different programming languages, service-oriented reconfiguration of functional modules, adjusting data interaction protocols, optimizing resource scheduling strategies, and other technical improvements. Any implementation scheme derived from reasonable modifications to the data processing flow, service call chain, or system architecture level that does not depart from the core technology of the present invention shall be deemed to be within the scope of protection defined by the claims of the present invention.

Claims

1. A power grid fault prediction system integrating digital twins, characterized by: The system comprises: The state perception module is used to obtain the strain gradient of the transmission tower steel structure through distributed fiber optic sensors, collect the displacement trajectory of the transmission tower base through the Beidou positioning unit, calculate the angular velocity based on the six-axis inertial unit, perform a spatiotemporal alignment algorithm on the multi-source data, combine them into a structural state tensor, and pass the structural state tensor to the path deduction module; a path deduction module, configured to receive the structural state tensor, perform Laplace operation on the strain gradient, input the result into a stress wave propagation model, extract non-rigid deformation components, generate a load transfer path map, and transmit the load transfer path map to a twin mapping module; a twin mapping module, configured to receive the load transfer path map, establish a finite element mesh model, input stress wave parameters into a damage accumulation model, modify the mesh Young's modulus, generate a structural degradation coefficient matrix, and transmit the structural degradation coefficient matrix to a prediction execution module; a prediction execution module, configured to receive the structural degradation coefficient matrix, perform time series difference on the degradation coefficients, input the sequence into a long short-term memory network, perform a Kolmogorov test, generate a fault spatiotemporal coordinate code, and enter the fault spatiotemporal coordinate code into a power grid dispatch decision; The structural state tensor includes strain gradient distribution, displacement trajectory coordinates, and angular velocity parameters. The load transfer path map specifically includes the stress wave attenuation path corrected by the displacement trajectory, the non-rigid deformation area identified based on the angular velocity parameters, and the main load conduction direction obtained through strain gradient analysis. The structural degradation coefficient matrix specifically refers to material fatigue, modulus correction value, and grid unit degradation index. The fault spatiotemporal coordinate encoding includes the prediction period sequence, confidence interval threshold, and fault geographic coordinates.

2. The power grid fault prediction system integrating digital twin according to claim 1 is characterized in that: When the path deduction module executes the stress wave propagation model, the formula is used: ； in, It represents the longitudinal wave propagation speed of stress wave in the steel structure of transmission tower and its unit is m / s. It represents the benchmark Young's modulus of the steel structure measured by the distributed optical fiber sensor and the unit is GPa. The dimensionless parameter representing the Poisson's ratio of the transmission tower steel, Indicates the density of steel and the unit is kg / m 3 ; When calculating the stress wave propagation velocity, the strain gradient distribution is used as an initial input parameter, and the velocity field is corrected in combination with the displacement trajectory coordinates.

3. The power grid fault prediction system integrating digital twin according to claim 2 is characterized in that: The twin mapping module uses the damage accumulation model to correct the mesh Young's modulus: ； in, represents the modified mesh element Young's modulus and is in GPa. Indicates the initial Young's modulus of the steel material as specified by the factory and the unit is GPa. represents the dimensionless damage factor generated by the qth load cycle, is the load cycle number, represents the total number of load cycles, pass Calculated, It represents the amplitude attenuation rate of the stress wave on the propagation path in the qth cycle and is calculated as (peak amplitude - trough amplitude) / peak amplitude. It represents the energy dissipation rate in the main load conduction direction during the qth cycle and is calculated as 1-(output energy / input energy).

4. The power grid fault prediction system integrating digital twins according to claim 3 is characterized in that: The long short-term memory network in the predictive execution module includes = 128 hidden units, whose gating mechanism uses sigmoid activation function; The significance level of the Kolmogorov test is set at 0.05 to verify the distribution consistency between the predicted period sequence in the fault spatiotemporal coordinate encoding and the historical fault data; in, represents the number of neurons in the hidden layer and is optimized on the training set by the gradient descent method. 0.05 represents the upper limit of the probability of the first type of error allowed in the statistical test.

5. The power grid fault prediction system integrating digital twin according to claim 4 is characterized in that: The spatiotemporal alignment algorithm uses a dynamic time warping method to unify the sampling frequency of the strain gradient, displacement trajectory, and angular velocity parameters to 100 Hz; When extracting the non-rigid deformation component, the curvature change threshold is set to 0.15 rad / m 2 ; Among them, 100 Hz represents the sampling frequency of collecting 100 data points per second and is set according to the maximum sampling capacity of the distributed optical fiber sensor, 0.15 rad / m 2 It represents the critical judgment value of the surface curvature change of the transmission tower structure and is obtained through the Q235 steel yield test calibration.

6. The power grid fault prediction system integrating digital twins according to claim 5 is characterized in that: The unit size of the finite element mesh model is adaptively adjusted according to the main load transmission direction; A 5mm×5mm grid is used in the load conduction dense area, and a 10mm×10mm grid is used in the conduction sparse area; Among them, 5mm represents the side length of the square grid unit and is applicable to areas where the stress value in the main load transmission direction exceeds 50MPa; 10mm represents the side length of the square grid unit and is applicable to areas where the stress value in the main load transmission direction is less than 50MPa.

7. The power grid fault prediction system integrating digital twin according to claim 6 is characterized in that: The update period of the structural degradation coefficient matrix is ​​set to 30 minutes; When the grid unit degradation index exceeds 0.7, the real-time warning mechanism of the prediction execution module is triggered; Among them, 30 minutes represents the minimum time interval for the system to update the structural degradation state, 0.7 represents the critical threshold of the grid unit safety factor and is determined according to the failure critical value of the 85% confidence interval in the tensile test of metal materials in GB / T228.1-2021.

8. A power grid fault prediction method integrating digital twins, characterized in that: The method is used to implement the power grid fault prediction system integrating digital twins according to any one of claims 1 to 7, comprising the following steps: S1: The strain gradient of the transmission tower steel structure is acquired through distributed fiber optic sensors. The displacement trajectory of the transmission tower base is collected through the Beidou positioning unit. The angular velocity is calculated based on the six-axis inertial unit. The spatiotemporal alignment algorithm is performed on the multi-source data to combine them into a structural state tensor. S2: performing a Laplace operation on the strain gradient in the structural state tensor, inputting the result into a stress wave propagation model, extracting non-rigid deformation components, and generating a load transfer path map; S3: establishing a finite element mesh model based on the load transfer path map, inputting stress wave parameters into a damage accumulation model, modifying the mesh Young's modulus, and generating a structural degradation coefficient matrix; S4: Performing time series difference on the degradation coefficients in the structural degradation coefficient matrix, inputting the sequence into a long short-term memory network, performing a Kolmogorov test, generating a fault spatiotemporal coordinate code and entering it into a power grid dispatch decision.

9. The method for predicting power grid faults integrating digital twins according to claim 8, characterized in that: The establishment of the stress wave propagation model in S2 includes: S201, calculating the reference propagation velocity according to the material property parameters ; S202, constructing a three-dimensional velocity field distribution based on the displacement trajectory coordinates; S203, using the finite difference method to solve the stress wave propagation partial differential equation; The Young's modulus correction in S3 includes: S301, calculating the cumulative damage factor of each grid unit ; S302, performing modulus reduction according to the damage factor; S303. Update the finite element model stiffness matrix.

Images