Power cable joint defect detection method and system based on digital twinborn model

Abstract

The invention provides a power cable joint defect detection method and system based on a digital twin model. The method comprises the following steps: acquiring performance parameters, temperature data and ultrasonic parameters of a cable joint; constructing a cable joint multi-field coupling simulation model according to the physical size material parameters of the cable joint; constructing an objective function according to the performance parameters and the temperature data, interacting data of the multi-field coupling simulation model with data under actual operation conditions through the objective function, and updating a pre-constructed cable joint digital twinborn simulation model to obtain a real-time updated cable joint digital twinborn simulation model; and training and testing the radial basis function neural network based on the output data of the cable joint digital twin simulation model updated in real time to obtain a fault diagnosis model of cable joint temperature and ultrasonic data based on deep learning, and detecting the real-time operation state of the cable joint according to the fault diagnosis model. And predicting a potential fault of the cable joint.

Description

Technical Field

[0001] The present invention relates to the technical field of cable operation and maintenance, and in particular to a method and system for detecting defects in power cable joints based on a digital twin model. Background Art

[0002] In the power system, the joints of high-voltage transmission cables are key components that connect cable segments, and their reliability directly affects the stability and safety of the entire power grid. Cable joints are prone to failures due to their high-voltage environment and complex geographical layout, such as insulation aging and poor contact, which may lead to serious power accidents. Therefore, accurate monitoring and evaluation of the status of power cable joints has become an important means to improve power grid reliability and prevent power accidents. Summary of the invention

[0003] In view of the above technical problems, the present invention provides a power cable joint defect detection method based on a digital twin model, comprising:

[0004] Obtain the performance parameters of cable connectors under various actual operating conditions;

[0005] Acquiring temperature data and ultrasonic information of the cable joint in operation in real time through a sensor network;

[0006] According to the physical size and material parameters of the cable joint entity, a multi-field coupling simulation model of the cable joint is constructed;

[0007] Constructing an objective function according to the performance parameters and temperature data, interacting the data of the cable joint multi-field coupling simulation model with the data under actual operating conditions through the objective function, updating the pre-constructed cable joint digital twin simulation model, and obtaining a real-time updated cable joint digital twin simulation model;

[0008] Based on the output data of the real-time updated digital twin simulation model of the cable joint, the radial basis function neural network is trained and tested to obtain a fault diagnosis model of the cable joint temperature and ultrasonic data based on deep learning. The real-time operating status of the cable joint is detected according to the fault diagnosis model.

[0009] Furthermore, the performance parameters of the cable connector under various actual operating conditions are obtained, including:

[0010] Build a cable physical test platform for the digital twin system of high-voltage cable joints;

[0011] The performance of the cable joint under various operating conditions is simulated through the test platform to obtain the performance parameters of the cable joint.

[0012] Furthermore, the temperature data and ultrasonic information of the cable joint in operation are obtained in real time through the sensor network, including:

[0013] Construct a sensor network consisting of temperature sensing and ultrasonic sensing node arrays;

[0014] The sensor network comprehensively monitors various parts of the cable, and obtains temperature data and ultrasonic information of the cable joint in operation in real time.

[0015] Furthermore, the cable joint multi-field coupling simulation model also analyzes the impact of different operating conditions and fault defects on the operating status of the cable joint by adopting the control variable method.

[0016] Furthermore, the cable joint multi-field coupling simulation model also uses the control variable method to analyze the impact of different operating conditions and fault defects on the operating status of the cable joint, including:

[0017] Simulate and analyze the impact of different load currents on the temperature distribution of cable joints and changes in electromagnetic field distribution;

[0018] Simulate and analyze the impact of excessive contact resistance of cable connectors on temperature distribution and electromagnetic field distribution changes;

[0019] Simulate and analyze the impact of different dielectric constants of cable insulation materials on the temperature distribution of cable joints and changes in electromagnetic field distribution;

[0020] The waveform signal of ultrasonic waves in cable joints under different defect conditions is simulated and analyzed.

[0021] Furthermore, the physical dimensions and material parameters of the cable connector entity include:

[0022] Get the physical size of the cable connector entity;

[0023] Set electromagnetic field boundary conditions;

[0024] Set the boundary conditions of the cable joint temperature field;

[0025] Set the acoustic field boundary conditions for the cable joint.

[0026] Furthermore, the data of the cable joint multi-field coupling simulation model is interacted with the data under actual operating conditions through the objective function, and the pre-built cable joint digital twin simulation model is updated to obtain a real-time updated cable joint digital twin simulation model, including:

[0027] The least square method must be used to construct the fitness function of the finite element simulation model data and the actual data for data interaction;

[0028] Based on the adaptive particle swarm algorithm, the parameters of the cable joint digital twin simulation model are corrected, the cable joint finite element simulation model is updated in real time, and the real-time updated cable joint digital twin simulation model is obtained.

[0029] Furthermore, it also includes:

[0030] Based on the cable joint digital twin simulation model and the fault diagnosis model, the operating status of the cable joint is displayed in real time through a visualization platform.

[0031] The present invention also provides a power cable joint defect detection system based on a digital twin model, which is characterized by comprising:

[0032] Performance parameter acquisition module, used to obtain the performance parameters of the cable connector under various actual operating conditions;

[0033] A temperature data and ultrasonic information acquisition module, used to acquire temperature data and ultrasonic information of the cable joint in operation in real time through a sensor network;

[0034] A multi-field coupling simulation model building module is used to build a multi-field coupling simulation model of a cable joint according to the physical size and material parameters of the cable joint entity;

[0035] A twin simulation model construction module is used to construct an objective function according to the performance parameters and temperature data, and to interact the data of the cable joint multi-field coupling simulation model with the data under actual operating conditions through the objective function, so as to update the pre-constructed cable joint digital twin simulation model and obtain a real-time updated cable joint digital twin simulation model;

[0036] The fault diagnosis module is used to train and test the radial basis function neural network based on the output data of the real-time updated digital twin simulation model of the cable joint, obtain a fault diagnosis model of the cable joint temperature and ultrasonic data based on deep learning, and detect the real-time operating status of the cable joint according to the fault diagnosis model.

[0037] Furthermore, it also includes:

[0038] The real-time display module is used to display the operating status of the cable joint in real time through a visualization platform based on the cable joint digital twin simulation model and the fault diagnosis model.

[0039] The present invention provides a method and system for detecting defects in power cable joints based on a digital twin model, which can monitor the temperature, ultrasonic and other operating parameters of the cable joints in real time, and predict the potential failures and performance degradation of the cable joints in the future through these data. This system can provide accurate maintenance and repair guidance for engineering and technical personnel, so as to prevent and deal with potential failures in advance and ensure the stable and reliable operation of the power transmission system. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a flow chart of a method for detecting defects in power cable joints based on a digital twin model provided by an embodiment of the present invention;

[0041] Figure 2 It is a basic flow chart of the implementation measure module provided by the embodiment of the present invention;

[0042] Figure 3 It is a cable joint physical test platform provided by an embodiment of the present invention;

[0043] Figure 4 It is a structural schematic diagram of a power cable joint defect detection system based on a digital twin model provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0044] Many specific details are set forth in the following description to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the connotation of the present invention, so the present invention is not limited to the specific implementation disclosed below.

[0045] Example 1

[0046] To achieve the above object, the present invention provides a power cable joint defect detection method based on a digital twin model, such as Figure 1 As shown, the method comprises the following steps:

[0047] Step S101, obtaining performance parameters of the cable connector under various actual operating conditions.

[0048] Build a cable entity test platform for the digital twin system of high-voltage cable joints, so that it can accurately reflect the performance of cable joints under actual operating conditions. Through the test platform, simulate the performance of cable joints under various operating conditions and obtain the performance parameters of cable joints. Provide necessary verification and calibration data for building the digital twin model of cable joints.

[0049] Step S102, obtaining temperature data and ultrasonic information of the cable joint in operation in real time through a sensor network.

[0050] Develop sensors for real-time monitoring of temperature data and ultrasonic information during the operation of cable joints, including temperature and ultrasonic sensor network devices that are adapted to the special operating environment of the cable (harsh on-site operating environment, complex electromagnetic environment interference, etc.) based on indicators such as accuracy, sensitivity, and measurement range; optimize the layout of the sensor network composed of temperature sensors and ultrasonic sensor node arrays to ensure that the sensor network nodes of the cable joint digital twin entity can fully cover all parts of the cable, including difficult-to-monitor joints and hidden parts, thereby enhancing the comprehensiveness and representativeness of the data, and transmitting and managing temperature, ultrasonic and other data collected from the sensor network that can reflect the actual operating status of the cable.

[0051] The layout optimization of the sensor network composed of temperature sensors and ultrasonic sensor node arrays is performed, and the sensor network layout of the cable joint digital twin system is optimized based on a genetic algorithm (GA), including:

[0052] Collect actual operating data of cable joints (temperature data and ultrasonic information);

[0053] Evaluate data to determine preliminary requirements for sensor placement;

[0054] Design preliminary sensor network layout;

[0055] Optimize network design and training cycles, adjust and optimize sensor locations to cover all critical areas;

[0056] Implement the sensor network and test its coverage, and adjust the sensor layout based on the test results to ensure optimal coverage and data collection quality.

[0057] The transmission and management of data collected from the sensor network includes:

[0058] Adopt efficient network transmission protocols and use 5G Internet of Things technology to upload data in real time and transmit data collected by sensor networks;

[0059] The collected data completes data governance consisting of three major parts: abnormal sensor device evaluation, data cleaning, and data selection, to construct a cable operation data set and provide supporting data for the cable joint digital twin model.

[0060] Step S103, constructing a cable joint multi-field coupling simulation model according to the physical dimensions and material parameters of the cable joint entity.

[0061] The physical dimensions and material parameters of the cable connector entity include:

[0062] Get the physical size of the cable connector entity;

[0063] Set the electromagnetic field boundary conditions of the cable joint;

[0064] Set the boundary conditions of the cable joint temperature field;

[0065] Set the acoustic field boundary conditions of the cable joint;

[0066] The constructed cable joint multi-physics field simulation model provides a virtual twin model for the cable digital twin system.

[0067] The cable joint multi-field coupling simulation model also uses the control variable method to analyze the impact of different operating conditions and fault defects on the operating status of the cable joint, including:

[0068] Simulate and analyze the impact of different load currents on the temperature distribution of cable joints and changes in electromagnetic field distribution;

[0069] Simulate and analyze the impact of excessive contact resistance of cable connectors on temperature distribution and electromagnetic field distribution changes;

[0070] Simulate and analyze the impact of different dielectric constants of cable insulation materials on the temperature distribution of cable joints and changes in electromagnetic field distribution;

[0071] The waveform signal of ultrasonic waves in cable joints under different defect conditions is simulated and analyzed.

[0072] The temperature and ultrasonic simulation data of cable joints under different operating conditions and fault defects are obtained by analyzing the controlled variable method, which provides training data samples for building a cable joint defect fault assessment model based on radial basis function neural network (RBFNN).

[0073] Step S104, constructing an objective function according to the performance parameters and temperature data, interacting the data of the cable joint multi-field coupling simulation model with the data under actual operating conditions through the objective function, updating the pre-constructed cable joint digital twin simulation model, and obtaining a real-time updated cable joint digital twin simulation model.

[0074] Based on the data interaction between the cable joint entity experimental sensor data and the simulation model data, a cable joint digital twin model update strategy is proposed, including:

[0075] It is proposed to construct the objective function based on the cable sensor monitoring temperature data, and use the least squares method to construct the fitness function of the finite element simulation model data and the actual data for data interaction;

[0076] Based on the adaptive particle swarm algorithm (APSO), the cable simulation model parameters are identified and corrected, the cable joint finite element simulation model is updated in real time, and the real-time updated cable joint digital twin simulation model is obtained.

[0077] Step S105, based on the output data of the real-time updated cable joint digital twin simulation model, the radial basis function neural network is trained and tested to obtain a fault diagnosis model of the cable joint temperature and ultrasonic data based on deep learning, and the real-time operating status of the cable joint is detected according to the fault diagnosis model.

[0078] Based on simulation data, the radial basis function neural network (RBFNN) is trained to build a deep learning fault diagnosis model for cable joint temperature and ultrasonic data, including:

[0079] Acquire and organize simulation data of cables with different defects;

[0080] Normalize the data and divide it into experimental set and test set;

[0081] Determine the RBFNN network framework and preliminarily build a radial basis function neural network (RBFNN) model;

[0082] Determine the optimal network structure and optimal number of cycle training;

[0083] Train a deep learning cable joint fault diagnosis model;

[0084] Use the test set to verify the accuracy and effectiveness of the network model;

[0085] Evaluate the real-time operating status of the cable joint according to the cable joint operating guidelines and fault diagnosis model.

[0086] Based on QT and GPU multi-threaded parallel processing technology, the GUI software interface design is completed, and a cable joint operation status visualization platform is developed to facilitate on-site application. Based on the cable joint digital twin simulation model and the fault diagnosis model, the operation status of the cable joint is displayed in real time through the visualization platform.

[0087] Example 2

[0088] In order to make the purpose, technical solutions and advantages of the present application more clear, the present invention is further described in detail.

[0089] like Figure 2 As shown, a cable joint defect detection method based on a digital twin model proposed in the present invention comprises the following steps:

[0090] Build a cable entity test platform for the digital twin system of high-voltage cable joints, such as Figure 3As shown. It can accurately reflect the performance of cable joints under actual operating conditions, study the simulation and test the performance of cable joints under various operating conditions, and provide necessary verification and calibration data for building the digital twin model of cable joints. Specifically, it includes: according to the standards and specifications of my country's power grid cable transmission lines, build a detailed model of cable joints, covering key parameters such as joint type, structural size, and materials used (such as insulation and shielding materials), and build a cable entity test platform for the digital twin system of high-voltage cable joints;

[0091] Simulate the performance of cable connectors under different voltage and current loads, as well as load fluctuations and other electrical performance test conditions through simulation equipment;

[0092] Simulate various environmental conditions that the cable connector may be exposed to, such as high and low temperatures, humidity changes, chemical corrosion or mechanical stress.

[0093] Develop sensors for real-time monitoring of temperature data and ultrasonic information during cable joint operation, including:

[0094] Select high-precision and high-sensitivity temperature and ultrasonic sensors to ensure accurate monitoring data even with slight parameter changes. The temperature sensor should be able to cover the range of -50℃ to 150℃, and the ultrasonic sensor should be able to adapt to the detection needs of different densities and materials.

[0095] At the same time, the sensor should be designed to withstand harsh environments, including being waterproof, dustproof, chemically resistant, and resistant to electromagnetic interference, to ensure stable operation in complex on-site operating environments.

[0096] Optimize the layout of the sensor network composed of temperature sensors and ultrasonic sensor node arrays to ensure that the sensor network nodes of the cable joint digital twin entity can fully cover all parts of the cable, including joints and hidden parts that are difficult to monitor, thereby enhancing the comprehensiveness and representativeness of the data, including:

[0097] The multi-source sensing network of a cable joint defect detection method based on a digital twin model proposed in the present invention is composed of a wireless sensor network (WSN) composed of temperature and ultrasonic sensors, which collects state quantities that can reflect the operation status of the cable system. These sensors collect data from multiple points of the cable system to monitor and analyze the operation status of the cable;

[0098] The temperature distribution of the cable joint is monitored by a sensor composed of a three-dimensional temperature measurement array. As shown in the figure, the three-dimensional temperature measurement array is used to measure the temperature at different positions of the cable. At the same time, the ultrasonic sensor is placed at the best detection position of the middle joint of the cable to achieve a comprehensive detection of internal defects of the cable joint.

[0099] Optimize the layout of the sensor network to ensure that the sensor network nodes can fully cover all parts to achieve the optimal layout distribution. Genetic Algorithm (GA) is used to optimize the layout of the sensor network (temperature, ultrasonic sensor) to optimize the coverage and improve data quality. First, the framework of the genetic algorithm is defined, including:

[0100] The input is the data collected by the sensor (including temperature information and ultrasonic data), the sensor configuration (including the location, type (temperature or ultrasonic), and quantity of the sensor);

[0101] The output is the optimized sensor layout (sensor location, quantity and type optimized according to the fitness function), coverage, and data quality.

[0102] Next is the implementation of the algorithm steps. First, define the objective function:

[0103] Objective Function=w1·Coverage(C)+w2·Data Quality(C)-w3·SensorCount(C)

[0104] Coverage (C) is the coverage rate calculated by the sensor network, that is, the proportion of the monitored area that is effectively covered.

[0105] Data Quality (C) evaluates the quality of the data collected by the sensor network, and Sensor Count (C) is the total number of sensors used in the sensor network. C represents the configuration parameters of the sensor layout. w1, w2, and w3 are weight factors used to balance the importance of these indicators.

[0106] The implementation of the genetic algorithm first encodes each sensor to represent the location, type and quantity of the sensor; then initializes the population and randomly generates a set of possible sensor location configuration schemes. After defining the fitness function (objective function), the algorithm performs selection, crossover, mutation and iteration operations, and finally selects the individual with the highest fitness from the final population as the optimal sensor layout scheme;

[0107] Genetic algorithms can effectively find the optimal sensor layout, which not only improves monitoring coverage and data quality, but also minimizes the number of sensors required.

[0108] Transmission and management collects temperature, ultrasonic and other data from the sensor network that can reflect the actual operating status of the cable, including:

[0109] The efficient network integrated transmission protocol Message Queuing Telemetry Transport (MQTT) is used, which is suitable for data transmission in the Internet of Things (IoT) environment. The data collected by the sensor network is uploaded and transmitted in real time through 5G IoT technology. The collected data is managed by three parts: abnormal sensor device evaluation, data cleaning and data selection, so as to build a cable operation data set and provide supporting data for the digital twin model of the cable joint, including:

[0110] First, the evaluation of abnormal sensor devices adopts a data processing method based on sliding window technology. By setting a fixed time window W and step size S in the continuous data stream, the data set in the window is periodically updated and the relevant statistical indicators are recalculated. The size of the window W determines the amount of data analyzed each time, and the step size S determines the frequency of window updates;

[0111] By using a sliding window to obtain an evaluation data set from the real-time monitoring data stream, the evaluation is conducted by considering abnormal values, consecutive identical values, data change rates, and overall and segmental coefficients of variation. The severity of the evaluation is adjusted by setting a tolerance. The effectiveness of the cable state quantity sensor device can be evaluated without exiting operation, and abnormal sensor devices can be discovered in a timely manner:

[0112] For monitoring of outliers, a statistical threshold method is used to identify data points that are beyond the normal range. Based on the detection of statistical thresholds (mean value plus or minus three times the standard deviation), the mean (μ) and standard deviation (σ) of the data set are used to define the outlier threshold. Usually, data points falling outside μ±3σ are identified as outliers:

[0113] Upper threshold>μ+3σ

[0114] Lower threshold<μ-3σ

[0115] For consecutive identical values, the number of consecutive occurrences of the same value in the window is analyzed. If the number of consecutive occurrences exceeds the preset threshold, it indicates that the sensor may fail.

[0116] For the abnormal detection of change rate (ChangeRate), the dynamic change of data can be evaluated by calculating the change rate between adjacent data points. The calculation formula of ChangeRate is:

[0117]

[0118] where x t and x t+1 are the values ​​of consecutive data points and Δt is the time interval between them.

[0119] When the absolute value of the change rate ChangeRate of adjacent data points exceeds twice the standard deviation (2σ) of the mean μ, that is, the absolute value of the change rate exceeds the range of μ±2σ, the change rate is considered to be abnormal.

[0120] The coefficient of variation (CV) is a statistic that measures the relative dispersion of a data set, and is defined as the ratio of the standard deviation σ of a data set to its mean μ. The calculation formula for CV is:

[0121]

[0122] Among them, μ is the mean and σ is the standard deviation.

[0123] When the CV value is low, it means that the data has a small dispersion and the change is relatively stable; when the CV value is high, it means that the data has a large dispersion and the fluctuation is relatively violent. If the CV value is greater than the threshold of 0.3, it can be considered that the data has a large variation, which may be a problem with data transmission, quality or sensor sensitivity.

[0124] The tolerance can be adjusted by adjusting the threshold or the sensitivity of the counter to adjust the strictness of the evaluation. The threshold of each feature can be adjusted according to historical data and the operating characteristics of the equipment, including the thresholds of outliers, continuous values, rates of change, and coefficients of variation. The lower the tolerance, the more sensitive the system is to anomalies.

[0125] The second step is data cleaning. Data cleaning is to monitor abnormal data in the data stream, including missing values, vacant values, out-of-range values, and singular values. Missing values ​​can be found by searching the timestamp, and vacant values ​​and out-of-range values ​​can be detected by setting simple judgment conditions. The Laida criterion can be used to detect abnormal values. Based on the standard deviation and mean, if a data point differs from the mean by more than 3 standard deviations, it can be regarded as a singular value:

[0126]

[0127] Where z is an intermediate calculation parameter. If z>k (usually k is 3), then x is a singular value, where μ is the mean, σ is the standard deviation, x is a single data point, and k is the sensitivity threshold.

[0128] After removing the outliers, there are null values ​​in the generated or monitored data. The Markov process is used to describe the state transition of the state quantity in the time dimension to complete the missing data. Using the known state transition probability matrix, the next state can be predicted, thereby estimating and completing the missing data points. Each state represents a possible measurement value. Regarding the state transition process, a state transition probability matrix is ​​constructed to learn the state transition frequency from historical data. First, a state transition probability matrix P is constructed:

[0129]

[0130] Where P ij Indicates the probability of transitioning from state i to state j. In the data completion process, the state transition process is used to describe the change process in the time series. The state transition process in both the forward and reverse directions is mined. The forward and reverse transition rules are integrated through dynamic assignment to achieve accurate completion of missing data. The forward and reverse state transition results are integrated, and the final completion value is determined using weighted average or other statistical methods:

[0131] x filled =ω×x forward +(1-ω)×x backward

[0132] where x filled is the completed data point, x forward and x backward are the predicted values ​​obtained from the forward and reverse Markov chain processes, respectively, and ω is the fusion weight, which is assigned according to the confidence of the prediction. Finally, according to the data quality and system feedback, the sensitivity threshold k and weight ω can be dynamically adjusted;

[0133] Then the data is smoothed and enveloped. The low-pass filter with a filter coefficient equal to the inverse of the span can achieve data smoothing. First, the size of the window N, that is, the strength of the filter, is determined. The larger the window, the more significant the smoothing effect, but important signal details may be lost. Calculate the moving average:

[0134]

[0135] where y t is the output value at time point t, x t-i is the value of the input sequence at time point ti;

[0136] Data Envelopment Processing, based on Hilbert transform (HT) transform, is used to extract analytical signals from raw data, thereby generating the envelope of the data and characterizing the overall trend. The analytical signal generated by the Hilbert transform contains the real and imaginary parts of the original signal (the result of the Hilbert transform). Calculate the Hilbert transform:

[0137]

[0138] in It is the result of Hilbert transform, that is, the orthogonal signal of the original signal x(t). The analytical signal is formed:

[0139]

[0140] Where i is the imaginary unit and z(t) is the analytical signal. Calculate the envelope:

[0141]

[0142] Where A(t) represents the signal envelope at time point t, which can effectively represent the overall amplitude trend of the signal.

[0143] Finally, data selection is performed. The nonlinear dynamic characteristics of the equipment state quantity are used to mine the time series change law of the equipment state quantity in the chaotic phase space to obtain the optimal data set for establishing the digital model of the cable joint. First, the state quantity time series data (temperature information, ultrasonic information) are collected from the sensor network to form the original state quantity data set {x i}, where i is the index of the time series;

[0144] Using the embedding theorem to transform time series data into trajectories in phase space involves selecting an embedding dimension m and a delay time τ to determine the dimension and time delay of the reconstructed phase space. The embedding dimension m can be determined by the pseudo nearest neighbor method, and the pseudo nearest neighbor dimension m can be determined according to the following formula:

[0145]

[0146] The delay time τ can be calculated by the mutual information method. The delay time τ calculated by the mutual information method can be determined according to the following formula:

[0147]

[0148] In the constructed phase space, the Lyapunov index, a nonlinear dynamics analysis tool, is used to measure the sensitivity of the system to the initial conditions. It is a key indicator for determining whether the system dynamics is chaotic. By calculating the Lyapunov index of the time series data, we can determine whether the system exhibits chaotic behavior. i} data, the Lyapunov exponent λ can be estimated using the following formula:

[0149]

[0150] Among them, δx(t) is the change of a small perturbation under the initial conditions after time t. This index represents the sensitivity of the system to the initial conditions. Then the Lyapunov exponent is determined. If λ<0, the system tends to be stable. The original data set {x i} is the optimal data set. If λ ≥ 0, the system may exhibit chaotic behavior. In this case, it is necessary to select an appropriate time window length C and calculate the time series data set {x i}, the optimal dataset length B is determined by the following formula:

[0151] B=2 m (m+1) m

[0152] Then evaluate whether the selected data set is sufficient, that is, determine whether the length B of the selected data set is greater than the required minimum number of data points N. If B>N, then the data set {x i} can build a cable digital twin model. If B≤N, the optimal data set is {x i}, where i=N-B+1,...,N.

[0153] The electric-thermal-acoustic multi-field coupling simulation model of the cable joint is constructed, and the control variable method is used to analyze the influence of different operating conditions and fault defects on the operating state of the cable joint, including:

[0154] Construct a 3D geometric model of the cable joint and an electric-thermal-acoustic multi-physics coupling simulation model based on the physical dimensions and material parameters of the cable joint entity;

[0155] Complete the material parameter setting in the simulation model according to the material parameters of the actual cable joint physical entity so that it has the same physical parameters as the cable joint physical entity;

[0156] Set the boundary conditions of the electro-magnetic field, set the electro-magnetic field with "current (ec)" and "magnetic field (mf)", set the excitation voltage of the cable with "potential", and set the grounding layer of the cable with "grounding". The main functional expressions of the electro-magnetic field are:

[0157]

[0158] Where J is the total current density, σ is the conductivity, E is the electric field intensity, and J e is the applied current density, Q V is the body charge density, H is the magnetic field intensity, B is the magnetic induction intensity, and A is the magnetic vector potential;

[0159] Set the boundary conditions of the temperature field, and set the initial temperature to 293.15K. Define the entire cable joint domain as a generalized heat source, which is generated by the electromagnetic loss in the insulating material; at the same time, the copper conductor connection point in the cable joint is also designated as a heat source. The equivalent conductivity of the crimping tube at the copper conductor connection at the joint is different from the conductivity of the copper conductor of the cable body. The temperature field mainly includes the function expression:

[0160]

[0161] Where ρ is the density of the material; C pis the constant pressure specific heat capacity of the material; u is the velocity vector. If the material itself is moving, a convection term is introduced; is the temperature gradient; q is the heat flux vector; Q is the heat generated inside the unit volume; Q ted The amount of heat added or removed from the outside per unit volume

[0162] Set the sound field boundary conditions. The sound field mainly includes the function expression:

[0163]

[0164] The above steps have built a finite element simulation model of the electric, thermal and acoustic fields of the cable joint. The control variable method is used to analyze the influence of different operating conditions and fault defects on the operating state of the cable joint. The specific simulation includes:

[0165] In actual operation, the cable will also cause differences in the distribution of electric field and temperature field under different load currents. The load current is set to the range of 800A-1200A, with a change gradient of 10A. The impact of different load currents on the temperature distribution of the cable joint and the change of the electromagnetic field distribution is simulated and analyzed.

[0166] Considering the situation where the cable joint contact resistance is too large, resulting in a cable joint fault defect, the equivalent conductivity of different cable conductors and crimping tube connections is set in the finite element simulation model to characterize this fault defect, and the influence of different conductivities at the joint connection on the temperature distribution of the cable joint and the change of the electromagnetic field distribution is simulated and analyzed;

[0167] In practical applications, the failure defects of cable joints due to different degrees of aging can be simulated by adjusting the dielectric constant of the cable insulation material in the finite element simulation model. Specifically, different insulation layer materials, such as polyethylene and semiconductor compounds, can be set, and their dielectric constant values ​​and combinations can be adjusted to represent different aging states. The simulation analysis explores the changes in the temperature distribution and electromagnetic field distribution of the cable joint under different dielectric constant conditions of the cable insulation material, so as to better understand the impact of aging on the performance of the cable joint;

[0168] Cable joints are subjected to long-term high-voltage operating conditions or incorrect installation processes, which may lead to defects of varying degrees inside the joints. When conducting ultrasonic simulation analysis of cable joints, by setting the defect sizes at different locations in the model, it is possible to explore in detail how ultrasonic signals propagate under various fault conditions. This involves accurately configuring various defects inside the finite element model of the cable joint, including cracks, holes, or inhomogeneities in dielectric materials. By setting these simulation conditions, ultrasonic waveform simulation signals under different defect configurations are obtained, and the waveform signals of ultrasonic waves in cable joints under different defect conditions are analyzed;

[0169] Finally, the temperature and ultrasonic simulation data of the cable joint under different operating conditions and fault defects are obtained according to the control variable method, providing training data samples for building a cable joint defect fault assessment model based on a radial basis function neural network (RBFNN) in step S7.

[0170] Based on the temperature data collected by the sensor network, an objective function is constructed, and data interaction between the cable joint simulation model data and the actual real-time operation data is performed. Based on the deep learning algorithm, the identification and correction of the model physical property parameters are realized, so that the cable joint digital twin simulation model is updated and corrected, and a real-time updated cable joint digital twin model is obtained. The cable joint digital twin model update strategy is proposed based on the data interaction between the cable joint entity experimental sensor data and the simulation model data, including:

[0171] The least squares method is used to construct the fitness function of the finite element simulation model temperature data and the actual temperature data for data interaction. The least squares method is used to optimize the model parameters so that the predicted output of the model is as close to the actual measured value as possible:

[0172]

[0173] where y i is the actual temperature measurement value, f(x i ,θ) is the model temperature prediction value, x i are the input data points and θ are the model parameters.

[0174] By comparing the temperature calculation value of the finite element simulation model with the actual temperature value, the accuracy of the simulation model prediction is evaluated. The next step is to apply the fitness function to help quantify the accuracy of the simulation model prediction and adjust the model to improve its prediction accuracy. The fitness function is used to evaluate the above-constructed objective function, that is, the degree of agreement between the cable simulation model prediction value and the actual data collected by the actual cable entity sensor network. The statistical method of R square value is used to quantify the accuracy of the model prediction:

[0175]

[0176] in It is the average value of actual measurement. According to the fitness function, when the data change exceeds the preset threshold, the model update is automatically triggered. The update judgment strategy aims to dynamically determine when to update the model based on the timeliness and accuracy of the data, define the maximum difference allowed between the model parameters and the actual data, and then set a detection mechanism to compare the model output with the real-time data. When the difference exceeds the preset threshold, it is determined that the model needs to be updated, that is, the determination of the update strategy and the setting of the model update conditions:

[0177]

[0178] where x model is the temperature prediction value of the cable simulation model, x real is the actual measured value, and threshold is the set difference threshold.

[0179] The trigger-based update mechanism algorithm sets up a mechanism to automatically trigger updates. The implementation process is to monitor the rate of change of key parameters and determine when to update model parameters through preset rules:

[0180]

[0181] in is the sensitivity of the model to parameter p, Δp is the parameter change, and ε is the sensitivity threshold that triggers the update;

[0182] Then, it is proposed to identify and correct the parameters of the cable simulation model based on the adaptive particle swarm algorithm (APSO). The equivalent conductivity of the conductor connection and the dielectric constant of the insulating material are selected as the parameters to be identified by the algorithm, and the finite element simulation model of the cable joint is updated in real time, so as to obtain a real-time updated cable digital twin model. The adaptive particle swarm optimization algorithm (APSO) is used in the parameter update of the cable simulation model to minimize the error between the real-time data and the model output. It is a population-based optimization technology that imitates the social behavior of bird flocks and is used to solve continuous optimization problems. In the scenario of cable simulation model parameter identification, APSO can help determine the optimal parameter set so that the output of the model is as close to the actual measured data as possible. The implementation points of the adaptive particle swarm algorithm (APSO) mainly include the following:

[0183] Initialization operation, each particle in the particle swarm represents a set of possible solutions, that is, a set of model parameters. And each particle has its own position (x i ) and speed (v i ), the position corresponds to the value of the model parameter, N particles are generated, and the initial position x of each particle i is i and speed v i Randomly generated, each particle represents a potential solution;

[0184] Iterative update, for each iteration and each particle i, calculate its fitness f(x i )(Fitness is usually the value of the objective function, which is the error between real-time data and model output, that is, the quality of the solution). Update the individual optimal p best,i and the global optimal g best , if the current fitness f(x i ) is better than the historical optimal fitness of particle i f(pbest,i ), then update p best,i Adaptively adjust the parameters according to the historical search information, and adaptively adjust the parameters in the speed update formula, such as the inertia weight ω, the individual learning factor c1 and the social learning factor c2. Update the position and speed of the particles, and each particle is based on the individual optimal position p best,i and the global optimal position g best Update its own speed and position. For each particle i, update its speed v i and position x i :

[0185]

[0186] in, is the current speed, is the updated speed, w (t) is the current inertia weight, c1 (t) and

[0187] c2 (t) is the current individual and sociological factor, r1 and r2 are random number vectors in the interval [0,1], and p best,i is the best historical position of the particle, g best is the global optimal position;

[0188]

[0189] in is the current location, is the updated position;

[0190] Adaptive adjustment of parameters. In standard particle swarm optimization, the speed and position updates of particles are controlled by fixed learning factors. Adaptive adjustment of parameters is performed based on historical search information. The parameters in the speed update formula, such as inertia weight ω, individual learning factor c1, and social learning factor c2, are adaptively adjusted. For example, when the individual optimal position of a particle has not improved for several iterations, the inertia weight ω is increased to enhance the global search capability; when the individual optimal positions of multiple particles are similar or the same, the inertia weight ω is reduced to enhance the local search capability.

[0191] Termination condition: when the maximum number of iterations is reached or the quality of the solution meets the predetermined standard, the algorithm stops and returns the parameters corresponding to the global optimal position as the parameters of the model;

[0192] The adaptive particle swarm algorithm (APSO) is used to identify the parameters and evaluate the fitness of the cable simulation model, and verify the physical parameter update and fitness of the model. The APSO is used to optimize the objective function, and the APSO algorithm is used to optimize the model parameters in each test scenario to make the predicted output of the model as close to the actual measured data as possible. The fitness function root mean square error (MSE) is used to evaluate the error between the model output and the actual measured value. The parameter identification formula is:

[0193]

[0194] Among them, y i is the actual measurement value of the ith value, is the i-th model prediction value, and N is the total number of measured values. At the same time, comprehensive performance evaluation tests are carried out, including data synchronization and update speed, evaluating the speed of synchronous data update between the data layer and the simulation layer, measuring the average speed and maximum delay of synchronous data update between the data layer and the simulation layer under different test scenarios; and the fitness of the finite element simulation model, through the determination coefficient R 2 Statistical methods such as the following are used to evaluate the degree of agreement between the simulation model output and the actual value:

[0195]

[0196] Using R 2 Evaluate how well the model output fits the actual measurements. A high R 2 Values ​​imply a better fit;

[0197] In other words, the cable sensor monitoring temperature data is selected as the objective function, the fitness function of the finite element model calculation value and the actual value is constructed using the least squares method, the finite element model material parameters are calculated using the adaptive particle swarm algorithm, the cable physical parameters are identified, and the equivalent conductivity of the conductor connection and the dielectric constant of the insulating material are selected as the parameters to be identified by the algorithm. Based on the real-time operation data, it is interacted with the cable simulation model, and the cable simulation model parameters are identified with the help of the adaptive particle swarm algorithm. The finite element simulation model can be updated in real time, thereby obtaining a real-time updated cable digital twin model, so as to fully and real-time perceive the internal state of the cable. At the same time, the equivalent conductivity of the conductor connection and the dielectric constant of the insulating material obtained by inversion based on the adaptive particle swarm algorithm are used to evaluate the overheating and insulation state of the cable joint.

[0198] Based on simulation data, the radial basis function neural network (RBFNN) is trained to build a deep learning fault diagnosis model of cable joint temperature and ultrasonic data. The real-time operation status of the cable joint is evaluated according to the cable joint operation guidelines and fault diagnosis model. Specifically, it includes:

[0199] In step S5, simulation data under different cable defects are acquired and sorted;

[0200] Normalize the data and divide the experimental set and the test set. The commonly used method is Min-Max Scaling. The calculation formula is:

[0201]

[0202] Among them, x is the original data, and x' is the normalized data;

[0203] A radial basis function neural network (RBFNN) model is constructed, and the RBFNN network structure is determined. The input vector dimension of the input layer is equal to the number of features (temperature and ultrasonic features). These data are used to diagnose the fault type of the cable joint. For the hidden layer (radial basis function layer), the commonly used radial basis function is the Gaussian function, and its expression is:

[0204]

[0205] Among them, c j is the center of the jth radial basis function, σ j is the width parameter, x is the input feature vector, φ j (x) is the output after the activation function. The output layer usually contains one or more neurons, and the number of output neurons depends on the number of categories in the classification task or the output dimension of the regression task.

[0206] When training the fault diagnosis model, the objective function is set to minimize the weighted loss function, combining the fault type and the diagnosis accuracy to ensure that the model can effectively distinguish different fault states. The loss function can be expressed as:

[0207]

[0208] Among them, y i is the true label of the i-th sample, y^ i is the predicted output of the model, λ1 and λ2 are weighting coefficients, Penalty(C i ) is the penalty term based on the cable joint fault type, C i is the characteristic value or configuration of the i-th sample.

[0209] Finally, the real-time collected data (temperature, ultrasonic information) are input into the model, and the model outputs the diagnosis result of the fault state. An effective cable joint fault diagnosis model based on RBFNN is constructed, and the real-time operation state of the cable joint is evaluated according to the cable joint operation guide and fault diagnosis model.

[0210] Based on the application of the constructed cable joint digital twin system and operation status evaluation model, a cable joint operation status visualization platform is developed. Specifically, it includes:

[0211] A cable status fault diagnosis model based on deep learning, based on the cable digital twin model and actual operation data, is proposed. The above implementation scheme designs and trains a deep learning model based on radial basis function neural network (RBFNN) to analyze and diagnose potential faults of cable joints. By integrating the data obtained from the digital twin model and the sensor data collected in actual operation, the model is trained to improve the accuracy and real-time performance of the diagnosis. In order to obtain information such as cable operation data and status evaluation online in real time;

[0212] A GUI software interface design based on QT and GPU multi-threaded parallel processing technology is proposed to develop a cable joint operation status visualization platform with a user-friendly interface and intuitive operation. The platform displays the key parameters of the cable joint in real time, such as temperature, current, voltage, and ultrasonic signals, and provides real-time charts and dashboards to display data changes in an intuitive way, such as trend charts, heat maps, and multi-dimensional analysis charts. A cable joint operation status visualization platform with real-time data monitoring, status display, and fault alarm is realized, which is convenient for on-site application. At the same time, the collected data is processed in real time through the fault diagnosis model, and the visualization platform provides instant feedback to the operator, such as fault warning and operation status evaluation.

[0213] Based on the same inventive concept, the present invention also provides a power cable joint defect detection system 400 based on a digital twin model, such as Figure 4 As shown, including:

[0214] The performance parameter acquisition module 410 is used to obtain the performance parameters of the cable connector under various actual operating conditions;

[0215] The temperature data and ultrasonic information acquisition module 420 is used to acquire the temperature data and ultrasonic information of the cable joint in operation in real time through a sensor network;

[0216] The multi-field coupling simulation model building module 430 is used to build a multi-field coupling simulation model of the cable joint according to the physical size and material parameters of the cable joint entity;

[0217] A twin simulation model construction module 440 is used to construct an objective function according to the performance parameters and temperature data, and to interact the data of the cable joint multi-field coupling simulation model with the data under actual operating conditions through the objective function, so as to update the pre-constructed cable joint digital twin simulation model and obtain a real-time updated cable joint digital twin simulation model;

[0218] The fault diagnosis module 450 is used to train and test the radial basis function neural network based on the output data of the real-time updated digital twin simulation model of the cable joint, obtain a fault diagnosis model of the cable joint temperature and ultrasonic data based on deep learning, and detect the real-time operating status of the cable joint according to the fault diagnosis model.

[0219] Furthermore, it also includes:

[0220] The real-time display module is used to display the operating status of the cable joint in real time through a visualization platform based on the cable joint digital twin simulation model and the fault diagnosis model.

[0221] The power cable joint defect detection method and system based on the digital twin model provided by the present invention mainly have the following technical advantages compared with the prior art:

[0222] (1) A digital twin system of high-voltage cable joints and its physical cable test method were constructed to accurately simulate and test the performance of cable joints under various operating conditions. By integrating a sensor network that monitors the temperature and ultrasonic parameters of cable joints in real time, the sensor layout was optimized to fully cover the cable joints, including joints that are difficult to monitor and hidden parts. The system not only collects and processes data reflecting the actual operating status of the cable, but also constructs an electro-thermal-acoustic multi-field coupling simulation model, and uses the control variable method to analyze the impact of different operating conditions and fault defects on the cable joints.

[0223] (2) Furthermore, the objective function is constructed using the data collected by the sensor network to achieve effective interaction between the simulation model data and the actual operation data, and the physical property parameters of the model are identified and corrected through a deep learning algorithm to ensure the continuous updating and accurate correction of the digital twin model. In addition, this study trained a temperature and ultrasonic fault diagnosis model for a cable joint based on a radial basis function neural network (RBFNN) and simulation data to evaluate the operating status of the cable joint in real time. Based on the developed digital twin system and status assessment model, a visualization platform for the operating status of the cable joint was developed to provide support for field applications.

[0224] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0225] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0226] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0227] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0228] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalents that do not depart from the spirit and scope of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A power cable joint defect detection method based on a digital twin model, characterized in that: include: Obtain the performance parameters of cable connectors under various actual operating conditions; Acquiring temperature data and ultrasonic information of the cable joint in operation in real time through a sensor network; According to the physical size and material parameters of the cable joint entity, a multi-field coupling simulation model of the cable joint is constructed; Constructing an objective function according to the performance parameters and temperature data, interacting the data of the cable joint multi-field coupling simulation model with the data under actual operating conditions through the objective function, updating the pre-constructed cable joint digital twin simulation model, and obtaining a real-time updated cable joint digital twin simulation model; Based on the output data of the real-time updated digital twin simulation model of the cable joint, the radial basis function neural network is trained and tested to obtain a fault diagnosis model of the cable joint temperature and ultrasonic data based on deep learning. The real-time operating status of the cable joint is detected according to the fault diagnosis model.

2. The method according to claim 1, characterized in that Obtain performance parameters of cable connectors under various actual operating conditions, including: Build a cable physical test platform for the digital twin system of high-voltage cable joints; The performance of the cable joint under various operating conditions is simulated through the test platform to obtain the performance parameters of the cable joint.

3. The method according to claim 1, characterized in that The temperature data and ultrasonic information of the cable joint in operation are obtained in real time through the sensor network, including: Construct a sensor network consisting of temperature sensing and ultrasonic sensing node arrays; The sensor network comprehensively monitors various parts of the cable, and obtains temperature data and ultrasonic information of the cable joint in operation in real time.

4. The method according to claim 1, characterized in that: The cable joint multi-field coupling simulation model also uses the control variable method to analyze the impact of different operating conditions and fault defects on the operating status of the cable joint.

5. The method according to claim 4, characterized in that The cable joint multi-field coupling simulation model also uses the control variable method to analyze the impact of different operating conditions and fault defects on the operating status of the cable joint, including: Simulate and analyze the impact of different load currents on the temperature distribution of cable joints and changes in electromagnetic field distribution; Simulate and analyze the impact of excessive contact resistance of cable joints on temperature distribution and electromagnetic field distribution changes; Simulate and analyze the impact of different dielectric constants of cable insulation materials on the temperature distribution of cable joints and changes in electromagnetic field distribution; The waveform signal of ultrasonic waves in cable joints under different defect conditions is simulated and analyzed.

6. The method according to claim 1, characterized in that The physical dimensions and material parameters of the cable connector entity include: Get the physical size of the cable connector entity; Set electromagnetic field boundary conditions; Set the boundary conditions of the cable joint temperature field; Set the acoustic field boundary conditions for the cable joint.

7. The method according to claim 1, characterized in that The data of the cable joint multi-field coupling simulation model is interacted with the data under actual operating conditions through the objective function, and the pre-built cable joint digital twin simulation model is updated to obtain a real-time updated cable joint digital twin simulation model, including: The least square method must be used to construct the fitness function of the finite element simulation model data and the actual data for data interaction; Based on the adaptive particle swarm algorithm, the parameters of the cable joint digital twin simulation model are corrected, the cable joint finite element simulation model is updated in real time, and the real-time updated cable joint digital twin simulation model is obtained.

8. The method according to claim 1, characterized in that Also includes: Based on the cable joint digital twin simulation model and the fault diagnosis model, the operating status of the cable joint is displayed in real time through a visualization platform.

9. A power cable joint defect detection system based on a digital twin model, characterized in that: include: Performance parameter acquisition module, used to obtain the performance parameters of the cable connector under various actual operating conditions; A temperature data and ultrasonic information acquisition module, used to acquire temperature data and ultrasonic information of the cable joint in operation in real time through a sensor network; A multi-field coupling simulation model building module is used to build a multi-field coupling simulation model of a cable joint according to the physical size and material parameters of the cable joint entity; A twin simulation model construction module is used to construct an objective function according to the performance parameters and temperature data, and to interact the data of the cable joint multi-field coupling simulation model with the data under actual operating conditions through the objective function, so as to update the pre-constructed cable joint digital twin simulation model and obtain a real-time updated cable joint digital twin simulation model; The fault diagnosis module is used to train and test the radial basis function neural network based on the output data of the real-time updated digital twin simulation model of the cable joint, obtain a fault diagnosis model of the cable joint temperature and ultrasonic data based on deep learning, and detect the real-time operating status of the cable joint according to the fault diagnosis model.

10. The system according to claim 9, characterized in that Also includes: The real-time display module is used to display the operating status of the cable joint in real time through a visualization platform based on the cable joint digital twin simulation model and the fault diagnosis model.

Images