Method and System for Constructing Digital Twin Model of Transmission Line Equipment Based on Improved IDEF

Abstract

A method and system for constructing a digital twin model of transmission line equipment based on improved IDEF. The method first divides the construction stage of the transmission line and establishes subsystems, then adds intermediate nodes between adjacent data nodes and sets a data exchange trigger mechanism, then performs data fusion based on the Kalman filter fusion method, then uses IDEF0, IDEF3, IDEF5, modular IDEF1X, and integrates the subsystems to jointly construct a digital twin comprehensive model, and finally conducts simulation and judges whether to update parameters based on the Euclidean distance; The present invention specifically considers different stages in the construction of transmission line equipment, takes into account multi-angle system modeling, reduces the complexity of data communication through intermediate nodes, improves the transmission efficiency, and at the same time proposes a modular IDEF1X information modeling method to improve the stability of the model, and judges whether to update parameters based on the Euclidean distance, intuitively and accurately completing the digital twin evolution, making the model have high reliability and accuracy.

Description

Technical Field

[0001] The present invention relates to digital twin model means, belonging to the field of digital construction of transmission lines, and particularly relates to a method and system for constructing a digital twin model of transmission line equipment based on improved IDEF. Background Technique

[0002] With the continuous development and modernization of the power system, transmission equipment plays a crucial role in power transmission; in the traditional power system, the monitoring and maintenance of transmission equipment mainly rely on regular inspections and the judgment of personnel experience, and it is impossible to monitor the equipment status in real time. Moreover, due to the inconsistent experience of each person, the judgment and prediction of transmission line faults often deviate from the actual situation to a certain extent; while digital twin technology combines physical entities with virtual mappings, and provides comprehensive monitoring, prediction and optimization functions for transmission line power equipment by simulating and analyzing the operating conditions of actual equipment, so as to realize more intelligent and accurate management of the power system.

[0003] In the current research of digital twins, there are various construction methods, such as constructing a conceptual model of product life cycle management, a five-dimensional model, a model based on meta-action theory, etc. However, these methods often face specific application scenarios and lack consideration of the functional extensibility of digital twins, resulting in an island phenomenon among multiple digital twins for different application scenarios. Considering that transmission lines have characteristics such as large scale, many objects, complex functional structures, numerous elements, massive data, and multi-source heterogeneity, the construction process of digital twin models is complex. The traditional modeling method with a single design structure lacks systematic and structured modeling analysis, resulting in content duplication or omission when constructing a digital twin model of a transmission line, and it is difficult to achieve the systematic modeling requirements of the digital twin model of the transmission line in terms of the reliability and accuracy of the model. Summary of the Invention

[0004] The purpose of the present invention is to overcome the above-mentioned defects and problems in the prior art, and provide a method and system for constructing a digital twin model of transmission line equipment based on improved IDEF with better reliability and accuracy.

[0005] To achieve the above purpose, the technical solution of the present invention is: A method for constructing a digital twin model of transmission line equipment based on improved IDEF, including:

[0006] S1. Divide the construction stage of transmission line equipment into a foundation construction stage, a tower erection construction stage, and a wire stringing construction stage, and establish respective subsystems for the above three stages; the subsystems include a three-dimensional static model subsystem, a three-dimensional visualization monitoring subsystem, and a simulation prediction subsystem;

[0007] S2. Add intermediate nodes between adjacent data nodes in the subsystem, simultaneously set the trigger mechanism for data exchange between adjacent data nodes, and fuse the multi-source heterogeneous data generated during the construction of transmission line equipment based on the distributed Kalman filtering fusion method;

[0008] S3. Use IDEF0 to construct a function model, IDEF3 to construct a network workflow model, IDEF5 to construct an ontology model, and modular IDEF1X to construct a data information model, and integrate the subsystems to jointly construct a digital twin comprehensive model;

[0009] S4. Perform simulation on the construction stage of transmission line equipment based on the digital twin comprehensive model, and judge whether to update the parameters of the digital twin comprehensive model based on the Euclidean distance between the simulation results and the physical entity status monitoring data; if so, use the variational Bayesian model based on the Euclidean distance to update the parameters of the digital twin comprehensive model.

[0010] The specific steps of step S2 include:

[0011] S21. Represent the state of each data node in the multi-source heterogeneous data through information pairs;

[0012] The information pair includes an information matrix and an information vector; the expression of the information matrix is as follows:

[0013]

[0014] Where: is the information matrix of the data node, is the estimated covariance at time k;

[0015] The expression of the information vector is as follows:

[0016]

[0017] Where: is the information vector of the data node, is the state estimate value at time k;

[0018] S22. Add an intermediate node i between every two adjacent data nodes, and introduce a binary variable and the covariance value u as a threshold during the data exchange process of each data node to determine whether to trigger data exchange;

[0019] The intermediate node i sends its own estimated covariance to the adjacent data nodes, and fuses the estimated covariance of the adjacent data nodes to obtain a fused information pair Then, the fusion of multi-source heterogeneous data is completed through iterative loops; the expression of the fused information pair is as follows:

[0020]

[0021] Wherein: is the information vector of the data node after fusion, is the information matrix of the data node after fusion;

[0022] The expression of the trigger mechanism is as follows:

[0023]

[0024] If then trigger data exchange,

[0025] If then do not trigger data exchange,

[0026] Wherein: is the demand covariance of data node i at time k, is the updated value of data node i at time k, indicates that data node i communicates with adjacent data nodes for data exchange at time k, indicates that data node i does not communicate with adjacent data nodes for data exchange at time k.

[0027] In step S3, the modular IDEF1X constructs a data information model, specifically including:

[0028] S31. For the construction equipment in the subsystem, establish a DSM matrix; both the rows and columns in the matrix represent the construction equipment;

[0029] S32. Based on the genetic algorithm, perform modular processing on the DSM matrix to obtain DSM modules;

[0030] S33. In the DSM module, the information of construction equipment with the same attributes within the same module is grouped into the same set, represented by a box in IDEF1X, and then coupled through the connection of construction equipment within the same module to establish the connection between the same modules, completing the construction of the data information model.

[0031] Step S32 specifically includes:

[0032] S321. Define a fitness function to divide the DSM matrix into modules; the expression of the fitness function is as follows:

[0033] max(F) = I / E;

[0034] Where: maximizing the value of the fitness function F, I is the average cohesion of the DSM module, and E is the average coupling of the DSM module;

[0035]

[0036] Where: N all is the total number of DSM modules, I m is the cohesion of module m, N m is the number of elements in module m, C m (i, j) is the connection strength between element i and element j in module m, C max is the maximum value of the connection strength between elements, E p，q is the coupling degree between module p and module q;

[0037] S322. Map the DSM module partitioning result to a two-dimensional coding matrix and use it as the chromosome of the genetic algorithm for operation;

[0038] S323. Based on the chromosome, perform population initialization and crossover mutation operations in sequence, and perform iterative operations repeatedly to determine the optimal population of the genetic algorithm, and use the optimal population as the best partitioning scheme for the DSM module.

[0039] In step S4, it is judged whether to update the parameters of the digital twin comprehensive model, which specifically includes:

[0040] S41. Compare the Euclidean distance between the simulation result of the digital twin comprehensive model and the physical entity state monitoring data under the same working condition to obtain the Euclidean distance of the historical data set;

[0041] The simulation result is X = (x1, x2, x3... x n ), and the state monitoring data is Y = (y1, y2, y3... y n ); the Euclidean distance between the simulation result and the state monitoring data is as follows:

[0042]

[0043] S42. Set the maximum Euclidean distance As as the update trigger threshold, then add a new sample set for simulation, and compare the simulation result of the new sample set with the historical state monitoring data to obtain the Euclidean distance of the new sample set;

[0044] S43. Compare the Euclidean distance of the historical data set with the Euclidean distance of the new sample set;

[0045] If both the Euclidean distance of the historical data set and the Euclidean distance of the new sample set are greater than the maximum Euclidean distance As, update the parameters of the digital twin comprehensive model.

[0046] In step S4, the parameters of the digital twin integrated model are updated, specifically including:

[0047] S44. Construct the input of Bayesian inference based on the likelihood function and the prior distribution;

[0048] The real-time state data of the physical entity is X = [x1, x2,..., x n , and the simulation data of the digital twin integrated model is X(θ i ) = [x1(θ i ), x2(θ i ),..., x n (θ i )]. The relationship between the real-time state data X and the simulation data X(θ i ) is X = X(θ i ) + ε; where: ε = [ε1, ε2,..., ε m is an m-dimensional vector of errors;

[0049] Then the expression of the likelihood function P(x|θ) is as follows:

[0050]

[0051] where: μ i , σ i are the mean and standard deviation of the error respectively;

[0052] S45. Use the beta distribution to describe the change process of the physical entity parameters; the expression of the beta distribution is as follows:

[0053] f(θ; α, β) = θ α-1 (1 - θ) β-1 ;

[0054] where: θ is the parameter to be updated, and both α and β are parameters greater than zero;

[0055] S46. Calculate the values of α and β according to the change data of the prior distribution of the parameters; the calculation formula is as follows:

[0056]

[0057] where: is the parameter mean, and s θ is the parameter standard deviation;

[0058] Then the output of Bayesian inference is expressed as follows:

[0059]

[0060] Where: p(θ|x) is the posterior distribution of parameter θ, that is, the parameter of the updated digital twin comprehensive model; p(x|θ) is the likelihood function, representing the probability of observing data x when the parameter value is θ; p(θ) is the prior distribution of parameter θ, the same as the prior probability distribution

[0061] S47. Approximate the posterior distribution p(θ|x) as a simple distribution q(θ) = q φ (θ);

[0062] S48. Use the KL divergence to quantify the mismatch between the posterior distribution p(θ|x) and the simple distribution q φ (θ); The expression of the KL divergence is as follows:

[0063]

[0064] Where: E φ is the expectation of the logarithmic function in the parentheses;

[0065] S49. Transform the problem of solving the posterior distribution into the problem of solving the maximum value of the evidence lower bound ELBO, and confirm the objective function to complete the parameter update of the digital twin comprehensive model;

[0066] The expression of the evidence lower bound ELBO is as follows:

[0067]

[0068] Where: p(x|θ)p(θ) is the joint distribution of θ and x;

[0069] The objective function is as follows:

[0070] F(θ) = max[ELBO];

[0071] Where: F(θ) is the initial value of the objective function.

[0072] A transmission line equipment digital twin model construction system based on improved IDEF, the system includes:

[0073] A digital twin subsystem construction module, used to divide the construction stage of transmission line equipment into a foundation construction stage, a tower erection construction stage, and a wire stringing construction stage, and establish respective subsystems for the above three stages; the subsystem includes a three-dimensional static model subsystem, a three-dimensional visualization monitoring subsystem, and a simulation prediction subsystem;

[0074] A digital twin data fusion module is used to add intermediate nodes between adjacent data nodes in a subsystem, set a trigger mechanism for data exchange between adjacent data nodes, and fuse multi-source heterogeneous data generated during the construction of transmission line equipment based on the distributed Kalman filter fusion method;

[0075] A digital twin model construction module is used to construct a functional model using IDEF0, a network workflow model using IDEF3, an ontology model using IDEF5, and a data information model using modular IDEF1X, and integrate the subsystem to jointly construct a digital twin comprehensive model;

[0076] A digital twin model update module is used to simulate the construction stage of transmission line equipment based on the digital twin comprehensive model, and judge whether to update the parameters of the digital twin comprehensive model based on the Euclidean distance between the simulation result and the physical entity state monitoring data; if so, use the variational Bayesian model based on the Euclidean distance to update the parameters of the digital twin comprehensive model.

[0077] The digital twin data fusion module performs data fusion according to the following steps:

[0078] S21. Represent the state of each data node in the multi-source heterogeneous data through an information pair;

[0079] The information pair includes an information matrix and an information vector; the expression of the information matrix is as follows:

[0080]

[0081] Where: is the information matrix of the data node, is the estimated covariance at time k;

[0082] The expression of the information vector is as follows:

[0083]

[0084] Where: is the information vector of the data node, is the state estimate value at time k;

[0085] S22. Add an intermediate node i between two adjacent data nodes, and introduce a binary variable and the covariance value u as a threshold during the data exchange process of each data node to determine whether to trigger data exchange;

[0086] The intermediate node i sends its own estimated covariance to the adjacent data nodes, and fuses the estimated covariance of the adjacent data nodes to obtain a fused information pair Then, the fusion of multi-source heterogeneous data is completed through loop iteration; the expression of the fusion information pair is as follows:

[0087]

[0088]

[0089] Where: is the information vector of the data node after fusion, is the information matrix of the data node after fusion;

[0090] The expression of the triggering mechanism is as follows:

[0091]

[0092] If then data exchange is triggered,

[0093] If then data exchange is not triggered,

[0094] Where: is the demand covariance of data node i at time k, is the update value of data node i at time k, indicates that data node i communicates with adjacent data nodes for data exchange at time k, indicates that data node i does not communicate with adjacent data nodes for data exchange at time k.

[0095] The digital twin model construction module constructs a digital twin model according to the following steps:

[0096] S31. For the construction equipment in the subsystem, establish a DSM matrix; both the rows and columns in the matrix represent construction equipment;

[0097] S32. Based on the genetic algorithm, modularize the DSM matrix to obtain DSM modules;

[0098] S33. In the DSM module, the information of construction equipment with the same attributes within the same module is grouped into the same set, represented by a box in IDEF1X, and then coupled through the connections of construction equipment within the same module to establish connections between the same modules, completing the construction of the data information model;

[0099] The specific steps of S32 include:

[0100] S321. Define a fitness function to partition the DSM matrix; the expression of the fitness function is as follows:

[0101] max(F) = I / E;

[0102] Where: max(F) is to maximize the value of the fitness function F, I is the average cohesion of the DSM module, and E is the average coupling of the DSM module;

[0103]

[0104]

[0105] Where: N all is the total number of DSM modules, I m is the cohesion of module m, N m is the number of elements in module m, C m (i, j) is the connection magnitude between element i and element j in module m, C max is the maximum value of the connection magnitude between elements, E p，q is the coupling degree between module p and module q;

[0106] S322. Map the DSM module partitioning result to a two-dimensional coding matrix and perform operations as the chromosome of the genetic algorithm;

[0107] S323. Based on the chromosome, perform population initialization and crossover mutation operations in sequence, and perform iterative operations repeatedly to determine the optimal population of the genetic algorithm, and use the optimal population as the best partitioning scheme of the DSM module.

[0108] The digital twin model update module updates the parameters of the model according to the following steps:

[0109] S44. Construct the input of Bayesian inference based on the likelihood function and the prior distribution;

[0110] The real-time state data of the physical entity is X = [x1, x2,..., x n , and the simulation data of the digital twin comprehensive model is X(θ i ) = [x1(θ i ), x2(θ i ),..., x n (θ i )]. The relationship between the real-time state data X and the simulation data X(θ i ) is X = X(θ i ) + ε; where: ε = [ε1, ε2,..., ε m is the m-dimensional vector of errors;

[0111] Then the expression of the likelihood function p(x|θ) is as follows:

[0112]

[0113] where: μ i , σ i are the mean and standard deviation of the error respectively;

[0114] S45. Describe the change process of the physical entity parameters using the beta distribution; the expression of the beta distribution is as follows:

[0115] f(θ; α, β) = θ α-1 (1 - θ) β-1 ;

[0116] where: θ is the parameter to be updated, and both α and β are parameters greater than zero;

[0117] S46. Calculate the values of α and β according to the prior distribution change data of the parameters; the calculation formula is as follows:

[0118]

[0119] where: is the parameter mean, and s θ is the parameter standard deviation;

[0120] Then the output of Bayesian inference is expressed as follows:

[0121]

[0122] where: p(θ|x) is the posterior distribution of the parameter θ, that is, the parameter of the updated digital twin comprehensive model; p(θ|x) is the likelihood function, representing the probability of observing the data x when the parameter value is θ; p(θ) is the prior distribution of the parameter θ, the same as the prior probability distribution

[0123] S47. Approximate the posterior distribution p(θ|x) as a simple distribution q(θ) = q φ (θ);

[0124] S48. Quantify the mismatch between the posterior distribution p(θ|x) and the simple distribution q φ (θ) using the KL divergence; the expression of the KL divergence is as follows:

[0125]

[0126] where: E φ is the expectation of the logarithmic function in the parentheses;

[0127] S49. Transform the problem of solving the posterior distribution into the problem of solving the maximum value of the evidence lower bound ELBO, and confirm the objective function to complete the parameter update of the digital twin comprehensive model;

[0128] The expression of the evidence lower bound ELBO is as follows:

[0129]

[0130] Where: p(x|θ)p(θ) is the joint distribution of θ and x;

[0131] The objective function is as follows:

[0132] F(θ) = max[ELBO];

[0133] Where: F(θ) is the initial value of the objective function.

[0134] Compared with the prior art, the beneficial effects of the present invention are:

[0135] In the method and system for constructing a digital twin model of transmission line equipment based on improved IDEF of the present invention, the method first divides the construction stage of transmission line equipment and establishes their respective subsystems, then adds intermediate nodes between adjacent data nodes and sets a data exchange trigger mechanism, and then performs data fusion based on the Kalman filter fusion method. Next, IDEF0 is used to construct functions, IDEF3 is used to construct network work processes, IDEF5 is used to construct ontologies, and modular IDEF1X is used to construct data information, and the subsystems are integrated to jointly construct a digital twin comprehensive model. Finally, simulation is carried out, and based on the Euclidean distance between the simulation result and the state monitoring data, it is judged whether to update the parameters. In the application of the present invention, different stages in the construction process of transmission line equipment are considered specifically, multi-angle system modeling is taken into account, and the complexity of data communication is reduced by adding intermediate nodes, improving data transmission efficiency. At the same time, a modular IDEF1X information modeling method is proposed to improve the stability of the digital twin model. Finally, based on the Euclidean distance, it is judged whether to update the parameters, and the digital twin evolution is completed intuitively and accurately, making the digital twin model of transmission line equipment have high reliability and accuracy. Description of the Drawings

[0136] Figure 1 is the flow chart of the method steps in the present invention.

[0137] Figure 2 is the schematic diagram of the composition of the digital twin comprehensive model in the present invention.

[0138] Figure 3 is the update flow chart of the digital twin comprehensive model in the present invention.

[0139] Figure 4 is the modeling flow chart of the digital twin comprehensive model in the present invention.

[0140] Figure 5It is the real-time monitoring data graph of the environmental temperature of the transmission line in the present invention.

[0141] Figure 6 It is the identification graph of the potential hazard area in the present invention.

[0142] Figure 7 It is the schematic diagram of the system structure in the present invention.

[0143] Figure 8 It is the schematic diagram of the equipment structure in the present invention.

[0144] In the figure: Digital twin stage division module 1, digital twin data fusion module 2, digital twin model construction module 3, digital twin model update module 4, processor 5, memory 6, computer program code 61. Specific implementation mode

[0145] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific implementation modes.

[0146] Embodiment 1:

[0147] See Figure 1 , a method for constructing a digital twin model of transmission line equipment based on improved IDEF, including:

[0148] S1. Divide the construction stage of the transmission line equipment into the foundation construction stage, tower erection construction stage, and wire stringing construction stage, and establish respective subsystems for the above three stages; the subsystems include a three-dimensional static model subsystem, a three-dimensional visual monitoring subsystem, and a simulation prediction subsystem;

[0149] See Figure 2 , the three-dimensional static model subsystem refers to: mapping the physical attributes of the physical entities in the subsystem to the virtual environment to construct a virtual model, forming a three-dimensional static model subsystem; the physical entities include towers, construction machinery, and construction tools, and the virtual models include tower models, construction machinery models, and construction tool models;

[0150] The three-dimensional visual monitoring subsystem refers to: integrating the storage function, analysis function, transmission function, and display function when processing digital twin body data in the subsystem, forming a three-dimensional visual monitoring subsystem; the digital twin body data includes physical entity data and operator data;

[0151] The simulation prediction subsystem refers to: performing simulation analysis on the states of the physical entities in the subsystem, simulating and predicting the state changes of the physical entities in the virtual environment, forming a simulation prediction subsystem; the simulation prediction includes physical entity fault warning and remaining working hours prediction.

[0152] S2. Add intermediate nodes between adjacent data nodes in the subsystem, and at the same time set the triggering mechanism for data exchange between adjacent data nodes, and fuse the multi-source heterogeneous data generated during the construction of transmission line equipment based on the distributed Kalman filter fusion method;

[0153] Further, it specifically includes:

[0154] S21. Represent the state of each data node in the multi-source heterogeneous data through information pairs;

[0155] The information pair includes an information matrix and an information vector; the expression of the information matrix is as follows:

[0156]

[0157] Where: is the information matrix of the data node, is the estimated covariance at time k;

[0158] The expression of the information vector is as follows:

[0159]

[0160]

[0161] Where: is the information vector of the data node, is the state estimate value at time k;

[0162] S22. Add an intermediate node i between every two adjacent data nodes, and introduce a binary variable and the covariance value u as a threshold in the data exchange process of each data node to determine whether to trigger data exchange;

[0163] The intermediate node i sends its own estimated covariance to the adjacent data nodes, and fuses the estimated covariance of the adjacent data nodes to obtain the fused information pair Then complete the fusion of multi-source heterogeneous data through iterative loops; the expression of the fused information pair is as follows:

[0164]

[0165] Where: is the information vector of the fused data node, is the information matrix of the fused data node;

[0166] The expression of the triggering mechanism is as follows:

[0167]

[0168] If then trigger data exchange,

[0169] If then do not trigger data exchange,

[0170] Wherein: is the demand covariance of data node i at time k, is the updated value of data node i at time k, represents that data node i communicates with adjacent data nodes for data exchange at time k, represents that data node i does not communicate with adjacent data nodes for data exchange at time k.

[0171] S3. Use IDEF0 to construct a function model, IDEF3 to construct a network workflow model, IDEF5 to construct an ontology model, and modular IDEF1X to construct a data information model, and integrate subsystems to jointly construct a digital twin comprehensive model;

[0172] See Figure 3 , in this embodiment, IDEF is used to achieve systematic and structured modeling and analysis of transmission line construction equipment oriented to top-level design, including IDEF0 function modeling, IDEF3 network workflow modeling, and IDEF5 ontology modeling.

[0173] The IDEF0 function modeling refers to: when constructing a transmission line equipment model, it is first necessary to analyze the function and mutual relationship of subsystems, etc., and strictly analyze the digital twin model of transmission line equipment from the top layer to the lower layer, from the system to the equipment. Through function modeling, the function and structural composition of transmission line equipment can be deeply analyzed, the main equipment categories can be determined, including towers, construction machinery, tools, etc., the model library can be constructed, and finally, the IDEF0 model can be constructed to clarify the hierarchical structure and information flow direction of the function block diagram.

[0174] The IDEF3 network workflow modeling refers to: using process modeling to decompose the workflow of each stage of transmission line construction equipment, and using the PFN model diagram of the IDEF3 method to construct the flow process of data in the digital twin model of transmission line construction equipment, and recording the priority and causal relationship of events occurring and state changes in the actual production process of manufacturing transmission line equipment in graphical language.

[0175] The IDEF5 ontology modeling mentioned above refers to: through ontology modeling, using the IDEF5 graphical ontology acquisition method to construct the conceptual model of the digital twin model of transmission line construction equipment, providing an interactive, shared, structured and consistent conceptual framework for the development of the ontology of the digital twin model of transmission line construction equipment.

[0176] The modular IDEF1X construction data information model mentioned above refers to: establishing the design structure matrix (DSM) of construction equipment, carrying out "high cohesion, low coupling" module division for the equipment, using the genetic algorithm to solve the DSM module division problem, and finally establishing a modular IDEF1X information model on this basis, which is beneficial to improving the stability of the digital twin model.

[0177] Furthermore, it specifically includes:

[0178] S31. For the construction equipment in the subsystem, establish an n×n DSM matrix; both the rows and columns in the matrix represent construction equipment.

[0179] S32. Based on the genetic algorithm, perform modular processing on the DSM matrix to obtain DSM modules;

[0180] S33. In the DSM modules, classify the information of construction equipment with the same attributes within the same module into the same set, represent it with a box in IDEF1X, and then couple through the connections of construction equipment within the same module to establish the connections between the same modules, completing the construction of the data information model.

[0181] Furthermore, the step S32 specifically includes:

[0182] S321. Define the fitness function and perform module division on the DSM matrix; the expression of the fitness function is as follows:

[0183] max(F) = I / E;

[0184] Where: max(F) is the maximization of the value of the fitness function F, I is the average cohesion of the DSM module, and E is the average coupling of the DSM module;

[0185]

[0186] Where: N all is the total number of DSM modules, I m is the cohesion of module m, N m is the number of elements in module m, C m (i, j) is the connection magnitude between element i and element j in module m, C max is the maximum value of the connection magnitude between elements, E p，q is the coupling degree between module p and module q;

[0187] S322. Map the DSM module division result to a two-dimensional coding matrix and use it as the chromosome of the genetic algorithm for operation;

[0188] The two-dimensional coding matrix is an n×m Boolean matrix, where n is the number of modules into which the elements in the DSM matrix are divided, m is the dimension of the DSM matrix, and the matrix element a i,j has a value of "1" or "0". A value of "1" means that the part element j is divided into the j-th module, and a value of "0" means that the part element j is not divided into the i-th module.

[0189] S323. Based on the chromosome, perform population initialization and crossover mutation operations in sequence, and perform iterative operations repeatedly to determine the optimal population of the genetic algorithm, and use the optimal population as the best division plan for the DSM module.

[0190] S4. Perform simulation on the construction stage of the transmission line equipment based on the digital twin comprehensive model, and judge whether to update the parameters of the digital twin comprehensive model based on the Euclidean distance between the simulation result and the physical entity status monitoring data; if so, use the variational Bayesian model based on the Euclidean distance to update the parameters of the digital twin comprehensive model.

[0191] Further, judging whether to update the parameters of the digital twin comprehensive model specifically includes:

[0192] S41. Compare the Euclidean distances between the simulation results of the digital twin comprehensive model and the physical entity status monitoring data under the same working conditions to obtain the Euclidean distances of the historical data set;

[0193] The simulation result is X=(x1, x2, x3……x n ), and the status monitoring data is Y=(y1, y2, y3……y n ); the Euclidean distance between the simulation result and the status monitoring data is as follows:

[0194]

[0195] S42. Set the maximum Euclidean distance As as the update trigger threshold, then add a new sample set for simulation, and compare the simulation result of the new sample set with the historical status monitoring data to obtain the Euclidean distance of the new sample set;

[0196] S43. Compare the Euclidean distance of the historical data set with the Euclidean distance of the new sample set;

[0197] If both the Euclidean distance of the historical data set and the Euclidean distance of the new sample set are greater than the maximum Euclidean distance As, update the parameters of the digital twin comprehensive model.

[0198] See Figure 4 , further, update the parameters of the digital twin comprehensive model, specifically including:

[0199] S44. Construct the input of Bayesian inference based on the likelihood function and the prior distribution;

[0200] The real-time state data of the physical entity is X = [x1, x2,..., x n , and the simulation data of the digital twin comprehensive model is X(θ i ) = [x1(θ i ), x2(θ i ),..., x n (θ i )]. The relationship between the real-time state data X and the simulation data X(θ i ) is X = X(θ i ) + ε; where: ε = [ε1, ε2,..., ε m is an m-dimensional vector of errors;

[0201] Then the expression of the likelihood function p(x|θ) is as follows:

[0202]

[0203] where: μ i , σ i are the mean and standard deviation of the error respectively;

[0204] S45. Use the beta distribution to describe the change process of the physical entity parameters; the expression of the beta distribution is as follows:

[0205] f(θ; α, β) = θ α-1 (1 - θ) β-1 ;

[0206] where: θ is the parameter to be updated, and both α and β are parameters greater than zero;

[0207] S46. Calculate the values of α and β according to the change data of the prior distribution of the parameters; the calculation formula is as follows:

[0208]

[0209] where: is the parameter mean, and S θ is the parameter standard deviation;

[0210] Then the output of Bayesian inference is expressed as follows:

[0211]

[0212] Among them: p(θ|x) is the posterior distribution of parameter θ, that is, the parameter of the updated digital twin comprehensive model; p(x|θ) is the likelihood function, representing the probability of observing data x when the parameter value is θ; p(θ) is the prior distribution of parameter θ, the same as the prior probability distribution

[0213] S47. Approximate the posterior distribution p(θ|x) as a simple distribution q(θ) = q φ (θ);

[0214] S48. Use the KL divergence to quantify the mismatch between the posterior distribution p(θ|x) and the simple distribution q φ (θ); the expression of the KL divergence is as follows:

[0215]

[0216] Among them: E φ is the expectation of the logarithmic function in the brackets;

[0217] S49. Transform the problem of solving the posterior distribution into the problem of solving the maximum value of the evidence lower bound ELBO, and confirm the objective function to complete the parameter update of the digital twin comprehensive model;

[0218] The expression of the evidence lower bound ELBO is as follows:

[0219]

[0220] Among them: p(x|θ)p(θ) is the joint distribution of θ and x;

[0221] The objective function is as follows:

[0222] F(θ) = max[ELBO];

[0223] Among them: F(θ) is the initial value of the objective function.

[0224] Transform the problem of solving the posterior distribution into the problem of solving the maximum value of ELBO. According to the real-time state data and the parameter prior distribution, the parameter posterior distribution can be calculated, thereby completing an update of the parameters. Combining multiple real-time state data, the entire twin model can be updated.

[0225] See Figure 5, in this embodiment, the real-time performance and reliability of data interaction of the digital twin model are verified through examples. It can be seen from the figure that the prediction error of the line temperature is controlled within ±0.4°C. The maximum value appears in the 21st group, the actual ambient temperature is 44.3°C, the lower limit of the monitored temperature is 44.1°C, and the upper limit is 44.7°C. This is mainly because the wind speed changed significantly at that time, and it was difficult for the model to fully adapt. Generally speaking, the prediction accuracy is high, which proves that the visual monitoring of the adaptive transmission line equipment digital twin model based on the improved IDEF is effective, and the modeling accuracy can be further improved through incremental learning in the future.

[0226] See Figure 6 , in this embodiment, the accuracy of the digital twin model in predicting potential hazard areas is verified through examples. The simulation results in the figure show that the misdiagnosis numbers of the safe area, machine potential hazard area, floating object potential hazard area, ice area, galloping area, polluted area, lightning area, and bird damage area (the area categories are numbered from 1 to 8 in sequence) are 15, 15, 6, 11, 12, 13, 12, and 13 respectively; the recognition accuracy rates of the eight areas are all above 85%, and the overall recognition accuracy rate reaches 88%. This verifies the ability of the model to identify potential hazard areas and proves that the potential hazard area recognition model of the adaptive transmission line equipment digital twin model based on the improved IDEF is reliable.

[0227] Embodiment 2:

[0228] See Figure 7 , a system for constructing a digital twin model of transmission line equipment based on the improved IDEF, the system includes:

[0229] The digital twin subsystem construction module 1 is used to divide the construction stage of the transmission line equipment into the foundation construction stage, tower erection construction stage, and wire stringing construction stage, and establish respective subsystems for the above three stages; the subsystems include a three-dimensional static model subsystem, a three-dimensional visual monitoring subsystem, and a simulation prediction subsystem;

[0230] Further, the subsystems divided by the digital twin subsystem construction module 1 are as follows:

[0231] The three-dimensional static model subsystem refers to mapping the physical attributes of the physical entities in the subsystem to the virtual environment to construct a virtual model, which forms a three-dimensional static model subsystem; the physical entities include towers, construction machinery, and construction tools, and the virtual models include tower models, construction machinery models, and construction tool models;

[0232] The three-dimensional visual monitoring subsystem refers to integrating the storage function, analysis function, transmission function, and display function when processing digital twin data in the subsystem, which forms a three-dimensional visual monitoring subsystem; the digital twin data includes physical entity data and operator data;

[0233] The simulation and prediction subsystem refers to: simulating and analyzing the states of physical entities in the subsystem, simulating and predicting the state changes of physical entities in the virtual environment, and forming the simulation and prediction subsystem; the simulation and prediction include physical entity fault warning and remaining working hours prediction.

[0234] The digital twin data fusion module 2 is used to add intermediate nodes between adjacent data nodes in the subsystem, set a trigger mechanism for data exchange between adjacent data nodes, and fuse multi-source heterogeneous data generated during the construction of transmission line equipment based on the distributed Kalman filter fusion method;

[0235] Further, the digital twin data fusion module 2 performs data fusion according to the following steps:

[0236] S21. Represent the state of each data node in the multi-source heterogeneous data through an information pair;

[0237] The information pair includes an information matrix and an information vector; the expression of the information matrix is as follows:

[0238]

[0239]

[0240] Where: is the information matrix of the data node, is the estimated covariance at time k;

[0241] The expression of the information vector is as follows:

[0242]

[0243] Where: is the information vector of the data node, is the state estimate value at time k;

[0244] S22. Add an intermediate node i between every two adjacent data nodes, and introduce a binary variable and a covariance value u as a threshold during the data exchange process of each data node to determine whether to trigger data exchange;

[0245] The intermediate node i sends its own estimated covariance to the adjacent data nodes, fuses the estimated covariance with that of the adjacent data nodes, and obtains a fused information pair Then, the fusion of multi-source heterogeneous data is completed through iterative loops; the expression of the fused information pair is as follows:

[0246]

[0247] Wherein: is the information vector of the data node after fusion, is the information matrix of the data node after fusion;

[0248] The expression of the trigger mechanism is as follows:

[0249]

[0250] If then trigger data exchange,

[0251] If then do not trigger data exchange,

[0252] Wherein: is the demand covariance of data node i at time k, is the updated value of data node i at time k, indicates that data node i communicates with adjacent data nodes for data exchange at time k, indicates that data node i does not communicate with adjacent data nodes for data exchange at time k.

[0253] The digital twin model construction module 3 is used to construct a functional model by IDEF0, a network workflow model by IDEF3, an ontology model by IDEF5, a data information model by modular IDEF1X, and integrate subsystems to jointly construct a digital twin comprehensive model;

[0254] Furthermore, the digital twin model construction module 3 constructs a digital twin model according to the following steps:

[0255] S31. For the construction equipment in the subsystem, establish a DSM matrix; both the rows and columns in the matrix represent construction equipment;

[0256] S32. Based on the genetic algorithm, modularize the DSM matrix to obtain DSM modules;

[0257] S33. In the DSM module, classify the information of construction equipment with the same attributes in the same module into the same set, represent it by a box in IDEF1X, and then couple through the connection of construction equipment in the same module to establish the connection between the same modules, completing the construction of the data information model;

[0258] The step S32 specifically includes:

[0259] S321. Define a fitness function to partition the DSM matrix; the expression of the fitness function is as follows:

[0260] max(F) = I / E;

[0261] Where: max(F) is the maximization of the value of the fitness function F, I is the average cohesion of the DSM module, and E is the average coupling of the DSM module;

[0262]

[0263] Where: N all is the total number of DSM modules, I m is the cohesion of module m, N m is the number of elements in module m, C m (i, j) is the connection size between element i and element j in module m, C max is the maximum value of the connection size between elements, E p，q is the coupling degree between module p and module q;

[0264] S322. Map the DSM module partitioning result to a two-dimensional coding matrix and perform operations as the chromosome of the genetic algorithm;

[0265] S323. Based on the chromosome, perform population initialization and crossover mutation operations in sequence, and repeatedly perform iterative operations to determine the optimal population of the genetic algorithm, and use the optimal population as the best partitioning scheme of the DSM module.

[0266] The digital twin model update module 4 is used to simulate the construction stage of the transmission line equipment based on the digital twin comprehensive model, and judge whether to update the parameters of the digital twin comprehensive model based on the Euclidean distance between the simulation result and the physical entity status monitoring data; if so, use the variational Bayesian model based on the Euclidean distance to update the parameters of the digital twin comprehensive model;

[0267] Furthermore, the digital twin model update module 4 updates the parameters of the model according to the following steps:

[0268] S44. Construct the input of Bayesian inference based on the likelihood function and the prior distribution;

[0269] The real-time status data of the physical entity is X = [x1, x2,..., x n , the simulation data of the digital twin comprehensive model is X(θ i ) = [x1(θ), x2(θ i ),..., x n (θ i )], and the relationship between the real-time status data X and the simulation data X(θ i ) is X = X(θ i ) + ε; where: ε = [ε1, ε2,... εm is an m-dimensional vector of errors;

[0270] Then the expression of the likelihood function P(x|θ) is as follows:

[0271]

[0272] where: μ i , σ i are the mean and standard deviation of the error respectively;

[0273] S45. Describe the change process of the physical entity parameters using the beta distribution; the expression of the beta distribution is as follows:

[0274] f(θ; α, β) = θ α-1 (1 - θ) β-1 ;

[0275] where: θ is the parameter to be updated, and both α and β are parameters greater than zero;

[0276] S46. Calculate the values of α and β according to the prior distribution change data of the parameters; the calculation formula is as follows:

[0277]

[0278] where: is the parameter mean, and S θ is the parameter standard deviation;

[0279] Then the output of Bayesian inference is expressed as follows:

[0280]

[0281] where: p(θ|x) is the posterior distribution of the parameter θ, that is, the parameter of the updated digital twin comprehensive model; p(x|θ) is the likelihood function, representing the probability of observing the data x when the parameter takes the value of θ; p(θ) is the prior distribution of the parameter θ, the same as the prior probability distribution

[0282] S47. Approximate the posterior distribution p(θ|x) as a simple distribution q(θ) = q φ (θ);

[0283] S48. Quantify the mismatch between the posterior distribution p(θ|x) and the simple distribution q φ (θ) using the KL divergence; the expression of the KL divergence is as follows:

[0284]

[0285] where: E φ is the expectation of the logarithmic function in the parentheses;

[0286] S49. Transform the problem of solving the posterior distribution into the problem of solving the maximum value of the evidence lower bound ELBO, and confirm the objective function to complete the parameter update of the digital twin comprehensive model;

[0287] The expression of the evidence lower bound ELBO is as follows:

[0288]

[0289] Among them: p(x|θ)p(θ) is the joint distribution of θ and x;

[0290] The objective function is as follows:

[0291] F(θ) = max[ELBO];

[0292] Among them: F(θ) is the initial value of the objective function.

[0293] Embodiment 3:

[0294] See Figure 8 , a device for constructing a digital twin model of transmission line equipment based on improved IDEF, the device includes a processor 5 and a memory 6;

[0295] The memory 6 is used to store the computer program code 61 and transmit the computer program code 61 to the processor 5;

[0296] The processor 5 is used to execute the method for constructing a digital twin model of transmission line equipment based on improved IDEF described in Embodiment 1 according to the instructions in the computer program code 61.

[0297] In this embodiment, there is also a computer-readable storage medium, and computer-executable instructions are stored in the computer-readable storage medium. When the computer-executable instructions are executed on a computer, the method for constructing a digital twin model of transmission line equipment based on improved IDEF described in Embodiment 1 is implemented.

[0298] Although the embodiments of the present invention have been shown and described above, it should be understood that the above embodiments are exemplary and cannot be construed as limiting the present invention. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for constructing a digital twin model of power transmission line equipment based on improved IDEF, characterized in that: include: S1. Divide the construction phases of the transmission line equipment into the foundation construction phase, the tower assembly construction phase, and the line stringing construction phase, and establish respective subsystems for the above three phases; the subsystems include a three-dimensional static model subsystem, a three-dimensional visualization monitoring subsystem, and a simulation prediction subsystem; S2. Add intermediate nodes between adjacent data nodes in the subsystem, set a trigger mechanism for data exchange between adjacent data nodes, and fuse multi-source heterogeneous data generated in the construction of transmission line equipment based on a distributed Kalman filter fusion method; S3. Use IDEF0 to build the functional model of the subsystem, use IDEF3 to build the network workflow model of the subsystem, use IDEF5 to build the ontology model of the subsystem, and use modular IDEF1X to build the data information model of the subsystem. Associate and map the functional model, network workflow model, ontology model, data information model with the subsystem to jointly build a digital twin comprehensive model. The data information model of the subsystem constructed by modular IDEF1X specifically includes: S31, establishing a DSM matrix; the rows and columns in the matrix represent the construction equipment in the construction stage; S32, based on the genetic algorithm, modularize the DSM matrix to obtain a DSM module; S33, in the DSM module, the information of the construction equipment with the same attributes in the same module is classified into the same set, which is represented by a box in IDEF1X, and then coupled through the connection of the construction equipment in the same module, the connection between the same modules is established, and the construction of the data information model is completed; S4. Based on the digital twin comprehensive model and the integrated multi-source heterogeneous data, simulate the construction stage of the transmission line equipment, and determine whether to update the parameters of the digital twin comprehensive model based on the Euclidean distance between the simulation results and the physical entity status monitoring data; if so, use the variational Bayesian model based on the Euclidean distance to update the parameters of the digital twin comprehensive model; The determination of whether to update the parameters of the digital twin comprehensive model specifically includes: S41, comparing the Euclidean distance between the simulation results of the digital twin comprehensive model and the physical entity state monitoring data to obtain the Euclidean distance of the historical data set; The simulation result is X=(x1, x2, x3...x n ), the state monitoring data is Y=(y1, y2, y3...y n ); The Euclidean distance between the simulation result and the state monitoring data is as follows: S42, setting the maximum Euclidean distance As as the update trigger threshold, then selecting new data samples from multi-source heterogeneous data as new sample sets for simulation, and comparing the simulation results of the new sample sets with the historical state monitoring data to obtain the Euclidean distance of the new sample sets; S43, comparing the Euclidean distance of the historical data set with the Euclidean distance of the new sample set; If the Euclidean distance of the historical data set and the Euclidean distance of the new sample set are both greater than the maximum Euclidean distance As, the parameters of the digital twin comprehensive model are updated.

2. According to a method for constructing a digital twin model of power transmission line equipment based on improved IDEF according to claim 1, it is characterized by: The step S2 specifically includes: S21, representing the state of each data node in the multi-source heterogeneous data through information pairs; The information pair includes an information matrix and an information vector; the expression of the information matrix is ​​as follows: in: is the information matrix of the data node, is the estimated covariance at time k; The expression of the information vector is as follows: in: is the information vector of the data node, is the estimated value of the state at time k; S22. Add an intermediate node i between two adjacent data nodes, and introduce a binary variable for each data node during the data exchange process of each data node. The covariance value u is used as a threshold to determine whether to trigger data exchange; The expression of the trigger mechanism is as follows: like The data exchange is triggered. like No data exchange is triggered. in: is the demand covariance of data node i at time k, is the updated value of data node i at time k, Indicates that data node i communicates with adjacent data nodes to exchange data at time k. Indicates that data node i does not communicate with adjacent data nodes for data exchange at time k; S23, based on the distributed Kalman filter fusion method, the intermediate node i sends its own estimated covariance to the adjacent data nodes, and performs data fusion with the estimated covariance of the adjacent data nodes to obtain the fusion information pair The fusion of multi-source heterogeneous data is completed through loop iteration; the expression of the fusion information pair is as follows: in: is the information vector of the fused data node, is the information matrix of the fused data nodes.

3. The method for constructing a digital twin model of power transmission line equipment based on improved IDEF according to claim 2, characterized in that: The step S32 specifically includes: S321, define a fitness function and divide the DSM matrix into modules; the fitness function is expressed as follows: max(F) = I / E; Where: max(F) is the maximization of the value of the fitness function F, I is the average cohesion of the DSM module, and E is the average coupling of the DSM module; Where: N all is the total number of DSM modules, I m is the cohesion of module m, N m is the number of elements in module m, C m (i, j) is the size of the connection between element i and element j in module m, C max is the maximum value of the connection size between elements, E p，q is the coupling degree between module p and module q; S322, based on the genetic algorithm, mapping the DSM module division result to a two-dimensional coding matrix as a chromosome of the genetic algorithm to participate in the operation; S323, performing population initialization and crossover mutation operations on chromosomes in sequence, and repeatedly iterating operations to determine the optimal population of the genetic algorithm, and using the optimal population as the best partitioning scheme for the DSM module.

4. The method for constructing a digital twin model of power transmission line equipment based on improved IDEF according to claim 1, characterized in that: In step S4, updating the parameters of the digital twin comprehensive model specifically includes: S44, constructing the input of Bayesian inference based on likelihood function and prior distribution; Assume that the real-time state data of the physical entity is X = [x1, x2, ..., x n ], the simulation data of the digital twin comprehensive model is X(θ i )=[x1(θ i ),x2(θ i ),...,x n (θ i )], real-time state data X and simulation data X(θ i ) is X=X(θ i )+ε; where: ε=[ε1, ε2,...,ε m ] is the m-dimensional vector of error; Then the expression of the likelihood function P(x|θ) is as follows: Where: μ i , σ i are the mean and standard deviation of the errors respectively; S45. Beta distribution is used to describe the change process of physical entity parameters; the expression of Beta distribution is as follows: f(θ;α,β)=θ a-1 (1-θ) β-1 ; Where: θ is the physical entity parameter to be updated, α and β are both physical entity parameters greater than zero; S46. Calculate the values ​​of α and β according to the prior distribution change data of the physical entity parameters; the calculation formula is as follows: in: is the mean value of the physical entity parameter, s θ is the parameter standard deviation; The output of Bayesian inference is as follows: Among them: P(θ|x) is the posterior distribution of θ, that is, the parameters of the updated digital twin comprehensive model; p(x|θ) is the likelihood function, which represents the probability of observing data x when the parameter value is θ; p(θ) is the prior distribution of θ, which is the same as the prior probability distribution S47. Approximate the posterior distribution p(θ|x) to a simple distribution q(θ)=q φ (θ); S48. Use KL divergence to quantify the posterior distribution p(θ|x) and the simple distribution q φ The mismatch between (θ); the expression of the KL divergence is as follows: Where: E φ is the expectation of the logarithmic function in brackets; S49, transform the problem of solving the posterior distribution into the problem of solving the ELBO maximum value of the evidence lower bound, confirm the objective function, and complete the parameter update of the digital twin comprehensive model; The expression of the evidence lower bound ELBO is as follows: Where: p(x|θ)p(θ) is the joint distribution of θ and x; The objective function is as follows: F(θ)=max[ELBO]; Where: F(θ) is the initial value of the objective function.

5. A digital twin model construction system for power transmission line equipment based on improved IDEF, characterized in that: The system comprises: A digital twin subsystem construction module (1) is used to divide the construction phase of the power transmission line equipment into a foundation construction phase, a tower assembly construction phase, and a line stringing construction phase, and to establish respective subsystems for the above three phases; the subsystems include a three-dimensional static model subsystem, a three-dimensional visualization monitoring subsystem, and a simulation prediction subsystem; A digital twin data fusion module (2) is used to add intermediate nodes between adjacent data nodes in the subsystem, set a trigger mechanism for data exchange between adjacent data nodes, and fuse multi-source heterogeneous data generated during the construction of transmission line equipment based on a distributed Kalman filter fusion method; A digital twin model construction module (3) is used to construct a functional model of the subsystem using IDEF0, a network workflow model of the subsystem using IDEF3, an ontology model of the subsystem using IDEF5, and a data information model of the subsystem using modular IDEF1X, and associate and map the functional model, network workflow model, ontology model, data information model and subsystem to jointly construct a digital twin comprehensive model; The digital twin model construction module (3) constructs the data information model of the subsystem according to the following steps: S31, establishing a DSM matrix; the rows and columns in the matrix represent the construction equipment in the construction stage; S32, based on the genetic algorithm, modularize the DSM matrix to obtain a DSM module; S33, in the DSM module, the information of the construction equipment with the same attributes in the same module is classified into the same set, which is represented by a box in IDEF1X, and then coupled through the connection of the construction equipment in the same module, the connection between the same modules is established, and the construction of the data information model is completed; The digital twin model update module (4) is used to simulate the construction phase of the transmission line equipment based on the digital twin comprehensive model and the fused multi-source heterogeneous data, and determine whether to update the parameters of the digital twin comprehensive model based on the Euclidean distance between the simulation results and the physical entity state monitoring data; if so, the variational Bayesian model based on the Euclidean distance is used to update the parameters of the digital twin comprehensive model; The digital twin model updating module (4) determines whether to update the parameters of the model according to the following steps: S41, comparing the Euclidean distance between the simulation results of the digital twin comprehensive model and the physical entity state monitoring data to obtain the Euclidean distance of the historical data set; The simulation result is X=(x1, x2, x3...x n ), the state monitoring data is Y=(y1, y2, y3...y n ); The Euclidean distance between the simulation result and the state monitoring data is as follows: S42, setting the maximum Euclidean distance As as the update trigger threshold, then selecting new data samples from multi-source heterogeneous data as new sample sets for simulation, and comparing the simulation results of the new sample sets with the historical state monitoring data to obtain the Euclidean distance of the new sample sets; S43, comparing the Euclidean distance of the historical data set with the Euclidean distance of the new sample set; If the Euclidean distance of the historical data set and the Euclidean distance of the new sample set are both greater than the maximum Euclidean distance As, the parameters of the digital twin comprehensive model are updated.

6. The digital twin model construction system for power transmission line equipment based on improved IDEF according to claim 5 is characterized by: The digital twin data fusion module (2) performs data fusion according to the following steps: S21, representing the state of each data node in the multi-source heterogeneous data through information pairs; The information pair includes an information matrix and an information vector; the expression of the information matrix is ​​as follows: in: is the information matrix of the data node, is the estimated covariance at time k; The expression of the information vector is as follows: in: is the information vector of the data node, is the estimated value of the state at time k; S22. Add an intermediate node i between two adjacent data nodes, and introduce a binary variable for each data node during the data exchange process of each data node. The covariance value u is used as a threshold to determine whether to trigger data exchange; The expression of the trigger mechanism is as follows: like The data exchange is triggered. like No data exchange is triggered. in: is the demand covariance of data node i at time k, is the updated value of data node i at time k, Indicates that data node i communicates with adjacent data nodes to exchange data at time k. Indicates that data node i does not communicate with adjacent data nodes for data exchange at time k; S23, based on the distributed Kalman filter fusion method, the intermediate node i sends its own estimated covariance to the adjacent data nodes, and performs data fusion with the estimated covariance of the adjacent data nodes to obtain the fusion information pair The fusion of multi-source heterogeneous data is completed through loop iteration; the expression of the fusion information pair is as follows: in: is the information vector of the fused data node, is the information matrix of the fused data nodes.

7. The digital twin model construction system for power transmission line equipment based on improved IDEF according to claim 6 is characterized by: The digital twin model construction module (3) constructs the DSM module according to the following steps: The step S32 specifically includes: S321, define a fitness function and divide the DSM matrix into modules; the fitness function is expressed as follows: max(F) = I / E; Where: max(F) is the maximization of the value of the fitness function F, I is the average cohesion of the DSM module, and E is the average coupling of the DSM module; Where: N all is the total number of DSM modules, I m is the cohesion of module m, N m is the number of elements in module m, C m (i, j) is the size of the connection between element i and element j in module m, C max is the maximum value of the connection size between elements, E p,q is the coupling degree between module p and module q; S322, based on the genetic algorithm, mapping the DSM module division result to a two-dimensional coding matrix as a chromosome of the genetic algorithm to participate in the operation; S323, performing population initialization and crossover mutation operations on chromosomes in sequence, and repeatedly iterating operations to determine the optimal population of the genetic algorithm, and using the optimal population as the best partitioning scheme for the DSM module.

8. The digital twin model construction system for power transmission line equipment based on improved IDEF according to claim 7 is characterized in that: The digital twin model updating module (4) updates the parameters of the model according to the following steps: S44, constructing the input of Bayesian inference based on likelihood function and prior distribution; Assume that the real-time state data of the physical entity is X = [x1, x2, ..., x n ], the simulation data of the digital twin comprehensive model is X(θ i )=[x1(θ i ), x2(θ i ), ..., x n (θ i )], real-time state data X and simulation data X(θ i ) is X=X(θ i )+ε; where: ε=[ε1, ε2,...,ε m ] is the m-dimensional vector of error; Then the expression of the likelihood function P(x|θ) is as follows: Where: μ i , σ i are the mean and standard deviation of the errors respectively; S45. Beta distribution is used to describe the change process of physical entity parameters; the expression of Beta distribution is as follows: f(θ;α,β)=θ α-1 (1-θ) β-1 ; Where: θ is the physical entity parameter to be updated, α and β are both physical entity parameters greater than zero; S46. Calculate the values ​​of α and β according to the prior distribution change data of the physical entity parameters; the calculation formula is as follows: in: is the mean value of the physical entity parameter, s θ is the parameter standard deviation; The output of Bayesian inference is as follows: Among them: p(θ|x) is the posterior distribution of θ, that is, the parameters of the updated digital twin comprehensive model; p(x|θ) is the likelihood function, which represents the probability of observing data x when the parameter value is θ; p(θ) is the prior distribution of θ, which is the same as the prior probability distribution S47. Approximate the posterior distribution p(θ|x) to a simple distribution q(θ)=q φ (θ); S48. Use KL divergence to quantify the posterior distribution p(θ|x) and the simple distribution q φ The mismatch between (θ); the expression of the KL divergence is as follows: Where: E φ is the expectation of the logarithmic function in brackets; S49, transform the problem of solving the posterior distribution into the problem of solving the ELBO maximum value of the evidence lower bound, confirm the objective function, and complete the parameter update of the digital twin comprehensive model; The expression of the evidence lower bound ELBO is as follows: Where: p(x|θ)p(θ) is the joint distribution of θ and x; The objective function is as follows: F(θ)=max[ELBO]; Where: F(θ) is the initial value of the objective function.

Images