A method for predicting the state of high-voltage switchgear based on digital model

Abstract

The present invention discloses a method for predicting the state of a high-voltage switch cabinet based on a digital model, and relates to the technical field of switch cabinets. The method comprises constructing a digital model of a switch cabinet coupled with multiple physical fields and a digital twin model of the switch cabinet; dynamically deducing the performance of key components of the switch cabinet according to the digital twin model of the switch cabinet to obtain an overall dynamic twin model of the switch cabinet; obtaining measurable parameters of the switch cabinet, importing the measurable parameters into the overall dynamic twin model of the switch cabinet, and obtaining various working conditions of the switch cabinet through the overall dynamic twin model of the switch cabinet; establishing a state evaluation method of multi-parameter fusion, and evaluating the state of the switch cabinet through the state evaluation method of multi-parameter fusion; and establishing a switch cabinet state prediction method based on various parameters and historical data of the digital twin. The present invention can evaluate and predict the state of the switch cabinet according to the real-time data of the switch cabinet, and solves the disadvantage that there is no existing state prediction method of a high-voltage switch cabinet based on a digital model.

Description

Technical Field

[0001] The present invention belongs to the technical field of switch cabinets, and in particular relates to a high-voltage switch cabinet state prediction method based on a digital model. Background Art

[0002] The main function of the switchgear is to open, close, control and protect electrical equipment in the process of power generation, transmission, distribution and power conversion in the power system. The components in the switchgear mainly include circuit breakers, disconnectors, load switches, operating mechanisms, transformers and various protective devices. There are many ways to classify switchgear. For example, according to the installation method of the circuit breaker, it can be divided into removable switchgear and fixed switchgear; or according to the different cabinet structures, it can be divided into open switchgear, metal-enclosed switchgear, and metal-enclosed armored switchgear; according to the voltage level, it can be divided into high-voltage switchgear, medium-voltage switchgear and low-voltage switchgear. It is mainly suitable for various occasions such as power plants, substations, petrochemicals, metallurgy and steel rolling, light industry and textiles, factories and mines, residential areas, high-rise buildings, etc.

[0003] In recent years, switchgear accidents have occurred from time to time, and once they occur, they will cause economic losses, casualties and other adverse social impacts. Its accident hazards and inherent defects are mainly concentrated in the wiring method, internal arc release capacity, internal insulation, heating and anti-mislocking, etc., which can be predicted and judged in advance to reduce the occurrence of accidents. However, there is currently no high-voltage switchgear state prediction method based on digital models. Therefore, a high-voltage switchgear state prediction method based on digital models is needed. Summary of the invention

[0004] The purpose of the present invention is to provide a high-voltage switch cabinet state prediction method based on a digital model, thereby overcoming the disadvantage that there is no existing high-voltage switch cabinet state prediction method based on a digital model.

[0005] To achieve the above object, the present invention provides a method for predicting the state of a high-voltage switch cabinet based on a digital model, comprising the following steps:

[0006] Constructing a digital model of a switch cabinet coupled with multiple physical fields, and establishing a digital twin model of the switch cabinet according to the digital model of the switch cabinet;

[0007] Dynamically deriving the performance of key components of the switch cabinet according to the switch cabinet digital twin model to obtain an overall dynamic twin model of the switch cabinet;

[0008] Obtain measurable parameters of the switch cabinet, import the measurable parameters into the overall dynamic twin model of the switch cabinet, and obtain various operating conditions of the switch cabinet through the overall dynamic twin model of the switch cabinet; establish a multi-parameter fusion state evaluation method based on the sensitivity relationship between the measurable parameters of the switch cabinet and various operating conditions of the switch cabinet, and evaluate the state of the switch cabinet through the multi-parameter fusion state evaluation method;

[0009] The constantly changing simulated working conditions are input into the overall dynamic twin model of the switchgear to obtain the characteristic parameter change characteristics of different components of the switchgear equipment under different modes; according to the characteristic parameter change characteristics, a switchgear state prediction method based on various parameters and historical data of the digital twin is established.

[0010] Preferably, the real-time characteristic parameters of the physical sensor are acquired, and the state evaluation method of parameter fusion is used to evaluate the state of the switch cabinet.

[0011] Preferably, the real-time characteristic parameters of the physical sensor are obtained, and the state of the switch cabinet is predicted by combining the real-time characteristic parameters with the switch cabinet state prediction method.

[0012] Preferably, dynamically deducing the performance of key components of the switch cabinet according to the switch cabinet digital twin model comprises the following steps:

[0013] Establish the correlation between typical faults and status parameters of the switch cabinet;

[0014] Establishing a fault model of the digital twin of the switch cabinet through the digital twin model of the switch cabinet;

[0015] Performing fault simulation on the fault model to obtain fault model detection data, and analyzing the characteristics of the fault type based on the fault model detection data, real-time detection data, characteristic parameters and historical data;

[0016] Combined with the analysis results, the association relationship between different fault modes and characteristic parameters of the switch cabinet is mined using the data mining algorithm based on association rules.

[0017] According to the correlation between different failure modes and characteristic parameters of the switch cabinet, the state parameters under the simulated working conditions are queried and imported into the twin model for calculation, so as to deduce the change law of component performance under the working condition.

[0018] Preferably, the multi-parameter fusion state assessment method comprises the following steps:

[0019] Establish an evaluation index system for switchgear operation status;

[0020] Acquire multiple characteristic parameters from the measurable parameters through an extraction network, wherein the characteristic parameters are parameters that can reflect the state of the switch cabinet;

[0021] The weights of characteristic parameters are calculated using the analytic hierarchy process;

[0022] Establish the relationship between multiple characteristic parameters and switchgear status, and establish a switchgear operation status evaluation model based on the weights of the characteristic parameters;

[0023] Real-time measurable parameters are obtained, and the real-time measurable parameters are input into the switch cabinet operation status evaluation model, and the switch cabinet operation status evaluation model outputs the real-time status of the switch cabinet.

[0024] Preferably, the switch cabinet state prediction method based on various parameters and historical data of the digital twin comprises the following steps:

[0025] According to the constantly changing simulated working conditions, the overall dynamic twin model of the switch cabinet is input to obtain the characteristic parameter change characteristics of different components of the switch cabinet equipment in different modes, and the long short-term memory network LSTM is established as the basic network according to the transformation relationship between the working conditions and the characteristic quantities;

[0026] Acquire real-time parameters of measurable parameters and input them into a basic network, and the basic network outputs a switch cabinet state prediction, thereby training optimal model parameters of the basic network;

[0027] The switch cabinet state prediction model is obtained according to the optimal model parameters.

[0028] Preferably, the switch cabinet digital twin model is constructed by adopting a switch cabinet modeling method based on digital twin and error adaptive optimization.

[0029] Preferably, the virtual sensor setting of measurable parameters in the digital twin model of the switch cabinet includes: constructing signal strength, signal delay and signal measurement error evaluation indicators of virtual sensors at different positions through numerical optimization methods, and conducting a comprehensive evaluation in combination with the detection range of the virtual sensors to determine the optimal position distribution of the virtual sensors.

[0030] Compared with the existing technology, the present invention has the following beneficial effects:

[0031] The digital model-based high-voltage switchgear state prediction method provided by the present invention constructs a digital model of the switchgear coupled with multiple physical fields, and establishes a switchgear digital twin model according to the digital model of the switchgear; dynamically deduce the performance of key components of the switchgear according to the digital twin model of the switchgear to obtain an overall dynamic twin model of the switchgear; obtain measurable parameters of the switchgear, import the measurable parameters into the overall dynamic twin model of the switchgear, and obtain various operating conditions of the switchgear through the overall dynamic twin model of the switchgear; establish a multi-parameter fusion state evaluation method based on the sensitivity relationship between the measurable parameters of the switchgear and the various operating conditions of the switchgear, and evaluate the state of the switchgear through the multi-parameter fusion state evaluation method; input the constantly changing simulated working conditions into the overall dynamic twin model of the switchgear to obtain the characteristic parameter change characteristics of different components of the switchgear equipment under different modes; and establish a switchgear state prediction method based on various parameters and historical data of the digital twin according to the characteristic parameter change characteristics. That is, the present invention can evaluate and predict the state of the switch cabinet according to the real-time data of the switch cabinet, which solves the shortcoming that there is no existing high-voltage switch cabinet state prediction method based on digital model. The switch cabinet can be inspected and maintained in advance according to the prediction results, the number of accidents of switch cabinets and ring network cabinets can be reduced, and the reliability of power grid operation can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solution of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only one embodiment of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0033] Figure 1 It is a flow chart of a method for predicting a high-voltage switch cabinet state based on a digital model of the present invention;

[0034] Figure 2 is a flow chart of a switch cabinet modeling method based on digital twin and error adaptive optimization of the present invention;

[0035] Figure 3 is a flow chart of the inner loop adaptive calculation error control method of the present invention;

[0036] Figure 4 It is a flow chart of the outer loop adaptive dynamic excitation control method of the present invention. DETAILED DESCRIPTION

[0037] The following is a clear and complete description of the technical solutions in the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0038] like Figure 1 As shown, the high-voltage switch cabinet state prediction method based on the digital model provided by the present invention comprises the following steps:

[0039] S1. constructing a digital model of a switch cabinet coupled with multiple physical fields, and establishing a digital twin model of the switch cabinet according to the digital model of the switch cabinet;

[0040] S2. Dynamically deduce the performance of key components of the switch cabinet according to the switch cabinet digital twin model to obtain an overall dynamic twin model of the switch cabinet;

[0041] S3. Obtain measurable parameters of the switch cabinet, import the measurable parameters into the overall dynamic twin model of the switch cabinet, and obtain various operating conditions of the switch cabinet through the overall dynamic twin model of the switch cabinet; establish a multi-parameter fusion state evaluation method according to the sensitivity relationship between the measurable parameters of the switch cabinet and the various operating conditions of the switch cabinet, and evaluate the state of the switch cabinet through the multi-parameter fusion state evaluation method;

[0042] S4. The constantly changing simulated working conditions are input into the overall dynamic twin model of the switch cabinet to obtain the characteristic parameter change characteristics of different components of the switch cabinet equipment under different modes; according to the characteristic parameter change characteristics, a switch cabinet state prediction method based on various parameters and historical data of the digital twin is established.

[0043] The above-mentioned high-voltage switchgear state prediction method based on digital model constructs a digital model of the switchgear coupled with multiple physical fields, and establishes a switchgear digital twin model according to the digital model of the switchgear; dynamically deduce the performance of key components of the switchgear according to the digital twin model of the switchgear to obtain the overall dynamic twin model of the switchgear; obtain the measurable parameters of the switchgear, import the measurable parameters into the overall dynamic twin model of the switchgear, and obtain various operating conditions of the switchgear through the overall dynamic twin model of the switchgear; establish a multi-parameter fusion state evaluation method based on the sensitivity relationship between the measurable parameters of the switchgear and the various operating conditions of the switchgear, and evaluate the state of the switchgear through the multi-parameter fusion state evaluation method; input the constantly changing simulated working conditions into the overall dynamic twin model of the switchgear to obtain the characteristic parameter change characteristics of different components of the switchgear equipment under different modes; according to the characteristic parameter change characteristics, establish a switchgear state prediction method based on various parameters and historical data of the digital twin. That is, the present invention can evaluate and predict the status of the switch cabinet according to the real-time data of the switch cabinet, which solves the shortcoming that there is no existing high-voltage switch cabinet status prediction method based on digital model. The switch cabinet can be inspected and maintained in advance according to the prediction results, the number of accidents of switch cabinets and ring network cabinets can be reduced, and the reliability of power grid operation can be improved.

[0044] One of the embodiments further includes step S5, acquiring real-time characteristic parameters of the physical sensor, and using the parameter fusion state assessment method to assess the state of the switch cabinet.

[0045] One of the embodiments further includes step S6, obtaining real-time characteristic parameters of the physical sensor, and using the switch cabinet state prediction method in combination with the real-time characteristic parameters to predict the state of the switch cabinet.

[0046] In one embodiment, in step S2, dynamically deducing the performance of key components of the switch cabinet according to the switch cabinet digital twin model includes the following steps:

[0047] S21, establishing a correlation relationship between typical faults of the switch cabinet and state parameters;

[0048] S22. Establishing a fault model of the digital twin of the switch cabinet through the digital twin model of the switch cabinet;

[0049] Specifically, discharge sources and hot spots are set inside the digital twin model of the switch cabinet to simulate faults such as short circuit and overheating of the switch cabinet equipment; according to the virtual sensor position distribution optimization strategy, virtual sensors for detecting various parameter signals are placed inside the digital twin model of the switch cabinet. The advantage of using virtual sensors is that internal parameter signals can be fully and accurately detected;

[0050] S23, performing fault simulation on the fault model to obtain fault model detection data, and analyzing the characteristics of the fault type based on the fault model detection data, real-time detection data, characteristic parameters and historical data;

[0051] Specifically, the data detected by the virtual sensor is combined with the detection data of the physical sensor, the operation data of the switch cabinet, the characteristic parameters and the historical data to study and analyze the change characteristics of the internal parameter signals of the switch cabinet equipment under different fault modes. Therefore, since a specific fault type may only be directly related to some characteristic parameters, and the correlation with other characteristic parameters is not obvious;

[0052] S24. Combined with the analysis results of step S23, in the process of dynamic deduction, too much input data will increase the complexity of the model, reduce the efficiency of the algorithm, and even cause a decrease in prediction accuracy. In order to reduce the complexity of state deduction caused by the increase in the dimension of the feature quantity, the data mining algorithm based on association rules such as the Apriori method is used to mine the association relationship between different fault modes of the switch cabinet and the characteristic parameters;

[0053] S25. According to the correlation between different failure modes and characteristic parameters of the switch cabinet, query the state parameters under the simulated working conditions, import them into the twin model for calculation, and deduce the change law of component performance under the working condition. The core of dynamic deduction model calculation lies in establishing the constitutive model of the material and dynamic iterative calculation. The constitutive model of the material should reflect the fatigue of the material at the current moment, the degradation law of the insulation and mechanical properties of the current material, etc. The dynamic iterative algorithm should take into account the degradation law of the material, and adopt nonlinear expansion, dynamic convergence criteria, adaptive time discretization and other methods for numerical calculation, so as to realize the dynamic deduction of the performance of key components.

[0054] In one embodiment, in step S3, the multi-parameter fusion state assessment method comprises the following steps:

[0055] S31. Establish an evaluation index system for the switchgear operation status;

[0056] Among them, the switchgear operating status evaluation index system includes characteristic parameters of the switchgear and predicted values ​​of the characteristic parameters.

[0057] S32, acquiring a plurality of characteristic parameters from the measurable parameters through an extraction network, wherein the characteristic parameters are parameters that can reflect the state of the switch cabinet, the data of the measurable parameters include: virtual sensor data, physical sensor data, operation data and historical data, and the types of measurable parameters include: input voltage, output voltage, input current, output current, partial discharge current, temperature and humidity, etc.;

[0058] Step S32 also includes fusing multiple feature parameters, which specifically includes the following steps:

[0059] S321, according to the type of the measurable parameter, the data of the measurable parameter is divided into: voltage data, current data and temperature data, and the data of the measurable parameter is preprocessed to unify the data;

[0060] S322, extracting the voltage data of the fault from the voltage data as a voltage feature, extracting the current data of the fault from the current data as a current feature, and extracting the temperature data of the fault from the temperature data as a temperature feature;

[0061] S323, inputting the voltage feature, the current feature, and the temperature feature into a fusion analysis method for fusion to obtain a fusion feature, wherein the fusion analysis method uses an XD-Markov machine to generate a state transition probability association matrix for different types of data;

[0062] S33, calculating the weight of characteristic parameters using the analytic hierarchy process;

[0063] S34, establishing the relationship between multiple characteristic parameters and the switch cabinet state, and establishing a switch cabinet operation state evaluation model in combination with the weights of the characteristic parameters;

[0064] Specifically, establishing a switch cabinet operation status evaluation model includes the following steps:

[0065] S341. Determine the network model to be used. The switch cabinet operation status assessment model is constructed using a residual neural network.

[0066] S342, using multiple characteristic parameters as inputs of the network model; establishing relationships between the multiple characteristic parameters and the switch cabinet state according to the weights of the characteristic parameters, and setting the relationships between the multiple characteristic parameters and the switch cabinet state as the hidden layer of the network model; the relationship between the switch cabinet state is the output of the network model;

[0067] S343. Use the pre-processed measurable parameters to train and verify the network model information to obtain a switch cabinet operation status evaluation model.

[0068] S35. Obtain real-time measurable parameters and input the real-time measurable parameters into the switch cabinet operation status evaluation model. The switch cabinet operation status evaluation model outputs the real-time status of the switch cabinet. The status of the switch cabinet includes: normal, performance degradation, abnormal fault, and abnormal fault can be further subdivided according to the severity. Of course, the switch cabinet operation status evaluation model also needs to be reset at this time.

[0069] In one embodiment, in step S4, the switch cabinet state prediction method based on various parameters and historical data of the digital twin includes the following steps:

[0070] S41. Input the overall dynamic twin model of the switch cabinet according to the constantly changing simulated working conditions, obtain the characteristic parameter change characteristics of different components of the switch cabinet equipment in different modes, and establish a long short-term memory network LSTM as the basic network according to the transformation relationship between the working conditions and the characteristic quantities;

[0071] S42, obtaining real-time parameters of measurable parameters and inputting them into a basic network, and the basic network outputs a switch cabinet state prediction, thereby training optimal model parameters of the basic network;

[0072] S43. Obtain a switch cabinet state prediction model according to the optimal model parameters.

[0073] In one of the embodiments, the virtual sensor setting of the measurable parameters in the digital twin model of the switch cabinet includes: constructing evaluation indicators such as signal strength, signal delay, and signal measurement error of virtual sensors at different positions through numerical optimization methods, and conducting a comprehensive evaluation in combination with the detection range of the virtual sensors to determine the optimal position distribution of the virtual sensors.

[0074] In one of the embodiments, a switch cabinet modeling method based on digital twin and error adaptive optimization is adopted to construct the switch cabinet digital twin model.

[0075] Specifically, Figure 2 As shown in FIG, the switch cabinet modeling method based on digital twin and error adaptive optimization includes the following steps:

[0076] S11. Acquire physical parameters of the internal insulation structure of the switch cabinet, construct a digital model of the switch cabinet coupled with multiple physical fields according to the physical parameters, and establish a connection relationship between the digital model and the connection point of the digital twin;

[0077] S12. Establish a measurable parameter interface, store the measurable parameters of the switch cabinet in a database through the measurable parameter interface, and interact with the digital model data through the database. The stability and reliability of the data are the prerequisites for accurate twin modeling;

[0078] S13, establishing a logical model according to the digital model, and performing twin visualization processing on the logical model through a state perception algorithm and a scene generation algorithm to obtain a simulation model;

[0079] S14, optimizing the simulation model using the data in the database, and returning the optimized data to the digital model;

[0080] S15. According to the monitoring requirements of the switch cabinet, a data model is constructed by using a digital storage method, and the data model is monitored in real time and an operation control strategy is determined;

[0081] S16, integrating the digital model, logic model, simulation model and data model to construct a digital twin model of the switch cabinet in a visualized virtual space;

[0082] S17. Optimize the switch cabinet digital twin model using an adaptive adoption and error optimization control method to obtain a final switch cabinet digital twin model.

[0083] The above-mentioned switch cabinet modeling method based on digital twin and error adaptive optimization obtains the physical parameters of the internal insulation structure of the switch cabinet, and constructs a digital model of the switch cabinet with multi-physical field coupling according to the physical parameters, and establishes a connection relationship between the numerical model and the digital twin interface; establishes a measurable parameter interface, and stores the measurable parameters of the switch cabinet in the database through the measurable parameter interface, and interacts with the digital model data through the database. The stability and reliability of the data are the premise for accurate twin modeling and access to external data; optimizes the simulation model using the data in the database, and returns the optimized data to the digital model; constructs a data model in a digital storage manner according to the monitoring requirements of the switch cabinet, and monitors the data model in real time and determines the operation control strategy; integrates the digital model, logical model, simulation model and data model to construct a digital twin model of the switch cabinet in a visualized virtual space; finally, optimizes the digital twin model of the switch cabinet using the adaptive adoption and error optimization control method to obtain the final digital twin model of the switch cabinet. By constructing a digital model of a switch cabinet with multi-physical field coupling through physical parameters, the final digital twin model of the switch cabinet can be measured in many aspects, and the adaptive error optimization control method is used to optimize the digital twin model of the switch cabinet, which can make the final digital twin model of the switch cabinet more accurate, thereby solving the existing method of modeling the switch cabinet without using digital twin technology. The final digital twin model of the switch cabinet can provide a reliable model for the state perception and performance prediction of the switch cabinet.

[0084] In one embodiment, in step S11, constructing a digital model of a switch cabinet coupled with multiple physical fields according to the physical parameters comprises the following steps:

[0085] S111, dividing the physical field inside the switch cabinet to obtain multiple physical fields such as the internal structure field, temperature field, electromagnetic field, force field and temperature field of the switch cabinet;

[0086] S112, confirming the coupling relationship between the physical fields according to the direct relationship between the internal insulation material of the switch cabinet and the physical field and the relationship between the physical fields;

[0087] Among them, the equation parameters of the temperature field are functions of other physical fields. The mutual influence between physical fields must be considered when calculating the temperature field. In the calculation of the electric field inside the switchgear, the conductivity of the internal insulating material is a function of temperature. When solving the Poisson's equation of the electric field, it is necessary to consider the influence of temperature on the electric field. Changes in the field quantities of the temperature field will also affect the specific heat capacity C and thermal conductivity λ.

[0088] S113. According to the coupling relationship and sensitivity relationship between the physical field quantities inside the switch cabinet, a digital model of the multi-physics coupled switch cabinet and corresponding multi-physics field coupling control equations are established.

[0089] By dividing the physical field inside the switchgear, determining the coupling relationship of the physical field based on the relationship between the internal insulating material of the switchgear and the physical field and the relationship between the physical fields, and establishing a multi-physics coupled high-voltage switchgear digital model and corresponding multi-physics field coupling control equations based on the coupling relationship and sensitivity relationship, the digital model of the switchgear is made more reliable and effective, thereby making the final digital twin model of the switchgear more effective and reliable.

[0090] In one embodiment, the measurable parameters include data in a power system database and data measured by on-site detection instruments.

[0091] Among them, the data in the power system database include current and voltage data.

[0092] In one of the embodiments, since the data measured by the on-site detection instrument is relatively complex and is not a digital quantity, the obtained data needs to be digitized and then transferred to the database. Therefore, the measurable parameter interface converts the measurable parameter into data of a unified standard.

[0093] In order to ensure data security and classify data management, the data storage types are divided into: structured data, unstructured data and historical / real-time data, and the database is established in a high-performance server. Historical / real-time data is the original data, and structured data and unstructured data are converted data types.

[0094] In one of the embodiments, in step S17, the adaptive error optimization control method includes an inner loop adaptive calculation error control method and an outer loop adaptive dynamic excitation control method.

[0095] Specifically, Figure 3 The inner loop adaptive calculation error control method is shown, comprising the following steps:

[0096] The inner loop adaptive calculation is mainly aimed at optimizing the error calculated by the switch cabinet digital twin model. The error e of the twin numerical calculation mainly includes interpolation error and discrete error. The value of error e is equal to the difference between the true value and the value calculated by the twin. However, the true value is generally unavailable and is an unknown constant, which is difficult to use as an error indicator. Therefore, the inner loop adaptive calculation error control method includes the following steps:

[0097] S1711, based on twin data h , the gradient reconstruction method is used to obtain a higher convergence order reconstruction result σ * (u h );

[0098] S1712, according to σ * (u h ) Calculate the reconstruction error e * , reconstruction error e * Equal to the reconstruction result σ * (u h ) and numerical results (twin data σ h )

[0099] S1713: Use the reconstructed error as a new error indicator to replace or approximate the real error. The new error indicator should satisfy:

[0100]

[0101] In formula (1), h is the interpolation spacing, that is, as the number of interpolation points increases, the new error indicator asymptotically equals the true error;

[0102] S1714, calculating an energy norm error and an error factor in a local domain, wherein the energy norm error and the error factor are used as a basis for encrypting interpolation points;

[0103] The calculation formula of energy norm error is:

[0104]

[0105] The error factor is calculated as:

[0106]

[0107] S1715. Control the number of interpolation points in the subdomain according to the size of the error factor on the subdomain, establish effective interpolation modes of different orders, input them into the digital twin model of the switch cabinet for recalculation, and repeat this cycle until the calculation accuracy is met.

[0108] Figure 3 An outer loop adaptive dynamic excitation control method is shown, comprising the following steps:

[0109] The outer loop adaptive calculation is mainly aimed at optimizing the error between the overall image of the twin and the physical device.

[0110] S1721. First, according to the fault mode and structural characteristics of the switch cabinet itself, the switch cabinet is divided into several subdomains. A distributed computing structure is adopted to use the LBM equation to perform a rough calculation on the overall architecture of the switch cabinet and a rough calculation of the fault point. When an abnormality is detected, the subdomain solver is activated to perform a detailed calculation of the roughly calculated fault point.

[0111] S1722, for each subdomain solution, monitor the corresponding data f(x 1 ,x 2 …x n ) and twin calculation corresponding data f′(x 1 ,x 2 …x n ) to construct the target optimization function g(x 1 ,x 2 …x n )=f(x 1 ,x 2 …x n )-f(x 1 ,x 2 …x n ), analyze the sensitivity relationship between the objective function and each influencing parameter s(x n ):

[0112]

[0113] In formula (4), x n represents the nth optimization parameter;

[0114] S1723, build a parameter optimization system to find g(x 1 ,x 2 …x n ) minimum value;

[0115] S1724, according to g(x 1 ,x 2 …x n ) is input into the twin model for iterative calculation. If the convergence criterion is met, the calculation is completed. Otherwise, the sampling points of the excitation conditions are increased and the optimization model is input into the cycle again until the convergence criterion is met to complete the adaptive adoption and error optimization control.

[0116] What is disclosed above is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or modifications within the technical scope disclosed in the present invention, which should be covered within the protection scope of the present invention.

Claims

1. A method for predicting the state of a high-voltage switchgear based on a digital model, characterized in that: The following steps are involved: Constructing a digital model of a switch cabinet coupled with multiple physical fields, and establishing a digital twin model of the switch cabinet according to the digital model of the switch cabinet; Dynamically deriving the performance of key components of the switch cabinet according to the switch cabinet digital twin model to obtain an overall dynamic twin model of the switch cabinet; Obtain measurable parameters of the switch cabinet, import the measurable parameters into the overall dynamic twin model of the switch cabinet, and obtain various operating conditions of the switch cabinet through the overall dynamic twin model of the switch cabinet; According to the sensitivity relationship between the measurable parameters of the switch cabinet and the various working conditions of the switch cabinet, a multi-parameter fusion state evaluation method is established, and the state of the switch cabinet is evaluated by the multi-parameter fusion state evaluation method; The multi-parameter fusion state assessment method comprises the following steps: Establish an evaluation index system for switchgear operation status; Acquire multiple characteristic parameters from the measurable parameters through an extraction network, wherein the characteristic parameters are parameters that can reflect the state of the switch cabinet, and the data of the measurable parameters include: virtual sensor data, physical sensor data, operation data and historical data; A plurality of characteristic parameters are obtained from the measurable parameters through an extraction network. The characteristic parameters are parameters that can reflect the state of the switch cabinet. The data of the measurable parameters include: virtual sensor data, physical sensor data, operation data and historical data, including: According to the type of measurable parameters, the data of measurable parameters are divided into: voltage data, current data and temperature data; Extracting fault voltage data from the voltage data as voltage features, extracting fault current data from the current data as current features, and extracting fault temperature data from the temperature data as temperature features; The voltage feature, current feature and temperature feature are input into the fusion analysis method for fusion to obtain the fusion feature. The fusion analysis method uses the XD-Markov machine to generate a state transition probability association matrix for different types of data. The weights of characteristic parameters are calculated using the analytic hierarchy process; Establish the relationship between multiple characteristic parameters and switchgear status, and establish a switchgear operation status evaluation model based on the weights of the characteristic parameters; The establishment of the switchgear operation status assessment model includes the following steps: Determine the network model to be used. The switch cabinet operation status assessment model is constructed using a residual neural network; Taking multiple characteristic parameters as input of the network model; establishing the relationship between multiple characteristic parameters and the switch cabinet state according to the weight of the characteristic parameters, setting the hidden layer of the network model with the relationship between multiple characteristic parameters and the switch cabinet state; the relationship between the switch cabinet state is the output of the network model; The pre-processed measurable parameters are used to train and verify the network model information, and the switch cabinet operation status evaluation model is obtained; Obtain real-time measurable parameters, input the real-time measurable parameters into the switch cabinet operation status evaluation model, and the switch cabinet operation status evaluation model outputs the real-time status of the switch cabinet. The status of the switch cabinet includes: normal, performance degradation, and abnormal failure; The constantly changing simulated working conditions are input into the overall dynamic twin model of the switchgear to obtain the characteristic parameter change characteristics of different components of the switchgear equipment under different modes; according to the characteristic parameter change characteristics, a switchgear state prediction method based on various parameters and historical data of the digital twin is established.

2. The method for predicting the state of a high-voltage switch cabinet based on a digital model according to claim 1 is characterized in that: The real-time characteristic parameters of the physical sensor are obtained, and the state evaluation method of the parameter fusion is used to evaluate the state of the switch cabinet.

3. The method for predicting the state of a high-voltage switch cabinet based on a digital model according to claim 1 is characterized in that: The real-time characteristic parameters of the physical sensor are obtained, and the switch cabinet state prediction method is used in combination with the real-time characteristic parameters to predict the state of the switch cabinet.

4. The method for predicting the state of a high-voltage switch cabinet based on a digital model according to claim 1 is characterized in that: Dynamically deducing the performance of key components of the switch cabinet according to the switch cabinet digital twin model includes the following steps: Establish the correlation between typical faults and status parameters of the switch cabinet; Establishing a fault model of the digital twin of the switch cabinet through the digital twin model of the switch cabinet; Performing fault simulation on the fault model to obtain fault model detection data, and analyzing the characteristics of the fault type based on the fault model detection data, real-time detection data, characteristic parameters and historical data; Combined with the analysis results, the association relationship between different fault modes and characteristic parameters of the switch cabinet is mined using the data mining algorithm based on association rules. According to the correlation between different failure modes and characteristic parameters of the switch cabinet, the state parameters under the simulated working conditions are queried and imported into the twin model for calculation, so as to deduce the change law of component performance under the working condition.

5. The method for predicting the state of a high-voltage switch cabinet based on a digital model according to claim 1 is characterized in that: The switch cabinet state prediction method based on various parameters and historical data of the digital twin comprises the following steps: According to the constantly changing simulation conditions, the overall dynamic twin model of the switch cabinet is input to obtain the characteristic parameter change characteristics of different components of the switch cabinet equipment in different modes, and the long short-term memory network LSTM is established as the basic network according to the transformation relationship between the working conditions and the characteristic quantities; Acquire real-time parameters of measurable parameters and input them into a basic network, and the basic network outputs a switch cabinet state prediction, thereby training optimal model parameters of the basic network; The switch cabinet state prediction model is obtained according to the optimal model parameters.

6. The method for predicting the state of a high-voltage switch cabinet based on a digital model according to claim 1 is characterized in that: The switch cabinet digital twin model is constructed by adopting a switch cabinet modeling method based on digital twin and error adaptive optimization.

7. The method for predicting the state of a high-voltage switch cabinet based on a digital model according to claim 1 is characterized in that: The virtual sensor setting of measurable parameters in the switch cabinet digital twin model includes: constructing signal strength, signal delay and signal measurement error evaluation indicators of virtual sensors at different positions through numerical optimization methods, and performing comprehensive evaluation in combination with the detection range of the virtual sensors to determine the optimal position distribution of the virtual sensors.

Images