Oil-immersed current transformer short circuit arc voltage calculation method and model based on GA-BP neural network parameter inversion

Abstract

The invention discloses an oil-immersed current transformer short circuit arc voltage calculation method and model based on GA-BP neural network parameter inversion. The calculation method comprises the following steps: constructing a Cassie arc model of an oil-immersed current transformer; taking Matlab as a simulation platform, and establishing a simulation circuit based on a Cassie arc model; testing system voltage of the simulation circuit is set, and a GA-BP neural network is used for inversion to obtain simulation parameters related to a Cassie arc model; inputting the simulation parameters into a simulation circuit, and debugging the simulation circuit to determine working parameters of the simulation circuit; and calculating the arc voltage of the oil-immersed current transformer under the target system voltage by using the simulation circuit of which the working parameters are determined, and reading arc voltage data through an oscilloscope of the simulation circuit. According to the invention, inversion is carried out on the Cassie arc model through the GA-BP neural network algorithm to obtain related parameters, and a basis is provided for calculating the short-circuit arc voltage of the oil-immersed current transformer.

Description

Technical Field

[0001] The invention relates to the field of current transformers, and in particular to a method and a model for calculating short-circuit arc voltage of an oil-immersed current transformer based on GA-BP neural network parameter inversion. Background Art

[0002] Oil-immersed current transformer is a device widely used in power systems. Its main function is to convert large current into a small current of a certain proportion to facilitate current measurement, automatic control and relay protection.

[0003] Internal short-circuit arcing faults of current transformers (CTs) can pose a great threat to the safe operation of power systems, so it is necessary to closely monitor the operating parameters of oil-immersed current transformers.

[0004] Arc energy is the basis for numerical calculation of explosion pressure effect of short-circuit arcing inside CT. The parameters that affect arc energy are mainly arc current and arc voltage. Arc current is generally known, while arc voltage is more random and has more influencing factors. However, the internal arc fault test of 220kV and above oil-immersed current transformers is a destructive test. During the test, explosion, fire, oil leakage, flying debris and other phenomena may occur. The test technology is extremely complex. Therefore, it is currently difficult to measure arc voltage data through arc fault tests.

[0005] On the other hand, it has been proposed to obtain arc voltage data through magnetohydrodynamic finite element simulation, but this takes a lot of time and the calculation efficiency often cannot meet application requirements.

[0006] The disclosure of the above background technology content is only used to assist in understanding the concept and technical solution of the present application. It does not necessarily belong to the prior art of the present application, nor does it necessarily provide technical guidance. In the absence of clear evidence that the above content has been disclosed before the filing date of the present application, the above background technology should not be used to evaluate the novelty and creativity of the present application. Summary of the invention

[0007] The purpose of the present invention is to provide a method for calculating the short-circuit arc voltage of an oil-immersed current transformer based on GA-BP neural network parameter inversion. The Cassie arc model is inverted by a GA-BP neural network algorithm to obtain relevant parameters, so that the short-circuit arc voltage of a high voltage level can be calculated efficiently and conveniently.

[0008] To achieve the above object, the technical solution adopted by the present invention is as follows:

[0009] A method for calculating short-circuit arc voltage of an oil-immersed current transformer based on GA-BP neural network parameter inversion comprises the following steps:

[0010] The Cassie arc equation is used as the Cassie arc channel model of the oil-immersed current transformer: Where g is the arc conductance per unit length, d g / d t is the differential of arc conductance per unit length with respect to time, τ c is the time constant of the Cassie arc, e is the arc voltage per unit length, i.e., the field strength in the arc column; e0 is the field strength in the static arc column, which is a constant;

[0011] Using Matlab as a simulation platform and Simulink to establish a simulation circuit based on the Cassie arc channel model, the simulation circuit is equipped with an oscilloscope;

[0012] The test system voltage of the simulation circuit based on the Cassie arc channel model is set, and the simulation parameters are obtained by GA-BP neural network inversion: the field intensity e0 in the static arc column and the time constant τ of the Cassie arc c ;

[0013] Inputting the simulation parameters into the simulation circuit, and debugging the simulation circuit to determine the operating parameters of the simulation circuit;

[0014] The arc voltage of the oil-immersed current transformer under the target system voltage is calculated by using the simulation circuit after the working parameters are determined, and the arc voltage data is read out through the oscilloscope of the simulation circuit.

[0015] Further, based on any one of the above-mentioned technical solutions or a combination of multiple technical solutions, the GA-BP neural network inversion includes the following steps:

[0016] Constructing a BP neural network, which includes designing the number of nodes of an input layer, a hidden layer, and an output layer of the network, wherein the input layer receives a known target, and the output layer generates a predicted value;

[0017] Use genetic algorithm to initialize the weights and biases of BP neural network, and evaluate the performance of BP neural network through fitness function;

[0018] Generate new populations through genetic algorithm selection, crossover, and mutation, update the weights and biases of the network, and obtain the GA-BP neural network model;

[0019] The GA-BP neural network model is used to invert the Cassie arc channel model: the field strength e0 in the static arc column and the time constant τ of the Cassie arc are adjusted using the back propagation method. c , until the difference between the corresponding model output and the known target is less than a preset range.

[0020] Further, based on any one of the above-mentioned technical solutions or a combination of multiple technical solutions, inverting the Cassie arc channel model using the GA-BP neural network model includes the following steps:

[0021] Under the conditions of given current, arc gap and voltage of the test system, a short-circuit test of the oil-immersed current transformer is carried out;

[0022] Extracting multiple parameters from the data of the short-circuit test as the known parameters, one of which is the test arc voltage;

[0023] The test arc voltage is used as the output of the Cassie arc channel model, and the remaining known parameters are used as inputs to train the Cassie arc channel model to obtain the field intensity e0 in the static arc column and the time constant τ of the Cassie arc. c .

[0024] Further, according to any one of the above-mentioned technical solutions or a combination of multiple technical solutions, the set value of the test system voltage is less than the target system voltage;

[0025] After debugging the simulation circuit, the method further includes verifying the simulation circuit:

[0026] Reading out the simulated arc voltage from the oscilloscope of the simulation circuit, and obtaining the test arc voltage from the data of the short-circuit test;

[0027] Comparing the simulated arc voltage with the test arc voltage, if the difference between the two is less than a preset arc voltage difference threshold, the simulation circuit verification result is passed, otherwise the simulation circuit verification result is failed;

[0028] If the simulation circuit verification result is not passed, changing the working parameters of the simulation circuit and re-debugging the simulation circuit;

[0029] When the simulation circuit verification result is passed, the arc voltage of the oil-immersed current transformer under the target system voltage is calculated.

[0030] Further, based on any one of the technical solutions or a combination of multiple technical solutions described above, the simulation circuit of the Cassie arc channel model includes an AC power supply, an inductor, a Cassie arc module package, a voltage adder, a current adder and a first oscilloscope, wherein the AC power supply provides a system voltage, the inductor is configured to discretely change a current time constant, and the Cassie arc module package includes a Cassie calculation module package of the Cassie arc channel model;

[0031] The voltage adder is configured to aggregate output voltages, the current adder is configured to aggregate output currents, and the first oscilloscope is configured to display a voltage waveform aggregated output by the voltage adder and a current waveform aggregated output by the current adder.

[0032] Further, based on any one of the above-mentioned technical solutions or a combination of multiple technical solutions, the Cassie arc module package also includes an input voltage node, a power frequency voltage generator, a step voltage generator, a voltage-controlled current source and an output voltage node;

[0033] The output ends of the power frequency voltage generator and the step voltage generator are connected to the input end of the Cassie calculation module package, and the output end of the Cassie calculation module package is connected to the voltage-controlled current source.

[0034] Further, based on any one of the above-mentioned technical solutions or a combination of multiple technical solutions, the Cassie calculation module package includes a first input voltage node, a second input voltage node, a function unit, an integration unit, a summing unit, an output unit and an output node, wherein the first input voltage node is used to connect the power frequency voltage generator and the summing unit, and the second input voltage node is used to connect the step voltage generator and the summing unit;

[0035] The function unit, the integration unit, and the summation unit are connected in sequence to form a closed loop;

[0036] The output unit and the output node are used to connect the summing unit and the voltage-controlled current source.

[0037] According to another aspect of the present invention, a short-circuit arc voltage calculation model for an oil-immersed current transformer based on GA-BP neural network parameter inversion is provided, comprising a simulation circuit established by Simulink based on a Matlab platform, wherein the simulation circuit comprises:

[0038] an AC power source configured to provide a system voltage;

[0039] an inductor configured to discretely vary a current time constant;

[0040] A Cassie arc module package, which includes a Cassie calculation module package of a Cassie arc channel model, which is configured to simulate low resistance, high current arcing, wherein the Cassie arc channel model: Where g is the arc conductance per unit length, d g / d t is the differential of arc conductance per unit length with respect to time, τ cis the time constant of the Cassie arc, e is the arc voltage per unit length, i.e., the field strength in the arc column; e0 is the field strength in the static arc column, which is a constant;

[0041] a voltage adder configured to sum the output voltages;

[0042] a current adder configured to sum the output currents;

[0043] A first oscilloscope configured to display a voltage waveform of the voltage adder and a current waveform of the current adder;

[0044] Among them, the parameters e0 and τ of the Cassie arc channel model are c Obtained by inversion using GA-BP neural network.

[0045] Further, based on any one of the above-mentioned technical solutions or a combination of multiple technical solutions, the GA-BP neural network inversion includes the following steps:

[0046] Constructing a BP neural network, which includes designing the number of nodes of an input layer, a hidden layer, and an output layer of the network, wherein the input layer receives a known target, and the output layer generates a predicted value;

[0047] Use genetic algorithm to initialize the weights and biases of BP neural network, and evaluate the performance of BP neural network through fitness function;

[0048] Generate new populations through genetic algorithm selection, crossover, and mutation, update the weights and biases of the network, and obtain the GA-BP neural network model;

[0049] The GA-BP neural network model is used to invert the Cassie arc channel model:

[0050] Under the conditions of given current, arc gap and voltage of the test system, a short-circuit test of the oil-immersed current transformer is carried out;

[0051] Extracting multiple parameters from the data of the short-circuit test as the known parameters, one of which is the test arc voltage;

[0052] The test arc voltage is used as the output of the Cassie arc channel model, and the remaining known parameters are used as inputs to train the Cassie arc channel model;

[0053] Using the back-propagation method, adjust the field strength e0 in the static arc column and the time constant τ of the Cassie arc c , until the difference between the corresponding model output and the known target is less than a preset range.

[0054] Further, based on any one of the above-mentioned technical solutions or a combination of multiple technical solutions, the Cassie arc module package also includes an input voltage node, a power frequency voltage generator, a step voltage generator, a voltage-controlled current source and an output voltage node, wherein the input voltage node is connected to the inductor, and the output voltage node is connected to the voltage adder and the current adder respectively;

[0055] The Cassie calculation module package includes a first input voltage node, a second input voltage node, a function unit, an integration unit, a summing unit, an output unit and an output node;

[0056] The power frequency voltage generator is connected to the summing unit via the first input voltage node, and the step voltage generator is connected to the summing unit via the second input voltage node; the function unit, the integration unit, and the summing unit are connected in sequence to form a closed loop;

[0057] The summing unit is connected to the voltage-controlled current source through the output unit and an output node.

[0058] The beneficial effects brought by the technical solution provided by the present invention are as follows:

[0059] a. Establish a black box model to characterize the arc, which only describes the external characteristics of the arc (i.e., the relationship between voltage and current), but does not study the complex physical process inside the arc, simplify the model, and improve the calculation efficiency;

[0060] b. Establish the Cassie arc channel model, which is more suitable for simulating and calculating the short-circuit arc voltage of large current oil-immersed current transformers;

[0061] c. The parameters of the Cassie arc channel model can be obtained by inverting the Cassie arc channel model through the GA-BP neural network algorithm, and the reliability of the model can be determined through experiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0063] Figure 1 A basic flow chart of a method for calculating short-circuit arc voltage of an oil-immersed current transformer based on GA-BP neural network parameter inversion provided as an exemplary embodiment of the present invention;

[0064] Figure 2A simulation circuit diagram based on a Cassie arc channel model established by using Simulink is provided as an exemplary embodiment of the present invention;

[0065] Figure 3 for Figure 2 Internal circuit diagram of the Cassie Arc module package in the simulation circuit;

[0066] Figure 4 for Figure 3 Internal circuit diagram of the Cassie computing module package;

[0067] Figure 5 A flowchart of the GA-BP neural network inversion steps provided for an exemplary embodiment of the present invention;

[0068] Figure 6 An operation flow chart of verifying a simulation circuit provided for an exemplary embodiment of the present invention;

[0069] Figure 7 A waveform diagram of the simulated arc voltage calculated in the 220kV system voltage embodiment;

[0070] Figure 8 A waveform diagram of a simulated current calculated in an embodiment of a 220 kV system voltage;

[0071] Fig. 9 This is a waveform diagram comparing the 220kV simulated arc voltage with the test arc voltage;

[0072] Fig.10 This is a waveform diagram comparing the effective value of the 220kV simulated arc voltage with the effective value of the test arc voltage;

[0073] Fig.11 This is a waveform diagram for calculating the short-circuit arc voltage of a 500kV oil-immersed current transformer;

[0074] Fig.12 This is a waveform diagram for calculating the short-circuit current of a 500kV oil-immersed current transformer;

[0075] Among them, the figure marks include: 1-AC power supply, 2-inductor, 3-Cassie arc module package, 4-voltage adder, 5-current adder, 6-second oscilloscope, 7-third oscilloscope, 8-first oscilloscope, 9-input voltage node, 10-power frequency voltage generator, 11-step voltage generator, 12-Cassie calculation module package, 13-voltage controlled current source, 14-output voltage node, 15-first input voltage node, 16-second input voltage node, 17-function unit, 18-integral unit, 19-summing unit, 20-output unit, 21-output node. DETAILED DESCRIPTION

[0076] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below 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 should fall within the scope of protection of the present invention.

[0077] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, device, product or equipment that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0078] The invention provides a method for calculating the short-circuit arc voltage of an oil-immersed current transformer based on the Cassie "black box" arc mathematical model, ignoring the internal short-circuit arc of the oil-immersed current transformer as an electrical component, only describing the external characteristics of the arc, that is, the relationship between voltage and current, without studying the complex physical process inside the arc, establishing the Cassie "black box" arc mathematical model, using power system simulation technology, establishing the Cassie "black box" arc simulation model in Simulink, using the GA-BP neural network algorithm to invert the Cassie arc channel model to obtain its related parameters, inputting the system voltage and related parameters, and outputting arc voltage and current data. The simulated arc voltage data is compared with the experimental arc voltage data to verify the correctness of the model, and finally the arc voltage data of the internal short-circuit arc of the high-voltage oil-immersed current transformer is calculated.

[0079] In one embodiment of the present invention, a method for calculating short-circuit arc voltage of an oil-immersed current transformer based on GA-BP neural network parameter inversion is provided. Figure 1 As shown, the calculation method mainly includes the following five steps:

[0080] The first step is to use the Cassie arc equation as the Cassie arc channel model of the oil-immersed current transformer: Where g is the arc conductance per unit length, d g / d tis the differential of arc conductance per unit length with respect to time, τ c is the time constant of the Cassie arc, e is the arc voltage per unit length, that is, the field strength of the arc column; e0 is the field strength in the static arc column, which is a constant; where the time constant τ of the Cassie arc c The electric field strength e0 in the static arc column is an unknown constant in the Cassie arc channel model.

[0081] The black box model is a pure mathematical model. It regards the arc as an electrical component and only describes the external characteristics of the arc, that is, the relationship between voltage and current, without studying the complex physical process inside the arc. The arc is usually represented by an equivalent circuit (black box method). This model is available because the goal of establishing an arc model is to determine the arc current, power and energy.

[0082] The arc test in oil in a closed container shows that the high-energy arc violently vaporizes the insulating oil and forms pulsating bubbles, so the arc in oil is essentially a gas arc. The gas arc model derives the corresponding time constant τ based on different assumptions. arc and dissipated power P loss At present, the commonly used arc models include Cassie arc model, Mayr arc model, Schwarz arc model, cybernetics model and so on.

[0083] The general form of the arc mathematical model is as follows:

[0084]

[0085] Where: g is the arc conductivity per unit length; e·i is the input power per unit length of the arc; e is the arc voltage per unit length, that is, the field intensity in the arc column; i is the arc current; τ arc is the time constant of the arc; P loss is the power dissipated per unit length of the arc. Various arc models derive the corresponding time constant τ based on different assumptions. arc and dissipated power P loss .

[0086] The Mayr model is suitable for breaking at low currents; the Cassie model is more suitable and accurate at high currents; the Schwarz arc model is more complicated to calculate. Since the current is large during the arc fault of the 500kV oil-immersed CT, the Cassie model is selected as the CT arc fault simulation model. The Cassie arc model is mainly applicable to the arcing period of large current before the current passes through zero. Its arc model equation expresses the relationship between arc voltage, arc current, arc conductivity, time constant, and energy loss. According to the assumptions of the Cassie arc model, the arc equation of the Cassie arc channel model can be derived. Calculating the external characteristics of the short-circuit arc inside the oil-immersed current transformer, that is, the relationship between voltage and current, ignores the study of the complex physical process inside the arc, greatly reduces the calculation complexity, and can provide a reference for the calculation of the short-circuit arc voltage of the oil-immersed current transformer.

[0087] Step 2: In the power system simulation software, for example, using Matlab as the simulation platform, using Simulink to establish a simulation circuit based on the Cassie arc channel model, wherein the simulation circuit is equipped with an oscilloscope;

[0088] See the simulation circuit Figure 2 , insert power system components into the simulink module, and build a simulation circuit based on the Cassie arc channel model: including an AC power supply 1, an inductor 2, a Cassie arc module package 3, a voltage adder 4, a current adder 5 and a first oscilloscope 8, wherein the AC power supply 1 provides a system voltage, the inductor 2 is configured to discretely change a current time constant, and the Cassie arc module package 3 includes a Cassie calculation module package 12 of the Cassie arc channel model;

[0089] The voltage adder 4 is configured to collect output voltages, the current adder 5 is configured to collect output currents, and the first oscilloscope 8 is configured to display the voltage waveform collected and output by the voltage adder 4 and the current waveform collected and output by the current adder 5. In addition to the first oscilloscope 8, a second oscilloscope 6 can be provided to separately display the voltage waveform collected and output by the voltage adder 4, and a third oscilloscope 7 can be provided to display the current waveform collected and output by the current adder 5.

[0090] In addition to the Cassie calculation module package 12, the Cassie arc module package 3 also includes an input voltage node 9, a power frequency voltage generator 10, a step voltage generator 11, a voltage-controlled current source 13 and an output voltage node 14; Figure 3As shown, the output ends of the power frequency voltage generator 10 and the step voltage generator 11 are connected to the input end of the Cassie calculation module package 12 , and the output end of the Cassie calculation module package 12 is connected to the voltage-controlled current source 13 .

[0091] Specific as Figure 4 As shown, the Cassie calculation module package 12 includes a first input voltage node 15, a second input voltage node 16, a function unit 17, an integration unit 18, a summing unit 19, an output unit 20 and an output node 21, wherein the first input voltage node 15 is used to connect the industrial frequency voltage generator 10 and the summing unit 19, and the second input voltage node 16 is used to connect the step voltage generator 11 and the summing unit 19; the function unit 17, the integration unit 18, and the summing unit 19 are connected in sequence to form a closed loop; the output unit 20 and the output node 21 are used to connect the summing unit 19 and the voltage-controlled current source 13.

[0092] The third step is to set the test system voltage of the simulation circuit based on the Cassie arc channel model, and use GA-BP neural network inversion to obtain simulation parameters: the field intensity e0 in the static arc column and the time constant τ of the Cassie arc c ;

[0093] When the external conditions, current and arc gap are determined, the parameters of the Cassie arc model are determined. Based on the obtained arc voltage and current output data of the oil-immersed current transformer test, the parameters of the Cassie arc model are inverted using the GA-BP neural network algorithm. Figure 2 The AC power supply 1 in the simulation circuit provides the system voltage. In the inversion stage, it is set to a test system voltage lower than the target system voltage. For example, if the arc voltage of a 500kV oil-immersed current transformer is to be calculated, the target system voltage is 500kV, and a test system voltage lower than this value is selected. For example, the output of the AC power supply 1 is set to a voltage of 220kV.

[0094] The GA-BP neural network inversion steps are as follows: Figure 5As shown: construct a BP neural network, which includes designing the number of nodes of the input layer, hidden layer and output layer of the network, wherein the input layer receives a known target and the output layer generates a predicted value; use a genetic algorithm to initialize the weights and biases of the BP neural network, and evaluate the performance of the BP neural network through a fitness function; generate a new population through genetic algorithm selection, crossover, and mutation, update the weights and biases of the network, and obtain a GA-BP neural network model; use the GA-BP neural network model to invert the Cassie arc channel model: use the back propagation method to adjust the field strength e0 in the static arc column and the time constant τ of the Cassie arc c , until the difference between the corresponding model output and the known target is less than a preset range, specifically:

[0095] Under the conditions of given current, arc gap and voltage of the test system, a short-circuit test of the oil-immersed current transformer is performed; multiple parameters are extracted from the data of the short-circuit test as the known parameters, one of which is the test arc voltage, and other parameters such as current and test conductivity are also extracted; the test conductivity is used as the input of the Cassie arc channel model, the test arc voltage is used as the output of the Cassie arc channel model, and the Cassie arc channel model is trained to obtain the field strength e0 in the static arc column and the time constant τ of the Cassie arc c .

[0096] Step 4: input the simulation parameters into the simulation circuit, debug the simulation circuit, and verify the simulation circuit to determine the working parameters of the simulation circuit;

[0097] Specific as Figure 6 As shown, the simulated arc voltage is read out from the oscilloscope of the simulation circuit, and the test arc voltage is obtained from the data of the short-circuit test; the simulated arc voltage and the test arc voltage are compared, if the difference between the two is less than the preset arc voltage difference threshold, the simulation circuit verification result is passed, otherwise the simulation circuit verification result is failed; when the simulation circuit verification result is failed, the working parameters of the simulation circuit are changed and the simulation circuit is re-debugged; when the simulation circuit verification result is passed, the working parameters of the simulation circuit are determined to proceed to the next step.

[0098] Step 5. Using the simulation circuit after determining the working parameters, set the output of the AC power supply 1 to 500 kV (i.e., the target system voltage) in the simulation circuit. The Cassie calculation module package 12 of the Cassie arc module package 3 calculates the arc voltage of the oil-immersed current transformer, and reads the arc voltage data through the oscilloscope of the simulation circuit.

[0099] In a specific embodiment, the (test) system voltage is first set to 220 kV, and the GA-BP neural network is used to invert the simulation parameters. The field strength e0 in the static arc column is 800 V, and the time constant τ of the Cassie arc is c The calculated waveform of the simulated arc voltage is 38μs. Figure 7 As shown, the waveform of the simulated current is calculated as Figure 8 shown.

[0100] The results of comparing the 220kV simulated arc voltage with the test arc voltage are as follows: Fig. 9 As shown in the figure, the comparison between the 220kV simulated arc voltage effective value and the test arc voltage effective value is shown in the figure Fig.10 shown; from Fig.10 It can be seen that in the stable arcing stage after 32ms, the arc voltage effective values ​​of the simulation and the test are basically consistent, indicating that the model and the simulation circuit test have passed. At this time, the short-circuit arc voltage and current data of the 500kV oil-immersed current transformer can be calculated. The arc voltage calculation results are as follows: Fig.11 As shown, the current calculation results are as follows Fig.12 shown.

[0101] In one embodiment of the present invention, a short-circuit arc voltage calculation model for an oil-immersed current transformer based on GA-BP neural network parameter inversion is provided, including a simulation circuit established based on Simulink of a Matlab platform, wherein the simulation circuit is as follows: Figure 2 As shown, including:

[0102] an AC power source 1 configured to provide a system voltage;

[0103] Inductor 2, which is configured to discretely change a current time constant;

[0104] The Cassie arc module package 3 includes a Cassie calculation module package 12 of a Cassie arc channel model, which is configured to simulate arcing with low resistance and high current. The Cassie arc channel model: Where g is the arc conductance per unit length, d g / d t is the differential of arc conductance per unit length with respect to time, τ c is the time constant of the Cassie arc, e is the arc voltage per unit length, i.e., the field strength in the arc column; e0 is the field strength in the static arc column, which is a constant;

[0105] a voltage adder 4 configured to aggregate the output voltages;

[0106] a current adder 5 configured to aggregate the output current;

[0107] A first oscilloscope 8, which is configured to display the voltage waveform of the voltage adder 4 and the current waveform of the current adder 5;

[0108] Among them, the parameters e0 and τ of the Cassie arc channel model are c Obtained by inversion using GA-BP neural network.

[0109] Further, based on any one of the above-mentioned technical solutions or a combination of multiple technical solutions, the GA-BP neural network inversion includes: Figure 5 Steps shown:

[0110] Constructing a BP neural network, which includes designing the number of nodes of an input layer, a hidden layer, and an output layer of the network, wherein the input layer receives a known target, and the output layer generates a predicted value;

[0111] Use genetic algorithm to initialize the weights and biases of BP neural network, and evaluate the performance of BP neural network through fitness function;

[0112] Generate new populations through genetic algorithm selection, crossover, and mutation, update the weights and biases of the network, and obtain the GA-BP neural network model;

[0113] The GA-BP neural network model is used to invert the Cassie arc channel model:

[0114] Under the conditions of given current, arc gap and voltage of the test system, a short-circuit test of the oil-immersed current transformer is carried out;

[0115] Extracting multiple parameters from the data of the short-circuit test as the known parameters, one of which is the test arc voltage;

[0116] The test arc voltage is used as the output of the Cassie arc channel model, and the remaining known parameters are used as inputs to train the Cassie arc channel model;

[0117] Using the back-propagation method, adjust the field strength e0 in the static arc column and the time constant τ of the Cassie arc c , until the difference between the corresponding model output and the known target is less than a preset range.

[0118] Further references Figure 3 and Figure 4, the Cassie arc module package 3 also includes an input voltage node 9, a power frequency voltage generator 10, a step voltage generator 11, a voltage-controlled current source 13 and an output voltage node 14, wherein the input voltage node 9 is connected to the inductor 2, and the output voltage node 14 is connected to the voltage adder 4 and the current adder 5 respectively;

[0119] The Cassie calculation module package 12 includes a first input voltage node 15, a second input voltage node 16, a function unit 17, an integration unit 18, a summing unit 19, an output unit 20 and an output node 21;

[0120] The power frequency voltage generator 10 is connected to the summing unit 19 via the first input voltage node 15, and the step voltage generator 11 is connected to the summing unit 19 via the second input voltage node 16; the function unit 17, the integration unit 18, and the summing unit 19 are connected in sequence to form a closed loop;

[0121] The summing unit 19 is connected to the voltage-controlled current source 13 via the output unit 20 and an output node 21 .

[0122] It should be noted that the oil-immersed current transformer short-circuit arc voltage calculation model based on GA-BP neural network parameter inversion provided in this embodiment and the oil-immersed current transformer short-circuit arc voltage calculation method based on GA-BP neural network parameter inversion provided in the above embodiment belong to the same inventive concept. Here, the entire content of the oil-immersed current transformer short-circuit arc voltage calculation method based on GA-BP neural network parameter inversion is incorporated into the present calculation model embodiment by citing the full text, and no further details are given.

[0123] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0124] The above is only a specific implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for calculating short-circuit arc voltage of oil-immersed current transformer based on GA-BP neural network parameter inversion, characterized in that: The following steps are involved: The Cassie arc equation is used as the Cassie arc channel model of the oil-immersed current transformer: Where g is the arc conductance per unit length, d g / d t is the differential of arc conductance per unit length with respect to time, τ c is the time constant of the Cassie arc, e is the arc voltage per unit length, i.e., the field strength in the arc column; e0 is the field strength in the static arc column, which is a constant; Using Matlab as a simulation platform and Simulink to establish a simulation circuit based on the Cassie arc channel model, the simulation circuit is equipped with an oscilloscope; The test system voltage of the simulation circuit based on the Cassie arc channel model is set, and the simulation parameters are obtained by GA-BP neural network inversion: the field intensity e0 in the static arc column and the time constant τ of the Cassie arc c ; Inputting the simulation parameters into the simulation circuit, and debugging the simulation circuit to determine the operating parameters of the simulation circuit; The arc voltage of the oil-immersed current transformer under the target system voltage is calculated by using the simulation circuit after the working parameters are determined, and the arc voltage data is read out through the oscilloscope of the simulation circuit.

2. The method for calculating short-circuit arc voltage of oil-immersed current transformer based on GA-BP neural network parameter inversion according to claim 1 is characterized in that: The GA-BP neural network inversion includes the following steps: Constructing a BP neural network, which includes designing the number of nodes of an input layer, a hidden layer, and an output layer of the network, wherein the input layer receives a known target, and the output layer generates a predicted value; Use genetic algorithm to initialize the weights and biases of BP neural network, and evaluate the performance of BP neural network through fitness function; Generate new populations through genetic algorithm selection, crossover, and mutation, update the weights and biases of the network, and obtain the GA-BP neural network model; The GA-BP neural network model is used to invert the Cassie arc channel model: the field strength e0 in the static arc column and the time constant τ of the Cassie arc are adjusted using the back propagation method. c , until the difference between the corresponding model output and the known target is less than a preset range.

3. The method for calculating short-circuit arc voltage of oil-immersed current transformer based on GA-BP neural network parameter inversion according to claim 2 is characterized in that: Inverting the Cassie arc channel model using the GA-BP neural network model includes the following steps: Under the conditions of given current, arc gap and voltage of the test system, a short-circuit test of the oil-immersed current transformer is carried out; Extracting multiple parameters from the data of the short-circuit test as the known parameters, one of which is the test arc voltage; The test arc voltage is used as the output of the Cassie arc channel model, and the remaining known parameters are used as inputs to train the Cassie arc channel model to obtain the field intensity e0 in the static arc column and the time constant τ of the Cassie arc. c .

4. The method for calculating short-circuit arc voltage of oil-immersed current transformer based on GA-BP neural network parameter inversion according to claim 3 is characterized in that: The set value of the test system voltage is less than the target system voltage; After debugging the simulation circuit, the method further includes verifying the simulation circuit: Reading out the simulated arc voltage from the oscilloscope of the simulation circuit, and obtaining the test arc voltage from the data of the short-circuit test; Comparing the simulated arc voltage with the test arc voltage, if the difference between the two is less than a preset arc voltage difference threshold, the simulation circuit verification result is passed, otherwise the simulation circuit verification result is failed; If the simulation circuit verification result is not passed, changing the working parameters of the simulation circuit and re-debugging the simulation circuit; When the simulation circuit verification result is passed, the arc voltage of the oil-immersed current transformer under the target system voltage is calculated.

5. The method for calculating short-circuit arc voltage of oil-immersed current transformer based on GA-BP neural network parameter inversion according to claim 1, characterized in that: The simulation circuit of the Cassie arc channel model comprises an AC power supply (1), an inductor (2), a Cassie arc module package (3), a voltage adder (4), a current adder (5) and a first oscilloscope (8), wherein the AC power supply (1) provides a system voltage, the inductor (2) is configured to discretely change a current time constant, and the Cassie arc module package (3) comprises a Cassie calculation module package (12) of the Cassie arc channel model; The voltage adder (4) is configured to collect output voltages, the current adder (5) is configured to collect output currents, and the first oscilloscope (8) is configured to display a voltage waveform collected and output by the voltage adder (4) and a current waveform collected and output by the current adder (5).

6. The method for calculating short-circuit arc voltage of oil-immersed current transformer based on GA-BP neural network parameter inversion according to claim 5, characterized in that: The Cassie arc module package (3) further comprises an input voltage node (9), a power frequency voltage generator (10), a step voltage generator (11), a voltage-controlled current source (13) and an output voltage node (14); The output ends of the power frequency voltage generator (10) and the step voltage generator (11) are connected to the input end of the Cassie calculation module package (12), and the output end of the Cassie calculation module package (12) is connected to the voltage-controlled current source (13).

7. The method for calculating short-circuit arc voltage of oil-immersed current transformer based on GA-BP neural network parameter inversion according to claim 5 or 6, characterized in that: The Cassie calculation module package (12) comprises a first input voltage node (15), a second input voltage node (16), a function unit (17), an integration unit (18), a summing unit (19), an output unit (20) and an output node (21), wherein the first input voltage node (15) is used to connect the power frequency voltage generator (10) and the summing unit (19), and the second input voltage node (16) is used to connect the step voltage generator (11) and the summing unit (19); The function unit (17), the integration unit (18), and the summation unit (19) are connected in sequence to form a closed loop; The output unit (20) and the output node (21) are used to connect the summing unit (19) and the voltage-controlled current source (13).

8. A short-circuit arc voltage calculation model for oil-immersed current transformer based on GA-BP neural network parameter inversion, characterized in that: The invention comprises a simulation circuit established by Simulink based on a Matlab platform, wherein the simulation circuit comprises: an AC power source (1) configured to provide a system voltage; an inductor (2) configured to discretely vary a current time constant; A Cassie arc module package (3) includes a Cassie calculation module package (12) of a Cassie arc channel model, which is configured to simulate arcing with low resistance and high current, wherein the Cassie arc channel model: Where g is the arc conductance per unit length, d g / d t is the differential of arc conductance per unit length with respect to time, τ c is the time constant of the Cassie arc, e is the arc voltage per unit length, i.e., the field strength in the arc column; e0 is the field strength in the static arc column, which is a constant; a voltage adder (4) configured to aggregate the output voltages; a current adder (5) configured to aggregate the output currents; A first oscilloscope (8), configured to display a voltage waveform of the voltage adder (4) and a current waveform of the current adder (5); Among them, the parameters e0 and τ of the Cassie arc channel model are c Obtained by inversion using GA-BP neural network.

9. The oil-immersed current transformer short-circuit arc voltage calculation model based on GA-BP neural network parameter inversion according to claim 8, characterized in that: The GA-BP neural network inversion includes the following steps: Constructing a BP neural network, which includes designing the number of nodes of an input layer, a hidden layer, and an output layer of the network, wherein the input layer receives a known target, and the output layer generates a predicted value; Use genetic algorithm to initialize the weights and biases of BP neural network, and evaluate the performance of BP neural network through fitness function; Generate new populations through genetic algorithm selection, crossover, and mutation, update the weights and biases of the network, and obtain the GA-BP neural network model; The GA-BP neural network model is used to invert the Cassie arc channel model: Under the conditions of given current, arc gap and voltage of the test system, a short-circuit test of the oil-immersed current transformer is carried out; Extracting multiple parameters from the data of the short-circuit test as the known parameters, one of which is the test arc voltage; The test arc voltage is used as the output of the Cassie arc channel model, and the remaining known parameters are used as inputs to train the Cassie arc channel model; Using the back-propagation method, adjust the field strength e0 in the static arc column and the time constant τ of the Cassie arc c , until the difference between the corresponding model output and the known target is less than a preset range.

10. The oil-immersed current transformer short-circuit arc voltage calculation model based on GA-BP neural network parameter inversion according to claim 8, characterized in that: The Cassie arc module package (3) further comprises an input voltage node (9), an industrial frequency voltage generator (10), a step voltage generator (11), a voltage-controlled current source (13) and an output voltage node (14), wherein the input voltage node (9) is connected to the inductor (2), and the output voltage node (14) is connected to the voltage adder (4) and the current adder (5), respectively; The Cassie calculation module package (12) includes a first input voltage node (15), a second input voltage node (16), a function unit (17), an integration unit (18), a summing unit (19), an output unit (20) and an output node (21); The power frequency voltage generator (10) is connected to the summing unit (19) via the first input voltage node (15), and the step voltage generator (11) is connected to the summing unit (19) via the second input voltage node (16); the function unit (17), the integration unit (18), and the summing unit (19) are connected in sequence to form a closed loop; The summing unit (19) is connected to the voltage-controlled current source (13) via the output unit (20) and the output node (21).

Images