Voltage transformer insulation state evaluation and dielectric self-repairing method and system

Abstract

The invention discloses a voltage transformer insulation state evaluation and dielectric self-repairing method and system. The method comprises the steps that a quantum dot array sensor is embedded in the surface of a voltage dividing capacitor unit of a CVT to obtain the capacitance value of a CVT insulation layer; a high-frequency scanning signal is injected into an end screen input port of the CVT, a local oscillator and a quantum mixer are arranged at an output port, and a dielectric loss factor is obtained; carrying out circuit analysis on the CVT insulation circuit, converting an insulation layer capacitance value and a dielectric loss factor into a capacitor breakdown number and a dielectric loss abnormal characteristic parameter, respectively taking the parameters as inputs of a WOA-based optimized DBA model, and obtaining a capacitor breakdown error variable quantity and a dielectric loss error variable quantity; and when the capacitance breakdown error variable quantity or the dielectric loss error variable quantity is not 0, judging that the insulation state of the CVT is abnormal, and triggering a self-repairing material feedback control mechanism. According to the invention, on-line evaluation of the internal insulation state of the CVT is realized by on-line monitoring whether the capacitor of the CVT is broken down and whether the dielectric loss is abnormally increased.

Description

Technical Field

[0001] The present invention relates to the technical field of online monitoring of capacitive voltage transformers, and in particular to a method and system for evaluating the insulation state of a voltage transformer and for dielectric self-repair. Background Art

[0002] Capacitive voltage transformer (CVT) is mainly composed of voltage-dividing capacitor, dielectric layer and shielding layer, and realizes voltage measurement through the principle of capacitive voltage division. However, the capacitance value and dielectric loss factor of traditional CVT are easily affected by environmental factors such as temperature, humidity and equipment aging, resulting in increased measurement errors. Existing CVT insulation status assessment methods mainly rely on offline verification technology, such as evaluating insulation performance through periodic replacement or laboratory testing. This method is not only time-consuming, but also unable to achieve real-time monitoring of the CVT operating status, resulting in potential insulation faults that cannot be discovered and repaired in time.

[0003] In power equipment, the performance of dielectric materials directly affects the insulation and operational stability of the equipment, but the existing technology lacks a means of dynamically monitoring the loss factor of dielectric materials, and the application of self-healing materials is mainly concentrated in low-voltage or low-frequency scenarios, which is difficult to meet the needs of high-voltage and high-frequency environments. Quantum sensing technology has significant advantages in the sensitivity of capacitance values, but its application in power equipment is still in its early stages. Quantum computing technologies such as quantum impedance spectroscopy have potential in complex model parameter extraction and optimization, but existing technologies have not yet combined it with online monitoring of power equipment.

[0004] The invention patent with the patent publication number CN110082698A discloses a capacitive voltage transformer comprehensive operation status evaluation simulation system. By collecting zero-sequence current, it can realize online monitoring of CVT, reduce the need for offline calibration, and improve maintenance efficiency. The system structure is simple, only the current transformer and the measuring instrument are required, easy to install and maintain, and the change of zero-sequence current directly reflects the fault status of the equipment. The judgment logic is simple and clear, and easy to implement. However, the collection of zero-sequence current may be affected by external environmental factors such as load changes, interference current, etc., resulting in insufficient monitoring accuracy. This method only realizes fault monitoring and cannot actively maintain or repair the equipment.

[0005] The invention patent with the patent publication number CN118393420A discloses a capacitive voltage transformer error assessment method, medium and terminal. This method decomposes the voltage data through Fourier transform, combines environmental factors, establishes a multi-dimensional matrix, and uses the mean shift algorithm for classification and matching, which effectively improves the accuracy of error assessment, cleans and preprocesses the secondary side voltage data, and reduces the interference of external factors such as load, temperature and humidity on error judgment. The steps are clear and suitable for error assessment in different scenarios, with high practicality. However, this method is mainly aimed at error assessment and cannot realize real-time monitoring of the CVT operating status. It involves multiple technologies such as Fourier transform and mean shift algorithm, which is complex to implement and may increase system costs. It only focuses on error assessment and fails to realize active maintenance or repair of equipment. Summary of the invention

[0006] The technical problem to be solved by the present invention is to provide an online monitoring system for whether the capacitor of a CVT is broken down and whether the dielectric loss is abnormally increased under a high voltage and high frequency environment so as to realize an online evaluation of the insulation state inside the CVT.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] A voltage transformer insulation state assessment and dielectric self-repair method, comprising:

[0009] A quantum dot array sensor is embedded on the surface of the voltage-dividing capacitor unit of the CVT to obtain the high-voltage dielectric capacitance and medium-voltage dielectric capacitance of the CVT insulation layer; a high-frequency scanning signal is injected into the input port of the end screen of the CVT, and a local oscillator and a quantum mixer are set at the output port to obtain the dielectric loss factor;

[0010] Conduct circuit analysis on the CVT insulation circuit, and convert the high-voltage dielectric capacitance and medium-voltage dielectric capacitance of the insulation layer and the dielectric loss factor into characteristic parameters of the capacitance breakdown quantity and dielectric loss abnormality;

[0011] The characteristic parameters of the capacitor breakdown quantity and the abnormal characteristic parameters of the dielectric loss are respectively used as the input of the DBA model optimized based on WOA to obtain the change of the capacitor breakdown error and the change of the dielectric loss error;

[0012] When the capacitance breakdown error change or the dielectric loss error change is not 0, the CVT insulation state is determined to be abnormal, and the self-healing material feedback control mechanism is triggered.

[0013] In one embodiment of the present invention, obtaining the dielectric loss factor includes: a quantum mixer mixes an injected high-frequency scanning signal with a signal of a local oscillator to generate a difference frequency signal, and uses quantum phase-locked amplification technology to improve the signal ratio of the difference frequency signal, and then performs fast Fourier transform on the difference frequency signal after quantum phase-locked amplification to convert the time domain signal into a frequency domain signal, extract complex impedance, and obtain the dielectric loss factor.

[0014] In one embodiment of the present invention, obtaining a characteristic parameter of a capacitor breakdown quantity includes:

[0015] After simplifying the equivalent circuit model of CVT, the output voltage of the capacitor voltage divider unit is simplified according to the dielectric loss factor; based on the simplified output voltage of the capacitor voltage divider unit and the transformation ratio of the intermediate transformer, the secondary output voltage of CVT is obtained.

[0016] Obtain the additional ratio difference Δf generated in the CVT secondary output signal, and according to the voltage division ratio K of the CVT capacitor voltage division unit 0 , simplify the additional ratio difference to obtain the simplified additional ratio difference Δf';

[0017] Obtain an additional phase difference Δδ generated in the CVT secondary output signal, simplify the additional phase difference Δδ according to the voltage division ratio of the CVT capacitor voltage division unit before and after the change, and obtain the simplified additional phase difference Δδ;

[0018] Assume that the CVT has N H A high voltage dielectric capacitor and N M When the high-voltage dielectric capacitor of the CVT is broken down, the H At this time, the high-voltage dielectric capacitor C ΔH ; Similarly, obtain the medium voltage dielectric capacitance C ΔM ;

[0019] According to the high voltage dielectric capacitance C ΔH and medium voltage dielectric capacitor C ΔM , again substitute the additional ratio difference Δf' to obtain the additional ratio difference Δf with characteristic parameters v ;

[0020] Obtain the relationship between the secondary output voltages of different groups of the same-phase CVT at the same voltage level Where U GI is the secondary output voltage of the G-th group CVT measured for the first time, U JI is the secondary output voltage of the Jth group of CVT measured for the Ith time;

[0021] According to the additional ratio difference Δf v , CVT secondary output voltage relationship Simplify and obtain the simplified CVT secondary output voltage relationship

[0022] According to the simplified CVT secondary output voltage relationship Obtain the characteristic parameters of the capacitor breakdown quantity.

[0023] In one embodiment of the present invention, the capacitance breakdown quantity characteristic parameter T is obtained by the following formula:

[0024]

[0025] Where U n is the secondary output voltage of the nth group of CVT.

[0026] In one embodiment of the present invention, obtaining the dielectric loss abnormal characteristic parameter includes:

[0027] The phase relationship of the secondary output voltages of the same-phase CVTs of different groups at the same voltage level is substituted into the simplified additional phase difference formula to obtain the abnormal characteristic parameters of the dielectric loss.

[0028] In one embodiment of the present invention, the dielectric loss abnormal characteristic parameter θ is obtained by the following formula:

[0029]

[0030] In the formula, is the voltage phase of the nth group of CVT secondary outputs.

[0031] In one embodiment of the present invention, obtaining the capacitance breakdown error variation includes:

[0032] Standardize and pre-process the characteristic parameters of the capacitor breakdown quantity;

[0033] Set the initialization parameters of the DBA model: number of hidden layers, number of hidden layer units, and learning rate;

[0034] Each hidden layer includes one visible layer and one hidden layer, and the energy function of each hidden layer is constructed;

[0035] When the state of the visible layer is determined, the probability of the hidden layer nodes being activated by the visible layer nodes is calculated;

[0036] When the hidden layer state is determined, the probability of the visible layer node being activated by the hidden layer node is calculated;

[0037] Parameter update: Update the connection weights in the energy function, the bias of the hidden layer, and the bias of the visible layer;

[0038] The DBA model is trained by inputting samples composed of characteristic parameters of the number of capacitor breakdowns. After the training is completed, the output of the last layer of energy function is the change in capacitor breakdown error.

[0039] In one embodiment of the present invention, the energy function of each hidden layer is for:

[0040]

[0041] In the formula, v o is the state of the oth neuron in the visible layer, h g is the state of the g-th hidden layer neuron, w og is the visible node v o and hidden layer nodes h g The connection weight, a g and b o is the visible node v o and hidden layer nodes h g The bias of , q and s are the number of neurons in the visible layer and hidden layer, v is the visible layer, h is the hidden layer, It is the characteristic parameter of the capacitance breakdown quantity after standardized preprocessing;

[0042] Calculate the probability that the hidden layer nodes are activated by the visible layer nodes Obtained by the following formula:

[0043]

[0044] In the formula, r g is a random number on [0,1], is the hidden node h of the 0th hidden layer g , is the characteristic parameter of the capacitor breakdown quantity after standardized preprocessing of the 0th hidden layer, is the state of the oth neuron in the hidden layer of the 0th layer, v is the state of the neuron in the visible layer, and v 0 is the state of the initial visible layer neurons;

[0045] Calculate the probability that the visible layer node is activated by the hidden layer node Obtained by the following formula:

[0046]

[0047] In the formula, r o is a random number on [0,1], h 0 is the initial hidden layer.

[0048] In one embodiment of the present invention, the DBA model based on WOA optimization includes optimizing the number of neurons and the learning rate of the DBA model:

[0049] Represent the number of neurons in the hidden layer and the learning rate as a vector;

[0050] Randomly generate the individual positions of each vector to form an initial solution set;

[0051] Bring the vectors in each initial solution set into the constructed solution set, train and calculate the loss function value;

[0052] Among all the initial solution sets, find the solution that minimizes the loss function value;

[0053] The activation probability is determined according to the state of the previous hidden layer, and according to the activation probability, the target encirclement method and the target search method are selected to update the position of each individual alternately, or the spiral hunting method is selected to update the position of each individual;

[0054] Calculate the loss function of each individual's new position and compare its loss function with the loss function of the historical position. If the loss function of the current position is better, update it to the optimal position, and find the optimal number of neurons and learning rate corresponding to the optimal position and replace them in the DBN model.

[0055] In one embodiment of the present invention, different methods are selected to update the location of each individual, including:

[0056]

[0057] In the formula, A=a(2r 1 -1), C = 2r 2 , a=2-2t / L, r 1 and r 2 is a random number in [0,1], t is the number of iterations, L is the maximum number of iterations, a is the convergence factor, X(t+1) is the individual at the t+1th position, X * (t) is the optimal individual at the tth position, are alternating signs, X(t) rand is an arbitrary value of the initial solution set, b is the spiral constant, l is a random number in [0,1], P is the activation probability, X * (t)-A·|C·X * (t)-X(t)| is the target surround update, X(t) rand -A·|C·X(t) rand -X(t)| is the target search method update, |X * (t)-X(t)|·e bl ·cos(2πl)+X * (t) Updated for selecting spiral hunting mode.

[0058] In one embodiment of the present invention, the method of obtaining the dielectric loss anomaly characteristic parameter is the same as the method of obtaining the capacitance breakdown error variation, the difference being that the training and input samples of the DBA model are different.

[0059] In one embodiment of the present invention, a microcapsule epoxy composite is embedded in the CVT insulating layer; and a self-healing material feedback control mechanism is triggered, including:

[0060] Locate the specific area of ​​defects in the insulation layer based on the change in capacitance breakdown error and dielectric loss error and their change trends;

[0061] If the capacitance breakdown error variation deviates from zero, it indicates that the high-voltage or medium-voltage capacitance unit has breakdown or partial discharge; if the dielectric loss error variation deviates from zero, it indicates that the dielectric loss factor has increased abnormally, and the insulation material has aged or partially deteriorated;

[0062] The system generates a repair command signal and transmits the signal to the self-repair material activation unit in the target area through a quantum phase-locked amplifier module;

[0063] Applying a local high-frequency electric field to the target area increases the dielectric loss of the microcapsule epoxy composite shell, causing the shell to rupture and release the repair agent;

[0064] The liquid siloxane in the microcapsule epoxy composite migrates to the defective area of ​​the insulating layer under the drive of the electric field gradient, fills the defects through chemical cross-linking reaction, and restores the insulation performance;

[0065] After the repair is completed, the quantum dot array sensor is restarted to collect the high-voltage dielectric capacitance, medium-voltage dielectric capacitance and dielectric loss factor of the repaired CVT. After calculation and update, they are input into the DBA model optimized based on WOA to calculate the updated capacitance breakdown error change and dielectric loss error change.

[0066] If the capacitance breakdown error variation and dielectric loss error variation return to zero, the repair is successful and the system returns to normal monitoring mode;

[0067] If there is still a significant deviation, it is determined that the repair has not been fully effective, the secondary repair process is triggered and an early warning signal is sent to the host computer, prompting manual intervention for maintenance.

[0068] The present invention also provides a voltage transformer insulation state assessment and dielectric self-repairing system, which uses the above-mentioned voltage transformer insulation state assessment and dielectric self-repairing method, including:

[0069] The data acquisition module is used to embed a quantum dot array sensor on the surface of the voltage-dividing capacitor unit of the CVT to obtain the high-voltage dielectric capacitance and medium-voltage dielectric capacitance of the CVT insulation layer; and inject a high-frequency scanning signal into the input port of the end screen of the CVT, and set a local oscillator and a quantum mixer at the output port to obtain the dielectric loss factor;

[0070] The characteristic parameter modeling module is used to perform circuit analysis on the CVT insulation circuit and convert the high-voltage dielectric capacitance and medium-voltage dielectric capacitance of the insulation layer and the dielectric loss factor into characteristic parameters of the capacitance breakdown quantity and abnormal dielectric loss;

[0071] The insulation state evaluation module is used to use the characteristic parameters of the capacitor breakdown quantity and the dielectric loss abnormality as the input of the DBA model optimized based on WOA, and obtain the capacitor breakdown error change and the dielectric loss error change;

[0072] The self-healing material feedback module determines that the CVT insulation state is abnormal and triggers the self-healing material feedback control mechanism when the capacitance breakdown error change or the dielectric loss error change is not zero.

[0073] Compared with the prior art, the present invention has the following beneficial effects:

[0074] The present invention realizes the online monitoring and active maintenance of the capacitive voltage transformer CVT by integrating quantum sensing and dielectric self-repairing collaborative technology, and has the following advantages: 1. Online monitoring and active maintenance: Through the quantum sensor array and deep learning model, real-time monitoring of the insulation state of the CVT is realized, and combined with the self-repairing material technology, active maintenance of the equipment is realized. 2. High precision and anti-interference: The quantum phase-locked amplification technology is adopted to significantly improve the signal-to-noise ratio and reduce the impact of external interference on the monitoring accuracy. 3. Comprehensive coverage of fault types: By constructing characteristic parameters, it can sensitively capture various fault types such as dielectric loss anomalies and capacitor breakdown, and realize comprehensive monitoring. 4. Self-repair function: By driving the release of self-repairing materials through electric fields, dynamic repair of insulation layer defects is realized, and the normal operation of the equipment is restored. 5. Wide applicability: It is suitable for power systems under high voltage and high frequency environments, and meets the monitoring and maintenance needs in complex scenarios.

[0075] By standardizing the characteristic parameters such as the number of capacitor breakdowns and the characteristics related to dielectric loss anomalies, a standardized characteristic parameter matrix is ​​formed. Subsequently, the characteristic parameters are trained using a deep belief network, and the number of neurons and learning rate of the DBN are optimized in combination with the whale optimization algorithm to construct an online monitoring model. Online data is collected and input into the trained DBN network model to calculate the online measurement error change. If the error change is not 0, the CVT insulation state is judged to be abnormal. This model realizes high-precision online monitoring of the CVT insulation state, providing a basis for subsequent self-repair control. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 The present invention is a flowchart of a method for evaluating the insulation status of a voltage transformer and for dielectric self-repairing according to an embodiment of the present invention.

[0077] Figure 2Schematic diagram of an equivalent circuit model of a CVT insulation layer according to an embodiment of the present invention.

[0078] Figure 3 It is an equivalent diagram of the insulation layer model of the CVT according to an embodiment of the present invention.

[0079] Figure 4 Schematic diagram of the RBM training process of an embodiment of the present invention.

[0080] Figure 5 The present invention is a block diagram of a voltage transformer insulation status assessment and dielectric self-repair system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0081] In order to facilitate those skilled in the art to understand the technical solution of the present invention, the technical solution of the present invention is further described in conjunction with the accompanying drawings of the specification.

[0082] The terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0083] See also Figure 1 As shown, the present invention provides a method for evaluating the insulation state of a voltage transformer and self-repairing a dielectric, comprising:

[0084] S10, embed a quantum dot array sensor on the surface of the voltage-dividing capacitor unit of the CVT to obtain the high-voltage dielectric capacitance and medium-voltage dielectric capacitance values ​​of the CVT insulation layer; and inject a high-frequency scanning signal into the input port of the end screen of the CVT, and set a local oscillator and a quantum mixer at the output port to obtain the dielectric loss factor.

[0085] In one embodiment of the present invention, a graphene quantum dot array sensor with a capacitance sensitivity of 0.1 pF is embedded in the surface of the voltage-dividing capacitor unit of the capacitive voltage transformer using quantum confinement effect. Distributed sensing units with a spacing of ≤5 mm are arranged to form a three-dimensional capacitance distribution monitoring network to measure the high-voltage dielectric capacitance C of the CVT insulation layer. H and medium voltage dielectric capacitor C M .

[0086] In this embodiment, obtaining the dielectric loss factor includes: a quantum mixer mixes the injected high-frequency scanning signal with a signal of a local oscillator to generate a difference frequency signal, and using a quantum phase-locked amplification technology to improve the signal ratio of the difference frequency signal.

[0087] The high-frequency sweep signal is 10kHz-1MHz. The signal of the local oscillator is used as a reference signal source in the CVT system to provide a stable signal, which is mixed with the injected sweep signal to generate a difference frequency signal, thereby extracting the low-frequency component of the original signal while suppressing high-frequency noise. The difference frequency signal uses quantum phase-locked amplification technology to increase the signal-to-noise ratio to ≥120dB, separate the signal component of interest from the weak signal, and minimize background noise and interference signals.

[0088] In this embodiment, the synchronous measurement of high-frequency dielectric response includes: performing fast Fourier transform on the difference frequency signal after quantum phase-locked amplification, converting the time domain signal into a frequency domain signal, extracting the complex impedance, and obtaining the dielectric loss factor.

[0089] In this embodiment, the complex impedance Z(w) is extracted:

[0090] Z(w)=Re+jXe;

[0091] Where Re is the real part of the complex impedance, Xe is the imaginary part of the complex impedance, j is the imaginary unit, and w is the frequency of the high-frequency scanning signal injected into the last screen input port of the CVT. According to the complex impedance calculation formula:

[0092]

[0093] The series resistance R s =Re, which indicates the loss caused by the resistance effect during the signal transmission of CVT; due to the parasitic inductance L p is small and can be ignored, so the overall capacitance of CVT is The dielectric loss factor can be calculated from the real and imaginary parts of the complex impedance: Further simplifying

[0094] S20, performing circuit analysis on the CVT insulation circuit, converting the high-voltage dielectric capacitance and the medium-voltage dielectric capacitance of the insulation layer and the dielectric loss factor into characteristic parameters of the capacitance breakdown quantity and abnormal dielectric loss.

[0095] In one embodiment of the present invention, by performing circuit analysis on the CVT insulation circuit, the high-voltage dielectric capacitance and medium-voltage dielectric capacitance values ​​of the CVT insulation layer and the dielectric loss factor obtained by measurement and calculation are converted into a capacitor breakdown quantity characteristic parameter T and a dielectric loss abnormality characteristic parameter θ. The capacitor breakdown quantity characteristic parameter T is related to the capacitor breakdown quantity, while the dielectric loss abnormality characteristic parameter θ is related to the dielectric loss change and has a high sensitivity. The online monitoring of the insulation state in the CVT can be achieved by real-time monitoring of the changes in the capacitor breakdown quantity characteristic parameter T and the dielectric loss abnormality characteristic parameter θ and their degree of change.

[0096] In one embodiment of the present invention, obtaining a characteristic parameter of a capacitor breakdown quantity includes:

[0097] S211, after simplifying the equivalent circuit model of the CVT, the output voltage of the capacitor voltage divider unit is simplified according to the dielectric loss factor; the secondary output voltage of the CVT is obtained according to the simplified output voltage of the capacitor voltage divider unit and the transformation ratio of the intermediate transformer.

[0098] See also Figure 2 , 3 As shown, in this embodiment, according to the physical structure of the CVT, the capacitor unit is equivalent to a model in which an ideal capacitor and a dielectric loss equivalent resistor are connected in parallel, and the intermediate transformer is equivalent to a T-type equivalent model, then an equivalent circuit model of the CVT can be obtained.

[0099] In this embodiment, the voltage borne by the electromagnetic unit is low, the internal insulation performance is stable, and it is not easy to fail. m , X m and the secondary load parameter R d , X d Compared with other equivalent parameters, it can be regarded as infinite, so the output of the CVT capacitor voltage divider is regarded as an open circuit. According to the Thevenin theorem, the capacitor voltage divider unit of the CVT is simplified. According to Kirchhoff's law, the voltage output of the equivalent model is:

[0100]

[0101] Among them, R M , R H For two different voltage divider resistors, are the output voltage of the capacitor voltage divider unit and the primary output voltage of the CVT respectively, and ω is the grid voltage frequency.

[0102] according to right Simplify to get:

[0103]

[0104] tanδ H , tanδ M They are the dielectric loss tangent values ​​of the high voltage and medium voltage of the capacitor voltage divider unit respectively.

[0105] In this embodiment, the CVT secondary output voltage is obtained By the following formula:

[0106]

[0107] In the formula, K Tis the transformation ratio of the intermediate transformer.

[0108] Internal insulation status abnormality judgment:

[0109] S212, obtaining the additional ratio difference Δf generated in the CVT secondary output signal, and calculating the voltage division ratio K of the CVT capacitor voltage division unit according to the voltage division ratio K of the CVT capacitor voltage division unit. 0 , simplify the additional ratio difference and obtain the simplified additional ratio difference Δf'.

[0110] In this embodiment, the value of the dielectric loss tangent is extremely small, and the additional ratio difference generated in the CVT secondary output signal is:

[0111]

[0112] In the formula, the high-voltage dielectric capacitance and medium-voltage dielectric capacitance are respectively H0 and C M0 Change to C ΔH and C ΔM The voltage division ratio K of the CVT capacitor voltage division unit 0 for:

[0113]

[0114] Substituting into the simplified additional ratio difference Δf' is

[0115]

[0116] S213, obtaining an additional phase difference Δδ generated in the CVT secondary output signal, simplifying the additional phase difference Δδ according to the voltage division ratio of the CVT capacitor voltage division unit before and after the change, and obtaining the simplified additional phase difference Δδ.

[0117] In this embodiment, the additional phase difference Δδ is obtained similarly, where the additional phase difference Δδ is obtained by the following formula:

[0118]

[0119] Where, tanδ ΔH , tanδ ΔM They are respectively the high voltage dielectric loss δ ΔH and medium voltage dielectric loss δ ΔM The tangent value of .

[0120] Substitute the voltage divider ratio K before the change 0 Simplifying the voltage divider ratio ΔK after the change, we can get:

[0121]

[0122] Where, tanδ H0 , tanδ M0They are respectively the high voltage dielectric loss δ M0 and medium voltage dielectric loss δ H0 The tangent value of .

[0123] Construct the characteristic parameters of the capacitor breakdown quantity:

[0124] S214, assuming that the CVT has N H A high voltage dielectric capacitor and N M When the high-voltage dielectric capacitor of the CVT is broken down, the H At this time, the high-voltage dielectric capacitor C ΔH ; Similarly, obtain the medium voltage dielectric capacitance C ΔM .

[0125] In one embodiment of the present invention, the high voltage dielectric capacitor C ΔH for:

[0126]

[0127] Similarly, the medium voltage dielectric capacitor C ΔM for:

[0128]

[0129] Where n M The number of medium-voltage dielectric capacitors of the CVT that are broken down

[0130] S215, according to the high voltage dielectric capacitor C ΔH and medium voltage dielectric capacitor C ΔM , again substitute the additional ratio difference Δf' to obtain the additional ratio difference Δf with characteristic parameters v .

[0131] In this embodiment, the additional ratio difference Δf v for:

[0132]

[0133] S216, obtaining the relationship between the secondary output voltages of the same-phase CVTs of different groups at the same voltage level Where U GI is the secondary output voltage of the G-th group CVT measured for the first time, U JI is the secondary output voltage measured for the Ith time for the Jth group of CVTs.

[0134] In this embodiment, the CVT secondary output voltage relationship is for:

[0135]

[0136] In the formula, f GIis the ratio difference of the G group CVT measured for the first time, U GI Δf is the secondary output voltage measured for the first time by the Gth group of CVTs. GI is the additional ratio difference of the G group CVT measured for the first time. Similarly, U JI 、f JI , Δf JI .

[0137] S217, according to the additional ratio difference Δf v , CVT secondary output voltage relationship Simplify and obtain the simplified CVT secondary output voltage relationship

[0138] In this embodiment, the simplified CVT secondary output voltage relationship is for:

[0139]

[0140] Where Δf GJ is the additional ratio difference of backward difference, is the forward difference additional ratio difference. Specifically, the formula is as follows:

[0141] Δf GJ =Δf G -Δf J ;

[0142] Δf GJ =Δf G +Δf J ;

[0143] Where Δf G is the additional ratio difference of group G, Δf J is the additional ratio difference of group K.

[0144] S218, according to the simplified CVT secondary output voltage relationship Obtain the characteristic parameters of the capacitor breakdown quantity.

[0145] In this embodiment, the final input yields:

[0146]

[0147] Where U G , U J are the secondary output voltages of the CVTs of the G and J groups, respectively, G0 、f J0 are the ratio differences of the initial measurements of the G and J groups of CVT, N GH 、N JHare the number of high-voltage dielectric capacitors in groups G and K, respectively, N GM 、N JM is the number of medium voltage dielectric capacitors in groups G and J, n GH 、n JH are the number of high-voltage dielectric capacitors that are broken down in groups G and J, respectively, n GM 、n JM is the number of medium voltage dielectric capacitors that are broken down in groups G and J, K G0 , K J0 are the voltage divider ratios in groups G and J respectively.

[0148] Then the characteristic parameter of the capacitor breakdown quantity is:

[0149]

[0150] Where U n is the secondary output voltage of the nth group of CVT.

[0151] In one embodiment of the present invention, obtaining the dielectric loss abnormal characteristic parameter includes: substituting the phase relationship of the secondary output voltages of the same-phase CVTs of different groups at the same voltage level into the simplified additional phase difference formula to obtain the dielectric loss abnormal characteristic parameter.

[0152] In this embodiment, the phase relationship of the secondary output voltages of the same-phase CVTs of different groups at the same voltage level is:

[0153]

[0154] In the formula, δ GI , δ JI The dielectric loss angle of the G and J groups of CVTs measured for the first time, Δδ GI , Δδ JI are the additional phase differences of the G and J groups of CVT measured for the first time, δ G0 , δ j0 are the dielectric loss angles of the G and J groups of CVTs initially measured, They are the secondary output voltage phases of the G and J groups of CVT measured for the Ith time respectively.

[0155] In this embodiment, the simplified additional phase difference formula is introduced to obtain the dielectric loss abnormal characteristic parameter:

[0156]

[0157] In the formula, K GI , K JI The voltage divider ratio of the G and J groups of CVT measured for the first time, tanδ GMI , tanδGHI are the tangent values ​​of the medium and high voltage dielectric losses measured for the first time in group G, tanδ GM0 , tanδ GH0 are the tangent values ​​of the medium and high voltage dielectric losses of the initial measurement of group G, tanδ JMI , tanδ JHI are the tangent values ​​of the medium and high voltage dielectric losses measured for the first time in the Jth group, tanδ JM0 , tanδ JH0 are the tangent values ​​of the medium and high voltage dielectric losses of the initial measurement of group J, K G0 It is the voltage division ratio initially measured for the G group CVT.

[0158] Then, the constructed dielectric loss anomaly characteristic parameter θ is obtained by the following formula:

[0159]

[0160] In the formula, is the voltage phase of the secondary output of the nth group of CVTs. The dielectric loss anomaly characteristic parameter θ is only related to the phase difference of the CVT and can sensitively capture the dielectric loss anomaly.

[0161] S30, using the characteristic parameter of the capacitor breakdown quantity and the characteristic parameter of the dielectric loss abnormality as inputs of the DBA model optimized based on WOA, respectively, to obtain the capacitor breakdown error change and the dielectric loss error change.

[0162] See also Figure 4 As shown, in one embodiment of the present invention, each layer of the restricted Boltzmann machine (RBM) of the deep belief network consists of a visible layer v and a hidden layer h, the neurons between the layers are unconnected, and the neurons between the visible layer v and the hidden layer h are fully connected, thereby forming an undirected graph model. The input layer input parameters are the standardized feature parameter matrix and The training process is for the feature parameter matrix and Similarly, the training process is based on the feature parameter matrix For example.

[0163] In this embodiment, obtaining the capacitance breakdown error variation includes:

[0164] S311, performing standardization preprocessing on the characteristic parameters of the capacitor breakdown quantity.

[0165] S312, set the initialization parameters of the DBA model: the number of hidden layers, the number of hidden layer units and the learning rate.

[0166] S313, each hidden layer includes a visible layer and a hidden layer, and an energy function of each hidden layer is constructed.

[0167] In this embodiment, each hidden layer includes a visible layer and a hidden layer. At the same time, the hidden layer of the previous hidden layer will serve as the visible layer of the next hidden layer. The energy function of each hidden layer is:

[0168]

[0169] In the formula, v o is the state of the oth neuron in the visible layer, h g is the state of the g-th hidden layer neuron, w og is the visible node v o and hidden layer nodes h g The connection weight, a g and b o is the visible node v o and hidden layer nodes h g The bias of , q and s are the number of neurons in the visible layer and hidden layer, v is the visible layer, h is the hidden layer, It is the characteristic parameter of the capacitance breakdown quantity after standardized preprocessing;

[0170] S314, when the state of the explicit layer is determined, the probability of the hidden layer node being activated by the explicit layer node is calculated.

[0171] In this embodiment, the probability of hidden layer nodes being activated by visible layer nodes is calculated Obtained by the following formula:

[0172]

[0173] In the formula, r g is a random number on [0,1], is the hidden node h of the 0th hidden layer g , is the characteristic parameter of the capacitor breakdown quantity after standardized preprocessing of the 0th hidden layer, is the state of the oth neuron in the hidden layer of the 0th layer, v is the state of the neuron in the visible layer, and v 0 is the initial state of the neurons in the visible layer.

[0174] S315, when the hidden layer state is determined, the probability of the visible layer node being activated by the hidden layer node is calculated.

[0175] In this embodiment, the probability of the visible layer node being activated by the hidden layer node is calculated Obtained by the following formula:

[0176]

[0177] In the formula, r o is a random number on [0,1], h 0is the initial hidden layer.

[0178] S316, parameter update: update the connection weights in the energy function, the bias of the hidden layer, and the bias of the visible layer.

[0179] In this embodiment, the connection weight is updated:

[0180]

[0181] Bias update of hidden layer:

[0182]

[0183] Update bias of the display layer:

[0184]

[0185] in is the initial input sample, is the first reconstructed sample, and γ is the learning rate. After repeated training x times, the last layer of energy function outputs the reconstructed sample Output the results for training.

[0186] S317, inputting samples composed of characteristic parameters of the number of capacitor breakdown to train the DBA model. After the training is completed, the output of the last layer of energy function is the change in capacitor breakdown error.

[0187] Whale Optimization Algorithm (WOA) parameter optimization:

[0188] In one embodiment of the present invention, compared with other algorithms, the WOA algorithm can efficiently find the optimal number of neurons and learning rate of the DBN model, reduce training time, improve the convergence speed of the model, and use global search to effectively avoid falling into the local optimal solution. For online monitoring, it can significantly improve the speed and accuracy of online monitoring evaluation. The number of neurons in the hidden layer h is α, and the learning rate is γ. Find the optimal number of neurons α of DBN through the WOA algorithm a and the learning rate γ a .

[0189] In this embodiment, the number of neurons and the learning rate of the DBA model are optimized, including:

[0190] S321, the number of neurons and the learning rate of the hidden layer are represented by a vector.

[0191] X = [α, γ];

[0192] S322, randomly generate the individual position of each vector to form an initial solution set.

[0193] In this embodiment, the initial solution set {X 1,X 2 ,…,X Q}, where Q is the number of initial solution sets constructed.

[0194] S323, bring each vector in the initial solution set into the constructed solution set, train and calculate the loss function value.

[0195] In this embodiment, the loss function L e for:

[0196]

[0197] S324, find the solution that minimizes the loss function value among all initial solution sets.

[0198] In this embodiment, the solution X that minimizes the loss function value is * =min{L 1 ,L 2 ,…,L Q}, L Q is the loss function of the Qth initial solution.

[0199] S325, determining the activation probability according to the state of the previous hidden layer, and according to the activation probability, selecting the target encirclement method and the target search method to update the position of each individual alternately or selecting the spiral hunting method to update the position of each individual.

[0200] In this embodiment, the activation probability is determined according to the state of the previous hidden layer. According to the activation probability, when it is the visible layer state, the activation probability is When it is a hidden layer state, the activation probability is

[0201] In this embodiment, different methods are selected to update the location of each individual, including:

[0202]

[0203] In the formula, A=a(2r 1 -1), C = 2r 2 , a=2-2t / L, r 1 and r 2 is a random number in [0,1], t is the number of iterations, L is the maximum number of iterations, a is the convergence factor, X(t+1) is the individual at the t+1th position, X * (t) is the optimal individual at the tth position, are alternating signs, X(t) ramd is an arbitrary value of the initial solution set, b is the spiral constant, l is a random number in [0,1], P is the activation probability, X * (t)-A·|C·X *(t)-X(t)| is the target surround update, X(t) rand -A·|C·X(t) rand -X(t)| is the target search method update, |X * (t)-X(t)|·e bl ·cos(2πl)+X * (t) Updated for selecting spiral hunting mode.

[0204] In this embodiment, the target encirclement method: at this time, the optimal position in the group is closest to the global optimal solution, so the remaining positions are updated according to the current optimal position to reduce the encirclement of the target position, and the t+1th position update is:

[0205] X(t+1)=X * (t)-A·|C·X * (t)-X(t)|;

[0206] In this embodiment, the target search method is: perform position optimization search based on the random walk mechanism, and autonomously update the position of each group according to the position information of the entire group. According to the result of the tth optimization search, the position update of the t+1th time is:

[0207] X(t+1)=X(t) rand -A·|C·X(t) rand -X(t)|

[0208] Where X(t) rand is an arbitrary value of the initial solution set.

[0209] In this embodiment, WOA simulates the behavior of a whale to surround and approach the optimal position along a spiral path to find the optimal position. The equation of the spiral path is as follows:

[0210] X(t+1)=|X * (t)-X(t)|·e bl ·cos(2πl)+X * (t);

[0211] S326, calculate the loss function of each individual's new position, and compare its loss function with the loss function of the historical position. If the loss function of the current position is better, update it to the optimal position, and find the optimal number of neurons and learning rate corresponding to the optimal position to replace them in the DBN model.

[0212] In this embodiment, the new position of each whale is calculated and its loss function is evaluated to compare the current position with the historical optimal position X * , if the loss function of the current position is better, update X * . The optimal α will be founda and γ a Replace it into the DBN model.

[0213] S40, when the capacitance breakdown error variation or the dielectric loss error variation is not zero, it is determined that the CVT insulation state is abnormal, and the self-healing material feedback control mechanism is triggered.

[0214] In this embodiment, the high-voltage dielectric capacitance and the medium-voltage dielectric capacitance values ​​are collected online and input into the trained DBN network model, and the capacitance breakdown error change and the dielectric loss error change are measured online. If the capacitance breakdown error change or the dielectric loss error change is not 0, that is, it exceeds the energy threshold, then it is determined that the CVT insulation state is abnormal.

[0215] In this embodiment, a microcapsule epoxy composite is pre-embedded in the CVT insulation layer to trigger a self-healing material feedback control mechanism, including:

[0216] S41, locating a specific area of ​​the defect in the insulation layer according to the capacitance breakdown error change and the dielectric loss error change and their change trends.

[0217] In this embodiment, according to the capacitance breakdown error variation ΔT o and dielectric loss error variation Δθ o The numerical value and changing trend of the dielectric breakdown quantity and dielectric loss anomaly in the CVT equivalent circuit model are combined to locate the specific area of ​​defects in the insulation layer.

[0218] S42, if the change in the capacitor breakdown error deviates from the zero value, it indicates that the high-voltage or medium-voltage capacitor unit has breakdown or partial discharge; if the change in the dielectric loss error deviates from the zero value, it indicates that the dielectric loss factor has increased abnormally and the insulating material has aged or partially degraded.

[0219] S43, the system generates a repair instruction signal and transmits the signal to the self-repair material activation unit in the target area through the quantum phase-locked amplifier module.

[0220] In this embodiment, a microcapsule epoxy composite is embedded in the CVT insulating layer, the capsule shell is made of a dielectric response sensitive material, and the complex command signal activates the material in the following manner.

[0221] S44, applying a local high-frequency electric field to the target area to increase the dielectric loss of the microcapsule epoxy composite shell, causing the shell to rupture and release the repair agent.

[0222] In this embodiment, a local high-frequency electric field is applied to the target area, with a frequency range of 1kHz-100kHz, so that the dielectric loss of the microcapsule shell increases, the temperature rises to a critical value of about 60°C, and the shell ruptures to release the repair agent.

[0223] S45, the liquid siloxane in the microcapsule epoxy composite migrates directionally to the defective area of ​​the insulating layer under the drive of the electric field gradient, fills the defects through chemical cross-linking reaction, and restores the insulating properties.

[0224] S46, after the repair is completed, restart the quantum dot array sensor, collect the high-voltage dielectric capacitance and medium-voltage dielectric capacitance values ​​and dielectric loss factor of the repaired CVT, calculate and update them, and input them into the DBA model optimized based on WOA to calculate the updated capacitance breakdown error change and dielectric loss error change.

[0225] S47, if the capacitance breakdown error variation and dielectric loss error variation return to zero, the repair is successful and the system returns to normal monitoring mode.

[0226] S48: If there is still a significant deviation, it is determined that the repair has not been fully effective, triggering the secondary repair process and sending an early warning signal to the upper computer, prompting manual intervention for maintenance.

[0227] In one embodiment of the present invention, the triggering self-repairing material feedback control mechanism of the present invention also includes a self-repairing material replenishment mechanism. The system has a built-in redundant microcapsule storage unit, and the repair material is replenished on demand through a piezoelectric micropump. After each repair, the three-dimensional capacitance distribution monitoring network updates the three-dimensional capacitance distribution map of the insulating layer in real time, dynamically adjusts the distribution density and position of the microcapsules, and ensures the accuracy and material utilization of subsequent repairs.

[0228] See also Figure 5 As shown, the present invention also provides a voltage transformer insulation state assessment and dielectric self-repairing system, which applies the above-mentioned voltage transformer insulation state assessment and dielectric self-repairing method, including:

[0229] The data acquisition module is used to embed a quantum dot array sensor on the surface of the voltage-dividing capacitor unit of the CVT to obtain the high-voltage dielectric capacitance and medium-voltage dielectric capacitance values ​​of the CVT insulation layer; and inject a high-frequency scanning signal into the input port of the end screen of the CVT, and set a local oscillator and a quantum mixer at the output port to obtain the dielectric loss factor.

[0230] The characteristic parameter modeling module is used to perform circuit analysis on the CVT insulation circuit, and convert the high-voltage dielectric capacitance and medium-voltage dielectric capacitance values ​​of the insulation layer and the dielectric loss factor into characteristic parameters of the capacitance breakdown quantity and abnormal dielectric loss.

[0231] The insulation state assessment module is used to use the characteristic parameters of the capacitor breakdown quantity and the abnormal characteristic parameters of the dielectric loss as the input of the DBA model optimized based on WOA, and obtain the change of the capacitor breakdown error and the change of the dielectric loss error.

[0232] The self-healing material feedback module determines that the CVT insulation state is abnormal and triggers the self-healing material feedback control mechanism when the capacitance breakdown error change or the dielectric loss error change is not zero.

[0233] It is obvious to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential features of the present invention. Therefore, the embodiments should be regarded as exemplary and non-limiting from any point of view, and the scope of the present invention is defined by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention, and any reference numerals in the claims should not be regarded as limiting the claims involved.

[0234] The above-described embodiments merely represent implementation methods of the invention. The protection scope of the present invention is not limited to the above-described embodiments. For those skilled in the art, several modifications and improvements may be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. A voltage transformer insulation state assessment and dielectric self-repair method, characterized in that: include: A quantum dot array sensor is embedded on the surface of the voltage-dividing capacitor unit of the CVT to obtain the high-voltage dielectric capacitance and medium-voltage dielectric capacitance values ​​of the CVT insulation layer; A high-frequency scanning signal is injected into the input port of the last screen of the CVT, and a local oscillator and a quantum mixer are set at the output port to obtain the dielectric loss factor; Conduct circuit analysis on the CVT insulation circuit, and convert the high-voltage dielectric capacitance and medium-voltage dielectric capacitance of the insulation layer and the dielectric loss factor into characteristic parameters of the capacitance breakdown quantity and dielectric loss abnormality; The characteristic parameters of the capacitor breakdown quantity and the dielectric loss abnormality are used as the input of the DBA model optimized based on WOA, respectively, to obtain the capacitor breakdown error change and the dielectric loss error change; When the capacitance breakdown error change or the dielectric loss error change is not 0, the CVT insulation state is determined to be abnormal, and the self-healing material feedback control mechanism is triggered.

2. The voltage transformer insulation state assessment and dielectric self-repair method according to claim 1, characterized in that: Obtaining the dielectric loss factor includes: a quantum mixer mixes the injected high-frequency scanning signal with the signal of the local oscillator to generate a difference frequency signal, and uses quantum phase-locked amplification technology to improve the signal ratio of the difference frequency signal, and then performs fast Fourier transformation on the difference frequency signal after quantum phase-locked amplification to convert the time domain signal into a frequency domain signal, extract the complex impedance, and obtain the dielectric loss factor.

3. The voltage transformer insulation state assessment and dielectric self-repair method according to claim 1, characterized in that: Obtain the characteristic parameters of the capacitor breakdown quantity, including: After simplifying the equivalent circuit model of CVT, the output voltage of the capacitor voltage divider unit is simplified according to the dielectric loss factor; based on the simplified output voltage of the capacitor voltage divider unit and the transformation ratio of the intermediate transformer, the secondary output voltage of CVT is obtained. Obtaining an additional ratio difference Δf generated in the CVT secondary output signal, simplifying the additional ratio difference according to the voltage division ratio K0 of the CVT capacitor voltage division unit, and obtaining a simplified additional ratio difference Δf'; Obtain an additional phase difference Δδ generated in the CVT secondary output signal, simplify the additional phase difference Δδ according to the voltage division ratio of the CVT capacitor voltage division unit before and after the change, and obtain the simplified additional phase difference Δδ; Assume that the CVT has N H A high voltage dielectric capacitor and N M When the high-voltage dielectric capacitor of the CVT is broken down, the H At this time, the high-voltage dielectric capacitor C ΔH ; Similarly, obtain the medium voltage dielectric capacitance C ΔM ; According to the high voltage dielectric capacitance C ΔH and medium voltage dielectric capacitor C ΔM , again substitute the additional ratio difference Δf' to obtain the additional ratio difference Δf with characteristic parameters v ; Obtain the relationship between the secondary output voltages of different groups of the same-phase CVT at the same voltage level Where U GI is the secondary output voltage of the G-th group CVT measured for the first time, U JI is the secondary output voltage of the Jth group of CVT measured for the Ith time; According to the additional ratio difference Δf v , CVT secondary output voltage relationship Simplify and obtain the simplified CVT secondary output voltage relationship According to the simplified CVT secondary output voltage relationship Obtain the characteristic parameters of the capacitor breakdown quantity.

4. The voltage transformer insulation status assessment and dielectric self-repair method according to claim 3, characterized in that: The characteristic parameter T of the capacitor breakdown quantity is obtained by the following formula: Where U n is the secondary output voltage of the nth group of CVT.

5. The voltage transformer insulation state assessment and dielectric self-repair method according to claim 3, characterized in that: Obtain dielectric loss abnormal characteristic parameters, including: The phase relationship of the secondary output voltages of the same-phase CVTs of different groups at the same voltage level is substituted into the simplified additional phase difference formula to obtain the abnormal characteristic parameters of the dielectric loss.

6. The voltage transformer insulation status assessment and dielectric self-repair method according to claim 5, characterized in that: The dielectric loss abnormal characteristic parameter θ is obtained by the following formula: In the formula, is the voltage phase of the nth group of CVT secondary outputs.

7. The voltage transformer insulation state assessment and dielectric self-repair method according to claim 1, characterized in that: Obtain the capacitance breakdown error change, including: Standardize and pre-process the characteristic parameters of the capacitor breakdown quantity; Set the initialization parameters of the DBA model: number of hidden layers, number of hidden layer units, and learning rate; Each hidden layer includes one visible layer and one hidden layer, and the energy function of each hidden layer is constructed; When the state of the visible layer is determined, the probability of the hidden layer nodes being activated by the visible layer nodes is calculated; When the hidden layer state is determined, the probability of the visible layer node being activated by the hidden layer node is calculated; Parameter update: Update the connection weights in the energy function, the bias of the hidden layer, and the bias of the visible layer; The DBA model is trained by inputting samples composed of characteristic parameters of the number of capacitor breakdowns. After the training is completed, the output of the last layer of energy function is the change in capacitor breakdown error.

8. The voltage transformer insulation status assessment and dielectric self-repair method according to claim 7, characterized in that: Energy function for each hidden layer for: In the formula, v o is the state of the oth neuron in the visible layer, h g is the state of the g-th hidden layer neuron, w og is the visible node v o and hidden layer nodes h g The connection weight, a g and b o is the visible node v o and hidden layer nodes h g The bias of , q and s are the number of neurons in the visible layer and hidden layer, v is the visible layer, h is the hidden layer, It is the characteristic parameter of the capacitance breakdown quantity after standardized preprocessing; Calculate the probability that the hidden layer nodes are activated by the visible layer nodes Obtained by the following formula: In the formula, r g is a random number on [0,1], is the hidden node h of the 0th hidden layer g , is the characteristic parameter of the capacitor breakdown quantity after standardized preprocessing of the 0th hidden layer, is the state of the oth neuron in the hidden layer of the 0th layer, v is the state of the neuron in the visible layer, and v 0 is the state of the initial visible layer neurons; Calculate the probability that the visible layer node is activated by the hidden layer node Obtained by the following formula: In the formula, r o is a random number on [0,1], h 0 is the initial hidden layer.

9. The voltage transformer insulation status assessment and dielectric self-repair method according to claim 7, characterized in that: The DBA model optimized based on WOA includes optimizing the number of neurons and learning rate of the DBA model: Represent the number of neurons in the hidden layer and the learning rate as a vector; Randomly generate the individual positions of each vector to form an initial solution set; Bring the vectors in each initial solution set into the constructed solution set, train and calculate the loss function value; Among all the initial solution sets, find the solution that minimizes the loss function value; The activation probability is determined according to the state of the previous hidden layer, and according to the activation probability, the target encirclement method and the target search method are selected to update the position of each individual alternately, or the spiral hunting method is selected to update the position of each individual; Calculate the loss function of each individual's new position and compare its loss function with the loss function of the historical position. If the loss function of the current position is better, update it to the optimal position, and find the optimal number of neurons and learning rate corresponding to the optimal position and replace them in the DBN model.

10. The voltage transformer insulation status assessment and dielectric self-repair method according to claim 9, characterized in that: Choose from different ways to update the location of each individual, including: Where A = a(2r1-1), C = 2r2, a = 2-2t / L, r1 and r2 are random numbers in [0,1], t is the number of iterations, L is the maximum number of iterations, a is the convergence factor, X(t+1) is the individual at the t+1th position, X * (t) is the optimal individual at the tth position, are alternating signs, X(t) rand is an arbitrary value of the initial solution set, b is the spiral constant, l is a random number in [0,1], P is the activation probability, X * (t)-A·|C·X * (t)-X(t)| is the target surround update, X(t) rand -A·|C·X(t) rand -X(t)| is the target search method update, |X * (t)-X(t)|·e bl ·cos(2πl)+X * (t) Updated for selecting spiral hunting mode.

11. The voltage transformer insulation state assessment and dielectric self-repair method according to claim 7, characterized in that: The method of obtaining the abnormal characteristic parameters of dielectric loss is the same as the method of obtaining the change in capacitor breakdown error. The difference is that the training and input samples of the DBA model are different.

12. The voltage transformer insulation state assessment and dielectric self-repair method according to claim 1, characterized in that: Pre-embed microcapsule epoxy composite in CVT insulation layer; trigger self-healing material feedback control mechanism, including: Locate the specific area of ​​defects in the insulation layer based on the change in capacitance breakdown error and dielectric loss error and their change trends; If the capacitance breakdown error variation deviates from zero, it indicates that the high-voltage or medium-voltage capacitance unit has breakdown or partial discharge; if the dielectric loss error variation deviates from zero, it indicates that the dielectric loss factor has increased abnormally, and the insulation material has aged or partially deteriorated; The system generates a repair command signal and transmits the signal to the self-repair material activation unit in the target area through a quantum phase-locked amplifier module; Applying a local high-frequency electric field to the target area increases the dielectric loss of the microcapsule epoxy composite shell, causing the shell to rupture and release the repair agent; The liquid siloxane in the microcapsule epoxy composite migrates to the defective area of ​​the insulating layer under the drive of the electric field gradient, fills the defects through chemical cross-linking reaction, and restores the insulation performance; After the repair is completed, the quantum dot array sensor is restarted to collect the high-voltage dielectric capacitance and medium-voltage dielectric capacitance values ​​and dielectric loss factor of the repaired CVT, and the updated values ​​are input into the DBA model based on WOA optimization to calculate the updated capacitance breakdown error change and dielectric loss error change. If the capacitance breakdown error variation and dielectric loss error variation return to zero, the repair is successful and the system returns to normal monitoring mode; If there is still a significant deviation, it is determined that the repair has not been fully effective, the secondary repair process is triggered and an early warning signal is sent to the host computer, prompting manual intervention for maintenance.

13. A voltage transformer insulation status assessment and dielectric self-repair system, characterized in that: The voltage transformer insulation state assessment and dielectric self-repair method according to any one of claims 1 to 12 comprises: The data acquisition module is used to embed a quantum dot array sensor on the surface of the voltage-dividing capacitor unit of the CVT to obtain the high-voltage dielectric capacitance and medium-voltage dielectric capacitance of the CVT insulation layer; and inject a high-frequency scanning signal into the input port of the end screen of the CVT, and set a local oscillator and a quantum mixer at the output port to obtain the dielectric loss factor; The characteristic parameter modeling module is used to perform circuit analysis on the CVT insulation circuit and convert the high-voltage dielectric capacitance and medium-voltage dielectric capacitance of the insulation layer and the dielectric loss factor into characteristic parameters of the capacitance breakdown quantity and abnormal dielectric loss; The insulation state evaluation module is used to use the characteristic parameters of the capacitor breakdown quantity and the dielectric loss abnormality as the input of the DBA model optimized based on WOA, and obtain the capacitor breakdown error change and the dielectric loss error change; The self-healing material feedback module determines that the CVT insulation state is abnormal and triggers the self-healing material feedback control mechanism when the capacitance breakdown error change or the dielectric loss error change is not zero.

Images