CN117390348B - Method and system for capacity attenuation treatment of metallized film capacitor - Google Patents
Method and system for capacity attenuation treatment of metallized film capacitor Download PDFInfo
- Publication number
- CN117390348B CN117390348B CN202311680354.0A CN202311680354A CN117390348B CN 117390348 B CN117390348 B CN 117390348B CN 202311680354 A CN202311680354 A CN 202311680354A CN 117390348 B CN117390348 B CN 117390348B
- Authority
- CN
- China
- Prior art keywords
- capacitor
- distribution
- electric field
- neural network
- capacitance
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 239000003990 capacitor Substances 0.000 title claims abstract description 115
- 238000000034 method Methods 0.000 title claims abstract description 63
- 239000011104 metalized film Substances 0.000 title claims abstract description 21
- 230000005684 electric field Effects 0.000 claims abstract description 48
- 238000013528 artificial neural network Methods 0.000 claims abstract description 37
- 238000012545 processing Methods 0.000 claims description 15
- 238000012360 testing method Methods 0.000 claims description 11
- 230000008569 process Effects 0.000 claims description 10
- 238000005259 measurement Methods 0.000 claims description 9
- 238000012549 training Methods 0.000 claims description 4
- 239000011159 matrix material Substances 0.000 claims description 2
- 230000000536 complexating effect Effects 0.000 claims 1
- 230000032683 aging Effects 0.000 abstract description 8
- 239000010408 film Substances 0.000 description 12
- 230000006870 function Effects 0.000 description 12
- 238000004364 calculation method Methods 0.000 description 7
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 238000009413 insulation Methods 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000002238 attenuated effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000013016 damping Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000005562 fading Methods 0.000 description 1
- 239000011810 insulating material Substances 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 229920000642 polymer Polymers 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 230000035882 stress Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000010409 thin film Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/11—Complex mathematical operations for solving equations, e.g. nonlinear equations, general mathematical optimization problems
- G06F17/13—Differential equations
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R27/00—Arrangements for measuring resistance, reactance, impedance, or electric characteristics derived therefrom
- G01R27/02—Measuring real or complex resistance, reactance, impedance, or other two-pole characteristics derived therefrom, e.g. time constant
- G01R27/26—Measuring inductance or capacitance; Measuring quality factor, e.g. by using the resonance method; Measuring loss factor; Measuring dielectric constants ; Measuring impedance or related variables
- G01R27/2605—Measuring capacitance
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/003—Environmental or reliability tests
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0499—Feedforward networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Abstract
The invention provides a method and a system for capacity attenuation treatment of a metallized film capacitor, wherein the method comprises the following steps: according to the actual use condition of the metallized film capacitor, firstly measuring the electric field distribution condition of the finite point of the capacitor, and then adopting the neural network PINNs inversion based on the physical information to obtain the continuous dielectric constantDistribution, again according to dielectric constantAnd geometry to obtain capacitance under practical conditionsC r Finally and initially hold valueC 0 Comparing to calculate the capacity attenuation rate of the capacitorδIts lifetime was evaluated. The invention solves the technical problems of low capacitor capacity attenuation, low prediction precision of the estimated capacitor life, and low safety and reliability of electronic equipment caused by the difficulty in identifying an aging device in advance.
Description
Technical Field
The invention relates to the field of fault monitoring of electronic equipment, in particular to a method and a system for capacity attenuation processing of a metallized film capacitor.
Background
The metallized film capacitor has the advantages of good self-healing property, no polarity, high insulation resistance, excellent frequency property, high stability and the like, and is widely applied to the fields of power systems, pulse power, aerospace and the like. However, the polymer insulating material can age under the action of various stresses such as electricity, heat, machinery and the like to cause the insulation performance to be reduced, and the capacitance value of the capacitor is gradually attenuated, so that the normal function is finally affected. Therefore, life assessment and prediction of thin film capacitors is critical to ensure high reliability applications and failure prediction thereof. The failure criterion is generally a capacitance loss of more than 5% of the initial capacitance. In order to ensure proper operation of the metallized film capacitor, it is necessary to be able to calculate the residual capacitance of the capacitor during the charge and discharge operations.
The traditional method for estimating the service life of the film capacitor generally carries out durability or reliability test on the film capacitor, and has the advantages of long time consumption, high cost and weak result pertinence. The life estimation method based on data driving does not need to carry out a large number of tests, can mine aging characteristics from historical data to carry out online monitoring or offline testing, greatly reduces cost and complexity, and has good application prospect.
In the prior patent application publication No. CN103543346A, a film capacitor prediction method is mainly used for predicting the service life of a film capacitor by simulating the actual use condition of the film capacitor by using a programmable temperature and humidity control box.
In the prior patent application publication No. CN109961171A, namely a capacitor fault prediction method based on machine learning and big data analysis, massive historical data are collected, processed, mined and learned, statistical analysis is carried out on faults of a capacitor, important correlation information is used in training of an artificial neural network, and the capacitor faults are predicted by a guiding deep learning method.
In the prior patent application publication No. CN112163391A, namely a method and a system for estimating the service life of a film capacitor under the influence of humidity, the relative humidity is set as a virtual variable, and is introduced into a failure model of the film capacitor in a qualitative variable mode, so that the influence of different humidity conditions on the capacitor is analyzed.
In the prior patent application document of the invention with publication number of CN114840991A, namely a method for predicting capacitance parameters by using a pulse capacitor on-line monitoring technology, the time Tm of the first zero crossing point of the voltage of the pulse capacitor is established as a function of damping ratio xi, and the concept of a capacitance failure constant lambda is provided, so that the residual capacitance is monitored in real time, and a failure capacitance parameter calculation scheme is formed.
However, the foregoing data-driven-based lifetime estimation method still has some problems to be improved. Moreover, the simulation technology based on the traditional method is often the simplification of the actual environment and the model, so that a large error exists between the simulation result and the actual measured value. The heterogeneity between individuals, the constantly changing operating environment, etc. can also introduce uncertainty into the determination of the capacitance of the capacitor. A method combining the measurement of limited data with a physical model in practical problems is established, the prediction precision of the residual capacity value of the film capacitor is improved, and the high-reliability application of the film capacitor in the fields of power systems and the like is ensured, so that the method becomes a research focus of the application technology of the metallized film capacitor.
In summary, the prior art has the technical problems that the capacitor capacity attenuation and the estimated capacitor life prediction precision are low, and the aged devices are difficult to identify in advance, so that the safety and reliability of the electronic equipment are low.
Disclosure of Invention
The technical problems to be solved by the invention are as follows: how to solve the technical problems of low capacitor capacity attenuation and estimated capacitor life prediction precision and low electronic equipment safety and reliability caused by difficulty in identifying an aging device in advance in the prior art.
The invention adopts the following technical scheme to solve the technical problems: the capacity attenuation treatment method of the metallized film capacitor comprises the following steps:
s1, measuring electric field distribution conditions according to field environment data and actual test data;
s2, establishing a loss function by using electric field intensity distribution according to electric field distribution conditions and using partial differential equation, preset initial condition and preset boundary condition as constraints according to Laplace equation and neural network PINNs based on physical information, and inverting to obtain continuous dielectric constants at different positions according to electric field intensity distributionAnd the distribution thereof;
s3, according to dielectric constantThe geometry of the capacitor to be measured, and the continuous capacitance valueC r ；
S4, comparing and processing capacitance valuesC r And initial capacitance valueC 0 Thereby obtaining the capacity attenuation rate of the capacitor to be measuredδ；
S5, capacity attenuation rate of the capacitor to be testedδAnd when the capacitance attenuation alarm signal is smaller than a preset threshold value, a capacity attenuation alarm signal is sent out, and the tested capacitor at the current position is tested and replaced accordingly.
According to the invention, the relative dielectric constant distribution of each capacitor is obtained by inversion of the actual measurement data of the limited electric field distribution, so that the capacitance distribution of the capacitor which is closer to the actual working is obtained, the capacity attenuation of the film capacitor can be effectively early warned, the aged capacitor can be identified in advance and replaced in time, and the safety and reliability of the electronic equipment are greatly improved.
In a more specific technical solution, step S2 includes:
s21, according to the observation potential corresponding to the x direction and the y direction in the electric field distribution conditionUIs determined by a difference methodTaking the electric field intensity distribution:
s22, solving a preset partial differential equation by utilizing the neural network PINNs of the physical information, wherein the neural network PINNs based on the physical information comprises the following steps: a feed-forward neural network FNN; processing time series in combination with feed-forward neural network FNNtPlanar sequencexPoint potential in the x-direction and input layer of (2)U(x) An output of (2);
s23, taking a preset partial differential equation, a preset boundary condition and a preset initial condition as constraints, and accordingly establishing a loss function so that the point potential in the x directionU(x) Satisfies a preset partial differential equation and a data observation valueuThereby obtaining the dielectric constantAnd the distribution thereof, wherein,x,yi.e. respectively expressedxA shaft(s),yThe coordinates of the direction of the axis,U(x,y)representation ofxDirection and directionyThe electric potentials corresponding to each other in the direction at the same time,U(x)representation ofxPoint potential in the direction.
The invention uses the PINN method based on physical information, inversion is carried out by utilizing limited electric field distribution actual measurement data to obtain the relative dielectric constant distribution of each capacitor, and further continuous capacitance distribution is obtained.
In a more specific embodiment, in step S21, the battery intensity distribution is obtained using the following logic:
（1）
wherein,εthe dielectric function is represented by a graph of the dielectric,uis a data observation of the potential.
In a more specific technical scheme, in step S21, the electric field component and the electric field strength are obtained by using the following logic:
（2）
（3）
（4）
in the method, in the process of the invention,E x 、E y 、Erespectively isx、yThe directional electric field component and the electric field strength,Uis the electrical potential at which the electrical potential,ε(x,y) The distribution of the relative dielectric constants is shown.
Is thatxSelecting the point interval in the direction; />Is thatyThe direction is selected at intervals of points,x i in (a)iFinger meansxDirection of the firstiThe point at which the current is to be measured,y i in (a)iFinger meansyDirection of the firstiThe point at which the current is to be measured,x i+1 finger meansxDirection of the firsti+1A point of the light-emitting diode is located,y i+1 finger meansyDirection of the firsti+1A point.
In a more specific embodiment, in step S22, the preset partial differential equation PDE is expressed using the following logic:
（5）
in the method, in the process of the invention,tis a time series.
In a more specific embodiment, in step S23, the loss function is defined using the following logicL(θ)：
（6）
（7）
（8）
（9）
In the method, in the process of the invention,α 1 、α 2 、α 3 as a loss functionL(θ) Is used for the weight of the (c),T f 、T i andT b is a point from the PDE, training data set and boundary/initial conditions,T f is a computational domainΩIs provided with a plurality of points in the matrix,θis a vector that contains all the weights and deviations in the neural network that need to be trained.
Is a physical equationfCorresponding loss function, < >>Is a loss function corresponding to the initial condition, +.>Is a loss function corresponding to boundary conditions, +.>Is the true value of the data.
The present invention utilizes a computational domainΩEvaluating compatibility between the neural network and the physical equation for a set of points in (a)T f The residual point distribution in the method is similar to the grid distribution in the finite element, so that the convergence of the neural network is fully considered, and the accuracy of a calculation result is ensured.
In a more specific aspect, the vectorθComprising the following steps: the neural network weights to be trained and the neural network biases to be trained.
In a more specific embodiment, in step S3, the dielectric constant is determinedThe geometry of the capacitor to be measured is determined by the following logic to obtain the continuous capacitance valueC r 。
（10）
In the method, in the process of the invention,is the dielectric constant, is related to the nature of the dielectric,Sis the right facing area of the capacitor plate,dis the distance between the capacitor plates,kthen it is a constant of electrostatic force.
In a more specific embodiment, in step S4, the capacity attenuation rate δ of the capacitor to be measured is obtained by using the following logic;
（11）
in the method, in the process of the invention,C 0 is the initial capacitance value.
According to the Laplace equation and the neural network inversion based on the physical information, the invention obtains the dielectric constants of different continuous positionsDistribution of (3); according to the dielectric constant->And geometry to determine continuous capacitanceC r The method comprises the steps of carrying out a first treatment on the surface of the Finally, the obtained capacitance valueC r And initial capacitance valueC 0 Comparing to calculate the capacity attenuation rate of the capacitorδAccording toδThe value is used for judging the state of the capacitor, so that the capacity attenuation of the capacitor and the prediction accuracy of the estimated service life of the capacitor are improved, and the method is beneficial to early identification of an aging device.
In a more specific aspect, a metallized film capacitor capacity fade processing system includes:
the electric field distribution measuring module is used for measuring electric field distribution according to field environment data and actual test data;
the dielectric constant and distribution inversion module is used for establishing a loss function by using the electric field intensity distribution according to the electric field distribution condition, using the Laplace equation and the neural network PINNs based on the physical information and adopting the partial differential equation, the preset initial condition and the preset boundary condition as constraints, thereby inverting to obtain the dielectric constants with continuous different positions according to the electric field intensity distributionAnd the distribution situation, the dielectric constant and the distribution inversion module are connected with the electric field distribution measurement module;
a continuous capacitance obtaining module for obtaining a dielectric constantThe geometry of the capacitor to be measured, and the continuous capacitance valueC r The continuous capacitance calculating module is connected with the dielectric constant and the distribution inversion module;
a capacity attenuation rate obtaining module for comparing and processing capacitance valuesC r And initial capacitance valueC 0 Thereby obtaining the capacity attenuation rate of the capacitor to be measuredδThe capacity attenuation rate obtaining module is connected with the continuous capacitance obtaining module;
a attenuation capacitor replacement module for replacing the attenuation rate of the capacity of the capacitor under testδWhen the attenuation capacity of the capacitor is smaller than the preset threshold value, a capacity attenuation alarm signal is sent out, and accordingly the tested capacitor at the current position is tested and replaced, and the attenuation capacitor replacement module is connected with the capacity attenuation rate obtaining module.
Compared with the prior art, the invention has the following advantages: according to the invention, the relative dielectric constant distribution of each capacitor is obtained by inversion of the actual measurement data of the limited electric field distribution, so that the capacitance distribution of the capacitor which is closer to the actual working is obtained, the capacity attenuation of the film capacitor can be effectively early warned, the aged capacitor can be identified in advance and replaced in time, and the safety and reliability of the electronic equipment are greatly improved.
The invention uses the PINN method based on physical information, inversion is carried out by utilizing limited electric field distribution actual measurement data to obtain the relative dielectric constant distribution of each capacitor, and further continuous capacitance distribution is obtained.
The present invention utilizes a computational domainΩEvaluating compatibility between the neural network and the physical equation for a set of points in (a)T f The residual point distribution in the method is similar to the grid distribution in the finite element, so that the convergence of the neural network is fully considered, and the accuracy of a calculation result is ensured.
According to the Laplace equation and the neural network inversion based on the physical information, the invention obtains the dielectric constants of different continuous positionsDistribution of (3); according to the dielectric constant->And geometry to determine continuous capacitanceC r The method comprises the steps of carrying out a first treatment on the surface of the Finally, the obtained capacitance valueC r And initial capacitance valueC 0 Comparing to calculate the capacity attenuation rate of the capacitorδAccording toδThe value is used for judging the state of the capacitor, so that the capacity attenuation of the capacitor and the prediction accuracy of the estimated service life of the capacitor are improved, and the method is beneficial to early identification of an aging device.
The invention solves the technical problems of low capacitor capacity attenuation and low prediction precision of the estimated capacitor life, and low safety and reliability of electronic equipment caused by the difficulty in identifying an aging device in advance in the prior art.
Drawings
FIG. 1 is a schematic diagram showing the basic steps of a method for capacity fade treatment for metallized film capacitors according to example 1 of the present invention;
FIG. 2 is a schematic diagram showing the electric field intensity calculation principle of the capacity fading treatment method of the metallized film capacitor according to the embodiment 1 of the present invention;
fig. 3 is a schematic diagram of the principles of the pins for solving the electric field according to embodiment 1 of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions in the embodiments of the present invention will be clearly and completely described in the following in conjunction with the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Example 1
As shown in FIG. 1, the capacity attenuation processing method of the metallized film capacitor provided by the invention comprises the following basic steps:
s1, measuring electric field distribution conditions according to the field environment and actual test data;
s2, obtaining continuous dielectric constants at different positions according to the Laplace equation and the inversion of a neural network (PINNs) based on physical informationDistribution conditions;
as shown in fig. 2, in the present embodiment, since the distribution of the relative dielectric constant is a continuous micro-functionable, it is not necessary to set boundary conditions for interfaces of different media. The potential needs to be discretized during the calculation. In the present embodiment, according to the electric potentialu(x,y) The distribution of the electric field intensity can be rapidly calculated by adopting a difference method:
（1）
Uis the electrical potential at which the electrical potential,ε(x,y) The distribution of the relative dielectric constants is shown.
（2）
（3）
（4）
In the middle ofE x 、E y 、ERespectively isx、yDirectional electric field components and electric field strength.
In this embodiment, the PINN uses partial differential equations, initial conditions, and boundary conditions as constraints to establish the loss function.
In this embodiment, the algorithm for solving PDE (partial differential equation) with PINNs is shown in FIG. 3. Time series using feed Forward Neural Network (FNN)tSpatial sequencexIs a potential and an input layer of (a)u(x) Wherein the input layer is composed of the computing domainΩDimension determination of (2) in two dimensionsx= (x,y) In three dimensionsx= (x,y,z)。θIs a vector that contains all the weights and biases in the neural network that need to be trained. PDE is expressed as:
（5）
in the present embodiment, a loss function is established with partial differential equation, boundary condition and initial condition as constraints, so that the solutionU(x) Satisfy partial differential equation and data observation valueu. Defining a loss functionL(θ)
（6）
（7）
（8）
（9）
Wherein,α 1 、α 2 、α 3 as a loss functionL(θ) Is used for the weight of the (c),T f 、T i andT b is a point from the PDE, training data set and boundary/initial conditions.T f Is a computational domainΩFor evaluating compatibility between the neural network and the physical equation. Although the principle is differentT f The distribution of the residual points in (c) is similar to the grid distribution in the finite element. The distribution of the residual points affects the convergence of the neural network and the accuracy of the calculation result.
S3, according to dielectric constantAnd geometry to determine continuous capacitanceC r ；
In the present embodiment, the dielectric constant is calculated according to equation 10Continuous capacitance value determination from capacitor geometryC r 。
（10）
Wherein,is the dielectric constant, is related to the nature of the dielectric,Sis the right facing area of the capacitor plate,dis the distance between the capacitor plates,kthen it is a constant of electrostatic force.
S4, obtaining the capacitance valueC r And initial capacitance valueC 0 By comparison, the capacity attenuation rate of the capacitor is calculatedδ；
In the present embodiment of the present invention, in the present embodiment,finally, the obtained capacitance valueC r And initial capacitance valueC 0 Comparing, calculating the capacity attenuation rate delta of the capacitor according to the formula 11;
（11）
s5, when the capacity attenuation rate of a certain capacitorδLess than 5%, a capacity fade warning signal is sent and then the capacitor at that location is tested, if a problem arises, even if the capacitor is replaced.
In summary, the invention obtains the relative dielectric constant distribution of each capacitor by inversion of the actual measurement data of the limited electric field distribution, thereby obtaining the capacitance distribution of the capacitor which is closer to the actual working, effectively early warning the capacity attenuation of the film capacitor, identifying the aged capacitor in advance and replacing in time, and greatly improving the safety and reliability of the electronic equipment.
The invention uses the PINN method based on physical information, inversion is carried out by utilizing limited electric field distribution actual measurement data to obtain the relative dielectric constant distribution of each capacitor, and further continuous capacitance distribution is obtained.
The present invention utilizes a computational domainΩEvaluating compatibility between the neural network and the physical equation for a set of points in (a)T f The residual point distribution in the method is similar to the grid distribution in the finite element, so that the convergence of the neural network is fully considered, and the accuracy of a calculation result is ensured.
According to the Laplace equation and the neural network inversion based on the physical information, the invention obtains the dielectric constants of different continuous positionsDistribution of (3); according to the dielectric constant->And geometry to determine continuous capacitanceC r The method comprises the steps of carrying out a first treatment on the surface of the Finally, the obtained capacitance valueC r And initial capacitance valueC 0 Comparing to calculate the capacity of the capacitorAttenuation RateδAccording toδThe value is used for judging the state of the capacitor, so that the capacity attenuation of the capacitor and the prediction accuracy of the estimated service life of the capacitor are improved, and the method is beneficial to early identification of an aging device.
The invention solves the technical problems of low capacitor capacity attenuation and low prediction precision of the estimated capacitor life, and low safety and reliability of electronic equipment caused by the difficulty in identifying an aging device in advance in the prior art.
The above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (9)
1. A method for capacity fade processing of a metallized film capacitor, said method comprising:
s1, measuring electric field distribution conditions according to field environment data and actual test data;
s2, establishing a loss function by using electric field intensity distribution according to the electric field distribution condition and using a partial differential equation, a preset initial condition and a preset boundary condition as constraints according to a Laplace equation and a neural network PINNs based on physical information, and inverting to obtain continuous dielectric constants at different positions according to the electric field intensity distributionAnd the distribution thereof;
wherein, the step S2 includes:
s21, according to the observation potential corresponding to the x direction and the y direction in the electric field distribution conditionU(x, y) And (3) obtaining the electric field intensity distribution by adopting a difference method:
s22, utilizing the neural network of the physical informationAnd (3) complexing PINNs, and solving the partial differential equation, wherein the neural network PINNs based on the physical information comprises the following steps: a feed-forward neural network FNN; combining processing time series by using the feed-forward neural network FNNtPlanar sequencexPoint potential in the x-direction and input layer of (2)U(x) An output of (2);
s23, taking the partial differential equation, the preset boundary condition and the preset initial condition as constraints, and accordingly establishing the loss function so that thexPoint potential in the directionU(x) Satisfying the partial differential equation and the data observationsuThereby obtaining the dielectric constantAnd the distribution thereof, wherein,x,yi.e. respectively expressedxA shaft(s),yThe coordinates of the direction of the axis,U (x,y)representation ofxDirection and directionyThe electric potentials corresponding to each other in the direction at the same time,U(x)representation ofxA dot potential in the direction;
s3, according to the dielectric constantThe geometry of the capacitor to be measured, and the continuous capacitance valueC r ；
S4, comparing and processing the capacitance valueC r And initial capacitance valueC 0 Thereby obtaining the capacity attenuation rate of the capacitor to be measuredδ；
S5, the capacity attenuation rate of the tested capacitorδAnd when the capacitance attenuation alarm signal is smaller than a preset threshold value, a capacitance attenuation alarm signal is sent out, and the tested capacitor at the current position is tested and replaced accordingly.
2. The method of capacity fade processing for a metallized film capacitor according to claim 1, wherein in said step S21, said spatial distribution of electric field intensity is determined using the following logic:
（1）
in the method, in the process of the invention,the dielectric function is represented by a graph of the dielectric,uis a data observation of the potential.
3. The method for capacity fade processing as defined in claim 1, wherein in said step S21,uis the electrical potential in the space and,ε(x, y) The distribution of the relative dielectric constants was expressed, and the electric field component and the electric field strength were obtained by the following logic:
（2）
（3）
（4）
in the method, in the process of the invention,Ex、Ey、Erespectively isx、yThe directional electric field component and the electric field strength,Uis the electrical potential in the space and,ε(x, y) A distribution of relative dielectric constants;
is thatxSelecting the point interval in the direction; />Is thatyDirection selection interval->Is the firstiPoint correspondencexCoordinates, & gt>Is the firstiCorresponding to +1 pointxCoordinates, & gt>Is the firstiPoint correspondenceyCoordinates, & gt>Is the firstiCorresponding to +1 pointyCoordinates.
4. The method of capacity fade processing for a metallized film capacitor according to claim 1, wherein in said step S22, said partial differential equation PDE is expressed by the following logic:
（5）
in the method, in the process of the invention,tis a time series.
5. The method of attenuating capacity of metallized film capacitor as set forth in claim 4, wherein said loss function is defined by logic comprisingL(θ)：
（6）
（7）
（8）
（9）
In the method, in the process of the invention,α 1 、α 2 、α 3 as a loss functionL(θ) Is used for the weight of the (c),T f 、T i andT b is a point from the PDE, training data set and boundary/initial conditions,T f is a computational domainΩIs provided with a plurality of points in the matrix,θin the form of a vector which is a vector,is a physical equationfCorresponding loss function, < >>Is a loss function corresponding to the initial condition, +.>Is a loss function corresponding to boundary conditions, +.>Is the true value of the data.
6. The metallized film capacitor capacity fade treatment method of claim 5, wherein said vector comprises: the neural network weights to be trained and the neural network biases to be trained.
7. The method of attenuating capacity of a metallized film capacitor according to claim 1, wherein in said step S3, the dielectric constant is determined byThe measured capacitor geometry, the following logic is used to determine the successive capacitance valuesC r ：
（10）
In the method, in the process of the invention,is the dielectric constant, is related to the nature of the dielectric,Sis the right facing area of the capacitor plate,dis the distance between the capacitor plates,kthen it is a constant of electrostatic force.
8. The method for capacity fade processing for a metallized film capacitor according to claim 1, wherein in said step S4, said capacity fade rate δ of said capacitor under test is determined by using the following logic:
（11）。
9. a metallized film capacitor capacity fade processing system, said system comprising:
the electric field distribution measuring module is used for measuring electric field distribution according to field environment data and actual test data;
the dielectric constant and distribution inversion module is used for establishing a loss function by using the electric field intensity distribution according to the electric field distribution condition, using a partial differential equation, a preset initial condition and a preset boundary condition as constraints according to the Laplace equation and the physical information-based neural network PINNs, and inverting to obtain continuous dielectric constants at different positions according to the electric field intensity distributionAnd the distribution condition of the dielectric constant and the distribution inversion module are connected with the electric field distribution measurement module;
the dielectric constant and distribution inversion module simultaneously corresponds to the observed potential in the x direction and the y direction in the electric field distribution conditionU(x, y) And (3) obtaining the electric field intensity distribution by adopting a difference method:
solving the partial differential equation by using the neural network PINNs of the physical information, wherein the neural network PINNs based on the physical information comprises: a feed-forward neural network FNN; combining processing time series by using the feed-forward neural network FNNtPlanar sequencexPoint potential in the x-direction and input layer of (2)U(x) An output of (2);
the partial differential equation, the preset boundary condition and the preset initial condition are taken as constraints, so that the loss function is built upxPoint potential in the directionU(x) Satisfying the partial differential equation and the data observationsuThereby obtaining the dielectric constantAnd the distribution thereof, wherein,x,yi.e. respectively expressedxA shaft(s),yThe coordinates of the direction of the axis,U(x,y)representation ofxDirection and directionyThe electric potentials corresponding to each other in the direction at the same time,U(x)representation ofxA dot potential in the direction;
a continuous capacitance obtaining module for obtaining the dielectric constantThe geometry of the capacitor to be measured, and the continuous capacitance valueC r The continuous capacitance calculating module is connected with the dielectric constant and distribution inversion module;
a capacity attenuation rate obtaining module for comparing and processing the capacitance valueC r And initial capacitance valueC 0 Thereby obtaining the capacity attenuation rate of the capacitor to be measuredδThe capacity attenuation rate obtaining module is connected with the continuous capacitance obtaining module;
a decay capacitor replacement module for replacing the capacity decay rate of the capacitor under testδAnd when the capacitance attenuation alarm signal is smaller than a preset threshold value, a capacitance attenuation alarm signal is sent out, the tested capacitor at the current position is tested and replaced according to the capacitance attenuation alarm signal, and the attenuation capacitor replacement module is connected with the capacitance attenuation rate obtaining module.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311680354.0A CN117390348B (en) | 2023-12-08 | 2023-12-08 | Method and system for capacity attenuation treatment of metallized film capacitor |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311680354.0A CN117390348B (en) | 2023-12-08 | 2023-12-08 | Method and system for capacity attenuation treatment of metallized film capacitor |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117390348A CN117390348A (en) | 2024-01-12 |
| CN117390348B true CN117390348B (en) | 2024-04-02 |
Family
ID=89437684
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311680354.0A Active CN117390348B (en) | 2023-12-08 | 2023-12-08 | Method and system for capacity attenuation treatment of metallized film capacitor |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117390348B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE10134517A1 (en) * | 2001-07-16 | 2003-02-06 | Captron Elect Gmbh | Method for quantitative measurement of fluid conductivity, providing more accurate measurements of its dielectric constant, for use in measuring the composition of alcohol water mixtures used in damping in offset printing |
| CN108593725A (en) * | 2018-04-26 | 2018-09-28 | 西北师范大学 | Capacitance chromatography imaging method based on Modified particle swarm optimization |
| CN111407272A (en) * | 2020-03-16 | 2020-07-14 | 北京航空航天大学 | Discrete capacitance tomography image reconstruction method based on closed-loop control principle |
| CN114414909A (en) * | 2021-11-26 | 2022-04-29 | 西安交通大学 | Method for predicting service life of direct-current self-healing metalized film capacitor |
| CN114966451A (en) * | 2022-06-06 | 2022-08-30 | 北京航空航天大学 | Power supply product residual service life online prediction method based on physical information neural network |
| CN115441466A (en) * | 2022-08-26 | 2022-12-06 | 南京大全自动化科技有限公司 | Capacitor capacity attenuation estimation method and device based on fusion terminal |
| CN116090526A (en) * | 2023-02-07 | 2023-05-09 | 东南大学 | Structural vibration noise inversion method based on embedded physical information neural network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP4195089A1 (en) * | 2021-12-06 | 2023-06-14 | Tata Consultancy Services Limited | Physics-informed neural network for inversely predicting effective material properties of metamaterials |
-
2023
- 2023-12-08 CN CN202311680354.0A patent/CN117390348B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE10134517A1 (en) * | 2001-07-16 | 2003-02-06 | Captron Elect Gmbh | Method for quantitative measurement of fluid conductivity, providing more accurate measurements of its dielectric constant, for use in measuring the composition of alcohol water mixtures used in damping in offset printing |
| CN108593725A (en) * | 2018-04-26 | 2018-09-28 | 西北师范大学 | Capacitance chromatography imaging method based on Modified particle swarm optimization |
| CN111407272A (en) * | 2020-03-16 | 2020-07-14 | 北京航空航天大学 | Discrete capacitance tomography image reconstruction method based on closed-loop control principle |
| CN114414909A (en) * | 2021-11-26 | 2022-04-29 | 西安交通大学 | Method for predicting service life of direct-current self-healing metalized film capacitor |
| CN114966451A (en) * | 2022-06-06 | 2022-08-30 | 北京航空航天大学 | Power supply product residual service life online prediction method based on physical information neural network |
| CN115441466A (en) * | 2022-08-26 | 2022-12-06 | 南京大全自动化科技有限公司 | Capacitor capacity attenuation estimation method and device based on fusion terminal |
| CN116090526A (en) * | 2023-02-07 | 2023-05-09 | 东南大学 | Structural vibration noise inversion method based on embedded physical information neural network |
Non-Patent Citations (3)
| Title |
|---|
| 一种金属化膜电容器边缘电场的计算方法;王鑫;张晓;鲁军勇;;数码设计;20170408(第07期);91-93 * |
| 基于COMSOL的电容层析成像系统灵敏度场的计算;吴新杰;胡晟;付荣荣;;辽宁大学学报(自然科学版);20110815(第03期);6-9 * |
| 基于改进粒子群优化的电容层析成像图像重建算法;严春满;陆根源;张道亮;董俊松;;计算机工程与科学;20190515(第05期);119-124 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117390348A (en) | 2024-01-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Xu et al. | Remaining useful life prediction of lithium-ion batteries based on wiener process under time-varying temperature condition | |
| Zhang et al. | Synchronous estimation of state of health and remaining useful lifetime for lithium-ion battery using the incremental capacity and artificial neural networks | |
| Kang et al. | A multi-fault diagnostic method based on an interleaved voltage measurement topology for series connected battery packs | |
| Liu et al. | Prognostics for state of health estimation of lithium-ion batteries based on combination Gaussian process functional regression | |
| CN111595485B (en) | Lithium ion battery online temperature distribution observer design method based on reduced order model | |
| Sun et al. | Remaining useful life prediction of aviation circular electrical connectors using vibration-induced physical model and particle filtering method | |
| Wang et al. | Perspectives and challenges for future lithium-ion battery control and management | |
| CN113204922B (en) | Transformer running state monitoring and predicting method | |
| Ahsan et al. | Data driven prognostics for predicting remaining useful life of IGBT | |
| Wang et al. | Methods for predicting the remaining useful life of equipment in consideration of the random failure threshold | |
| CN108051364A (en) | A kind of EPR nuclear energy cable residue lifetime estimation method and prediction EPR nuclear energy cable remaining life methods | |
| Ali et al. | A hybrid data-driven approach for multistep ahead prediction of state of health and remaining useful life of lithium-ion batteries | |
| Chen et al. | Remaining useful life prediction for complex systems with multiple indicators based on particle filter and parameter correlation | |
| CN117390348B (en) | Method and system for capacity attenuation treatment of metallized film capacitor | |
| Nenadic et al. | Estimation of state-of-charge and capacity of used lithium-ion cells | |
| Lu et al. | A novel method of prediction for capacity and remaining useful life of lithium-ion battery based on multi-time scale Weibull accelerated failure time regression | |
| Zhao et al. | Harmonic reducer performance prediction algorithm based on multivariate state estimation and largevis dimensionality reduction | |
| Cheng et al. | RUL Prediction Method for Electrical Connectors with Intermittent Faults Based on an Attention-LSTM Model | |
| Wang et al. | New forecasting method of closing time for aerospace relay in storage accelerated degradation testing | |
| Dong | Prediction of the Remaining Useful Life of Lithium-Ion Batteries Based on Dempster-Shafer Theory and the Support Vector Regression-Particle Filter | |
| Hu et al. | Health factor analysis and remaining useful life prediction for batteries based on a cross-cycle health factor clustering framework | |
| Rigamonti et al. | A particle filtering-based approach for the prediction of the remaining useful life of an aluminum electrolytic capacitor | |
| Mei et al. | A novel fusion prognostic approach for the prediction of the remaining useful life of a lithiumion battery | |
| CN114184211B (en) | Method for judging consistency of performance change mechanism in inertial navigation reliability test | |
| Dineva et al. | Insulation state assessment of rotating electrical machines by employing generalized additive models |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |