Method for testing local electric field of high-voltage power module

Abstract

A local electric field test method of a high-voltage power module comprises the following steps: acquiring surface potential distribution of packaging insulation of the power module, performing central difference processing to obtain surface electric field intensity, and inverting distribution of an interface electric field in combination with a second-order total variation regularization algorithm. The non-intrusive electrostatic probe is used for measuring the potential distribution of the packaging insulation surface of the power module, the second-order total variation regularization inversion algorithm is combined, the interface electric field distribution formed by the ceramic plate and the pouring sealant is measured, and effective detection of local electric field distribution under a complex structure is achieved.

Description

Technical Field

[0001] The present invention relates to a technology in the field of electrical engineering, in particular to a local electric field testing method for a high-voltage power module. Background Art

[0002] The high-voltage power module packaging has an uneven electric field distribution due to the complex electrode and insulation structure. Among them, the electric field concentration phenomenon is the most serious in the three-junction area formed by the ceramic substrate, copper cladding and potting glue. If the electric field intensity in a local area is too high, the material may be subjected to greater electrothermal stress, which may cause local aging, crack expansion or material degradation, affecting the performance, reliability and service life of the power module.

[0003] At present, the local electric field characteristics of power modules are mainly calculated by finite element numerical simulation technology, which analyzes the electric field distribution in complex structures through mathematical models. However, the numerical simulation method essentially relies on the assumptions and simplifications of material properties, boundary conditions and initial values, so the validity of the results is largely affected by the settings of these parameters. In addition, the results obtained by numerical simulation often reflect the electric field distribution under ideal conditions, but may not reflect the electric field conditions under actual working conditions. Therefore, an experimental research method that is closer to the actual environment is needed, and the experimental method can provide a means of verification for the effectiveness of the numerical simulation method. The existing technology can test the electric field strength in the medium through electric field sensors based on technologies such as the photoelectric effect, but this technology is still limited in its application to high-voltage power module packaging insulation, mainly because the sensor size is comparable to the size of the packaging insulation structure, so the sensor intervention is likely to destroy the original electric field distribution, making the test results inaccurate. Summary of the invention

[0004] Aiming at the problems that it is difficult to test the electric field strength of the interface formed by the potting compound and the ceramic substrate in the high-voltage power module in the prior art, the interference of the electric field distribution caused by the intervention of the existing sensor, and the weak nonlinear optical effect of the packaging material, the present invention proposes a local electric field testing method for the high-voltage power module, measures the potential distribution on the surface of the power module package insulation by a non-intrusive electrostatic probe, and measures the electric field distribution at the interface formed by the ceramic plate and the potting compound in combination with a second-order total variation regularized inversion algorithm, thereby realizing effective detection of the local electric field distribution under complex structures.

[0005] The present invention is achieved through the following technical solutions:

[0006] The present invention relates to a local electric field testing method for a high-voltage power module. After collecting the surface potential distribution of the power module package insulation, the surface electric field strength is obtained by central difference processing, and then the distribution of the interface electric field is inverted by combining a second-order total variation regularization algorithm.

[0007] The acquisition means: dividing the upper and lower surfaces of the potting glue layer, namely the surface S1 and the interface S2, into N = n × n units respectively, and testing the potential value of each unit on S1 under pressure by a non-invasive electrostatic probe method. i,j=1~N.

[0008] The central difference processing is to obtain the component of the electric field in the x direction according to the potential value of each unit. and the y-direction component Then the surface electric field strength on each unit is obtained Where: E i，j Constitute an n×n two-dimensional matrix.

[0009] The second-order total variation regularization algorithm specifically includes:

[0010] Step 1) According to the electrostatic field theory, for any point p on the surface S1 and its corresponding point q on the interface S2, i.e., the point perpendicular to the point in the direction of the interface normal vector, a surface electric field and interface electric field model is established. Where: (i, j) is the two-dimensional coordinates of the plane where point p and point q are located. and The interface electric field and surface electric field The magnitude of the electric field at point (i, j) corresponds to the surface electric field and interface electric field d is the insulation thickness between the surface S1 and the interface S2, ε is the dielectric constant of the material, q k is the volume charge density of the kth unit in the insulation, m is the total number of charge units in the insulation, R k Q k The distance from point p, θ k Q k The angle between the line connecting the point and point p and the tangent vector of the plane is, is the unit vector pointing from qk to point (i,j),

[0011] Step 2) Convert the two-dimensional surface electric field and interface electric field model into one-dimensional: According to the surface electric field intensity distribution, obtain the location of the maximum electric field intensity i=a, j=b. Further obtain a one-dimensional array E containing the maximum electric field intensity s1 [n], n is the length of the one-dimensional array.

[0012] Step 3) Use second-order total variation regularization to solve the ill-posed problem from surface electric field to interface electric field: construct the objective function min(||E S1 (x)-∫M(x,x')E S2 (x')dx'‖ 2 +λTV (2)(E S2 (x))), where: M(x,x') is the kernel function describing the transfer relationship between the surface electric field and the interface electric field, x is the surface electric field coordinate, x' is the interface electric field coordinate, λ is the regularization parameter, TV (2) (E S2 (x)) is the second-order total variation regularization term. The problem is converted to the frequency domain and discretized to obtain Where: E s1 [k] and E S2 [k] is the discretized one-dimensional surface electric field and interface electric field distribution, M[k] is the discretized one-dimensional kernel function or discretized transfer array, and then the gradient descent method is used to optimize the objective function, and the parameter update is: E S2 [k] (t+1) =E S2 [k] (t) +α t d t , where: E S2 [k] (t+1) is the estimated value of the interface electric field at the t+1th iteration, E S2 [k] (t) is the estimated value of the interface electric field at the tth iteration, α t is the iteration step length, d t For the objective function E S2 The gradient of [k].

[0013] Step 4) Iterate and optimize to obtain the interface electric field E in the frequency domain S2 [k] is then transformed to the time domain through inverse Fourier transform to finally obtain the interface electric field distribution.

[0014] The present invention relates to a system for realizing the above method, comprising: a surface potential testing unit, a data acquisition unit, an electric field calculation unit and an interface electric field inversion unit, wherein: the surface potential testing unit obtains two-dimensional potential distribution data of the upper surface of the power module potting glue layer according to a non-intrusive electrostatic probe testing method; the data acquisition unit receives the potential data obtained from the electrostatic probe test and performs preprocessing to eliminate the environmental noise interference in the measurement process; the electric field calculation unit performs central difference processing on the preprocessed potential data to obtain the two-dimensional distribution of the surface electric field; the interface electric field inversion unit converts the two-dimensional electric field distribution into a one-dimensional electric field distribution according to the two-dimensional distribution of the surface electric field and the position of the maximum value of the electric field intensity, and according to the surface electric field and interface electric field models, adopts second-order total variation regularization and gradient descent method to iteratively optimize to obtain the interface electric field, and then obtains the one-dimensional distribution of the interface electric field in the time domain through inverse Fourier transformation. Technical Effects

[0015] The present invention measures the surface potential through a non-intrusive electrostatic probe, and then combines the second-order total variation regularized inversion algorithm to reconstruct the local electric field distribution at the interface between the ceramic substrate and the potting glue. The present invention avoids the problem of electric field distribution interference caused by sensor intervention in traditional testing methods, while filling the gap of lack of experimental research in the existing power module local electric field research technology, and provides an effective tool for the study of local electric field distribution under complex structures. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flow chart of the present invention;

[0017] Figure 2 (a) is a power module sample in the embodiment;

[0018] Figure 2 (b) is a schematic diagram of the probe scanning measurement area in the non-intrusive surface potential test;

[0019] In the figure: electrostatic voltage probe 1, power module sample 2, potting glue layer 3, L-shaped copper cladding 4, square copper cladding 5, ceramic substrate 6, electrostatic voltage probe scanning area 7 in the embodiment, high voltage power supply 8;

[0020] Figure 3 Schematic diagram of surface and interface partitioning method;

[0021] Figure 4 Schematic diagram of the two-dimensional distribution of surface potential obtained by the test;

[0022] Figure 5 Schematic diagram of the two-dimensional distribution of the surface electric field obtained by central difference processing;

[0023] Figure 6 Schematic diagram of one-dimensional distribution of surface electric field;

[0024] Figure 7 is a schematic diagram of the inverse transfer function in the frequency domain;

[0025] Figure 8 Schematic diagram of the one-dimensional distribution of the interface electric field in the time domain. DETAILED DESCRIPTION

[0026] like Figure 1 As shown, this embodiment involves a local electric field testing method for a high-voltage power module. After collecting the surface potential distribution of the power module package insulation by a non-intrusive electrostatic probe, the surface electric field strength is obtained by central difference processing, and then the distribution of the interface electric field is inverted by combining the second-order total variation regularization algorithm.

[0027] like Figure 2As shown in (a), the power module comprises: a ceramic substrate 6, an L-shaped copper cladding layer 4 and a square copper cladding layer 5 arranged thereon, and a potting glue layer 3 that is entirely coated on the outside.

[0028] The thickness of the potting glue layer is d=1mm, the dielectric constant is 2.8, the thickness of the ceramic substrate 6 is 0.635mm, the distance between the L-shaped copper cladding 4 and the square copper cladding 5 is 5mm, the corner radius is 0.3mm, and the thickness of the copper cladding is 0.3mm. During the surface potential test, the high-voltage power supply of the surface potential test unit of the non-intrusive electrostatic probe 1 has a stable output voltage of 5kV, which is connected to the L-shaped copper cladding lead-out terminal of the power module, the square copper cladding is grounded, the non-intrusive electrostatic probe is 2mm from the upper surface of the potting layer, the probe test sensitivity is 10mV, the probe movement rate is 4mm / s, and the sampling rate is 100s -1 , the moving range is a square area with a side length of 16 mm, such as Figure 2 (b) as shown.

[0029] like Figure 3 The potential value of each unit on S1 under pressure is tested by non-intrusive electrostatic probe method. i,j=1~N. In this embodiment, n is taken as 400.

[0030] like Figure 4 As shown in FIG. 1 , the two-dimensional distribution of the surface potential of the power module encapsulation layer obtained through the above measurement; according to the two-dimensional distribution data of the surface potential, the central difference processing is adopted to obtain the two-dimensional distribution data of the surface electric field, as shown in FIG. Figure 5 As shown in FIG. 1 , the interface electric field distribution formed by ceramic and potting glue is obtained by combining the surface electric field data with the inversion algorithm; according to the two-dimensional distribution data of the surface potential, the one-dimensional distribution data of the electric field at the intersection where the maximum electric field intensity is located (i=a=75) is obtained, as shown in FIG. Figure 6 As shown in the figure, it is the surface electric field distribution at the intersection along the y direction; according to the surface electric field and interface electric field model, the discretized transfer array M[k] in the frequency domain is obtained, as Figure 7 As shown, it is M[k] in the frequency domain; according to the one-dimensional distribution data of the surface electric field obtained by the test and the transfer array, the interface electric field is obtained by iterative optimization using the second-order total variation regularization and gradient descent method. The number of iterations is 100 times, and then the one-dimensional distribution of the interface electric field in the time domain is obtained by inverse Fourier transformation, as shown in Figure 8 As shown, the interface electric field distribution along the y direction at the interface i=a=75.

[0031] Compared with the traditional sensor direct intervention electric field measurement, the present invention collects the surface potential distribution of the power module package insulation through a non-intrusive electrostatic probe, avoiding the interference of the sensor on the electric field distribution and ensuring the accuracy and reliability of the test results; by introducing the second-order total variation regularization algorithm and the gradient descent method, the interface electric field is inverted and optimized. Compared with the prior art, the present invention improves the reconstruction accuracy of the electric field distribution and solves the edge problem that the electric field may have sudden changes at the interface; it fills the gap in experimental research on local electric fields under complex structures.

[0032] The above-mentioned specific implementation can be partially adjusted in different ways by those skilled in the art without departing from the principle and purpose of the present invention. The protection scope of the present invention shall be based on the claims and shall not be limited by the above-mentioned specific implementation. Each implementation scheme within its scope shall be subject to the constraints of the present invention.

Claims

1. A local electric field testing method for a high voltage power module, characterized in that: After collecting the surface potential distribution of the power module package insulation, the surface electric field intensity is obtained through central difference processing, and then the distribution of the interface electric field is inverted by combining the second-order total variation regularization algorithm; The second-order total variation regularization algorithm specifically includes: Step 1) According to the electrostatic field theory, for any point p on the surface S1 and its corresponding point q on the interface S2, i.e., the point perpendicular to the point in the direction of the interface normal vector, a surface electric field and interface electric field model is established. Where: (i, j) is the two-dimensional coordinates of the plane where point p and point q are located. and The interface electric field and surface electric field The magnitude of the electric field at point (i, j) corresponds to the surface electric field and interface electric field d is the insulation thickness between the surface S1 and the interface S2, ε is the dielectric constant of the material, q k is the volume charge density of the kth unit in the insulation, m is the total number of charge units in the insulation, R k Q k The distance from point p, θ k Q k The angle between the line connecting the point and point p and the tangent vector of the plane is, is the unit vector pointing from qk to point (i, j); Step 2) Convert the two-dimensional surface electric field and interface electric field model into one-dimensional: According to the surface electric field intensity distribution, the position where the maximum electric field intensity is located is obtained i=a, j=b, and further a one-dimensional array E containing the maximum electric field intensity is obtained. s1 [n], n is the length of the one-dimensional array; Step 3) Use second-order total variation regularization to solve the ill-posed problem from surface electric field to interface electric field: construct the objective function min(||E S1 (x)-∫M(x,x')E S2 (x')dx'|| 2 +λTV (2) (E S2 (x))), where: M(x,x') is the kernel function describing the transfer relationship between the surface electric field and the interface electric field, x is the surface electric field coordinate, x' is the interface electric field coordinate, λ is the regularization parameter, TV (2) (E S2 (x)) is the second-order total variation regularization term. The problem is converted to the frequency domain and discretized to obtain Where: E s1 [k] and E S2 [k] is the discretized one-dimensional surface electric field and interface electric field distribution, M[k] is the discretized one-dimensional kernel function or discretized transfer array, and then the gradient descent method is used to optimize the objective function, and the parameter update is: E S2 [k] (t+1) =E S2 [k] (t) +α t d t , where: E S2 [k] (t+1) is the estimated value of the interface electric field at the t+1th iteration, E S2 [k] (t) is the estimated value of the interface electric field at the tth iteration, α t is the iteration step length, d t For the objective function E S2 The gradient of [k]; Step 4) Iterate and optimize to obtain the interface electric field E in the frequency domain S2 [k] is then transformed to the time domain through inverse Fourier transform to finally obtain the interface electric field distribution.

2. The local electric field testing method of a high voltage power module according to claim 1, characterized in that: The acquisition means: dividing the upper and lower surfaces of the potting glue layer, namely the surface S1 and the interface S2, into N = n × n units respectively, and testing the potential value of each unit on S1 under pressure by a non-invasive electrostatic probe method. i,j=1~N.

3. The local electric field testing method of a high voltage power module according to claim 1, characterized in that: The central difference processing is to obtain the component of the electric field in the x direction according to the potential value of each unit. and the y-direction component Then the surface electric field strength on each unit is obtained Where: E i，j Constitute an n×n two-dimensional matrix.

4. A local electric field testing system for a high voltage power module implementing the method described in any one of claims 1 to 3, characterized in that: include: A surface potential test unit, a data acquisition unit, an electric field calculation unit and an interface electric field inversion unit, wherein: the surface potential test unit obtains two-dimensional potential distribution data on the upper surface of the power module potting glue layer according to a non-invasive electrostatic probe test method; the data acquisition unit receives the potential data obtained from the electrostatic probe test and performs preprocessing to eliminate environmental noise interference during the measurement process; the electric field calculation unit performs central difference processing on the preprocessed potential data to obtain a two-dimensional distribution of the surface electric field; the interface electric field inversion unit converts the two-dimensional electric field distribution into a one-dimensional electric field distribution according to the two-dimensional distribution of the surface electric field and the position of the maximum electric field intensity, and according to the surface electric field and interface electric field model, adopts second-order total variation regularization and gradient descent method to iteratively optimize to obtain the interface electric field, and then obtains the one-dimensional distribution of the interface electric field in the time domain through inverse Fourier transform.

Images