An IGBT bonding wire aging monitoring method based on magnetic flux density combination features

Abstract

This invention discloses an IGBT bond wire aging monitoring method based on magnetic flux density combination characteristics. Specifically, it involves: extracting magnetic flux density distribution curves under different bond wire aging states using finite element simulation; arbitrarily extracting two magnetic flux density data points from the curves to form a magnetic flux density combination feature; progressively classifying these features into different categories and groups according to different aging states and load currents; and establishing a magnetic flux density combination evaluation model based on inter-class variance (BCV) and minimum inter-class distance (MICD) to select the optimal magnetic flux density combination feature extraction points, thereby achieving accurate monitoring of the IGBT bond wire health status. This invention allows for direct measurement outside the IGBT module, is applicable to actual converter operating conditions, and features easy implementation, low invasiveness, and strong robustness, overcoming many technical problems in existing technologies such as difficult parameter extraction and poor security.

Description

Technical Field

[0001] This invention belongs to the field of power device health status monitoring technology, and particularly relates to an IGBT bond wire aging monitoring method based on magnetic flux density combination characteristics. Background Technology

[0002] Insulated-gate bipolar transistors (IGBTs) offer advantages such as low on-state voltage drop, high power density, and fast switching speed, making them widely used in power converters in rail transportation, new energy power generation, and electric vehicles. However, under complex and variable operating conditions, IGBT modules become one of the most vulnerable links in power converter systems. Bond wire aging is a common failure mode for IGBT modules. With repeated power cycling and thermal stress, the bond wires gradually peel off and break, leading to increased on-resistance, localized overheating, and even complete failure, seriously threatening the reliability and safety of the power converter system. Therefore, real-time monitoring of the aging status of IGBT bond wires and timely detection and replacement of deteriorated modules are crucial for ensuring the stable operation of power electronic systems.

[0003] Currently, monitoring methods for bond wire aging mainly rely on the electrical characteristic parameters of the IGBT power circuit, such as on-state voltage drop and saturation voltage, to indirectly reflect the health status of the bond wire through changes in these parameters. However, these methods have significant limitations in practical applications: on the one hand, electrical characteristic parameters are easily affected by operating conditions such as module junction temperature and load current fluctuations, leading to a decrease in the accuracy of monitoring results; on the other hand, the extraction of certain parameters requires special drive circuits or the injection of test current (such as short-circuit current), which not only increases system complexity and implementation difficulty but may also affect the normal operation of the equipment and pose safety hazards. Furthermore, in actual converters, due to the compact packaging of the power circuit and the high requirements for electrical isolation, direct extraction of electrical signals is often difficult and highly intrusive, limiting the engineering applicability of such methods.

[0004] Therefore, there is an urgent need to develop a non-invasive, highly robust, and practically applicable method for monitoring the health status of bond wires, in order to overcome the problems of difficult extraction, large interference, and insufficient security faced by existing monitoring technologies based on electrical characteristic parameters, and to provide a reliable means for early fault warning and predictive maintenance of IGBT modules. Summary of the Invention

[0005] To address the limitations of existing IGBT bond wire aging status monitoring technologies, this invention aims to propose a bond wire aging monitoring method that can be implemented directly outside the IGBT module and is adapted to the actual operating environment of the converter. This method can accurately identify the health status of bond wires under different operating conditions, and in particular, it maintains the reliability of monitoring results even under load current fluctuations. This solves the problems faced by current monitoring methods based on electrical characteristic parameters, such as complex parameter acquisition, insufficient security, and susceptibility to interference from junction temperature and load current changes.

[0006] Therefore, this invention provides a method for monitoring the aging of IGBT bond wires based on the combined characteristics of magnetic flux density.

[0007] This invention discloses an IGBT bond wire aging monitoring method based on magnetic flux density combination characteristics. It utilizes the magnetic flux density above the bond wire as a monitoring parameter, constructs a magnetic flux density combination characteristic evaluation model to reduce the influence of objective factors, and completes the monitoring of the bond wire health status of IGBTs under different load current conditions. Specifically, it includes the following steps:

[0008] Step 1: Use finite element simulation to simulate the aging state of IGBTs with different bond wires, extract the magnetic flux density distribution curves perpendicular to the bond wire direction under different bond wire breakage numbers, and perform noise reduction filtering, normalization and aging state labeling on the curve data.

[0009] The data of the magnetic flux density distribution curve is perpendicular to the extension direction of the bonding line and parallel to the plane where the IGBT chip is located.

[0010] Step 2: Extract any two magnetic flux density data points from the magnetic flux density distribution curve and combine them to form a new combined magnetic flux density feature K. pq Furthermore, the magnetic flux density characteristics K of all combinations of different load currents under the same aging state are... pq They are grouped into one category; and all categories under different aging states are grouped together.

[0011] Combined magnetic flux density characteristics K pq The composition method is shown in the following formula:

[0012] (1)

[0013] Among them, B x1 and B x2 Let N be the magnetic flux density at any two points on the magnetic flux density distribution curve, and let N be the number of bond lines. i Let be the current distribution coefficient of each bond wire, and S be the part of the bond wire magnetic flux density calculation formula that is related to the position of the bond wire.

[0014] Step 3: Establish a magnetic flux density combination evaluation model using inter-class variance (BCV) and minimum inter-class distance (MICD) to evaluate classes and groups composed of data points with different magnetic flux densities and select the optimal magnetic flux density combination feature extraction points.

[0015] The goal of BCV is to determine the degree of difference between class data, specifically the inter-class variance S. B The calculation formula is:

[0016] (2)

[0017] Where, N B n p X p_mean and X _mean These represent the total number of samples, the number of samples in class p, the mean of class p, and the overall mean, respectively.

[0018] MICD is used to measure the degree of separation between the two closest groups among all class pairs, with the smallest inter-group distance D being the minimum inter-group distance. min The calculation formula is:

[0019] (3)

[0020] Where x p and x q It is category C a and C b Its characteristics.

[0021] Furthermore, the combined characteristic evaluation model for magnetic flux density measures the combined value of BCV and MICD using a comprehensive metric score. The formula for calculating the comprehensive metric score is as follows:

[0022] (4)

[0023] In the formula, α and β are both weight values.

[0024] Step 4: After selecting the optimal monitoring point through the evaluation model, extract the magnetic flux density of the two points to obtain the combined characteristics of magnetic flux density, and use the magnetic flux density extraction board to monitor different aging states of the bonding wire under actual operating conditions.

[0025] Furthermore, α and β in step 3 can be set according to actual needs. In this invention, the value is 0.5.

[0026] The beneficial technical effects of this invention compared to the prior art are as follows:

[0027] 1. This invention uses the combined characteristics of external magnetic flux density of IGBT modules to monitor the health status of bond wires. It can perform non-invasive measurement and extraction, effectively reducing the difficulty and danger of bond wire health status monitoring, and is beneficial for application in converter time-conditioning.

[0028] 2. This invention utilizes the combined characteristics of magnetic flux density to monitor the health status of bond wires, which can effectively reduce the impact of IGBT junction temperature on the monitoring of the number of broken bond wires.

[0029] 3. This invention uses inter-class variance (BCV) and minimum inter-class distance (MICD) to establish a magnetic flux density combined evaluation model, reducing the adverse effects of objective factors and realizing accurate estimation of the bond wire health status of IGBT when the load current condition changes. Attached Figure Description

[0030] Figure 1 This is a flowchart of the IGBT bond wire aging monitoring method based on magnetic flux density combination characteristics according to the present invention.

[0031] Figure 2 This is a schematic diagram of magnetic flux density distribution curve acquisition as defined in this invention.

[0032] Figure 3 It is the magnetic flux density distribution curve directly above the IGBT bond line.

[0033] Figure 4 These are the magnetic flux density measurement results under different aging conditions of the present invention 1.

[0034] Figure 5 These are the results of two magnetic flux density measurements under different aging conditions of the present invention.

[0035] Figure 6 This is the bonding wire aging monitoring result achieved by the present invention. Detailed Implementation

[0036] The present invention will be further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0037] This invention discloses an IGBT bond wire aging monitoring method based on magnetic flux density combination characteristics. It utilizes the magnetic flux density above the bond wire as a monitoring parameter, constructs a magnetic flux density combination characteristic evaluation model to reduce the influence of objective factors, and completes the monitoring of the bond wire health status of IGBTs under different load current conditions. The process is as follows: Figure 1 As shown, the specific steps include:

[0038] Step 1: Use finite element simulation to simulate the aging state of IGBTs with different bond wires, extract the magnetic flux density distribution curves perpendicular to the bond wire direction under different bond wire breakage numbers, and perform noise reduction filtering, normalization and aging state labeling on the curve data.

[0039] The magnetic flux density distribution curve data is perpendicular to the bonding line extension direction and parallel to the plane where the IGBT chip is located. A schematic diagram of the magnetic flux density distribution curve acquisition is shown below. Figure 2 As shown.

[0040] Step 2: Extract any two magnetic flux density data points from the magnetic flux density distribution curve and combine them to form a new combined magnetic flux density feature K. pq Furthermore, the magnetic flux density characteristics K of all combinations of different load currents under the same aging state are... pq They are grouped into one category; and all categories under different aging states are grouped together.

[0041] Combined magnetic flux density characteristics K pq The composition method is shown in the following formula:

[0042] (1)

[0043] Among them, B x1 and B x2 Let N be the magnetic flux density at any two points on the magnetic flux density distribution curve, and let N be the number of bond lines. i Let be the current distribution coefficient of each bond wire, and S be the part of the bond wire magnetic flux density calculation formula that is related to the position of the bond wire.

[0044] Step 3: Establish a magnetic flux density combination evaluation model using inter-class variance (BCV) and minimum inter-class distance (MICD) to evaluate classes and groups composed of data points with different magnetic flux densities and select the optimal magnetic flux density combination feature extraction points.

[0045] The goal of BCV is to determine the degree of difference between class data, specifically the inter-class variance S. B The calculation formula is:

[0046] (2)

[0047] Where, N B n p X p_mean and X _mean These represent the total number of samples, the number of samples in class p, the mean of class p, and the overall mean, respectively.

[0048] MICD is used to measure the degree of separation between the two closest groups among all class pairs, with the smallest inter-group distance D being the minimum inter-group distance. min The calculation formula is:

[0049] (3)

[0050] Where x p and x q It is category C a and C b Its characteristics.

[0051] Furthermore, the combined characteristic evaluation model for magnetic flux density measures the combined value of BCV and MICD using a comprehensive metric score. The formula for calculating the comprehensive metric score is as follows:

[0052] (4)

[0053] In the formula, α and β are both weight values, which can be set according to actual needs. In this invention, the value is taken as 0.5.

[0054] Step 4: After selecting the optimal monitoring point through the evaluation model, extract the magnetic flux density of the two points to obtain the combined characteristics of magnetic flux density, and use the magnetic flux density extraction board to monitor different aging states of the bonding wire under actual operating conditions.

[0055] Example 1: Finite element modeling of the IGBT module used in a three-phase inverter. First, a three-dimensional solid model is constructed based on the three-dimensional dimensions and material properties of the IGBT module package. Then, considering the physical boundary conditions of the IGBT module, an electro-magnetic-thermal multiphysics coupled finite element model of the IGBT module is constructed, and based on... Figure 2 The magnetic flux density distribution curve is schematically captured above the IGBT bond line, as shown in the image. Figure 3 As shown in the figure. Furthermore, combining the magnetic flux density information of any two points, a magnetic flux density combination evaluation model is established using the inter-class variance (BCV) and minimum inter-class distance (MICD) to select the optimal magnetic flux density acquisition location. The magnetic flux density acquisition location score and optimal location obtained based on finite element simulation data are shown in the figure. Figure 4 As shown.

[0056] Example 2: Bond wire aging monitoring was performed on the IGBT modules in a three-phase inverter. Two magnetic flux densities were obtained by determining the optimal magnetic flux density acquisition location, and a magnetic flux density combination feature was further constructed. The magnetic flux density combination features of the IGBT modules under different aging states were collected. The measurement results of the two magnetic flux densities under different aging states of this invention are as follows: Figure 4 and Figure 5 As shown in the figure. The fundamental frequency of the IGBT module is 10Hz, and the switching frequency is 2000Hz. Figure 6 The invention demonstrates its effectiveness in monitoring the aging of IGBT module bond wires in a three-phase inverter.

[0057] The bonding wire health status monitoring method provided by this invention addresses the difficulties in acquiring IGBT electrical parameters and the poor safety of such methods. It employs a combination of magnetic flux density characteristics to achieve simple and direct measurement of bonding wire health status. Furthermore, to address the issue of poor accuracy of bonding wire health status monitoring schemes based on IGBT electrical characteristic parameters when IGBT junction temperature changes and load currents occur, a magnetic flux density combination evaluation model is established using inter-class variance (BCV) and minimum inter-class distance (MICD) to reduce the adverse effects of objective factors and achieve accurate monitoring of bonding wire health status.

[0058] The monitoring method proposed in this invention can be implemented directly outside the IGBT module and is suitable for assessing the bond wire health status during actual converter operation. This method can accurately identify the degree of bond wire aging under different junction temperature conditions, and has the advantages of simple implementation, strong directness, and good robustness. Most importantly, this method can still ensure the accuracy of monitoring results when the load current fluctuates, thus effectively solving the problems of complex parameter extraction, low security, and susceptibility to interference from changes in junction temperature and load current in existing monitoring methods based on power circuit electrical characteristic parameters. The established bond wire health status assessment framework is not only applicable to computer-based IGBT bond wire health status simulation and physical experimental research, but can also be further extended to converter systems composed of other power devices.

Claims

1. A method for monitoring the aging of IGBT bond wires based on the combined characteristics of magnetic flux density, characterized in that, By using the magnetic flux density above the bond wire as a monitoring parameter, a magnetic flux density combined characteristic evaluation model is constructed to reduce the influence of objective factors, and the bond wire health status monitoring of IGBTs under different load current conditions is completed. The specific steps include: Step 1: Use finite element simulation to simulate the aging state of IGBTs with different bond wires, extract the magnetic flux density distribution curves perpendicular to the bond wire direction under different bond wire breakage numbers, and perform noise reduction filtering, normalization and aging state labeling on the curve data. The data of the magnetic flux density distribution curve is perpendicular to the extension direction of the bond line and parallel to the plane where the IGBT chip is located. Step 2: Extract any two magnetic flux density data points from the magnetic flux density distribution curve and combine them to form a new combined magnetic flux density feature K. pq Furthermore, the magnetic flux density characteristics K of all combinations of different load currents under the same aging state are... pq They are grouped into one category; and all categories under different aging states are grouped together. Combined magnetic flux density characteristics K pq The composition method is shown in the following formula: （1） Among them, B x1 and B x2 Let N be the magnetic flux density at any two points on the magnetic flux density distribution curve, and let N be the number of bond lines. i is the current distribution coefficient of each bond wire, and S is the part of the bond wire position related to the magnetic flux density calculation formula. Step 3: Establish a magnetic flux density combination evaluation model using inter-class variance (BCV) and minimum inter-class distance (MICD) to evaluate classes and groups composed of data points with different magnetic flux densities and select the optimal magnetic flux density combination feature extraction points. The goal of BCV is to determine the degree of difference between class data, specifically the inter-class variance S. B The calculation formula is: （2） Where, N B n p X p_mean and X _mean These represent the total number of samples, the number of samples in each of the p classes, the p-class mean, and the overall mean, respectively. MICD is used to measure the degree of separation between the two closest groups among all class pairs, with the smallest inter-group distance D being the minimum inter-group distance. min The calculation formula is: （3） Where x p and x q It is category C a and C b Features; Furthermore, the combined characteristic evaluation model for magnetic flux density measures the combined value of BCV and MICD using a comprehensive metric score. The formula for calculating the comprehensive metric score is as follows: （4） In the formula, α and β are both weight values; Step 4: After selecting the optimal monitoring point through the evaluation model, extract the magnetic flux density of the two points to obtain the combined characteristics of magnetic flux density, and use the magnetic flux density extraction board to monitor different aging states of the bonding wire under actual operating conditions.

2. The IGBT bond wire aging monitoring method based on magnetic flux density combination characteristics according to claim 1, characterized in that, In step 3, α and β are both 0.5.

Images