IGBT module aging decoupling monitoring method based on temperature quantitative index

Abstract

The invention discloses an IGBT module aging decoupling monitoring method based on a temperature quantitative index, and the method specifically comprises the steps: collecting the junction temperature and environment temperature of a module in real time when the IGBT module is in a thermal steady state operation condition; according to the junction temperature and the environment temperature, half-cycle node temperature difference and full-cycle node temperature difference are extracted; respectively calculating a first quantitative index for representing the aging degree of the welding layer and a second quantitative index for representing the aging degree of the bonding wire in combination with pre-acquired initial temperature difference data of the IGBT module in the healthy state; and comparing the first quantitative index and the second quantitative index obtained by calculation with a preset health threshold value, and identifying the aging mode of the IGBT module according to a comparison result. According to the invention, the influence of two kinds of aging on the junction temperature is decoupled from the mechanism, and the interference of temperature-dependent parameters is eliminated; the method has the advantages of being low in calculation complexity, free of hardware transformation, high in recognition precision, high in robustness and the like, and is suitable for real-time online maintenance of a converter system.

Description

Technical Field

[0001] This invention belongs to the field of power electronic device condition monitoring and reliability assessment technology, and particularly relates to an aging decoupling monitoring method for IGBT modules based on temperature quantification index. Background Technology

[0002] With the development of power electronics technology, IGBT modules are increasingly widely used in new energy power generation, electric vehicles, and rail transportation. However, as a core component in converters, IGBT modules are subjected to long-term electrical, thermal, and mechanical stresses, making them one of the most vulnerable devices in the system. The main failure modes of IGBT modules include bond-wire fatigue and solder layer degradation. Bond-wire fatigue typically manifests as a collector-emitter on-state voltage drop (Vf). ce An increase in the thermal resistance (R0) will directly lead to increased module power loss, which in turn causes an increase in junction temperature. Solder layer aging usually manifests as solder layer cracks or voids, resulting in decreased internal thermal resistance (R0) of the module. th Increased heat dissipation performance leads to decreased heat dissipation, which in turn causes a rise in junction temperature. Existing technologies face significant challenges in diagnosing these faults, primarily due to the strong thermal coupling effect between the two aging modes: increased losses caused by bond wire aging accelerate solder layer aging, while the temperature rise caused by solder layer aging accelerates bond wire fatigue. Furthermore, since both aging processes ultimately result in increased junction temperature, it is difficult to distinguish which part is causing the problem using only single temperature monitoring. Existing V... ce Online monitoring methods are easily affected by current and temperature fluctuations, making it difficult to guarantee accuracy; while thermal resistance monitoring methods usually require offline testing or complex hardware circuitry, making them difficult to implement during normal converter operation. Therefore, there is an urgent need for an online aging monitoring method that requires no additional hardware, is computationally simple, and can effectively decouple and distinguish between bond wire aging and solder layer aging. Summary of the Invention

[0003] To address the challenge of distinguishing between solder layer aging and bond line aging in existing technologies, this invention provides a decoupled monitoring method for IGBT module aging based on temperature quantification.

[0004] The present invention provides a method for decoupling monitoring of IGBT module aging based on temperature quantification, comprising the following steps:

[0005] Step 1: Under the condition that the IGBT module is in thermal steady-state operation, the junction temperature T of the module is collected in real time. j and ambient temperature T a .

[0006] Step 2: Based on the collected junction temperature T j and ambient temperature Ta Two key temperature difference features were extracted during the power loss cycle: the half-cycle node temperature difference ∆T. half-cycle and the temperature difference ∆T at all nodes throughout the cycle full-cycle .

[0007] Step 3: Using the temperature difference characteristic value obtained in Step 2, combined with the initial temperature difference data of the IGBT module in a healthy state obtained in advance, calculate the first quantitative index Ag1 used to characterize the degree of solder layer aging and the second quantitative index Ag2 used to characterize the degree of bond line aging.

[0008] Step 4: Compare the calculated first quantitative index Ag1 and the second quantitative index Ag2 with the preset health threshold respectively, and identify the aging mode of the IGBT module based on the comparison results. The aging mode includes health status, solder layer aging, bond line aging barrier, and mixed aging of solder layer and bond line.

[0009] Furthermore, the half-cycle node temperature difference ∆T extracted in step 2 half-cycle and the temperature difference ∆T at all nodes throughout the cycle full-cycle This refers to the periodic variation of power loss in an IGBT module until it reaches a thermal steady state, defined as the period over 20 consecutive power cycles during which ∆T... half-cycle and ∆T full-cycle When the rate of change is less than 2%, the positive zero-crossing moment of the fundamental frequency of the output current is taken as the starting point of the period, the fundamental frequency of the output current is taken as f, a power loss period is defined as 1 / f, and the difference between the junction temperature and the ambient temperature at time point t = 1 / (2f) is extracted as the half-cycle node temperature difference ∆T. half-cycle The difference between the junction temperature and the ambient temperature at time point t = 2 / (2f) is extracted as the nodal temperature difference ∆T over the entire cycle. full-cycle .

[0010] Furthermore, the specific formula for calculating the first quantitative indicator Ag1 in step 3 is as follows:

[0011]

[0012] Where, ∆T half-cycle-in and ∆T full-cycle-in These represent the half-cycle node temperature difference and the full-cycle node temperature difference of the IGBT module in its initial healthy state, respectively; ∆T half-cycle-de and ∆T full-cycle-de These represent the half-cycle node temperature difference and the full-cycle node temperature difference of the IGBT module under the current monitoring state; the first quantitative index, Ag1, characterizes the thermal resistance R of the IGBT module. th The change factor.

[0013] Furthermore, the specific formula for calculating the second quantitative indicator Ag2 in step 3 is as follows:

[0014]

[0015] Among them, g in and g de These are all correction factors, calculated as follows:

[0016]

[0017]

[0018] In the formula, f is the fundamental frequency. The thermal time constant under healthy initial conditions. The thermal time constant is the value under the current monitoring condition; the second quantitative index, Ag2, characterizes the power loss amplitude P of the IGBT module. M The change factor.

[0019] Furthermore, the thermal time constant is calculated using the following formula:

[0020]

[0021] The calculation was obtained by substituting the temperature difference data under the initial healthy state. The temperature difference data under the current monitoring status is used to calculate the result. .

[0022] Furthermore, step 4, which identifies the fault modes of the IGBT module based on the comparison results, specifically includes:

[0023] like , If the error between Ag1 and Ag2 and 1 does not exceed 2%, then the IGBT module is determined to be in a healthy state.

[0024] like , The IGBT module was determined to have experienced single-layer solder aging.

[0025] like , The IGBT module was determined to have experienced single bond wire aging.

[0026] like , The IGBT module was determined to have undergone mixed aging, consisting of solder layer aging and bond line aging.

[0027] The beneficial technical effects of this invention compared to the prior art are as follows:

[0028] 1. Strong decoupling: The physical mapping relationship between temperature difference characteristics and fault modes is established through mathematical derivation, eliminating the interference of thermal resistance changes on power loss estimation and realizing accurate separation of the two aging modes.

[0029] 2. Non-invasive: No additional voltage / current sensors or test current injection are required. It can be implemented using only existing temperature monitoring data and is easy to integrate into existing converter control systems.

[0030] 3. Low computational cost: The algorithm is based on analytical formulas and does not require complex iterations or large-scale training, making it suitable for real-time operation in embedded controllers. Attached Figure Description

[0031] Figure 1 This is a flowchart of the IGBT module aging decoupling monitoring method based on temperature quantification index of the present invention.

[0032] Figure 2 This is a circuit topology diagram of the three-phase inverter in the embodiment.

[0033] Figure 3 This is a physical diagram of the three-phase inverter in the embodiment.

[0034] Figure 4 This is a schematic diagram showing the relationship between power loss and junction temperature fluctuation of the IGBT module in this invention, which defines half-cycle and full-cycle nodes.

[0035] Figure 5 This is a comparison of temperature difference waveforms for different fault groups in the embodiment.

[0036] Figure 6 The temperature maps of half-cycle points and full-cycle points extracted from different fault groups in the embodiment are shown.

[0037] Figure 7 The image shows the monitoring results of weld layer aging indicators Ag1 and Ag2 in the example.

[0038] Figure 8 This is a graph used to verify the accuracy of an example. Detailed Implementation

[0039] The present invention will be further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0040] The flowchart of the IGBT module aging decoupling monitoring method based on temperature quantification index of the present invention is as follows: Figure 1 As shown, it includes the following steps:

[0041] Step 1: Under the condition that the IGBT module is in thermal steady-state operation, the junction temperature T of the module is collected in real time. j and ambient temperature T a .

[0042] Step 2: Based on the collected junction temperature T j and ambient temperature T a Two key temperature difference features were extracted during the power loss cycle: the half-cycle node temperature difference ∆T. half-cycle and the temperature difference ∆T at all nodes throughout the cycle full-cycle .

[0043] Extracted half-cycle node temperature difference ∆T half-cycle and the temperature difference ∆T at all nodes throughout the cycle full-cycle This refers to the periodic variation of power loss in an IGBT module until it reaches a thermal steady state, defined as the period over 20 consecutive power cycles during which ∆T... half-cycle and ∆T full-cycle When the rate of change is less than 2%, the positive zero-crossing moment of the fundamental frequency of the output current is taken as the starting point of the period, the fundamental frequency of the output current is taken as f, a power loss period is defined as 1 / f, and the difference between the junction temperature and the ambient temperature at time point t = 1 / (2f) is extracted as the half-cycle node temperature difference ∆T. half-cycle The difference between the junction temperature and the ambient temperature at time point t = 2 / (2f) is extracted as the nodal temperature difference ∆T over the entire cycle. full-cycle .

[0044] Step 3: Using the temperature difference characteristic value obtained in Step 2, combined with the initial temperature difference data of the IGBT module in a healthy state obtained in advance, calculate the first quantitative index Ag1 used to characterize the degree of solder layer aging and the second quantitative index Ag2 used to characterize the degree of bond line aging.

[0045] The specific formula for calculating the first quantitative indicator Ag1 is as follows:

[0046]

[0047] Where, ∆T half-cycle-in and ∆T full-cycle-in These represent the half-cycle node temperature difference and the full-cycle node temperature difference of the IGBT module in its initial healthy state, respectively; ∆T half-cycle-de and ∆T full-cycle-de These represent the half-cycle node temperature difference and the full-cycle node temperature difference of the IGBT module under the current monitoring state; the first quantitative index, Ag1, characterizes the thermal resistance R of the IGBT module. th The change factor.

[0048] The specific formula for calculating the second quantitative indicator Ag2 is as follows:

[0049]

[0050] Among them, g in and g de These are all correction factors, calculated as follows:

[0051]

[0052]

[0053] In the formula, f is the fundamental frequency. The thermal time constant under healthy initial conditions. The thermal time constant is the value under the current monitoring condition; the second quantitative index, Ag2, characterizes the power loss amplitude P of the IGBT module. M The change factor.

[0054] The thermal time constant is calculated using the following formula:

[0055]

[0056] The calculation was obtained by substituting the temperature difference data under the initial healthy state. The temperature difference data under the current monitoring status is used to calculate the result. .

[0057] Step 4: Compare the calculated first quantitative index Ag1 and the second quantitative index Ag2 with the preset health threshold respectively, and identify the aging mode of the IGBT module based on the comparison results. The aging mode includes health status, solder layer aging, bond line aging, and mixed aging of solder layer and bond line.

[0058] The aging modes of IGBT modules identified based on the comparison results specifically include:

[0059] like , If the error between Ag1 and Ag2 and 1 does not exceed 2%, then the IGBT module is determined to be in a healthy state.

[0060] like , The IGBT module was determined to have experienced single-layer solder aging.

[0061] like , The IGBT module was determined to have experienced single bond wire aging.

[0062] like , The IGBT module was determined to have undergone mixed aging, consisting of solder layer aging and bond line aging.

[0063] The preset health threshold is 1, and a preset error tolerance range is allowed in practical applications.

[0064] Principle of aging monitoring:

[0065] This invention is based on the electrothermal coupling model of IGBT modules. In inverter applications, the power loss P of the IGBT module... loss It exhibits a half-sine wave pattern. According to thermal network theory, junction temperature fluctuations depend not only on the amplitude P of the power loss. M It also depends on the module's thermal resistance R. th and heat capacity C th .

[0066] 1. Temperature response model

[0067] Under thermal steady-state conditions, the temperature difference response ∆T of the IGBT module within the power loss cycle (frequency f) can be expressed as the power loss amplitude P. M A function of thermal resistance. For example... Figure 4 As shown, the half-period node is defined as t = 1 / (2f), and the full-period node is defined as t = 2 / (2f).

[0068] Solving the heat transfer differential equation, the half-cycle node temperature difference ∆T is obtained. half-cycle and the temperature difference ∆T at all nodes throughout the cycle full-cycle The following relationship must be satisfied:

[0069]

[0070] in, The transient temperature rise caused by a single power pulse at half a cycle time is related to the power loss amplitude P. M and thermal resistance R th Positive correlation; is the thermal time constant. This represents the thermal decay factor over a complete cycle. The formula above represents the cooling attenuation factor over half a cycle (from peak to trough). It also indicates that the nodal temperature difference ∆T over the entire cycle... full-cycle It is the half-cycle node temperature difference ∆T half-cycle After half a cycle of natural cooling, the ratio of the two is directly determined by the thermal time constant. This decision provides a theoretical basis for the subsequent extraction of thermal resistance variation characteristics.

[0071] 2. Derivation of Ag1, a quantitative index for weld layer aging

[0072] Solder layer aging leads to a reduction in effective heat dissipation area, which is directly reflected in thermal resistance R. th Increase, while heat capacity C th It remains essentially unchanged. Therefore, the thermal time constant... Changes in temperature can directly reflect the aging of the weld layer. From the above temperature response formula, the formula for calculating the thermal time constant can be obtained:

[0073]

[0074] The solder layer aging index Ag1 is defined as the ratio of the thermal time constant of the aged state (subscript de) to the initial state (subscript in), which represents the change in thermal resistance.

[0075] (1)

[0076] When Ag1 > 1, the weld layer is determined to be aged.

[0077] 3. Derivation of Ag2, a quantitative index for bond line aging

[0078] Bond wire aging (breakage or peeling) leads to increased on-resistance, which in turn increases the on-voltage drop V. ce Increase. At the same current, this will result in a power loss magnitude P. M Add. Define the bond wire aging index Ag2 as the power loss amplitude aging state (P M-de ) and initial state (P M-in The ratio of ) to:

[0079] (2)

[0080] However, the extracted temperature difference characteristic ∆T is simultaneously affected by the power loss P M and thermal resistance R th Influence:

[0081]

[0082] In order to accurately extract P M The changes must eliminate R. th The impact of changes. Substituting and rearranging the temperature difference expression, we obtain the decoupled calculation formula:

[0083] (3)

[0084] Where the correction coefficient The calculation formula is:

[0085]

[0086] In the formula, f is the fundamental frequency. When calculating g... in Thermal time constant under healthy initial conditions In calculating g de Thermal time constant under current monitoring conditions The formula uses the first quantification index Ag1 to mathematically compensate for the effect of thermal resistance changes, so that the calculated Ag2 can only reflect the power loss caused by bond wire aging (i.e., the on-state voltage drop V). ce The change in Ag2 is considered. When Ag2 > 1 is detected, it can be determined that the module has experienced bond wire aging.

[0087] Example verification:

[0088] To verify the effectiveness of this invention, a three-phase inverter experimental platform was built (circuit topology as follows). Figure 2 As shown, the actual object is as follows Figure 3 As shown in the figure, the device under test is an FF75R12RT4 IGBT module.

[0089] 1. Experimental parameter settings:

[0090] The DC bus voltage is set to 60V, the switching frequency is 2kHz, and the fundamental output frequency is 0.1Hz. A fiber optic temperature sensor is used to collect the junction temperature T on the surface of the IGBT chip in real time. j .

[0091] 2. Fault simulation grouping:

[0092] Four experimental control groups were set up to simulate different aging states:

[0093] Group A (Healthy Group): Module intact.

[0094] Group B (Solder Layer Aging Group): Thermal grease is added under the module to simulate an increase in solder layer thermal resistance.

[0095] Group C (bond wire aging group): Part of the bond wire was cut to simulate an increase in on-state voltage drop.

[0096] Group D (Mixed Fault Group): Perform both Group B and Group C treatments simultaneously.

[0097] 3. Analysis of monitoring results

[0098] Temperature difference data:

[0099] The temperature difference waveforms of different fault groups are compared, and the temperature graphs of the extracted half-cycle points and full-cycle points are shown below. Figure 5 and Figure 6 As shown, compared with group A, the temperature difference of groups B, C and D all increased to varying degrees, with the mixed fault group D showing the largest temperature rise.

[0100] Indicator Calculation and Fault Identification:

[0101] For Ag1 (e.g.) Figure 7 As shown): The calculated values ​​of groups B and D are significantly greater than 1 (approximately 3), accurately reflecting the increase in thermal resistance; while the Ag1 of group C (only bond line breakage) remains near 1, indicating that the index is not affected by the increase in power loss, verifying the specificity of the solder layer aging.

[0102] For Ag2 (e.g.) Figure 7As shown): The calculated values ​​of groups C and D are significantly greater than 1 (approximately 1.2), accurately reflecting the increase in power loss; while the Ag2 of group B (only thermal resistance increases) remains near 1, indicating that the index successfully eliminates the influence of thermal resistance changes and achieves decoupling from bond line aging.

[0103] 4. Accuracy Verification

[0104] The Ag1 and Ag2 calculated by this invention are compared with the physical degradation ratios measured by precision instruments (k1 is the rate of change of thermal resistance, and k3 is the rate of change of loss). Figure 8 As shown in the results, the relative error of the method of the present invention is less than 2% in all test groups.

[0105] In summary, this invention can accurately and independently monitor the aging status of solder layers and bonding wires during normal inverter operation using simple temperature sampling data, and has extremely high engineering application value.

Claims

1. A decoupling monitoring method for aging IGBT modules based on temperature quantification, characterized in that, Includes the following steps: Step 1: Under the condition that the IGBT module is in thermal steady-state operation, the junction temperature T of the module is collected in real time. j and ambient temperature T a ; Step 2: Based on the collected junction temperature T j and ambient temperature T a Two key temperature difference features were extracted during the power loss cycle: the half-cycle node temperature difference ∆T. half-cycle and the temperature difference ∆T at all nodes throughout the cycle full-cycle ; Step 3: Using the temperature difference characteristic value obtained in Step 2, combined with the initial temperature difference data of the IGBT module in a healthy state obtained in advance, calculate the first quantitative index Ag1 used to characterize the degree of solder layer aging and the second quantitative index Ag2 used to characterize the degree of bond line aging. Step 4: Compare the calculated first quantitative index Ag1 and the second quantitative index Ag2 with the preset health threshold respectively, and identify the aging mode of the IGBT module based on the comparison results. The aging mode includes health status, solder layer aging, bond line aging, and mixed aging of solder layer and bond line.

2. The IGBT module aging decoupling monitoring method based on temperature quantification index according to claim 1, characterized in that, The half-cycle node temperature difference ∆T extracted in step 2 half-cycle and the temperature difference ∆T at all nodes throughout the cycle full-cycle This refers to the periodic variation of power loss in an IGBT module until it reaches a thermal steady state, defined as the period over 20 consecutive power cycles during which ∆T... half-cycle and ∆T full-cycle When the rate of change is less than 2%, the positive zero-crossing moment of the fundamental frequency of the output current is taken as the starting point of the period, the fundamental frequency of the output current is taken as f, a power loss period is defined as 1 / f, and the difference between the junction temperature and the ambient temperature at time point t = 1 / (2f) is extracted as the half-cycle node temperature difference ∆T. half-cycle The difference between the junction temperature and the ambient temperature at time point t = 2 / (2f) is extracted as the nodal temperature difference ∆T over the entire cycle. full-cycle .

3. The IGBT module aging decoupling monitoring method based on temperature quantification index according to claim 1, characterized in that, The specific formula for calculating the first quantitative index Ag1 in step 3 is as follows: ； Where, ∆T half-cycle-in and ∆T full-cycle-in These represent the half-cycle node temperature difference and the full-cycle node temperature difference of the IGBT module in its initial healthy state, respectively; ∆T half-cycle-de and ∆T full-cycle-de These represent the half-cycle node temperature difference and the full-cycle node temperature difference of the IGBT module under the current monitoring state; the first quantitative index, Ag1, characterizes the thermal resistance R of the IGBT module. th The change factor.

4. The IGBT module aging decoupling monitoring method based on temperature quantification index according to claim 3, characterized in that, The specific formula for calculating the second quantitative index Ag2 in step 3 is as follows: ； Among them, g in and g de These are all correction factors, calculated as follows: ； ； In the formula, f is the fundamental frequency. The thermal time constant under healthy initial conditions. The thermal time constant is the value under the current monitoring condition; the second quantitative index, Ag2, characterizes the power loss amplitude P of the IGBT module. M The change factor.

5. The IGBT module aging decoupling monitoring method based on temperature quantification index according to claim 4, characterized in that, The thermal time constant is calculated using the following formula: ； The calculation was obtained by substituting the temperature difference data under the initial healthy state. The temperature difference data under the current monitoring status is used to calculate the result. .

6. The IGBT module aging decoupling monitoring method based on temperature quantification index according to claim 1, characterized in that, Step 4, which identifies the aging mode of the IGBT module based on the comparison results, specifically includes: like , If the error between Ag1 and Ag2 and 1 does not exceed 2%, then the IGBT module is determined to be in a healthy state. like , The IGBT module was determined to have experienced single-layer solder aging. like , The IGBT module was determined to have experienced single bond wire aging. like , The IGBT module was determined to have undergone a mixed aging mode of solder layer and bonding wire.

Images