IGBT junction temperature prediction method and system based on dynamic thermal impedance correction

Abstract

The invention provides an IGBT junction temperature prediction method and system based on dynamic thermal impedance correction. The method comprises the following steps: acquiring temperature data in real time, reversely deducing IGBT loss in combination with a pre-constructed thermal network model, and storing the IGBT loss in historical data; obtaining an IGBT loss average value under the same working condition, and calculating a difference value between the IGBT loss average value and the IGBT loss and a specific value of the IGBT loss average value to obtain a loss offset coefficient; judging whether the loss deviation coefficient is continuously greater than a preset value in a continuous period, if so, judging that the heat-conducting silicone grease is aged, calculating an aging coefficient through loss data and temperature data in the continuous period, and correcting the thermal network model by adopting the aging coefficient; and obtaining the corrected internal transient thermal impedance of the IGBT and the loss of the IGBT, and further predicting the junction temperature of the IGBT. According to the method, the thermal network model is adopted to reversely deduce the real-time loss power of the IGBT, and electric parameter drift errors are avoided; and a dynamic thermal impedance correction mechanism is introduced to ensure long-term accuracy of junction temperature prediction.

Description

Technical Field

[0001] The present invention relates to the field of thermal management of power electronic devices, and particularly to an IGBT junction temperature prediction method and system based on dynamic thermal impedance correction. Background Art

[0002] As a core device of a power electronic system, the junction temperature of an insulated gate bipolar transistor (IGBT) is a key parameter affecting the reliability and lifespan of the system. Accurately predicting the IGBT junction temperature is crucial for optimizing the design of the thermal management system, improving system efficiency, and extending the lifespan of the device.

[0003] Traditional IGBT junction temperature prediction methods mainly rely on electrical models, calculating the loss power by measuring electrical parameters such as the saturation voltage drop and switching loss of the device, and then estimating the junction temperature. However, these methods have the following defects:

[0004] (1) Electrical parameter drift: The electrical parameters of the IGBT will drift due to factors such as device aging and temperature changes, resulting in the accumulation of calculation errors in the loss power and affecting the accuracy of junction temperature prediction.

[0005] (2) Sensor accuracy limitation: In harsh environments such as high temperature and strong electromagnetic interference, the accuracy and reliability of sensors will decline, making it difficult to achieve long-term stable junction temperature monitoring. Moreover, the packaging type of the IGBT and the layout of the printed circuit board (PCB) also greatly increase the difficulty of sensor detection of electrical parameters.

[0006] (3) Model simplification error: Traditional models usually adopt lumped parameter models and cannot reflect the temperature gradient inside the IGBT chip and within the packaging structure.

[0007] To solve the above problems, the present invention proposes an IGBT junction temperature prediction method based on dynamic thermal impedance correction. Summary of the Invention

[0008] Based on this, it is necessary to provide an IGBT junction temperature prediction method based on dynamic thermal impedance correction for the problem of existing deviation in calculating loss power relying on electrical parameters.

[0009] To solve the above technical problems, the technical solution adopted by the present invention is:

[0010] The IGBT junction temperature prediction method based on dynamic thermal impedance correction includes the following steps:

[0011] Obtain temperature data in real time, and inversely calculate the IGBT loss P in combination with a pre-constructed thermal network model loss and save it to historical data; the temperature data includes the IGBT case temperature T case , the heat sink substrate temperature T hs and the ambient temperature T amp ;

[0012] Obtain the average IGBT loss P under the same operating conditions loss-pre , and calculate the loss offset coefficient dP by calculating the ratio of the difference between it and the IGBT loss P loss ;

[0013] Judge whether the loss offset coefficient dP continues to be greater than the preset value within consecutive cycles. If so, it is determined that the thermal grease is aged, and the aging coefficient is calculated through the loss data and temperature data within consecutive cycles, and the thermal network model is corrected using the aging coefficient;

[0014] Obtain the corrected internal transient thermal impedance Z of the IGBT th and the IGBT loss P loss ’, and then predict the IGBT junction temperature T junction : T junction = P loss ’ * Z th + T case .

[0015] Furthermore, the specific derivation process of the IGBT loss P loss is as follows:

[0016] Obtain the ambient temperature T amp , the heat sink substrate temperature T hs in real time, and calculate the thermal grease temperature T of the IGBT in combination with the pre-constructed thermal network model TIM : where C hs and R hs represent the equivalent heat capacity and thermal resistance of the heat sink, t represents the time parameter, and R TIM represents the equivalent thermal resistance of the thermal grease;

[0017] Obtain the equivalent heat capacity C of the thermal grease TIM , and derive the IGBT loss P loss :

[0018] Furthermore, the IGBT case temperature T case and the heat sink substrate temperature T hs are obtained using an infrared thermopile and an NTC thermistor, and the ambient temperature T amp is obtained using an ambient temperature sensor. The three are synchronously sampled and data fusion and compensation are performed through the Kalman filter algorithm to obtain the required temperature data.

[0019] Furthermore, the calculation process of the aging coefficient is as follows:

[0020] Calculate the average IGBT loss P through the loss data within consecutive cycles loss-pre, introduce the aging coefficients α, β, and the average loss P loss-pre into the loss formula Solve the aging coefficients α and β through the temperature data within consecutive cycles.

[0021] Furthermore, when determining whether the loss offset coefficient dP continuously exceeds a preset value within consecutive cycles, the determination is made on the premise that the IGBT and heat sink structure and parameters are stable.

[0022] Furthermore, two groups of infrared thermopiles are respectively vertically installed 10 - 20 mm directly above the IGBT housing and the surface of the heat sink, and two groups of NTC thermistors are respectively arranged on the surface of the IGBT housing and the surface of the heat sink adjacent to the IGBT.

[0023] Furthermore, use a single-chip microcomputer to perform synchronous ADC sampling on the infrared thermopiles, NTC thermistors, and environmental temperature sensors.

[0024] The present invention also relates to an IGBT junction temperature prediction system based on dynamic thermal impedance correction, including an IGBT loss derivation module, a loss offset coefficient calculation module, a thermal network model correction module, and an IGBT junction temperature prediction module.

[0025] The IGBT loss derivation module is used to obtain temperature data in real time and inversely deduce the IGBT loss P in combination with a pre-constructed thermal network model loss and save it to historical data; the temperature data includes the IGBT housing temperature T case , the heat sink substrate temperature T hs , and the environmental temperature T amp .

[0026] The loss offset coefficient calculation module is used to obtain the average IGBT loss P under the same working conditions loss-pre , calculate the difference between it and the IGBT loss P loss and its ratio to obtain the loss offset coefficient dP.

[0027] The thermal network model correction module is used to determine whether the loss offset coefficient dP continuously exceeds a preset value within consecutive cycles. If so, it is determined that the thermal grease has aged, calculate the aging coefficient through the loss data and temperature data within consecutive cycles, and correct the thermal network model using the aging coefficient.

[0028] The IGBT junction temperature prediction module is used to obtain the corrected internal transient thermal impedance Z of the IGBT th and the IGBT loss P loss ’, and then predict the IGBT junction temperature T junction : T junction = P loss ’ * Z th + T case .

[0029] Compared with the prior art, the beneficial effects of the present invention include:

[0030] 1. By adopting a thermal network model to inversely deduce the real-time loss power of the IGBT, the drift error of electrical parameters is avoided; the calculation error of the loss power caused by the drift of the IGBT electrical parameters with aging and temperature change is avoided, and the long-term accuracy of prediction is improved;

[0031] 2. By introducing a dynamic thermal impedance correction mechanism, real-time compensation is carried out for the thermal impedance change caused by the aging of the thermal interface material (TIM), ensuring the long-term accuracy of the junction temperature prediction;

[0032] 3. Through the multi-source temperature data fusion, the accuracy, reliability and response speed of temperature measurement are improved, providing a high-quality data basis for the subsequent loss inversion and junction temperature prediction. Description of the Drawings

[0033] The disclosure of the present invention will be described with reference to the accompanying drawings. It should be understood that the drawings are only for illustrative purposes and are not intended to limit the scope of protection of the present invention. In the drawings, the same reference numerals are used to refer to the same components. Among them:

[0034] Figure 1 is the flow chart of the IGBT junction temperature prediction method based on dynamic thermal impedance correction introduced in Embodiment 1 of the present invention;

[0035] Figure 2 is the cascaded cauer model diagram of the IGBT including the heat capacity effect and thermal grease (TIM) and heat sink inside;

[0036] Figure 3 is based on Figure 2 Schematic diagram of the foster-cauer hybrid model. Detailed Embodiments

[0037] It is easy to understand that according to the technical solution of the present invention, without changing the essence of the present invention, those of ordinary skill in the art can propose various structural ways and implementation ways that can be mutually replaced. Therefore, the following detailed embodiments and the accompanying drawings are only illustrative descriptions of the technical solution of the present invention, and should not be regarded as all of the present invention or regarded as a limitation or restriction on the technical solution of the present invention.

[0038] Embodiment 1

[0039] This embodiment introduces an IGBT junction temperature prediction method based on dynamic thermal impedance correction, including the following steps:

[0040] Step S1. Obtain temperature data in real time, and inversely deduce the IGBT loss P by combining with the pre-constructed thermal network modelloss and save it into the historical data; the temperature data includes the IGBT case temperature T case , the radiator substrate temperature T hs and the ambient temperature T amp .

[0041] The temperature data is collected by an infrared thermopile, an NTC thermistor, and an ambient temperature sensor. The specific deployment is shown in Table 1:

[0042] Table 1: Temperature Data Acquisition Deployment Table

[0043]

[0044] The ambient temperature sensor is used to collect the ambient temperature T amp of the IGBT. The single-chip microcomputer is used for synchronous ADC sampling to ensure that the acquisition time difference between the IGBT case temperature and the heatsink temperature does not exceed 1 ms. A timestamp is established, and the high-precision temperature sampling data of the NTC thermistor is used to correct the temperature accuracy of the thermopile. At the same time, the Kalman filter is combined to reduce the sampling error caused by sensor errors and environmental disturbances.

[0045] Construct a cascaded cauer model for the IGBT that includes the heat capacity effect, thermal grease (TIM), and heatsink as shown in Figure 2 . Since Figure 1 part of the IGBT structure is encapsulated in the module and cannot be directly modeled and simulated using finite elements, the transient thermal impedance curve in the IGBT module manual is used for fitting to establish a hybrid model of the foster model and the cauer model, as shown in Figure 3 . The parameter calibration of the hybrid model of the foster model and the cauer model is shown in the following table:

[0046] Table 2: Parameter Calibration Table for the foster-cauer Hybrid Model

[0047]

[0048] Combine the obtained case temperature T cas and the radiator temperature T hs and the ambient temperature T amp with the foster-cauer hybrid model to obtain the IGBT loss P loss : where C hs and R hs represent the equivalent heat capacity and thermal resistance of the radiator, t represents the time parameter, and C TIM and R TIM represent the equivalent heat capacity and thermal resistance of the thermal grease.

[0049] Step S2. Obtain the average IGBT loss P under the same working conditions loss-pre , calculate the difference between it and the IGBT loss P loss , and then calculate the loss offset coefficient dP by dividing the difference by the ratio of the two.

[0050] To ensure full connection between the IGBT and the heat sink, thermal interface material (TIM) is applied between the two contact surfaces. Over time, the TIM will gradually age, manifested as a decrease in heat capacity C TIM and an increase in thermal resistance R TIM , resulting in a decline in the system's heat dissipation capacity and inaccurate calculation of P loss and the IGBT junction temperature T junction . Therefore, it is necessary to correct the thermal impedance of the TIM at regular intervals to obtain accurate loss P loss and the IGBT junction temperature T junction .

[0051] After the IGBT power loss P loss is obtained by back - calculation, it is saved in the historical data and compared with the average value of the historical data under the same working conditions. According to the following formula, the P loss loss offset coefficient dP is obtained:

[0052]

[0053] where P loss-pre is the average value of the back - calculated P loss under the same working conditions of the IGBT.

[0054] Step S3. Determine whether the loss offset coefficient dP continuously exceeds the preset value within a continuous period. If so, it is determined that the thermal interface material is aging. Calculate the aging coefficient based on the loss data and temperature data within the continuous period, and use the aging coefficient to correct the thermal network model.

[0055] If dP > 0.1 is satisfied continuously for multiple times, it indicates that the back - calculated P loss differs by more than 10% compared with the historical data. Assuming that the structure and parameters of the IGBT and the heat sink are stable, it is determined that the TIM is aging. Correct the TIM RC thermal resistance parameters:

[0056] According to Equation and replace P loss with P loss_pre , replace R TIM with αR TIM , and replace C TIM with βC TIM , to obtain:

[0057]

[0058] Among them, α and β are aging coefficients. By sampling temperature data at consecutive time points and loss history data, aging parameters are solved, the thermal network model is corrected, and the aging coefficients are saved. The internal transient thermal impedance Z of the IGBT is obtained again through the corrected thermal network model th and the IGBT loss P loss ’.

[0059] Step S4. Obtain the corrected internal transient thermal impedance Z of the IGBT th and the IGBT loss P loss ’, and then predict the IGBT junction temperature T junction : T junction = P loss ’ * Z th + T case .

[0060] Judging the working condition of the IGBT based on the calculated curve of the junction temperature changing with time, and judging whether overheat protection is needed, is more reasonable than the traditional method of detecting whether the IGBT is overheated by measuring the temperature of the heat sink connected to the IGBT, and can more effectively protect the normal operation of the IGBT. It should be noted that if the loss offset coefficient dP is not continuously greater than the preset value within consecutive cycles, it is considered that the thermal grease is not aged, and there is no need to correct the thermal network model. After the IGBT power loss is deduced, the junction temperature can be deduced through the foster-cauer hybrid model in Figure 3 , and the formula is as follows: T junction = P loss Z th + T case .

[0061] Therefore, this embodiment has the following technical effects:

[0062] 1. Inverse deduction method based on a high-order distributed thermal network model: The cauer thermal network model is used to describe the thermal characteristics of the IGBT and its heat dissipation system, and the measured IGBT case temperature, heat sink substrate temperature, and ambient temperature are used to inversely deduce the real-time loss power of the IGBT. This method avoids directly measuring the electrical parameters of the IGBT, thus eliminating the errors caused by electrical parameter drift.

[0063] 2. Dynamic thermal impedance correction based on multi-sensor data: Combining the Kalman filter algorithm, combining multi-source temperature information and historical data, dynamically estimating the thermal impedance and heat capacity of the thermal interface material (TIM), and real-time updating the thermal network model parameters. This method can effectively compensate for the model mismatch caused by TIM aging and improve the long-term accuracy of junction temperature prediction.

[0064] 3. Multi-source temperature fusion and intelligent compensation: Infrared thermopiles, NTC thermistors, and ambient temperature sensors are used for multi-source temperature measurement, and data fusion is carried out through hardware-level synchronization and intelligent compensation algorithms (Kalman filtering) to improve the accuracy and response speed of temperature measurement.

[0065] At the method level: Shift from traditional electrical models to thermal network models and use backstepping methods to calculate loss power.

[0066] At the model level: Introduce a dynamic thermal impedance correction mechanism to solve the model mismatch problem caused by TIM aging.

[0067] At the system level: Adopt a multi-modal temperature sensing scheme and intelligent data processing methods to improve the quality of temperature data and the robustness of the system.

[0068] This embodiment provides a more accurate, reliable, and robust IGBT junction temperature prediction method, which is especially suitable for high-power application scenarios such as electric vehicles and new energy converters with high requirements for long-term operation reliability. By using thermal network models and multi-sensor fusion technologies, the limitations of traditional electrical model methods are overcome, and the system's adaptability and economy are improved.

[0069] Embodiment 2

[0070] This embodiment introduces an IGBT junction temperature prediction system based on dynamic thermal impedance correction, including an IGBT loss derivation module, a loss offset coefficient calculation module, a thermal network model correction module, and an IGBT junction temperature prediction module.

[0071] The GBT loss derivation module is used to obtain temperature data in real time, and combined with a pre-constructed thermal network model, it back-calculates the IGBT loss P loss and saves it to historical data; the temperature data includes the IGBT case temperature T case , the heat sink substrate temperature T hs , and the ambient temperature T amp .

[0072] The loss offset coefficient calculation module is used to obtain the average IGBT loss P loss-pre under the same working conditions, and calculates the difference between it and the IGBT loss P loss and its ratio to obtain the loss offset coefficient dP.

[0073] The thermal network model correction module is used to determine whether the loss offset coefficient dP continuously exceeds a preset value within a continuous period. If so, it determines that the thermal grease is aging, calculates the aging coefficient through the loss data and temperature data within the continuous period, and uses the aging coefficient to correct the thermal network model.

[0074] The IGBT junction temperature prediction module is used to obtain the corrected internal transient thermal impedance Z of the IGBTth and the IGBT loss P loss ’, and then predict the IGBT junction temperature T junction : T junction = P loss ’ * Z th + T case .

[0075] Embodiment 3

[0076] This embodiment provides a computer terminal, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the steps of the IGBT junction temperature prediction method based on dynamic thermal impedance correction in Embodiment 1 are implemented.

[0077] When the IGBT junction temperature prediction method based on dynamic thermal impedance correction in Embodiment 1 is applied, it can be applied in the form of software. For example, it can be designed as an independently running program and installed on a computer terminal. The computer terminal can be a computer, a smart phone, etc. It can also be designed as an embedded running program and installed on a computer terminal, such as installed on a single-chip microcomputer.

[0078] Embodiment 4

[0079] This embodiment provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the steps of the IGBT junction temperature prediction method based on dynamic thermal impedance correction in Embodiment 1 are implemented.

[0080] When the IGBT junction temperature prediction method based on dynamic thermal impedance correction in Embodiment 1 is applied, it can be applied in the form of software. For example, it can be designed as an independently running program on a computer-readable storage medium. The computer-readable storage medium can be a USB flash drive, designed as a USB key, and designed as a program to start the whole method through external triggering through the USB flash drive.

[0081] The technical scope of the present invention is not limited to the content described above. Those skilled in the art can make various deformations and modifications to the above embodiments without departing from the technical idea of the present invention, and these deformations and modifications should all fall within the protection scope of the present invention.

Claims

1. A method for predicting the IGBT junction temperature based on dynamic thermal impedance correction, which is applied to an IGBT whose contact surface is connected to a radiator through thermal grease, is characterized in that It includes: Obtain temperature data in real time, and inversely deduce the IGBT loss P in combination with a pre-constructed thermal network model loss and save it into historical data; the temperature data includes the IGBT case temperature T case , the heat sink substrate temperature T hs and the ambient temperature T amp ; Obtain the average IGBT loss P under the same operating conditions loss-pre , and calculate the loss offset coefficient dP by calculating the ratio of the difference between it and the IGBT loss P loss ; Judging whether the loss offset coefficient dP continuously exceeds a preset value within consecutive periods. If so, it is determined that the thermal grease is aged, and the aging coefficient is calculated based on the loss data and temperature data within consecutive periods, and the thermal network model is corrected using the aging coefficient; Obtain the corrected internal transient thermal impedance Z of the IGBT th and the IGBT loss P loss ’, and then predict the IGBT junction temperature T junction : T junction = P loss ’ * Z th + T case .

2. The IGBT junction temperature prediction method based on dynamic thermal impedance correction according to claim 1, characterized in that IGBT loss P loss The specific derivation process is as follows: Obtain the ambient temperature T in real time amp and the radiator substrate temperature T hs , and calculate the thermal grease temperature T of the IGBT by combining with the pre-constructed thermal network model TIM : where C hs and R hs represent the equivalent heat capacity and thermal resistance of the radiator, t represents the time parameter, and R TIM represents the equivalent thermal resistance of the thermal grease; Obtain the equivalent heat capacity C of the thermal grease TIM , and derive the IGBT loss P loss :

3. The IGBT junction temperature prediction method based on dynamic thermal impedance correction according to claim 2, wherein IGBT case temperature T case and the radiator substrate temperature T hs are obtained by using an infrared thermopile and an NTC thermistor. The ambient temperature T amp is obtained by using an ambient temperature sensor. The three are synchronously sampled and the data is fused and compensated through the Kalman filtering algorithm to obtain the required temperature data.

4. The IGBT junction temperature prediction method based on dynamic thermal impedance correction according to claim 2, characterized in that The calculation process of the aging coefficient is as follows: Calculate the average IGBT loss P based on the loss data within consecutive periods loss-pre , and introduce the aging coefficients α, β and the average loss P loss-pre into the loss formula Solve for the aging coefficients α, β based on the temperature data within consecutive periods 5. The IGBT junction temperature prediction method based on dynamic thermal impedance correction according to claim 4, characterized in that When judging whether the loss offset coefficient dP continuously exceeds a preset value within consecutive periods, the judgment is made on the premise that the IGBT and radiator structures and parameters are stable.

6. The IGBT junction temperature prediction method based on dynamic thermal impedance correction according to claim 3, characterized in that Two groups of infrared thermopiles are respectively vertically installed 10 - 20 mm directly above the IGBT housing and the radiator surface, and two groups of NTC thermistors are respectively arranged on the surface of the IGBT housing and the surface of the radiator adjacent to the IGBT.

7. The IGBT junction temperature prediction method based on dynamic thermal impedance correction according to claim 3, characterized in that The single-chip microcomputer is used to perform synchronous ADC sampling on the infrared thermopile, NTC thermistor, and ambient temperature sensor.

8. The IGBT junction temperature prediction system based on dynamic thermal impedance correction is characterized in that It includes: IGBT loss derivation module, which is used to obtain temperature data in real time and inversely deduce IGBT loss P in combination with a pre-constructed thermal network model loss and save it into historical data; the temperature data includes the IGBT case temperature T case , the heat sink substrate temperature T hs and the ambient temperature T amp ; Loss offset coefficient calculation module, which is used to obtain the average IGBT loss P under the same working conditions loss-pre , and calculate the difference between it and the IGBT loss P loss . After calculating the ratio of the difference to obtain the loss offset coefficient dP; A thermal network model correction module, which is used to judge whether the loss offset coefficient dP continuously exceeds a preset value within consecutive periods. If so, it is determined that the thermal grease is aged, and the aging coefficient is calculated based on the loss data and temperature data within consecutive periods, and the thermal network model is corrected using the aging coefficient; IGBT junction temperature prediction module, which is used to obtain the corrected internal transient thermal impedance Z of the IGBT th and the IGBT loss P loss ’, and then predict the IGBT junction temperature T junction : T junction = P loss ’ * Z th + T case .

9. A computer terminal, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the IGBT junction temperature prediction method based on dynamic thermal impedance correction as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the IGBT junction temperature prediction method based on dynamic thermal impedance correction as described in any one of claims 1 to 7.

Images