A frequency domain analysis method for checking IGBT short circuit type 1 behavior simulation model

Abstract

This invention belongs to the field of semiconductor technology, specifically a frequency domain analysis method for verifying the simulation model of IGBT short-circuit type 1 behavior. It includes: S100: calculating the fundamental frequency of the DFT analysis based on the short-circuit duration; S200: calculating the voltage frequency domain evaluation under frequency domain analysis. P V S300: Current Frequency Domain Evaluation under Computational Frequency Domain Analysis P I S400: Evaluation based on voltage frequency domain P V Current frequency domain evaluation P I Calculate the overall frequency domain evaluation index. P This invention serves as a validity metric for simulation models of different short-circuit type 1 failures. It provides better support for the full-condition design and selection of devices such as frequency converters, improving design efficiency while reducing R&D cycle and cost.

Description

Technical Field

[0001] This invention belongs to the field of semiconductor technology, specifically a frequency domain analysis method for verifying the simulation model of IGBT short-circuit type 1 behavior. Background Technology

[0002] IGBTs, semiconductor devices, possess advantages such as fully controllable switching, high withstand voltage ratings, large current carrying capacity, high switching frequency, and low drive power, making them the most widely used power electronic switching devices in medium- and high-power applications such as industrial motor drives, medium- and high-voltage frequency converters, and smart grids. IGBT type 1 behavior refers to a load short circuit or shoot-through between two IGBTs on a bridge arm before the IGBT is turned on. Once such a fault occurs, it can lead to damage to the IGBT and even the entire system. To address this, numerous studies have conducted simulation modeling and analysis of IGBT short-circuit switching processes to demonstrate type 1 short-circuit behavior. However, when evaluating these models, only a rough comparison of short-circuit voltage and current characteristic curves is used, without the development of suitable quantitative indicators, making it impossible to effectively demonstrate the model's effectiveness.

[0003] The IGBT device short-circuit type 1 behavior failure simulation model is used to simulate the voltage and current stress changes of specific IGBT devices during a type 1 short circuit. More specifically, it tracks the current rise rate at the start of the short circuit, the voltage rise rate during forced turn-off, and even the relationship between junction temperature and time. These parameters can provide data for final device-level experiments. For example, in short-circuit tests of medium- and high-power frequency converters, the simulation model of the IGBT device model within the device can be used to simulate the electrical stress changes under stable and short-circuit conditions, preventing unexpected situations such as device damage in real experiments. Furthermore, the data conclusions from the type 1 short-circuit failure simulation model can even guide the design of frequency converters. However, current simulation models for short-circuit type 1 failure of IGBT devices lack suitable quantitative indicators. The effectiveness of the model is judged only by roughly comparing the peak values ​​and shapes of voltage and current curves. This comparison without quantitative indicators can be sufficient under the normal steady-state conditions of the inverter device. However, once a type 1 short-circuit failure occurs, it may cause irreversible damage to the equipment and facilities. A reliable simulation model is needed to evaluate this condition. Therefore, simulation models for short-circuit type 1 failure cannot be evaluated without quantitative indicators. Summary of the Invention

[0004] To address the lack of reliability quantification evaluation indicators in short-circuit type 1 failure simulation models, this invention employs DFT to perform frequency domain analysis on the voltage and current waveforms of specific devices during the short-circuit process, providing a frequency domain analysis method for verifying IGBT short-circuit type 1 behavior simulation models.

[0005] This invention adopts the following technical solution: a frequency domain analysis method for verifying the short-circuit behavior simulation model of IGBT type 1, comprising:

[0006] S100: Calculate the fundamental frequency for DFT analysis based on the short-circuit duration;

[0007] S200: Voltage Frequency Domain Evaluation under Computational Frequency Domain Analysis P V ;

[0008] S300: Current Frequency Domain Evaluation under Computational Frequency Domain Analysis P I ;

[0009] S400: Evaluation based on voltage frequency domain P V Current frequency domain evaluation P I Calculate the overall frequency domain evaluation index. P , as a measure of the effectiveness of different short-circuit type 1 failure simulation models.

[0010] In some embodiments, in step S100, the fundamental frequency f = 1 / T, where T is the duration of the short-circuit behavior waveform of type 1, and the highest frequency of the DFT analysis is set to 100 times the fundamental frequency.

[0011] In some embodiments, in step S100, the time from 0.5us before the start of the type1 short circuit behavior to 0.5us after the end of the type1 short circuit behavior is 0.5us.

[0012] In some embodiments, in step S200, voltage frequency domain evaluation P V The calculation is as follows:

[0013]

[0014] In the formula, where, n This is the frequency component index; specifically, when n=0, it represents the DC component. N v The number of frequency components after decomposition of a voltage sequence waveform; F v ( n ) is the first voltage waveform sequence after decomposition. n The amplitude of each frequency component; F ’ v (n) represents the nth digit of the simulated voltage waveform sequence after decomposition of the simulation model. n The amplitude of each frequency component, N + In mathematics, it is a symbolic representation of a set of numbers, meaning an integer.

[0015] In some embodiments, voltage frequency domain evaluation P V It is obtained through the following process:

[0016] According to DFT theory, waveform sequences f ( k ) and DFT transformed F ( n The relationship between them is:

[0017]

[0018] in, k For time sequence number, n It is a multiple of the fundamental frequency. N Number of moments;

[0019] in F ( n Since ) is a complex number and not a real number, F ( n This can be rewritten as:

[0020]

[0021] Among them, Re[ F [(n)] is the real part of F(n), Im[ F ( n )]for F ( n The imaginary part of ) | F ( n | represents the amplitude. For phase;

[0022] Combining the above formula, the waveform sequence can be rewritten as:

[0023]

[0024] Measured voltage waveform sequence f V ( n ) and simulated voltage waveform sequence f’ V ( n )have:

[0025]

[0026] The voltage frequency domain evaluation can be derived from the above formula. P V It can be calculated as follows:

[0027] .

[0028] in, n This is the frequency component index; specifically, when n=0, it represents the DC component. N v The number of frequency components after decomposition of a voltage sequence waveform; F v ( n ) is the first voltage waveform sequence after decomposition. n The amplitude of each frequency component; F ’ v (n) represents the nth digit of the simulated voltage waveform sequence after decomposition of the simulation model. n The amplitude of each frequency component, N + In mathematics, it is a symbolic representation of a set of numbers, meaning an integer.

[0029] In some embodiments, in step S200, N V Constrained by the following formula:

[0030]

[0031] in It is 0.95.

[0032] In some embodiments, in step S300, the current frequency domain evaluation is performed. P I for:

[0033]

[0034] In the formula, where, n This is the frequency component index; specifically, when n=0, it represents the DC component. N I The number of frequency components after decomposition of the current sequence waveform; F I ( n ) is the first digit after decomposition of the current waveform sequence. n The amplitude of each frequency component; F ’ I (n) represents the nth sequence of the simulated current waveform after decomposition of the simulation model. n The amplitude of each frequency component, N + In mathematics, it is a symbolic representation of a set of numbers, meaning an integer.

[0035] In some embodiments, in step S300, the current frequency domain evaluation is performed. P I It is obtained through the following process:

[0036] Measured current waveform sequencef I ( n and simulated current waveform sequence f’ I ( n )have:

[0037]

[0038] The current frequency domain evaluation can be derived from the above formula. P I for:

[0039] .

[0040] In some embodiments, in step S400, the frequency domain analysis index of the IGBT device short-circuit type 1 behavior failure simulation model is... P for:

[0041] .

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] 1. This invention provides a method for controlling the collector-emitter voltage V during short-circuit behavior of type 1. CE and collector current I C The frequency domain analysis indices formed by performing DFT analysis separately P As a quantitative indicator for evaluating the type 1 short-circuit failure model, it improves the evaluation capability of the short-circuit failure model.

[0044] 2. This invention provides better support for the design and selection of devices such as frequency converters under all operating conditions, improves design efficiency, and reduces the development cycle and cost. Attached Figure Description

[0045] Figure 1 The present invention provides voltage and current waveforms during the short-circuit behavior of an IGBT device type 1;

[0046] Figure 2 This invention provides an example of waveform selection range for a frequency domain evaluation method for the short-circuit behavior of an IGBT type 1 device.

[0047] Figure 3 This invention provides a DFT decomposition sequence diagram of the short-circuit voltage waveform of an IGBT device.

[0048] Figure 4 A comparison chart of the measured curves of an IGBT device provided by the present invention and the simulation curves of the short-circuit behavior of different model type 1;

[0049] Figure 5 The spectrum analysis diagram of the measured voltage of an IGBT and the simulated voltage of short-circuit behavior of different model type 1 is provided for the present invention.

[0050] Figure 6 The present invention provides a spectrum analysis diagram of the measured current of an IGBT and the simulated current of short-circuit behavior of different model type 1. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0053] Figure 1 The middle figure shows the voltage curve V during the short-circuit failure process of an IGBT device (type 1). CE and current curve I C To ensure the integrity of the frequency domain analysis method, it is agreed that the waveform for DFT analysis includes 0.5µs before the start of the type 1 short-circuit behavior and 0.5µs after the end of the type 1 short-circuit behavior, such as... Figure 2 As shown, the gray line represents the entire type 1 short-circuit behavior process, and the overlapping part of the gray and black lines is the part analyzed by DFT. Let the duration of this waveform be T, then the fundamental frequency f = 1 / T. Considering the degree of overlap of high-frequency oscillations during the short-circuit process, the highest frequency for DFT analysis is set to 100 times the fundamental frequency.

[0054] refer to Figure 3 , Figure 3 By gradually superimposing waveforms of specific frequency components, it can be seen from the figure that the superimposed waveforms of the first 49 frequency components already approximate the details of the original waveform, and the first 69 frequency components are sufficient to make the waveform evaluation criteria meet the constraints of claim 6 in the statement; (Reference) Figure 5 , Figure 5 The graph shows a comparison of the measured waveform, the simulated waveform of the first model, and the simulated waveform of the second model. As the frequency increases, the content gradually decreases and approaches 0. Therefore, a frequency higher than 100 times is an empirical value, and this limitation allows the algorithm to run stably and effectively.

[0055] According to DFT theory, waveform sequences f ( k ) and DFT transformed F (n The relationship between them is:

[0056] (1)

[0057] in, k For time sequence number, n It is a multiple of the fundamental frequency. N This represents the number of time points. F ( n Since ) is a complex number and not a real number, F ( n This can be rewritten as:

[0058] (2)

[0059] Among them, Re[ F [(n)] is the real part of F(n), Im[ F ( n )]for F ( n The imaginary part of ) | F ( n | represents the amplitude. For phase. Therefore, combining (1) and (2), the waveform sequence can be rewritten as:

[0060] (3)

[0061] Then the measured voltage waveform sequence f V ( n ) and simulated voltage waveform sequence f’ V ( n )have:

[0062] (4)

[0063] Then voltage frequency domain evaluation P V It can be calculated as follows:

[0064] (5)

[0065] P V and N V The number of elements is related; on the one hand, directly increasing N... V This will lead to P V Decrease; on the other hand, because when | F V ( n When decreasing, the accuracy is not high. FV ’ ( n )|make(| F V ( n )|-| F V ’ ( n )|) / | F V ( n The more it expands, the more it leads to... P V Not robust. Therefore, the agreement... N V Constrained by the following formula:

[0066] (6)

[0067] in The value is 0.95. This constraint is... N v The solution can only be achieved by first solving the expression itself. P v Solve for it.

[0068] Ensure P V The main spectral amplitudes were fully calculated, while avoiding the occurrence of robustness issues.

[0069] Similarly, the measured current waveform sequence f I ( n and simulated current waveform sequence f’ I ( n )have:

[0070] (7)

[0071] Similarly, current frequency domain evaluation P I for:

[0072] (8)

[0073] The frequency domain analysis parameters of the IGBT device short-circuit type 1 behavior failure simulation model are as follows: P for:

[0074] (9)

[0075] This invention is mainly aimed at evaluating the effectiveness of the short-circuit type 1 failure simulation model for semiconductor IGBT devices. For short-circuit failure models of other similar semiconductor devices, including semiconductor devices with different packaging forms, voltage levels and capacity levels, the method provided in this patent can be used as a reference.

[0076] Example:

[0077] Taking the short-circuit behavior of a certain type 1 IGBT as an example, its specific voltage and current curves are as follows: Figure 1 As shown, the length of the waveform being analyzed is selected simultaneously as follows: Figure 2 As shown, the gray line represents the entire type 1 short-circuit behavior process, and the overlapping part of the black and red lines is the part analyzed by DFT. The short-circuit type 1 behavior lasts for a total of 12µs, so the fundamental frequency is 83.4kHz and the highest frequency is 8.34MHz.

[0078] The voltage curve in the measured waveform can be subjected to DFT analysis, and as shown in equation (3), its original voltage waveform can be converted into a superposition of several sinusoidal waveforms, as shown in equation (3). Figure 3 As shown, the comparison between the original voltage waveform and the waveform with only the DC component and with the first 4, 9, 14, 19, 49, 69, and 99 frequency components respectively is clearly demonstrated. It can be found that as the number of superimposed components increases, the sequence represented by the DFT method in Equation (3) becomes more and more similar to the original waveform sequence, indicating the rationality of using DFT for frequency domain analysis.

[0079] Two type 1 short-circuit behavior failure simulation models were applied to model this type of IGBT. The first model is the classic Hefner model, and the second model, called the Improved Hefner model, is an improvement on the first. The second model has been verified in practice and better reflects the true characteristics of various IGBT devices. Figure 4 The figure shows a comparison of the voltage and current waveforms of the IGBT device under the actual type 1 short-circuit condition with the first and second type 1 short-circuit behavior failure simulation models. It can be found that the current curve of the second model is not only closer to the actual waveform, but also the voltage waveform oscillation at the turn-off moment is smaller, and the voltage drop rate is comparable to the actual voltage drop rate.

[0080] DFT analysis was performed on these three waveforms, and the amplitudes of their respective frequency components were compared. The voltage spectrum is shown below. Figure 5 As shown, the current spectrum is as follows Figure 6As shown in the diagram, in the voltage spectrum, the frequency components from the 1st to the 10th orders of the first model far exceed the frequency components of the original waveform compared to the second model. The trend for the frequency components from the 11th to the 14th orders is opposite to that of the 1st to the 10th orders. For the frequency components after the 15th order, the second model surpasses the first model and is closer to the frequency components of the true waveform. In the current spectrum, the amplitudes of almost all frequency components in the second model are closer to the true waveform than those in the first model.

[0081] To more accurately evaluate model effectiveness and compare the performance of evaluation metrics, the root mean square error (RMSE), mean absolute percentage error (MAPE), and spectral analysis metrics were calculated. P The comparisons are shown in Table 1.

[0082] Table 1

[0083]

[0084] Frequency domain analysis indicators P The percentage difference in frequency components between the measured and simulated sequences was calculated, no longer affected by the sequence dimensions. N V , N I The introduction of this method enhances the robustness of evaluating P, therefore the frequency domain analysis index... P It overcomes the shortcomings of RMSE in measuring the degree of waveform difference and MAPE in being unromantic due to the number of data points. Meanwhile... P Similar to RMSE and MAPE, in the problem of modeling the short-circuit state of IGBTs with short-circuit type 1 behavior, it is necessary to reflect the effectiveness of the model by the difference between measured and simulated waveform trends. Frequency domain analysis methods do not focus on data values, but instead refine waveform difference indicators from the perspective of frequency, which meets the needs of IGBT short-circuit failure model evaluation.

[0085] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A frequency domain analysis method for verifying the short-circuit behavior simulation model of an IGBT type 1, characterized in that, include: S100: Calculate the fundamental frequency for DFT analysis based on the short-circuit duration; S2 00: Voltage Frequency Domain Evaluation under Computational Frequency Domain Analysis P V ; In S200, voltage frequency domain evaluation P V The calculation is as follows: in, n This is the frequency component index; specifically, when n=0, it represents the DC component. N v The number of frequency components after decomposition of a voltage sequence waveform; F v ( n ) is the first voltage waveform sequence after decomposition. n The amplitude of each frequency component; F ’ v (n) represents the nth digit of the simulated voltage waveform sequence after decomposition of the simulation model. n The amplitude of each frequency component, N + In mathematics, this is the symbolic representation of a set of numbers, meaning integers; Voltage frequency domain evaluation P V It is obtained through the following process: According to DFT theory, waveform sequences f ( k ) and DFT transformed F ( n The relationship between them is: in, k For time sequence number, n It is a multiple of the fundamental frequency. N Number of moments; in F ( n Since ) is a complex number and not a real number, F ( n This can be rewritten as: Among them, Re[ F [(n)] is the real part of F(n), Im[ F ( n )]for F ( n The imaginary part of ) | F ( n | represents the amplitude. For phase; Combining the above formula, the waveform sequence can be rewritten as: Measured voltage waveform sequence f V ( n ) and simulated voltage waveform sequence f’ V ( n )have: The voltage frequency domain evaluation can be derived from the above formula. P V It can be calculated as follows: in, n This is the frequency component index; specifically, when n=0, it represents the DC component. N v The number of frequency components after decomposition of a voltage sequence waveform; F v ( n ) is the first voltage waveform sequence after decomposition. n The amplitude of each frequency component; F ’ v (n) represents the nth digit of the simulated voltage waveform sequence after decomposition of the simulation model. n The amplitude of each frequency component, N + In mathematics, this is the symbolic representation of a set of numbers, meaning integers; S300: Current Frequency Domain Evaluation under Computational Frequency Domain Analysis P I ; S400: Evaluation based on voltage frequency domain P V Current frequency domain evaluation P I Calculate the overall frequency domain evaluation index. P , as a measure of the effectiveness of different short-circuit type 1 failure simulation models.

2. The frequency domain analysis method for verifying the simulation model of IGBT short-circuit type 1 behavior according to claim 1, characterized in that, In step S100, the fundamental frequency f = 1 / T, where T is the duration of the short-circuit behavior waveform of type 1, and the highest frequency of the DFT analysis is set to 100 times the fundamental frequency.

3. The frequency domain analysis method for verifying the simulation model of IGBT short-circuit type 1 behavior according to claim 2, characterized in that, In step S100, the time from 0.5us before the start of the type1 short-circuit behavior to 0.5us after the end of the type1 short-circuit behavior.

4. The frequency domain analysis method for verifying the simulation model of IGBT short-circuit type 1 behavior according to claim 1, characterized in that, In step S200 N V Constrained by the following formula: in It is 0.

95.

5. The frequency domain analysis method for verifying the simulation model of IGBT short-circuit type 1 behavior according to claim 1, characterized in that, In step S300, current frequency domain evaluation P I for: in, n This is the frequency component index; specifically, when n=0, it represents the DC component. N I The number of frequency components after decomposition of the current sequence waveform; F I ( n ) is the first digit after decomposition of the current waveform sequence. n The amplitude of each frequency component; F ’ I (n) represents the nth sequence of the simulated current waveform after decomposition of the simulation model. n The amplitude of each frequency component, N + In mathematics, it is a symbolic representation of a set of numbers, meaning an integer.

6. The frequency domain analysis method for verifying the simulation model of IGBT short-circuit type 1 behavior according to claim 1, characterized in that, In step S300, current frequency domain evaluation P I It is obtained through the following process: Measured current waveform sequence f I ( n and simulated current waveform sequence f’ I ( n )have: The current frequency domain evaluation is derived from the above formula. P I for: 。 7. The frequency domain analysis method for verifying the simulation model of IGBT short-circuit type 1 behavior according to claim 1, characterized in that, In step S400, the frequency domain analysis index of the IGBT device short-circuit type 1 behavior failure simulation model is... P for: 。

Images