A method for co-predicting the lifetime of IGBTs and capacitors in a bidirectional coupling in an energy storage converter

By constructing an iterative correction algorithm based on the bidirectional asymmetric time-varying aging effect transfer function and hysteresis convergence criterion, bidirectional coupled lifetime co-prediction of IGBT and capacitor is realized, solving the coupling effect problem between devices in energy storage converter and improving the accuracy and reliability of lifetime prediction.

CN123115654APending Publication Date: 2026-11-17EAST GRP CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611261377.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-19
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In existing technologies, the lifetime prediction of IGBTs and DC bus capacitors in energy storage converters lacks consideration of inter-device coordination and ignores the bidirectional coupling effect, resulting in large prediction errors and difficulty in convergence.

Method used

By collecting the operating parameters of IGBTs and capacitors in real time, a bidirectional asymmetric time-varying aging influence transfer function is constructed. A bidirectional coupled iterative correction algorithm with hysteresis convergence criterion is adopted to realize the bidirectional coupled lifetime co-prediction of IGBTs and capacitors.

Benefits of technology

It significantly improves the accuracy and reliability of life prediction for core components of energy storage converters, providing key decision-making basis for preventive maintenance and economical operation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123115654A_ABST
    Figure CN123115654A_ABST
Patent Text Reader

Abstract

This invention relates to the field of power electronic equipment life prediction technology, and discloses a bidirectional coupled life prediction method for IGBTs and capacitors in energy storage converters. By collecting IGBT and capacitor operating parameters in real time and establishing independent life consumption benchmarks for each, a bidirectional asymmetric time-varying aging influence transfer function is innovatively constructed, achieving bidirectional quantitative coupling of the aging process of IGBTs and capacitors for the first time. Based on this, a bidirectional coupled iterative correction algorithm with hysteresis convergence criteria is used to effectively overcome the numerical oscillation and convergence problems caused by bidirectional positive feedback loops, enabling the model to maintain stable solutions in the later stages of capacitor aging. Finally, the system-level life is determined by minimizing the device-level remaining life after collaborative correction, significantly improving the accuracy and reliability of life prediction for core components of energy storage converters, and providing key decision-making basis for preventive maintenance and economical operation of energy storage systems.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of power electronic equipment life prediction technology, and in particular to a method for coordinating the life prediction of IGBTs and capacitors in a bidirectional coupling energy storage converter. Background Technology

[0002] The power conversion system (PCS) is the core energy conversion unit of a battery energy storage system, and its reliability is directly related to the safety and economy of the energy storage system.

[0003] In energy storage converters, IGBT modules and DC bus capacitors are the two core components with the highest failure rates and shortest lifespans. During long-term operation, they do not age in isolation but exhibit a significant bidirectional coupling effect. On one hand, capacitor aging leads to an increase in equivalent series resistance (ESR) and a decrease in capacitance, increasing DC bus voltage ripple and directly exacerbating IGBT switching losses and junction temperature fluctuations, thus accelerating IGBT aging. On the other hand, IGBT aging degrades switching characteristics, generating more high-frequency harmonic currents, increasing the capacitor ripple current burden, and further accelerating capacitor heating and aging.

[0004] Currently, the technologies for predicting the lifespan of core components in energy storage converters have the following main shortcomings: First, the prediction targets are singular, lacking consideration of inter-device synergy. Existing technologies only calculate lifetime consumption for the IGBT single device using junction temperature information and the Coffin-Manson model, or only for the DC bus capacitor single device using ESR estimation and electro-thermal coupling for lifetime prediction. Both of these approaches evaluate individual components in isolation, completely neglecting the aging interactions between the IGBT and the capacitor.

[0005] Second, even those modelers that do exist only employ one-way constraints, neglecting the reverse physical path. A few schemes construct constrained prediction models by collecting DC capacitor and IGBT data, incorporating IGBT operating characteristics into capacitor degradation calculations. However, this is merely a one-way mapping from IGBT to capacitor, ignoring the reverse excitation effect of capacitor aging on the IGBT. When the capacitor enters the mid-to-late stages of aging, with capacitance decay exceeding 15%, the reverse effect becomes increasingly significant; ignoring this path will lead to IGBT lifetime prediction errors exceeding 30%.

[0006] Third, the bidirectional coupling model lacks effective convergence solutions. Once the bidirectional positive feedback coupling between the IGBT and the capacitor is considered, the iterative solution for the remaining lifetime constitutes a positive feedback loop. The predicted values ​​of each device influence each other and change alternately, making conventional successive iterations prone to numerical oscillations and failure to converge. This technical problem simply does not exist in existing unidirectional or isolated models, and current technologies do not provide any solutions, making it difficult for the bidirectional coupling model to be stably applied in practical engineering.

[0007] Therefore, improvements to existing technologies are necessary. Summary of the Invention

[0008] This invention provides a method for coordinating the lifetime prediction of IGBTs and capacitors in a bidirectional coupling in an energy storage converter, in order to solve the problems existing in the prior art.

[0009] To achieve the above objectives, the present invention provides the following technical solution: A method for coordinating lifetime prediction of IGBT and capacitor bidirectional coupling in an energy storage converter, the method comprising: S1. Real-time acquisition of operating parameters of IGBT modules and DC bus capacitors in the energy storage converter; S2. Based on the collected data, calculate the independent lifetime consumption rate of the IGBT module and the independent lifetime consumption rate of the DC bus capacitor, respectively. S3. Construct a bidirectional asymmetric time-varying aging effect transfer function, wherein the transfer function includes a first transfer function for quantifying the effect of IGBT aging on capacitor aging acceleration, and a second transfer function for quantifying the effect of capacitor aging on IGBT aging acceleration. S4. Based on the independent lifetime consumption rate and the initial remaining lifetime, perform bidirectional coupled iterative correction with hysteresis convergence criterion. In each iteration, correct the lifetime of the DC bus capacitor according to the first transfer function and correct the lifetime of the IGBT module according to the second transfer function. Convergence is determined by the criterion that the relative change amount of multiple consecutive iterations is less than a preset threshold. Output the IGBT remaining lifetime and capacitor remaining lifetime after collaborative correction. S5. Based on the co-corrected remaining IGBT lifetime and the remaining capacitor lifetime, determine the system-level remaining lifetime of the energy storage converter by taking the minimum of the two.

[0010] Furthermore, in the bidirectional coupling lifetime co-prediction method of the energy storage converter between IGBT and capacitor, in step S1, the operating parameters of the IGBT module include the collector current I0. C , saturation conduction voltage drop V CE(sat) Switching frequency f sw and shell temperature T c ; The operating parameters of the DC bus capacitor include the terminal voltage V. c Ripple voltage ΔV rip Ripple current I rip and surface temperature T surf .

[0011] Furthermore, in the bidirectional coupling lifetime co-prediction method for IGBT and capacitor in the energy storage converter, in step S2, the independent lifetime consumption rate of the IGBT module is calculated through the following sub-steps: S211, based on the collected collector current I C , saturation conduction voltage drop V CE(sat) Given the duty cycle D, calculate the conduction loss P. con =V CE(sat) ×I C ×D; S212. Based on the collected switching frequency f sw and pre-stored activation energy loss E on and the energy loss during shutdown E off Calculate the switching loss P sw =(E on +E off )×f sw ; S213. Based on the Foster thermal network model, according to the collected shell temperature T c The sum of the conduction loss and the switching loss, and the pre-stored junction-to-case thermal resistance R th(j-c) Calculate the junction temperature T of the IGBT j =T c +(P con +P sw )×R th(j-c) The junction temperature curve as a function of time was obtained. S214. Extract the junction temperature fluctuation amplitude ΔT from the junction temperature curve. j and average junction temperature T jm ; S215. Using the improved Coffin-Manson-Arrhenius lifetime model, based on the junction temperature fluctuation amplitude ΔT j The average junction temperature T jm Pre-stored material constants A, fatigue index α, activation energy E a Given the Boltzmann constant k, calculate the failure cycle number N under this cycle. f =A×(ΔT j )-α×exp(E a / (k×T jm )); S216. The junction temperature curve is statistically analyzed using the rainflow counting method to obtain the number of cycles under different combinations of junction temperature fluctuation amplitude and average junction temperature. S217. Using Miner's linear cumulative damage rule, the ratio of the number of cycles under each combination to the corresponding number of failure cycles is accumulated to calculate the independent lifetime consumption rate λ of the IGBT module. IGBT =dD IGBT / dt.

[0012] Furthermore, in the bidirectional coupling lifetime co-prediction method for IGBTs and capacitors in the energy storage converter, in step S2, the independent lifetime consumption rate of the DC bus capacitor is calculated through the following sub-steps: S221. Based on the collected ripple voltage ΔV rip and ripple current I rip Calculate the equivalent series resistance ESR = ΔV rip / I rip And calculate the current capacitance value C; S222, according to the ripple current I rip Calculate the internal power dissipation P of the capacitor based on the equivalent series resistance ESR. cap =I rip 2 ×ESR; S223, Based on the collected surface temperature T surf The internal power consumption P of the capacitor cap and the pre-stored thermal resistance R from the core to the surface th(core-surf) Calculate the temperature T of the capacitor core. core =T surf +P cap ×R th(core-surf) ; S224. Based on the electrolytic capacitor lifespan model, according to the capacitor core temperature T core And voltage stress, calculate the independent lifetime consumption rate λ of the DC bus capacitor. cap =dD cap / dt.

[0013] Furthermore, in the bidirectional coupling lifetime co-prediction method for IGBT and capacitor in the energy storage converter, step S3 includes: S31. Define a first transfer function k1(t). This first transfer function is used to quantify the accelerating effect of IGBT aging on capacitor aging. The first transfer function is the junction temperature fluctuation amplitude ΔT of the IGBT. j With IGBT Health Index HI IGBT The function is denoted as k1=g1(ΔT) j HI IGBT ); S32. Define a second transfer function k2(t). This second transfer function is used to quantify the accelerating effect of capacitor aging on IGBT aging. The second transfer function is the capacitor voltage ripple ratio ε = ΔV. rip / V c With Capacitor Health Index HI cap The function is denoted as k2=g2(ε,HI). cap ); S33. Identify the specific functional forms of the first transfer function and the second transfer function through offline synchronous accelerated aging tests.

[0014] Furthermore, in the method for co-predicting the lifetime of IGBTs and capacitors in the energy storage converter, step S33 includes: S331. Build an accelerated aging platform, place IGBT modules and DC bus capacitors of the same batch in the same thermal environment, and apply rated electrical stress for synchronous aging. S332. Pause the test at preset time intervals and measure the junction temperature fluctuation ΔT of the IGBT at room temperature. j The voltage ripple ratio ε of the capacitor, as well as the change in IGBT lifespan rate and the change in capacitor lifespan rate. S333, with the junction temperature fluctuation ΔT j and the IGBT health index HI IGBT Using the capacitor lifetime consumption rate increment as input and the least squares support vector machine as output, a regression is performed to obtain the first transfer function k1=g1(ΔT). j HI IGBT ); S334. Using the voltage ripple ratio ε and the capacitor health index HIcap as inputs, and the IGBT lifetime consumption rate increment as output, a least squares support vector machine is used for regression to obtain the second transfer function k2=g2(ε,HIcap). cap ).

[0015] Furthermore, in the bidirectional coupling lifetime co-prediction method of the energy storage converter between IGBT and capacitor, in step S332, the preset time interval is 200 hours.

[0016] Furthermore, in the bidirectional coupling lifetime co-prediction method of the energy storage converter between IGBTs and capacitors, in S333 and S334, the goodness of fit R of the least squares support vector machine regression is... 2 All exceeded 0.96.

[0017] Furthermore, in the method for co-predicting the lifetime of IGBTs and capacitors in the energy storage converter, step S4 includes: S41. Initialization: Set the iteration number k=0, and the initial remaining lifespan of the IGBT RUL. IGBT (0) and the initial remaining life of the capacitor RUL cap (0) IGBT initial aging rate λ IGBT (0) and the initial aging rate λ of the capacitor cap (0) Set the convergence counter cnt=0 and the iteration step size Δt; S42. Perform iterative correction: Update the remaining IGBT lifetime and capacitor lifetime according to the following formula: RUL IGBT (k+1) =RUL IGBT (k) ×[1-k2 (k) λ cap (k) Δt]; RUL cap (k+1) =RUL cap (k) ×[1-k1 (k) λ IGBT (k) Δt]; Among them, k1 (k) Let k2 be the first transfer function in the k-th iteration. (k) This is the second transfer function at the k-th iteration; S43. State and Transfer Function Update: Calculate the new IGBT health index HI based on the updated remaining IGBT lifetime and capacitor lifetime. IGBT and Capacitor Health Index HI cap This allows us to obtain the corresponding junction temperature fluctuation amplitude ΔT. j Substituting the voltage ripple ratio ε into the S3 equation, we update the first transfer function k1. (k+1) Second transfer function k2 (k+1) And based on the model in S2, recalculate the current independent lifetime consumption rate λ of the IGBT module. IGBT (k+1) and the current independent lifetime consumption rate λ of the DC bus capacitor cap (k+1) ; S44. Convergence Criterion Calculation: Calculate the maximum relative change in this iteration: δ (k) =max(|RUL IGBT (k+1) -RUL IGBT(k) | / RUL IGBT (k) ,|RUL cap (k+1) -RUL cap (k) | / RUL cap (k) ); S45. Hysteresis Convergence Criterion: If δ (k) If the value is less than the preset threshold, the convergence counter cnt is incremented by 1; otherwise, the convergence counter cnt is cleared to zero. If the convergence counter cnt reaches the preset number of consecutive iterations, the iteration is considered to have reached stable convergence, the iteration is stopped, and the current RUL is output. IGBT and RUL cap As the IGBT remaining lifetime and capacitor remaining lifetime after co-correction; Otherwise, let k = k + 1, and return to S42 to continue the iteration.

[0018] Furthermore, in the method for co-predicting the lifetime of IGBTs and capacitors in the energy storage converter, step S5 includes: S51, the remaining IGBT lifetime RUL obtained after co-correction. IGBT and DC bus capacitor remaining life RUL cap The smaller value in the range is determined as the system-level remaining lifetime (RUL) of the energy storage converter. sys RUL sys =min(RUL IGBT RUL cap ); S52, based on the system-level remaining lifetime RUL sys With design life RUL design The ratio of the two values ​​is used to calculate the system health index HI. sys =RUL sys / RUL_ design ×100%, and output the current health status of the system as a percentage; S53, RUL based on the last N iterations in the bidirectional coupling iteration process IGBT and RUL cap The statistical distribution is used to calculate the confidence interval of the system-level remaining lifetime: the mean of the results of the last N iterations is used as the point estimate, and the standard deviation σ of the results of the last N iterations is multiplied by the preset confidence factor z as the interval half-width, to give the confidence interval of the system-level remaining lifetime at the confidence level.

[0019] Compared with the prior art, the present invention has the following beneficial effects: This invention provides a bidirectional coupled lifetime co-prediction method for IGBTs and capacitors in energy storage converters. By real-time acquisition of IGBT and capacitor operating parameters and establishing independent lifetime consumption benchmarks for each, it innovatively constructs a bidirectional asymmetric time-varying aging influence transfer function, achieving for the first time bidirectional quantitative coupling of the IGBT and capacitor aging process. Based on this, a bidirectional coupled iterative correction algorithm with hysteresis convergence criterion effectively overcomes the numerical oscillation and convergence problems caused by the bidirectional positive feedback loop, enabling the model to maintain stable solutions even in the later stages of capacitor aging. Finally, the system-level lifetime is determined by minimizing the remaining lifetime at the device level after co-correction, significantly improving the accuracy and reliability of lifetime prediction for core components of energy storage converters, and providing key decision-making basis for preventive maintenance and economical operation of energy storage systems. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart illustrating a method for co-predicting the lifetime of IGBTs and capacitors in a bidirectional coupling in an energy storage converter, as provided in an embodiment of the present invention. Figure 2 This is a further detailed flowchart of S5 provided in the embodiment of the present invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and 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.

[0023] Please refer to Figure 1 This is a flowchart illustrating a method for co-predicting the lifetime of IGBTs and capacitors in a bidirectional coupling process in an energy storage converter, as provided in Embodiment 1 of the present invention. The method specifically includes the following steps: S1. Real-time acquisition of operating parameters of IGBT modules and DC bus capacitors in the energy storage converter.

[0024] It should be noted that in this step, key operating parameters of the IGBT module and DC bus capacitor are synchronously and in real time collected by a sensor array deployed on the power module of the energy storage converter.

[0025] Preferably, in step S1, the operating parameters of the IGBT module include the collector current I0. C , saturation conduction voltage drop V CE(sat) Switching frequency f sw and shell temperature T c The collector current I C With saturation conduction voltage drop V CE(sat) For accurate calculation of subsequent conduction losses; the switching frequency f sw Used to combine pre-stored activation energy loss E on and the energy loss during shutdown E off Calculate switching losses; the case temperature T c As a boundary condition in the Foster thermal network model, it is used for the indirect estimation of junction temperature.

[0026] The operating parameters of the DC bus capacitor include the terminal voltage V. c Ripple voltage ΔV rip Ripple current I rip and surface temperature T surf The ripple voltage ΔV rip With ripple current I rip Used to calculate the equivalent series resistance (ESR) of a capacitor, and thus assess the capacitor's aging condition; the terminal voltage V c Used to calculate voltage ripple ratio; the surface temperature T surf As boundary conditions in the capacitor thermal model, they are used to calculate the temperature of the capacitor core.

[0027] The real-time acquisition of the above parameters provides a data foundation for subsequent calculation of independent lifespan consumption rates and bidirectional coupling iterative correction. It should be noted that all of these parameters can be obtained online through the sampling circuit of the existing converter control system or by additional temperature sensors, without requiring shutdown and disassembly, thus ensuring the feasibility of the method.

[0028] S2. Based on the collected data, calculate the independent lifetime consumption rate of the IGBT module and the independent lifetime consumption rate of the DC bus capacitor, respectively.

[0029] It should be noted that this step aims to establish independent aging benchmark models for the IGBT module and the DC bus capacitor, respectively, while ignoring the coupling effect between devices, so as to provide an initial lifetime consumption rate for subsequent bidirectional coupling correction.

[0030] Preferably, in S2, the independent lifetime consumption rate of the IGBT module is calculated through the following sub-steps: S211, based on the collected collector current I C , saturation conduction voltage drop V CE(sat) Given the duty cycle D, calculate the conduction loss P.con =V CE(sat) ×I C ×D; This loss is one of the main heat sources for the rise in IGBT junction temperature.

[0031] S212. Based on the collected switching frequency f sw and pre-stored activation energy loss E on and the energy loss during shutdown E off Calculate the switching loss P sw =(E on +E off )×f sw ; The Eo n and E off It can be obtained through pre-calibration of IGBT modules of the same model under standard test conditions using a double-pulse test.

[0032] S213. Based on the Foster thermal network model, according to the collected shell temperature T c The sum of the conduction loss and the switching loss, and the pre-stored junction-to-case thermal resistance R th(j-c) Calculate the junction temperature T of the IGBT j =T c +(P con +P sw )×R th(j-c) The junction temperature curve as a function of time was obtained. This model simplifies the IGBT module into a lumped thermal resistance network. Using the case temperature as a known boundary, it indirectly calculates the junction temperature curve of the chip junction temperature over time by superimposing the product of power loss and thermal resistance.

[0033] S214. Extract the junction temperature fluctuation amplitude ΔT from the junction temperature curve. j and average junction temperature T jm ; The junction temperature fluctuation amplitude ΔT j The average junction temperature T reflects the thermal stress intensity of the IGBT during power cycling and is a key input for fatigue life assessment; jm It reflects the long-term heat accumulation level of the device and affects the thermal activation aging rate of the material.

[0034] S215. Using the improved Coffin-Manson-Arrhenius lifetime model, based on the junction temperature fluctuation amplitude ΔT j The average junction temperature T jm Pre-stored material constants A, fatigue index α, activation energy E a Given the Boltzmann constant k, calculate the failure cycle number N under this cycle. f =A×(ΔTj )-α×exp(E a / (k×T jm )); This model takes into account both thermomechanical fatigue (reflected by the Coffin-Manson term) and thermally activated aging (reflected by the Arrhenius term), and can more accurately reflect the life consumption pattern of IGBTs under actual operating conditions than a single model.

[0035] S216. The junction temperature curve is statistically analyzed using the rainflow counting method to obtain the number of cycles under different combinations of junction temperature fluctuation amplitude and average junction temperature. Rainflow counting can effectively extract the full cycle and half cycle in the junction temperature curve, transforming the complex random junction temperature history into an accumulative fatigue damage unit.

[0036] S217. Using Miner's linear cumulative damage rule, the ratio of the number of cycles under each combination to the corresponding number of failure cycles is accumulated to calculate the independent lifetime consumption rate λ of the IGBT module. IGBT =dD IGBT / dt.

[0037] The Miner's Law assumes that damage at various stress levels can be linearly added, and the device fails when the cumulative damage reaches 1. The independent lifetime consumption rate λIGBT characterizes the rate at which the IGBT module's own lifetime is consumed, without considering the impact of capacitor aging feedback.

[0038] In step S2, the independent lifetime consumption rate of the DC bus capacitor is calculated through the following sub-steps: S221. Based on the collected ripple voltage ΔV rip and ripple current I rip Calculate the equivalent series resistance ESR = ΔV rip / I rip And calculate the current capacitance value C; ESR is a core indicator reflecting the aging degree of electrolytic capacitors, and it shows a significant upward trend as the aging process progresses. Meanwhile, based on the terminal voltage V... c The current capacitance value C is calculated based on the charge and discharge characteristics, serving as another reference for the capacitor's health status.

[0039] S222, according to the ripple current I rip Calculate the internal power dissipation P of the capacitor based on the equivalent series resistance ESR. cap =I rip 2 ×ESR; All of this power consumption is dissipated in the form of Joule heat within the capacitor core, making it the direct heat source for the temperature rise of the capacitor core.

[0040] S223, Based on the collected surface temperature T surf The internal power consumption P of the capacitor cap and the pre-stored thermal resistance R from the core to the surface th(core-surf) Calculate the temperature T of the capacitor core. core =T surf +P cap ×R th(core-surf) ; Since the capacitor core is the area where electrochemical reactions and aging degradation actually occur, the core temperature is a more accurate characterization of the capacitor's aging thermal stress level than the surface temperature.

[0041] S224. Based on the electrolytic capacitor lifespan model, according to the capacitor core temperature T core And voltage stress, calculate the independent lifetime consumption rate λ of the DC bus capacitor. cap =dD cap / dt.

[0042] The electrolytic capacitor lifetime model typically employs an extended form of the Arrhenius equation, using temperature stress and voltage stress as accelerating factors to estimate the rate of capacitor lifespan consumption under independent operating conditions.

[0043] S3. Construct a bidirectional asymmetric time-varying aging effect transfer function, wherein the transfer function includes a first transfer function for quantifying the effect of IGBT aging on capacitor aging acceleration, and a second transfer function for quantifying the effect of capacitor aging on IGBT aging acceleration.

[0044] It should be noted that this step aims to establish a quantitative model of the aging cross-effect between the IGBT module and the DC bus capacitor, which is the core link in achieving "bidirectional coupling" in this invention.

[0045] Preferably, step S3 can be further refined to include the following sub-steps: S31. Define a first transfer function k1(t). This first transfer function is used to quantify the accelerating effect of IGBT aging on capacitor aging. The first transfer function is the junction temperature fluctuation amplitude ΔT of the IGBT. j With IGBT Health Index HI IGBT The function is denoted as k1=g1(ΔT) j HI IGBT ); Among them, the junction temperature fluctuation amplitude ΔT j This reflects the intensity of high-frequency harmonic currents generated during IGBT switching. Injecting these harmonic currents into the DC bus increases the ripple current load on the capacitor. The IGBT health index (HI) reflects this. IGBT This reflects the aging depth of the IGBT. The more severe the aging, the more obvious the degradation of the switching characteristics and the richer the harmonic content.

[0046] S32. Define a second transfer function k2(t). This second transfer function is used to quantify the accelerating effect of capacitor aging on IGBT aging. The second transfer function is the capacitor voltage ripple ratio ε = ΔV. rip / V c With Capacitor Health Index HI cap The function is denoted as k2=g2(ε,HI). cap ); Among them, the voltage ripple ratio ε reflects the degree of fluctuation of the DC bus voltage. After capacitor aging leads to an increase in ESR and a decrease in capacitance, the DC bus voltage ripple increases significantly, directly aggravating the switching losses and junction temperature fluctuations of the IGBT; the capacitor health index HI cap This reflects the depth of capacitor aging.

[0047] It is important to note that the first transfer function k1 and the second transfer function k2 are not equal in form, reflecting the asymmetry of two physical mechanisms: the impact path of IGBT aging on the capacitor is "deterioration of switching characteristics → increase in harmonics → increase in capacitor ripple current," while the impact path of capacitor aging on the IGBT is "increase in ESR / decrease in capacitance → increase in voltage ripple → increase in IGBT switching losses." Their physical nature, intensity, and dynamic characteristics are all different. Furthermore, both functions are updated in real time with changes in the aging state, exhibiting time-varying characteristics.

[0048] S33. Identify the specific functional forms of the first transfer function and the second transfer function through offline synchronous accelerated aging tests.

[0049] S33 includes: S331. Build an accelerated aging platform, place IGBT modules and DC bus capacitors of the same batch in the same thermal environment, and apply rated electrical stress for synchronous aging. By controlling the ambient temperature and electrical stress level, the aging process of the device is accelerated to obtain sufficient aging data samples in a shorter time.

[0050] S332. Pause the test at preset time intervals and measure the junction temperature fluctuation ΔT of the IGBT at room temperature. j The voltage ripple ratio ε of the capacitor, as well as the change in IGBT lifespan rate and the change in capacitor lifespan rate. It should be further noted that the preset time interval is 200 hours. The selection of the preset time interval needs to take into account both the resolution of the aging data and the efficiency of the experiment. The 200-hour interval can ensure the validity of the data while avoiding interference from too frequent start-stop cycles on the aging process.

[0051] S333, with the junction temperature fluctuation ΔT jand the IGBT health index HI IGBT Using the capacitor lifetime consumption rate increment as input and the least squares support vector machine as output, a regression is performed to obtain the first transfer function k1=g1(ΔT). j HI IGBT ); LSSVM solves linear equations instead of quadratic programming problems, achieving high computational efficiency while maintaining fitting accuracy, making it suitable for engineering applications.

[0052] S334. Using the voltage ripple ratio ε and the capacitor health index HIcap as inputs, and the IGBT lifetime consumption rate increment as output, a least squares support vector machine is used for regression to obtain the second transfer function k2=g2(ε,HIcap). cap ).

[0053] The goodness of fit R of the least squares support vector machine regression 2 All values ​​exceeded 0.96, indicating that the established transfer function can accurately describe the nonlinear mapping relationship of the aging cross-effect.

[0054] S4. Based on the independent lifetime consumption rate and the initial remaining lifetime, perform bidirectional coupled iterative correction with hysteresis convergence criterion. In each iteration, correct the lifetime of the DC bus capacitor according to the first transfer function and correct the lifetime of the IGBT module according to the second transfer function. Convergence is determined by the criterion that the relative change in multiple consecutive iterations is less than a preset threshold. Output the IGBT remaining lifetime and capacitor remaining lifetime after collaborative correction.

[0055] It should be noted that this step is the core solution process of this invention, which aims to obtain the collaboratively corrected remaining lifetime after considering the cross-aging effect by iterative calculation based on the independent lifetime consumption rate and the bidirectional transfer function.

[0056] Preferably, step S4 can be further refined to include the following sub-steps: S41. Initialization: Set the iteration number k=0, and the initial remaining lifespan of the IGBT RUL. IGBT (0) and the initial remaining life of the capacitor RUL cap (0) IGBT initial aging rate λ IGBT (0) and the initial aging rate λ of the capacitor cap (0) Set the convergence counter cnt=0 and the iteration step size Δt; It should be further noted that the typical value of the iteration step size Δt can be 1 hour. This value ensures the accuracy of the calculation while taking into account the real-time requirements of online applications.

[0057] S42. Perform iterative correction: Update the remaining IGBT lifetime and capacitor lifetime according to the following formula: RUL IGBT (k+1) =RUL IGBT (k) ×[1-k2 (k) λ cap (k) Δt]; RUL cap (k+1) =RUL cap (k) ×[1-k1 (k) λ IGBT (k) Δt]; Among them, k1 (k) Let k2 be the first transfer function in the k-th iteration. (k) This is the second transfer function at the k-th iteration; The physical meaning of the above formula is that the remaining lifespan of an IGBT depends not only on its own independent aging rate λ. IGBT It is also affected by the capacitor aging rate λ cap The remaining lifespan of the capacitor is accelerated by the second transfer function k2; similarly, the remaining lifespan of the IGBT is also accelerated by the aging rate of the IGBT through the first transfer function k1. The two equations are coupled to form a bidirectional positive feedback loop.

[0058] S43. State and Transfer Function Update: Calculate the new IGBT health index HI based on the updated remaining IGBT lifetime and capacitor lifetime. IGBT and Capacitor Health Index HI cap This allows us to obtain the corresponding junction temperature fluctuation amplitude ΔT. j Substituting the voltage ripple ratio ε into the S3 equation, we update the first transfer function k1. (k+1) Second transfer function k2 (k+1) And based on the model in S2, recalculate the current independent lifetime consumption rate λ of the IGBT module. IGBT (k+1) and the current independent lifetime consumption rate λ of the DC bus capacitor cap (k+1) ; This step ensures that both the transfer function and the independent aging rate are dynamically updated as the iteration process progresses, reflecting the time-varying characteristics of the aging state.

[0059] S44. Convergence Criterion Calculation: Calculate the maximum relative change in this iteration: δ (k)=max(|RUL IGBT (k+1) -RUL IGBT (k) | / RUL IGBT (k) ,|RUL cap (k+1) -RUL cap (k) | / RUL cap (k) ); This metric quantifies the maximum relative change in the remaining lifetime of the two devices during this iteration, serving as a basis for determining the convergence state.

[0060] S45. Hysteresis Convergence Criterion: If δ (k) If the value is less than the preset threshold, the convergence counter cnt is incremented by 1; otherwise, the convergence counter cnt is cleared to zero. If the convergence counter cnt reaches the preset number of consecutive iterations, the iteration is considered to have reached stable convergence, the iteration is stopped, and the current RUL is output. IGBT and RUL cap As the IGBT remaining lifetime and capacitor remaining lifetime after co-correction; Otherwise, let k = k + 1, and return to S42 to continue the iteration.

[0061] It should be further noted that the preset number of times can be, for example, 3 times.

[0062] The core advantage of the hysteresis criterion lies in the fact that during the iterative process of a bidirectional positive feedback coupled system, a calculation may occasionally fall below a threshold due to a small numerical fluctuation, only to rebound above the threshold in the next iteration. Conventional single-threshold criteria are prone to misjudging convergence or causing the iteration to never stop. This invention requires that the maximum relative change in multiple consecutive iterations (e.g., 3 times) be less than a preset threshold before convergence is determined, effectively avoiding the risk of misjudgment in the coupled oscillation region. It ensures that even under strong coupled positive feedback, it can still stably output a unique convergent remaining lifetime value, providing a reliable numerical stability guarantee for online engineering applications.

[0063] S5. Based on the co-corrected remaining IGBT lifetime and the remaining capacitor lifetime, determine the system-level remaining lifetime of the energy storage converter by taking the minimum of the two.

[0064] Please refer to Figure 2 In one embodiment of this example, step S5 can be further refined to include the following sub-steps: S51, the remaining IGBT lifetime RUL obtained after co-correction. IGBT and DC bus capacitor remaining life RUL cap The smaller value in the range is determined as the system-level remaining lifetime (RUL) of the energy storage converter. sysRUL sys =min(RUL IGBT RUL cap ); This value selection principle is based on the "barrel effect" principle, where the system-level lifetime is determined by the weakest link that fails first, which meets the actual needs of energy storage converter reliability engineering.

[0065] S52, based on the system-level remaining lifetime RUL sys With design life RUL design The ratio of the two values ​​is used to calculate the system health index HI. sys =RUL sys / RUL_ design ×100%, and output the current health status of the system as a percentage; This index intuitively reflects the aging degree of the energy storage converter relative to its design life, making it easier for operation and maintenance personnel to quickly grasp the equipment status.

[0066] S53, RUL based on the last N iterations in the bidirectional coupling iteration process IGBT and RUL cap The statistical distribution is used to calculate the confidence interval of the system-level remaining lifetime: the mean of the results of the last N iterations is used as the point estimate, and the standard deviation σ of the results of the last N iterations is multiplied by the preset confidence factor z as the interval half-width, to give the confidence interval of the system-level remaining lifetime at the confidence level.

[0067] Here, z corresponds to the standard normal distribution quantile at the required confidence level (e.g., z = 1.96 at a 95% confidence level). The output of the confidence interval provides a quantified uncertainty boundary for preventive maintenance decisions of the energy storage system, avoiding the decision risks that may arise from point estimation.

[0068] Although this invention uses terms such as IGBT and capacitor frequently, the possibility of using other terms is not excluded. These terms are used merely for the convenience of describing and explaining the essence of this invention; interpreting them as any additional limitation would contradict the spirit of this invention.

[0069] The present invention provides a method for coordinating lifetime prediction of IGBT and capacitor bidirectional coupling in an energy storage converter, which has the following significant advantages: 1. Achieved for the first time bidirectional coupling lifetime prediction of IGBT and capacitor. Existing technologies either evaluate single devices in isolation or only establish unidirectional constraints. This invention constructs a bidirectional asymmetric time-varying aging effect transfer function, which fully characterizes the reverse physical path of "capacitor aging → increased voltage ripple → increased IGBT loss → increased junction temperature → accelerated IGBT aging", and the forward path of "IGBT aging → increased harmonics → increased capacitor ripple current → accelerated capacitor aging", realizing true bidirectional coupling collaborative prediction.

[0070] 2. Significantly improves prediction accuracy and eliminates serious biases caused by unidirectional models. This invention, through the introduction of time-varying transfer functions—namely, the first transfer function k1(t) (the acceleration factor of IGBT aging on the capacitor) and the second transfer function k2(t) (the acceleration factor of capacitor aging on the IGBT)—accurately quantifies the variation of the cross-aging effect with aging depth, thus significantly improving prediction accuracy.

[0071] 3. Successfully solves the numerical divergence problem caused by bidirectional positive feedback coupling. Addressing the iterative numerical oscillation problem that inevitably arises after introducing bidirectional paths, this invention, for the first time, introduces a hysteresis convergence criterion in the field of device lifetime prediction—requiring that the maximum relative change in multiple consecutive iterations is less than a preset threshold before convergence is determined. This criterion effectively avoids misjudgment of convergence in the fluctuation region, ensuring a stable and uniquely converged remaining lifetime value even under strongly coupled positive feedback conditions, providing a reliable guarantee for online engineering applications.

[0072] 4. Strong industrial feasibility and online deployment capability. This invention transforms the abstract bidirectional coupling relationship into a time-varying transfer function identification framework based on least squares support vector machine (LSSVM), allowing function calibration to be completed through offline synchronous accelerated aging tests. The prediction process relies solely on online measurable parameters such as voltage, current, and temperature, requiring no shutdown or offline testing. It can continuously output lifetime prediction results during the normal operation of the energy storage converter, greatly enhancing the practicality of the solution.

[0073] 5. This invention is fundamentally different from existing "multiphysics coupling" technologies and cannot be compared. Existing multiphysics coupling technologies all occur within a single device, targeting the interaction of multiple physical fields within the same device; the embodiments of this invention address the aging cross-influence transmission and coordinated degradation between two independent and dissimilar devices. The two are completely different in terms of technical dimensions, scope of application, and the problems they solve.

[0074] Finally, it should be noted that although the above embodiments have been described in the description and drawings of this invention, this should not limit the scope of patent protection of this invention. Any technical solutions that are based on the essential concept of this invention, utilize the content described in the description and drawings of this invention to make equivalent structural or procedural substitutions or modifications, as well as the direct or indirect application of the technical solutions of the above embodiments to other related technical fields, are all included within the scope of patent protection of this invention.

Claims

1. A method for co-predicting the lifetime of IGBTs and capacitors in a bidirectional coupling energy storage converter, characterized in that, The method includes: S1. Real-time acquisition of operating parameters of IGBT modules and DC bus capacitors in the energy storage converter; S2. Based on the collected data, calculate the independent lifetime consumption rate of the IGBT module and the independent lifetime consumption rate of the DC bus capacitor, respectively. S3. Construct a bidirectional asymmetric time-varying aging effect transfer function, wherein the transfer function includes a first transfer function for quantifying the effect of IGBT aging on capacitor aging acceleration, and a second transfer function for quantifying the effect of capacitor aging on IGBT aging acceleration. S4. Based on the independent lifetime consumption rate and the initial remaining lifetime, perform bidirectional coupled iterative correction with hysteresis convergence criterion. In each iteration, correct the lifetime of the DC bus capacitor according to the first transfer function and correct the lifetime of the IGBT module according to the second transfer function. Convergence is determined by the criterion that the relative change amount of multiple consecutive iterations is less than a preset threshold. Output the IGBT remaining lifetime and capacitor remaining lifetime after collaborative correction. S5. Based on the co-corrected remaining IGBT lifetime and the remaining capacitor lifetime, determine the system-level remaining lifetime of the energy storage converter by taking the minimum of the two.

2. The method for co-predicting the lifetime of IGBTs and capacitors in a bidirectional coupling in an energy storage converter according to claim 1, characterized in that, In S1, the operating parameters of the IGBT module include the collector current I0. C , saturation conduction voltage drop V CE(sat) Switching frequency f sw and shell temperature T c ; The operating parameters of the DC bus capacitor include the terminal voltage V. c Ripple voltage ΔV rip Ripple current I rip and surface temperature T surf .

3. The method for co-predicting the lifetime of IGBTs and capacitors in a bidirectional coupling in an energy storage converter according to claim 2, characterized in that, In step S2, the independent lifetime consumption rate of the IGBT module is calculated through the following sub-steps: S211, based on the collected collector current I C , saturation conduction voltage drop V CE(sat) Given the duty cycle D, calculate the conduction loss P. con =V CE(sat) ×I C ×D; S212. Based on the collected switching frequency f sw and pre-stored activation energy loss E on and the energy loss during shutdown E off Calculate the switching loss P sw =(E on +E off )×f sw ; S213. Based on the Foster thermal network model, according to the collected shell temperature T c The sum of the conduction loss and the switching loss, and the pre-stored junction-to-case thermal resistance R th(j-c) Calculate the junction temperature T of the IGBT j =T c +(P con +P sw )×R th(j-c) The junction temperature curve as a function of time was obtained. S214. Extract the junction temperature fluctuation amplitude ΔT from the junction temperature curve. j and average junction temperature T jm ; S215. Using the improved Coffin-Manson-Arrhenius lifetime model, based on the junction temperature fluctuation amplitude ΔT j The average junction temperature T jm Pre-stored material constants A, fatigue index α, activation energy E a Given the Boltzmann constant k, calculate the failure cycle number N under this cycle. f =A×(ΔT j )-α×exp(E a / (k×T jm )); S216. The junction temperature curve is statistically analyzed using the rainflow counting method to obtain the number of cycles under different combinations of junction temperature fluctuation amplitude and average junction temperature. S217. Using Miner's linear cumulative damage rule, the ratio of the number of cycles under each combination to the corresponding number of failure cycles is accumulated to calculate the independent lifetime consumption rate λ of the IGBT module. IGBT =dD IGBT / dt.

4. The method for co-predicting the lifetime of IGBTs and capacitors in a bidirectional coupling in an energy storage converter according to claim 3, characterized in that, In step S2, the independent lifetime consumption rate of the DC bus capacitor is calculated through the following sub-steps: S221. Based on the collected ripple voltage ΔV rip and ripple current I rip Calculate the equivalent series resistance ESR = ΔV rip / I rip And calculate the current capacitance value C; S222, according to the ripple current I rip Calculate the internal power dissipation P of the capacitor based on the equivalent series resistance ESR. cap =I rip 2 ×ESR; S223, Based on the collected surface temperature T surf The internal power consumption P of the capacitor cap and the pre-stored thermal resistance R from the core to the surface th(core-surf) Calculate the temperature T of the capacitor core. core =T surf +P cap ×R th(core-surf) ; S224. Based on the electrolytic capacitor lifespan model, according to the capacitor core temperature T core And voltage stress, calculate the independent lifetime consumption rate λ of the DC bus capacitor. cap =dD cap / dt.

5. The method for co-predicting the lifetime of IGBTs and capacitors in a bidirectional coupling in an energy storage converter according to claim 4, characterized in that, S3 includes: S31. Define a first transfer function k1(t). This first transfer function is used to quantify the accelerating effect of IGBT aging on capacitor aging. The first transfer function is the junction temperature fluctuation amplitude ΔT of the IGBT. j With IGBT Health Index HI IGBT The function is denoted as k1=g1(ΔT) j HI IGBT ); S32. Define a second transfer function k2(t). This second transfer function is used to quantify the accelerating effect of capacitor aging on IGBT aging. The second transfer function is the capacitor voltage ripple ratio ε = ΔV. rip / V c With Capacitor Health Index HI cap The function is denoted as k2=g2(ε,HI). cap ); S33. Identify the specific functional forms of the first transfer function and the second transfer function through offline synchronous accelerated aging tests.

6. The method for co-predicting the lifetime of IGBTs and capacitors in a bidirectional coupling in an energy storage converter according to claim 5, characterized in that, S33 includes: S331. Build an accelerated aging platform, place IGBT modules and DC bus capacitors of the same batch in the same thermal environment, and apply rated electrical stress for synchronous aging. S332. Pause the test at preset time intervals and measure the junction temperature fluctuation ΔT of the IGBT at room temperature. j The voltage ripple ratio ε of the capacitor, as well as the change in IGBT lifespan rate and the change in capacitor lifespan rate. S333, with the junction temperature fluctuation ΔT j and the IGBT health index HI IGBT Using the capacitor lifetime consumption rate increment as input and the least squares support vector machine as output, a regression is performed to obtain the first transfer function k1=g1(ΔT). j HI IGBT ); S334. Using the voltage ripple ratio ε and the capacitor health index HIcap as inputs, and the IGBT lifetime consumption rate increment as output, a least squares support vector machine is used for regression to obtain the second transfer function k2=g2(ε,HIcap). cap ).

7. The method for co-predicting the lifetime of IGBTs and capacitors in a bidirectional coupling in an energy storage converter according to claim 6, characterized in that, In S332, the preset time interval is 200 hours.

8. The method for co-predicting the lifetime of IGBTs and capacitors in a bidirectional coupling energy storage converter according to claim 7, characterized in that, In S333 and S334, the goodness of fit R of the least squares support vector machine regression is... 2 All exceeded 0.

96.

9. The method for co-predicting the lifetime of IGBTs and capacitors in a bidirectional coupling in an energy storage converter according to claim 8, characterized in that, S4 includes: S41. Initialization: Set the iteration number k=0, and the initial remaining lifespan of the IGBT RUL. IGBT (0) and the initial remaining life of the capacitor RUL cap (0) IGBT initial aging rate λ IGBT (0) and the initial aging rate λ of the capacitor cap (0) Set the convergence counter cnt=0 and the iteration step size Δt; S42. Perform iterative correction: Update the remaining IGBT lifetime and capacitor lifetime according to the following formula: NUMBER IGBT (k+1) =RUL IGBT (k) ×[1-k2 (k) λ cap (k) Δt]; NUMBER cap (k+1) =RUL cap (k) ×[1-k1 (k) λ IGBT (k) Δt]; Among them, k1 (k) Let k2 be the first transfer function in the k-th iteration. (k) This is the second transfer function at the k-th iteration; S43. State and Transfer Function Update: Calculate the new IGBT health index HI based on the updated remaining IGBT lifetime and capacitor lifetime. IGBT and Capacitor Health Index HI cap This allows us to obtain the corresponding junction temperature fluctuation amplitude ΔT. j Substituting the voltage ripple ratio ε into the S3 equation, we update the first transfer function k1. (k+1) Second transfer function k2 (k+1) And based on the model in S2, the current independent lifetime consumption rate λ of the IGBT module is recalculated. IGBT (k+1) and the current independent lifetime consumption rate λ of the DC bus capacitor cap (k+1) ; S44. Convergence Criterion Calculation: Calculate the maximum relative change in this iteration: δ (k) =max(|RUL IGBT (k+1) -RUL IGBT (k) | / RUL IGBT (k) ,|RUL cap (k+1) -RUL cap (k) | / RUL cap (k) ); S45. Hysteresis Convergence Criterion: If δ (k) If the value is less than the preset threshold, the convergence counter cnt is incremented by 1; otherwise, the convergence counter cnt is cleared to zero. If the convergence counter cnt reaches the preset number of consecutive iterations, the iteration is considered to have reached stable convergence, the iteration is stopped, and the current RUL is output. IGBT and RUL cap As the IGBT remaining lifetime and capacitor remaining lifetime after co-correction; Otherwise, let k = k + 1 and return to S42 to continue the iteration.

10. The method for co-predicting the lifetime of IGBTs and capacitors in a bidirectional coupling energy storage converter according to claim 9, characterized in that, S5 includes: S51, the remaining IGBT lifetime RUL obtained after co-correction. IGBT and DC bus capacitor remaining life RUL cap The smaller value in the range is determined as the system-level remaining lifetime (RUL) of the energy storage converter. sys RUL sys =min(RUL IGBT RUL cap ); S52, based on the system-level remaining lifetime RUL sys With design life RUL design The ratio of the two values ​​is used to calculate the system health index HI. sys =RUL sys / RUL_ design ×100%, and output the current health status of the system as a percentage; S53, RUL based on the last N iterations in the bidirectional coupling iteration process IGBT and RUL cap The statistical distribution is used to calculate the confidence interval of the system-level remaining lifetime: the mean of the results of the last N iterations is used as the point estimate, and the standard deviation σ of the results of the last N iterations is multiplied by the preset confidence factor z as the interval half-width, to give the confidence interval of the system-level remaining lifetime at the confidence level.