A new energy power equipment fault prediction method and system

Abstract

The application relates to the technical field of power equipment fault prediction, in particular to a new energy power equipment fault prediction method and system. Real-time pulse width modulation bridge arm real-time duty ratio, real-time direct current side current and collector real-time saturation voltage drop are acquired; a duty ratio deviation is calculated according to the pulse width modulation bridge arm real-time duty ratio; an original junction temperature estimation value is calculated through a temperature sensitive electrical parameter method; a main compensation item is calculated according to the real-time direct current side current and the duty ratio deviation; a temperature correction item is calculated according to the real-time temperature of a radiator and a preset calibration temperature; a compensation junction temperature is calculated according to the original junction temperature estimation value, the main compensation item and the temperature correction item; and a thermal fatigue cumulative damage value is obtained through a rain flow counting method and the thermal fatigue life is predicted. The application solves the problem that the thermal fatigue cumulative damage calculation deviation of a photovoltaic inverter under a maximum power point tracking mode is large, leading to inaccurate fault prediction.

Description

Technical Field

[0001] This invention relates to the field of power equipment fault prediction technology, specifically a method and system for predicting faults in new energy power equipment. Background Technology

[0002] With the large-scale development of the photovoltaic power generation industry, the photovoltaic grid-connected inverter, as the core equipment for photoelectric conversion and grid connection, directly determines the power generation efficiency and power supply stability of the photovoltaic power station. The insulated-gate bipolar transistor (IGBT) is the core power switching device inside the photovoltaic inverter responsible for power conversion. Periodic fluctuations in junction temperature can cause thermal fatigue failure, which is one of the main failure modes restricting the long-term stable operation of the inverter.

[0003] However, under cloudy conditions, photovoltaic inverters continuously operate in maximum power point tracking (MPPT) mode. The device duty cycle is frequently and significantly adjusted due to rapid fluctuations in sunlight intensity, resulting in a fast junction temperature change rate. Existing methods inject a fixed small current pulse during the device turn-off phase, but this requires a dedicated probe period in each switching cycle; frequent interruptions interfere with the normal adjustment of the MPPT. While not injecting an additional probe current avoids disturbances to the pulse width modulation output, the actual sampled current differs from the calibrated current, introducing a systematic bias. This leads to severely inaccurate thermal fatigue life predictions for insulated-gate bipolar transistors (IGBTs), affecting the reliability of equipment failure prediction. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for predicting faults in new energy power equipment, in order to solve the problem that under cloudy conditions, the duty cycle fluctuation of maximum power point tracking modulation causes systematic deviations in junction temperature estimation, resulting in serious inaccuracies in the prediction of thermal fatigue life of insulated gate bipolar transistors.

[0005] To achieve the above objectives, in one aspect, the present invention provides a method for predicting faults in new energy power equipment, the method comprising: Step S1: Real-time acquisition of the pulse width modulation arm duty cycle, real-time DC side current, and real-time collector-emitter saturation voltage drop of the photovoltaic inverter in maximum power point tracking mode.

[0006] Step S2: Calculate the duty cycle deviation based on the real-time duty cycle of the pulse width modulation bridge arm and the preset full-power duty cycle; calculate the original junction temperature estimate based on the collector-emitter saturation voltage drop using the temperature-sensitive electrical parameter method.

[0007] Step S3: Based on the duty cycle deviation and the calibration current of the temperature-sensitive electrical parameter method, determine whether there is a current deviation error in the operating condition; when there is a current deviation error in the operating condition, calculate the main compensation term for compensating the current deviation error in the operating condition based on the real-time DC side current and the duty cycle deviation, and use the main compensation term to compensate the original junction temperature estimate to obtain the compensated junction temperature; when there is no current deviation error in the operating condition, set the main compensation term to zero, and obtain the compensated junction temperature based on the zeroed main compensation term and the original junction temperature estimate.

[0008] Step S4: Based on the time series of the compensated junction temperature, calculate the cumulative thermal fatigue damage value using the rainflow counting method; based on the cumulative thermal fatigue damage value and the growth rate within a preset time window, calculate the predicted value of the remaining thermal fatigue life; when the predicted value of the remaining thermal fatigue life is lower than a preset warning time threshold, output the fault prediction result.

[0009] Furthermore, the method for determining whether there is an operating current deviation error based on the duty cycle deviation and the calibration current of the temperature-sensitive electrical parameter method includes: The constant calibration current used in the calibration of temperature-sensitive electrical parameters and the real-time DC side current at the current moment Based on duty cycle deviation and the real-time DC side current at the current moment The equivalent sampling current estimate was calculated. ,in The preset scaling factor is used to estimate the equivalent sampling current. With constant calibration current Compare and calculate the relative deviation ;when and If the operating current deviation error is found, it is determined that there is one; otherwise, it is not. To preset the duty cycle deviation threshold, This is the preset relative current deviation threshold.

[0010] Furthermore, the method for calculating the main compensation term for the current deviation error under compensation conditions based on the real-time DC-side current and duty cycle deviation includes: Based on real-time DC side current and duty cycle deviation The main compensation term was calculated. ,in For the current moment, This is the preset main compensation coefficient.

[0011] Furthermore, the method for obtaining the preset main compensation coefficient includes: Based on the output characteristic curves and collector-emitter saturation voltage drop temperature characteristic curves in the datasheets of the corresponding models of insulated-gate bipolar transistors, the differential resistance is extracted. Temperature coefficient of emitter saturation voltage drop The photovoltaic inverter was placed with multiple pulse width modulation arms in real time with constant DC current. The deviation between the equivalent sampling current and the calibration current at the actual sampling time was measured, and the sampling timing correction factor was obtained by linear regression fitting. The preset main compensation coefficient is calculated. .

[0012] Furthermore, the method for compensating the original junction temperature estimate with the main compensation term to obtain the compensated junction temperature includes: Real-time temperature of the radiator Based on the real-time temperature of the radiator With preset calibration temperature By setting a temperature correction coefficient Calculate the temperature correction term Based on the original junction temperature estimate Main compensation item Temperature correction item With preset static bias The calculated compensated junction temperature .

[0013] Furthermore, the online adaptive update method for the preset main compensation coefficient and the preset temperature correction coefficient includes: Within the update cycle that satisfies the preset quasi-static thermal equilibrium conditions, the reference junction temperature is calculated based on the real-time radiator temperature, real-time DC-side current, and the preset thermal network model. According to the reference junction temperature Compensation junction temperature with the current cycle The residual for the estimated junction temperature was calculated. Estimate residuals based on junction temperature The preset main compensation coefficients are calculated using the recursive least squares method. and preset temperature correction factor Perform parameter recursive updates; the recursive least squares method is as follows: ; ; .

[0014] in In order to be in The parameter estimation vector at time step [time]. In order to be in The regression vector at time step, It is the gain vector; Forgetting factor; In order to be in The covariance matrix at time step 1; the updated preset principal compensation coefficients and preset temperature correction coefficients are used to calculate the principal compensation term and temperature correction term at the next time step.

[0015] When there is no deviation error in the operating current, the current parameter estimation vector remains unchanged, and the covariance matrix remains unchanged. Unchanged; when the preset quasi-static thermal equilibrium condition is not met, the current parameter estimation vector remains unchanged, and the recursive update for this cycle is not performed; when At that time, among them The preset margin coefficient indicates a strong disturbance condition, and the current parameter estimation vector is maintained. The result remains unchanged, and the covariance matrix is ​​reset to a preset initial value. No recursive update is performed for this period.

[0016] Furthermore, the method for calculating the cumulative thermal fatigue damage value using the rainflow counting method based on the time series of the compensated junction temperature includes: Based on the time series of compensated junction temperatures, a rainflow counting algorithm is used to perform cyclic counting and extract the junction temperature swing for each thermal cycle. and average junction temperature ; average junction temperature After conversion to thermodynamic temperature, based on the junction temperature swing... and average junction temperature The characteristic lifetime corresponding to each thermal cycle is calculated. ,in For material constants, The lifetime index is the junction temperature swing amplitude. Boltzmann's constant, To activate the energy, the cumulative thermal fatigue damage value was calculated. ;in As of The set of thermal cycles at time t, where For the first The count value of each thermal cycle.

[0017] Based on the same inventive concept, this invention also provides a fault prediction system for new energy power equipment, the system comprising: The real-time data acquisition module is used to acquire the real-time duty cycle of the pulse width modulation bridge arm, the real-time DC side current, and the real-time saturation voltage drop of the collector and emitter of the photovoltaic inverter in maximum power point tracking mode.

[0018] The duty cycle deviation and junction temperature estimation module is used to calculate the duty cycle deviation based on the real-time duty cycle of the pulse width modulation bridge arm and the preset full-power duty cycle; and to calculate the original junction temperature estimation value with measurement error based on the collector-emitter saturation voltage drop using the temperature-sensitive electrical parameter method.

[0019] The operating current deviation error compensation module is used to determine whether there is an operating current deviation error based on the duty cycle deviation and the calibration current of the temperature-sensitive electrical parameter method. When there is an operating current deviation error, the module calculates the main compensation term for compensating the operating current deviation error based on the real-time DC side current and duty cycle deviation, and uses the main compensation term to compensate the original junction temperature estimate to obtain the compensated junction temperature. When there is no operating current deviation error, the main compensation term is set to zero, and the compensated junction temperature is obtained based on the zeroed main compensation term and the original junction temperature estimate.

[0020] The thermal fatigue life determination module is used to calculate the cumulative thermal fatigue damage value using the rainflow counting method based on the time series of the compensated junction temperature; calculate the predicted value of the remaining thermal fatigue life based on the cumulative thermal fatigue damage value and the growth rate within a preset time window; and output the fault prediction result when the predicted value of the remaining thermal fatigue life is lower than a preset warning time threshold.

[0021] Compared with the prior art, the beneficial effects of the present invention are: 1. By estimating the deviation of the equivalent sampling current from the calibration current based on the duty cycle deviation, and constructing a main compensation term accordingly, combined with heat sink temperature correction, the systematic error of the temperature-sensitive electrical parameter sampling system caused by duty cycle fluctuations under cloudy conditions can be effectively eliminated without interrupting the pulse width modulation output, thereby improving the estimation accuracy of the junction temperature of the insulated gate bipolar transistor.

[0022] 2. The main compensation coefficient and temperature correction coefficient are updated online adaptively by recursive least squares method. Combined with quasi-static thermal balance window judgment and covariance matrix reset mechanism under strong disturbance conditions, the compensation parameters can be continuously self-corrected as the device ages and the operating conditions change, avoiding reliance on manual offline recalibration and improving the long-term stability of compensation accuracy. Attached Figure Description

[0023] Figure 1 This is a flowchart of a fault prediction method for new energy power equipment according to the present invention; Figure 2 This is a block diagram of a fault prediction system for new energy power equipment according to the present invention. Detailed Implementation

[0024] 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.

[0025] Before providing examples, it is necessary to describe the application scenarios of the present invention. The present invention is applicable to the online estimation of junction temperature and prediction of thermal fatigue life of insulated gate bipolar transistors when photovoltaic grid-connected inverters are operating in maximum power point tracking mode under cloudy conditions.

[0026] Example 1: As Figure 1 As shown in the figure, this embodiment provides a method for predicting faults in new energy power equipment, the method including: Real-time acquisition of the pulse width modulation arm duty cycle, real-time DC side current, and real-time collector-emitter saturation voltage drop of the photovoltaic inverter in maximum power point tracking mode.

[0027] For example, when the photovoltaic grid-connected inverter is operating in maximum power point tracking mode, the real-time duty cycle of the pulse width modulation (PWM) arm, the real-time DC-side current, and the real-time collector-emitter saturation voltage drop are collected at a period of 1ms. The real-time duty cycle of the PWM arm is obtained by directly reading the currently active PWM comparator register inside the digital signal processor. For example, under the conditions of a switching frequency of 10kHz and a timer period register value of 5000, if the comparator register value is read as 3100, the real-time duty cycle is 0.62. The real-time DC-side current is measured by a closed-loop Hall current sensor (range 600A, accuracy ±0.1%) installed on the DC bus. The sensor outputs a 4-20mA signal, which is converted into a 0-5V effective sampling voltage by a conditioning circuit and then sent to a 12-bit analog-to-digital converter. When the ADC reading is 2948, the corresponding current value is 432A. The real-time saturation voltage drop of the collector and emitter is directly sampled by a temperature-sensitive electrical parameter sampling circuit on the gate driver board within the IGBT conduction range, at a fixed delay before the next turn-off. The device conduction current corresponding to the sampling time is estimated by equivalent estimation of the DC side current and duty cycle deviation. The 0-5V voltage output by the sampling circuit is isolated and amplified before being sent to a 12-bit analog-to-digital converter. When the ADC reading is 1334, the corresponding voltage value is 1.629V.

[0028] The duty cycle deviation is calculated based on the real-time duty cycle of the pulse width modulation bridge arm and the preset full-power duty cycle; the original junction temperature is estimated by calculating the collector-emitter saturation voltage drop using the temperature-sensitive electrical parameter method.

[0029] For example, the preset full-power duty cycle is calibrated to 0.82 after stable operation for 5 minutes under full-power conditions in sunny weather. This preset full-power duty cycle is the reference duty cycle of the same inverter under preset DC bus voltage range and grid connection voltage conditions. When the DC bus voltage range changes, the preset full-power duty cycle is taken from the pre-calibrated reference duty cycle table. The real-time duty cycle of the pulse width modulation bridge arm at the current moment... The duty cycle deviation is calculated to be 0.62. Real-time saturation voltage drop at the emitter The value is 1.629V. The Temperature Sensitive Electrical Parameter Method (TSEP method) is used at a constant calibration current. A calibration relationship between collector-emitter saturation voltage drop and junction temperature is pre-established, with the reference junction temperature in the calibration relationship. At 25°C, the reference collector-emitter saturation voltage drop The temperature coefficient of collector-emitter saturation voltage drop is 1.783V. The calculated original junction temperature is -1.8 mV / ℃. .

[0030] Based on the duty cycle deviation and the calibration current of the temperature-sensitive electrical parameter method, determine whether there is an operating current deviation error.

[0031] The method for determining whether there is an operating current deviation error based on the duty cycle deviation and the calibration current of the temperature-sensitive electrical parameter method includes: The constant calibration current used in the calibration of temperature-sensitive electrical parameters and the real-time DC side current at the current moment Based on duty cycle deviation and the real-time DC side current at the current moment The equivalent sampling current estimate was calculated. ,in The preset scaling factor is used to estimate the equivalent sampling current. With constant calibration current Compare and calculate the relative deviation ;when and If the operating current deviation error is found, it is determined that there is one; otherwise, it is not. To preset the duty cycle deviation threshold, This is the preset relative current deviation threshold.

[0032] When there is a deviation error in the operating current, the main compensation term for compensating the deviation error in the operating current is calculated based on the real-time DC side current and duty cycle deviation.

[0033] The method for calculating the main compensation term for compensating the current deviation error under the compensation condition based on the real-time DC side current and duty cycle deviation includes: Based on real-time DC side current and duty cycle deviation The main compensation term was calculated. ,in For the current moment, This is the preset main compensation coefficient.

[0034] The method for obtaining the preset main compensation coefficient includes: Based on the output characteristic curves and collector-emitter saturation voltage drop temperature characteristic curves in the datasheets of the corresponding models of insulated-gate bipolar transistors, the differential resistance is extracted. Temperature coefficient of emitter saturation voltage drop The photovoltaic inverter was placed with multiple pulse width modulation arms in real time with constant DC current. The deviation between the equivalent sampling current and the calibration current at the actual sampling time was measured, and the sampling timing correction factor was obtained by linear regression fitting. The preset main compensation coefficient is calculated. .

[0035] The original junction temperature estimate is compensated by the main compensation term to obtain the compensated junction temperature.

[0036] The method for compensating the original junction temperature estimate with the main compensation term to obtain the compensated junction temperature includes: Real-time temperature of the radiator Based on the real-time temperature of the radiator With preset calibration temperature By setting a temperature correction coefficient Calculate the temperature correction term Based on the original junction temperature estimate Main compensation item Temperature correction item With preset static bias The calculated compensated junction temperature .

[0037] The online adaptive update methods for the preset main compensation coefficient and the preset temperature correction coefficient include: Within the update cycle that satisfies the preset quasi-static thermal equilibrium conditions, the reference junction temperature is calculated based on the real-time radiator temperature, real-time DC-side current, and the preset thermal network model. According to the reference junction temperature Compensation junction temperature with the current cycle The residual for the estimated junction temperature was calculated. Estimate residuals based on junction temperature The preset main compensation coefficients are calculated using the recursive least squares method. and preset temperature correction factor Perform parameter recursive updates; the recursive least squares method is as follows: ; ; .

[0038] in In order to be in The parameter estimation vector at time step [time]. In order to be in The regression vector at time step, It is the gain vector; Forgetting factor; In order to be in The covariance matrix at time step 1; the updated preset principal compensation coefficients and preset temperature correction coefficients are used to calculate the principal compensation term and temperature correction term at the next time step.

[0039] When there is no deviation error in the operating current, the current parameter estimation vector remains unchanged, and the covariance matrix remains unchanged. Unchanged; when the preset quasi-static thermal equilibrium condition is not met, the current parameter estimation vector remains unchanged, and the recursive update for this cycle is not performed; when At that time, among them The preset margin coefficient indicates a strong disturbance condition, and the current parameter estimation vector is maintained. The result remains unchanged, and the covariance matrix is ​​reset to a preset initial value. No recursive update is performed for this period.

[0040] For example, the duty cycle deviation at the current moment absolute value Constant calibration current This refers to the equivalent load current level corresponding to the collector-emitter saturation voltage drop sampling in the temperature-sensitive electrical parameter method calibration experiment. Unlike the traditional method of injecting probe current during TSEP sampling, this invention... Measurements are taken directly when the power device is in normal conduction mode, without injecting additional independent probe current. In this embodiment, the value is preset during the factory calibration stage of the photovoltaic inverter, and is set to 40%~60% of the rated DC side current of the device, in order to balance the signal-to-noise ratio of the on-state voltage drop and the self-heating effect of the device. Obtain the real-time DC-side current at the current moment. During the factory calibration phase of the photovoltaic inverter, the DC-side current is kept constant, and the equivalent sampling current at the sampling time of temperature-sensitive electrical parameters is obtained under different pulse width modulation (PWM) arms with different real-time duty cycles. The equivalent sampling current obtained under each PWM arm's real-time duty cycle is compared with the constant calibration current. Using the offset of the PWM arm's real-time duty cycle relative to the full-power point as a reference, a linear fit is performed on the obtained data points. The slope of the fitted straight line is the preset proportional coefficient. The estimated equivalent sampling current is not required to be exactly the same as the instantaneous collector current of a single insulated-gate bipolar transistor, but is used to characterize the equivalent current deviation of the collector-emitter saturation voltage drop relative to the calibration current condition at the same sampling time.

[0041] In this embodiment, the preset scaling factor is 1.0. Substituting the numerical values, the estimated equivalent sampling current is obtained. The estimated equivalent sampling current is compared with the constant calibration current to calculate the relative deviation. During the factory calibration of the photovoltaic inverter, duty cycle deviation data under stable sunny conditions are continuously collected. The normal distribution characteristics are statistically analyzed, and the upper limit of the 99.7% confidence interval is taken as the preset duty cycle deviation threshold to eliminate misjudgments caused by normal control fluctuations. The preset relative current deviation threshold is determined based on the accuracy level of the DC-side current sensor, taking 1.5 to 2 times the sensor's nominal accuracy to eliminate interference from the sensor's own measurement noise. In this embodiment, the preset duty cycle deviation threshold is determined using the above method. Preset current relative deviation threshold .because and It was determined that there was a deviation error in the operating current. The sign of the symbol causes the equivalent sampled current estimate to automatically shift with changes in the duty cycle, thus affecting the relative accuracy of the TSEP sampled current to the calibration current. Both positive and negative directional deviations are adaptively compensated.

[0042] Based on the output characteristic curves and collector-emitter saturation voltage drop temperature characteristic curves in the datasheet of the corresponding Insulated Gate Bipolar Transistor (IGBT), the differential resistance is extracted. Temperature coefficient of emitter saturation voltage drop The photovoltaic inverter was placed with multiple pulse width modulation arms in real time with constant DC current. The deviation between the equivalent sampling current and the calibration current at the actual sampling time was measured, and the sampling timing correction factor was obtained by linear regression fitting. Substituting into the calculation yields When there is a deviation error in the operating current, the main compensation term is calculated by substituting the data. In this example, the deviation between the equivalent sampling current of 345.6A and the rated current of 200A, amplified by the differential resistor and temperature coefficient, yields a theoretical value of 91.73°C for the main compensation term. To prevent excessive correction of the compensation term under extreme duty cycle fluctuations, which could cause the estimated junction temperature to exceed the physically reasonable range of the device, an upper limit constraint is imposed on the main compensation term based on the maximum allowable junction temperature range specified in the IGBT datasheet. In this embodiment, we take The above-mentioned limiting process is used to constrain the value of the main compensation term in the calculation of the junction temperature. When the main compensation term reaches the limiting boundary, the recursive least squares parameter update is not performed in this cycle to avoid the parameter recursion deviation introduced by the limiting nonlinearity.

[0043] The real-time temperature of the heatsink is measured using an NTC thermistor embedded in the surface of the heatsink beneath the IGBT module, along with a voltage divider circuit. After analog-to-digital conversion, the temperature is calculated using the Steinhart-Hart formula. A reading of 2075 corresponds to a temperature of 48.5℃. (Preset calibration temperature) During the TSEP calibration experiment, the temperature of the heatsink reaching thermal steady state was measured and recorded using a temperature sensor, and determined to be 45℃. Under the condition that the DC-side current and duty cycle were constant, the heatsink temperature was varied, and the original junction temperature was recorded to estimate the residual. The slope of the residual with respect to the heatsink temperature difference was obtained through linear regression fitting. The offline initial value is And it is updated online during operation using recursive least squares method. Based on the real-time temperature of the radiator. Preset calibration temperature and preset temperature correction factor The calculated temperature correction term is as follows: .

[0044] To avoid relying on the actual physical junction temperature, which cannot be obtained online in real time, this embodiment limits the triggering time for parameter recursion updates to within a quasi-static thermal equilibrium window. Within N consecutive control cycles, if the real-time duty cycle fluctuation of the pulse-width modulation bridge arm is less than a preset steady-state threshold, and the rate of change of the DC-side current is less than a preset rate of change, the system determines that the preset quasi-static thermal equilibrium condition is met. Within the update cycle that meets the preset quasi-static thermal equilibrium condition, the current real-time radiator temperature and real-time DC-side current are substituted into the Foster fourth-order thermal network model pre-installed in the controller to calculate the reference junction temperature. The thermal resistance and thermal capacity parameters of the Foster fourth-order thermal network model are obtained by fitting the transient thermal impedance curves from the datasheet of the corresponding model of the insulated-gate bipolar transistor (IGBT), or by identifying the heat sink temperature and junction temperature response data from the power step experiment. The preset thermal network model pre-writes the current-loss mapping relationship of the corresponding model of IGBT. This current-loss mapping relationship is calibrated based on the device's on-state voltage drop characteristics, switching frequency, and drive parameters, and is used to convert the real-time DC-side current into the equivalent heat generation power required for junction temperature calculation.

[0045] During periods when the preset quasi-static thermal equilibrium condition is not met, the system freezes the parameter updates of the recursive least squares method, ensuring that the previous quasi-static thermal equilibrium window has been updated. , The original junction temperature remains unchanged, and the compensated junction temperature is calculated based on the current estimated junction temperature, the main compensation term, and the temperature correction term. This periodic reference junction temperature... The value of In the verification phase following TSEP calibration, the inverter is operated under a preset full-power duty cycle and a preset calibration current. The actual junction temperature on the IGBT chip surface is measured using an infrared thermal imager or fiber optic temperature sensor. Simultaneously, the original junction temperature estimate given by the TSEP method is recorded. The difference between the actual junction temperature and the original junction temperature estimate, plus the temperature correction term, is used as the preset static bias. This preset static bias, after calibration, is fixedly written into the controller as a constant term in the compensation formula. In this embodiment, the preset static bias... Substituting the data, the compensated junction temperature is calculated. .

[0046] Taking another update cycle that does not trigger the main compensation term limit and meets the preset quasi-static thermal equilibrium condition as an example, the recursive least squares parameter update process is explained. Regression vector The initial parameter estimate at the current time is The initial value of the covariance matrix is Forgetting factor .according to ; ; Calculate the gain vector Substituting the data, the residual for estimating the junction temperature is obtained. Substituting the initial parameter estimates, gain vector, and junction temperature estimation residuals into the recursive formula, we obtain... After the update , This is used to calculate the main compensation term and temperature correction term for the next moment, enabling adaptive parameter correction; when there is no operating current deviation error, the main compensation term is directly set... And based on the original junction temperature estimate Temperature correction items, preset static bias Calculate the compensation junction temperature using the main compensation term after setting it to zero. Parameter estimation vector Keep the value from the previous time step unchanged.

[0047] Statistical analysis of duty cycle fluctuation data of photovoltaic inverters under typical cloudy conditions shows that when the absolute value of the duty cycle deviation exceeds five times the preset duty cycle deviation threshold, it is determined to be a strong disturbance condition, requiring the covariance matrix to be reset to maintain algorithm stability. In this embodiment, a preset margin coefficient is used. The value is 5, when When the covariance matrix is ​​reset, Reset to preset initial value To prevent matrix degradation, when both the strong disturbance condition and the absence of operating current deviation error or the failure to meet the preset quasi-static thermal balance condition are met simultaneously, the parameter preservation and covariance matrix reset rules under the strong disturbance condition shall prevail.

[0048] Based on the time series of the compensated junction temperature, the cumulative thermal fatigue damage value is calculated using the rainflow counting method.

[0049] The method for calculating the cumulative thermal fatigue damage value using the rainflow counting method based on the time series of the compensated junction temperature includes: Based on the time series of compensated junction temperatures, a rainflow counting algorithm is used to perform cyclic counting and extract the junction temperature swing for each thermal cycle. and average junction temperature ; average junction temperature After conversion to thermodynamic temperature, based on the junction temperature swing... and average junction temperature The characteristic lifetime corresponding to each thermal cycle is calculated. ,in For material constants, The lifetime index is the junction temperature swing amplitude. Boltzmann's constant, To activate the energy, the cumulative thermal fatigue damage value was calculated. ;in As of The collection of thermal cycles at any given time. For the first The count value of each thermal cycle.

[0050] Based on the cumulative thermal fatigue damage value and the growth rate within a preset time window, the predicted value of the remaining thermal fatigue life is calculated; when the predicted value of the remaining thermal fatigue life is lower than the preset warning time threshold, the fault prediction result is output.

[0051] For example, the cumulative thermal fatigue damage value is calculated using the rainflow counting method based on the time series of the compensated junction temperature. The following uses the compensated junction temperature obtained during another continuous 10-minute runtime. Taking a time series as an example, the sampling frequency is 10Hz, with a total of 6000 data points. A rainflow counting algorithm conforming to ASTM E1049 is used to perform cyclic counting on the compensated junction temperature time series, extracting the junction temperature swing for each complete thermal cycle. and average junction temperature During a 10-minute cloudy event, the main thermal cycles extracted from rainflow counts included... , The cycle, , The cycle, , A total of 23 effective thermal cycles were performed. Based on the Coffin-Manson-Arrhenius lifetime model, the material constants of the insulated-gate bipolar transistor used in this embodiment were pre-obtained by fitting power cycling experimental data. Life expectancy index Activation energy Boltzmann constant The average junction temperature of each thermal cycle. After converting the temperature from Celsius to thermodynamic temperature, the corresponding characteristic lifetimes are calculated by substituting them into the characteristic lifetime model, and thermal fatigue damage is accumulated according to the cycle count values ​​obtained from rainflow counting. For example, , thermal cycle corresponding ; , thermal cycle corresponding ; , thermal cycle corresponding Substituting the aforementioned thermal cycle parameters into the characteristic life model, the damage values ​​for all 23 effective thermal cycles are summed to obtain the thermal fatigue damage increment for this 10-minute statistical window. .

[0052] According to Miner's linear cumulative damage criterion, the device fatigue life is exhausted when the sum of damage fractions reaches 1. Therefore, a preset lifetime threshold is set. The preset time window length is determined based on the statistical fluctuation characteristics of the damage rate under cloudy conditions. Using 5 periods strikes a balance between smoothing short-term fluctuations and tracking trend changes, covering the complete duration of typical cloudy events. This ensures that the window includes at least one complete shading-recovery cycle, allowing the fitted slope to reflect the actual operating trend. The preset time window is determined using the five most recent 10-minute statistical periods. The preset warning time threshold is determined based on the photovoltaic power plant's operation and maintenance response cycle, covering the time required for spare parts procurement (approximately 7 days), downtime window coordination (approximately 7 days), and replacement construction (approximately 3 days), with a safety margin of approximately 13 days; therefore, it is set to 30 days. In this embodiment, the inverter's daily effective operating time is calculated as 8 hours (MPPT continuously operates from sunrise to sunset), used to convert the damage rate from seconds to calendar days.

[0053] Increment of thermal fatigue damage in this statistical window This is added to the total historical damage. The equipment has previously accumulated approximately 6.7 years of operation; the total historical damage is updated after adding this window's increment. The historical total damage records for the most recent 5 consecutive 10-minute statistical periods are retrieved, in chronological order as follows: , , , , The time interval between adjacent cycles is 600s.

[0054] A linear fit was performed on the above five data points with time as the independent variable and historical total damage as the dependent variable. Let... , and order Corresponding in sequence , , , , The rate of damage growth Calculations yielded .

[0055] Predicted Remaining Thermal Fatigue Life Substitute , and The predicted value of residual thermal fatigue life was calculated. ; Converted to calendar days based on 8 hours of effective daily operation. If the remaining thermal fatigue life of the insulated gate bipolar transistor is less than 30 days below the preset warning time threshold, the warning condition will be triggered, and the fault prediction result will be output: The current predicted value of the remaining thermal fatigue life of the insulated gate bipolar transistor is 24.2 days. It is recommended that the device be repaired or replaced 24 days before the current time.

[0056] Example 2: Based on the same inventive concept, such as Figure 2 As shown, this embodiment also provides a fault prediction system for new energy power equipment, the system comprising: The real-time data acquisition module is used to acquire the real-time duty cycle of the pulse width modulation bridge arm, the real-time DC side current, and the real-time saturation voltage drop of the collector and emitter of the photovoltaic inverter in maximum power point tracking mode.

[0057] The duty cycle deviation and junction temperature estimation module is used to calculate the duty cycle deviation based on the real-time duty cycle of the pulse width modulation bridge arm and the preset full-power duty cycle; and to calculate the original junction temperature estimation value with measurement error based on the collector-emitter saturation voltage drop using the temperature-sensitive electrical parameter method.

[0058] The operating current deviation error compensation module is used to determine whether there is an operating current deviation error based on the duty cycle deviation and the calibration current of the temperature-sensitive electrical parameter method. When there is an operating current deviation error, the module calculates the main compensation term for compensating the operating current deviation error based on the real-time DC side current and duty cycle deviation, and uses the main compensation term to compensate the original junction temperature estimate to obtain the compensated junction temperature. When there is no operating current deviation error, the main compensation term is set to zero, and the compensated junction temperature is obtained based on the zeroed main compensation term and the original junction temperature estimate.

[0059] The thermal fatigue life determination module is used to calculate the cumulative thermal fatigue damage value using the rainflow counting method based on the time series of the compensated junction temperature; calculate the predicted value of the remaining thermal fatigue life based on the cumulative thermal fatigue damage value and the growth rate within a preset time window; and output the fault prediction result when the predicted value of the remaining thermal fatigue life is lower than a preset warning time threshold.

[0060] It should be noted that the specific methods by which each module performs operations in the system described in the above embodiments have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0061] Finally, it should be noted that although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for predicting faults in new energy power equipment, characterized in that, The method includes: Real-time acquisition of the pulse width modulation arm duty cycle, real-time DC side current, and real-time collector-emitter saturation voltage drop of the photovoltaic inverter in maximum power point tracking mode; The duty cycle deviation is calculated based on the real-time duty cycle of the pulse width modulation bridge arm and the preset full-power duty cycle; the original junction temperature is estimated by calculating the collector-emitter saturation voltage drop using the temperature-sensitive electrical parameter method. Based on the duty cycle deviation and the calibration current of the temperature-sensitive electrical parameter method, determine whether there is an operating current deviation error; when there is an operating current deviation error, calculate the main compensation term to compensate for the operating current deviation error based on the real-time DC side current and duty cycle deviation, and use the main compensation term to compensate for the original junction temperature estimate to obtain the compensated junction temperature; when there is no operating current deviation error, set the main compensation term to zero, and obtain the compensated junction temperature based on the zeroed main compensation term and the original junction temperature estimate. Based on the time series of the compensated junction temperature, the cumulative thermal fatigue damage value is calculated using the rainflow counting method; based on the cumulative thermal fatigue damage value and its growth rate within a preset time window, the predicted value of the remaining thermal fatigue life is calculated; when the predicted value of the remaining thermal fatigue life is lower than a preset warning time threshold, the fault prediction result is output.

2. The method for predicting faults in new energy power equipment according to claim 1, characterized in that, The method for determining whether there is an operating current deviation error based on the duty cycle deviation and the calibration current of the temperature-sensitive electrical parameter method includes: The constant calibration current used in the calibration of temperature-sensitive electrical parameters and the real-time DC side current at the current moment Based on duty cycle deviation and the real-time DC side current at the current moment The equivalent sampling current estimate was calculated. ,in The preset scaling factor is used to estimate the equivalent sampling current. With constant calibration current Compare and calculate the relative deviation ;when and If the operating current deviation error is found, it is determined that there is one; otherwise, it is not. To preset the duty cycle deviation threshold, This is the preset relative current deviation threshold.

3. The method for predicting faults in new energy power equipment according to claim 2, characterized in that, The method for calculating the main compensation term for compensating the current deviation error under the compensation condition based on the real-time DC side current and duty cycle deviation includes: Based on real-time DC side current and duty cycle deviation The main compensation term was calculated. ,in For the current moment, This is the preset main compensation coefficient.

4. The method for predicting faults in new energy power equipment according to claim 3, characterized in that, The method for obtaining the preset main compensation coefficient includes: Based on the output characteristic curves and collector-emitter saturation voltage drop temperature characteristic curves in the datasheets of the corresponding models of insulated-gate bipolar transistors, the differential resistance is extracted. Temperature coefficient of emitter saturation voltage drop The photovoltaic inverter was placed with multiple pulse width modulation arms in real time with different duty cycles while keeping the DC-side current constant. The deviation between the equivalent sampling current and the calibration current at the actual sampling time was measured, and the sampling timing correction factor was obtained by linear regression fitting. The preset main compensation coefficient is calculated. .

5. The method for predicting faults in new energy power equipment according to claim 3, characterized in that, The method for compensating the original junction temperature estimate with the main compensation term to obtain the compensated junction temperature includes: Real-time temperature of the radiator Based on the real-time temperature of the radiator With preset calibration temperature By setting a temperature correction coefficient Calculate the temperature correction term Based on the original junction temperature estimate Main compensation item Temperature correction item With preset static bias The calculated compensation junction temperature .

6. The method for predicting faults in new energy power equipment according to claim 5, characterized in that, The online adaptive update methods for the preset main compensation coefficient and the preset temperature correction coefficient include: Within the update cycle that satisfies the preset quasi-static thermal equilibrium conditions, the reference junction temperature is calculated based on the real-time radiator temperature, real-time DC-side current, and the preset thermal network model. According to the reference junction temperature Compensation junction temperature with the current cycle The residual for the estimated junction temperature was calculated. Estimate residuals based on junction temperature The preset main compensation coefficients are calculated using the recursive least squares method. and preset temperature correction factor Perform parameter recursive updates; the recursive least squares method is as follows: ； ； ； in In order to be in The parameter estimation vector at time step [time]. In order to be in The regression vector at time step, It is the gain vector; Forgetting factor; In order to be in The covariance matrix at time step 1; the updated preset principal compensation coefficients and preset temperature correction coefficients are used to calculate the principal compensation term and temperature correction term at the next time step; When there is no deviation error in the operating current, the current parameter estimation vector remains unchanged, and the covariance matrix remains unchanged. Unchanged; when the preset quasi-static thermal equilibrium condition is not met, the current parameter estimation vector remains unchanged, and the recursive update for this cycle is not performed; when At that time, among them The preset margin coefficient indicates a strong disturbance condition, and the current parameter estimation vector is maintained. The result remains unchanged, and the covariance matrix is ​​reset to a preset initial value. No recursive update is performed for this period.

7. The method for predicting faults in new energy power equipment according to claim 1, characterized in that, The method for calculating the cumulative thermal fatigue damage value using the rainflow counting method based on the time series of the compensated junction temperature includes: Based on the time series of compensated junction temperatures, a rainflow counting algorithm is used to perform cyclic counting and extract the junction temperature swing for each thermal cycle. and average junction temperature ; average junction temperature After conversion to thermodynamic temperature, based on the junction temperature swing... and average junction temperature The characteristic lifetime corresponding to each thermal cycle is calculated. ,in For material constants, The lifetime index is the junction temperature swing amplitude. Boltzmann's constant, To activate the energy, the cumulative thermal fatigue damage value was calculated. ;in As of The set of thermal cycles at time t, where For the first The count value of each thermal cycle.

8. A fault prediction system for new energy power equipment, characterized in that, The system includes: The real-time data acquisition module is used to acquire the real-time duty cycle of the pulse width modulation bridge arm, the real-time DC side current, and the real-time saturation voltage drop of the collector and emitter of the photovoltaic inverter in the maximum power point tracking mode. The duty cycle deviation and junction temperature estimation module is used to calculate the duty cycle deviation based on the real-time duty cycle of the pulse width modulation bridge arm and the preset full-power duty cycle; and to calculate the original junction temperature estimation value with measurement error based on the collector-emitter saturation voltage drop using the temperature-sensitive electrical parameter method. The operating current deviation error compensation module is used to determine whether there is an operating current deviation error based on the duty cycle deviation and the calibration current of the temperature-sensitive electrical parameter method. When there is an operating current deviation error, the module calculates the main compensation term for compensating the operating current deviation error based on the real-time DC side current and the duty cycle deviation, and uses the main compensation term to compensate the original junction temperature estimate to obtain the compensated junction temperature. When there is no operating current deviation error, the main compensation term is set to zero, and the compensated junction temperature is obtained based on the zeroed main compensation term and the original junction temperature estimate. The thermal fatigue life determination module is used to calculate the cumulative thermal fatigue damage value using the rainflow counting method based on the time series of the compensated junction temperature; calculate the predicted value of the remaining thermal fatigue life based on the cumulative thermal fatigue damage value and the growth rate within a preset time window; and output the fault prediction result when the predicted value of the remaining thermal fatigue life is lower than a preset warning time threshold.

Images