A super capacitor life intelligent prediction method and system

Abstract

The application relates to the technical field of life prediction, and discloses a super capacitor life intelligent prediction method and system, which acquires monitoring data of a to-be-measured super capacitor in a preset monitoring window, and obtains a standardized monitoring data set; a current capacity retention rate and a current internal resistance growth rate are calculated based on the standardized monitoring data set, and a health degree benchmark value is obtained; a stress accumulation index is constructed according to a duration of exceeding an upper limit of a rated voltage in an end voltage attenuation sequence, an integral heat value higher than a warning temperature in a shell surface temperature field distribution data and a single charge-discharge depth extreme value; a remaining life prediction value of the super capacitor is obtained by using a decay trend line of the health degree benchmark value changing with an equivalent cycle equivalent, and by adopting a double exponential smoothing algorithm to extrapolate and calculate the decay trend line; and through the health degree benchmark value and the stress accumulation index, combined with a double exponential smoothing time sequence extrapolation and a prediction error online correction mechanism, the comprehensive improvement of life prediction accuracy, reliability and adaptive capacity is realized.

Description

Technical Field

[0001] This invention relates to the field of lifespan prediction technology, and more specifically, to a method and system for intelligent prediction of supercapacitor lifespan. Background Technology

[0002] Supercapacitor life prediction refers to the technology of monitoring and analyzing the electrical parameters, temperature characteristics and cyclic operating conditions of supercapacitors during operation, assessing their current health status and estimating their remaining lifespan. It is a key support means to ensure the safe and reliable operation of energy storage systems and optimize operation and maintenance strategies.

[0003] Existing technologies often employ methods based on single-parameter threshold judgments or simple linear extrapolation to determine the degree of lifespan degradation by monitoring the voltage drop or internal resistance increase. However, the parameter monitoring dimensions in existing technologies are relatively singular, making it difficult to comprehensively reflect the complex degradation mechanism under the coupled effects of multiple stress factors such as voltage over-limit, temperature accumulation, and charge / discharge depth. Furthermore, simple linear models lack the ability to mine the temporal correlation of historical monitoring data, resulting in a significant decrease in prediction accuracy as operating conditions fluctuate. The lack of a reliability assessment mechanism for prediction results affects the accuracy of operation and maintenance decisions, and it is impossible to achieve closed-loop self-optimization based on prediction errors, making it difficult to adapt to the differentiated prediction needs of energy storage systems under various operating modes such as constant power and pulsed power. Summary of the Invention

[0004] To address the aforementioned technical issues, this application aims to provide a method and system for intelligent prediction of supercapacitor lifespan. By constructing a multi-level fusion evaluation system of health baseline value and stress accumulation index, and combining a double-exponential smoothing time series extrapolation and an online correction mechanism for prediction errors, the method achieves a comprehensive improvement in the accuracy, reliability, and adaptability of lifespan prediction.

[0005] To achieve the above objectives, this invention provides a smart prediction method for supercapacitor lifetime, comprising: Acquire monitoring data of the supercapacitor under test within a preset monitoring window. The monitoring data includes terminal voltage decay sequence, ripple current RMS value sequence, shell surface temperature field distribution data, and actual number of charge-discharge cycles. Perform time-series alignment and outlier removal on the monitoring data to obtain a standardized monitoring dataset. The current capacity retention rate and the current internal resistance growth rate of the supercapacitor under test are calculated based on the standardized monitoring dataset, and the weighted sum is used to obtain the health baseline value. The stress accumulation index is constructed based on the duration of the voltage decay sequence exceeding the rated voltage limit, the integral heat value above the warning temperature in the temperature field distribution data of the casing surface, and the extreme value of the single charge and discharge depth. By using historical monitoring data, a decay trend line is constructed to show the change of the health baseline value with the equivalent cyclic equivalence. The decay trend line is extrapolated using a double exponential smoothing algorithm to obtain the predicted remaining lifespan of the supercapacitor.

[0006] Furthermore, the monitoring data is subjected to time-series alignment and outlier removal to obtain a standardized monitoring dataset, including: Determine the time deviation between the sampling timestamp of the terminal voltage decay sequence and the sampling timestamp of the ripple current RMS value sequence, and perform interpolation resampling on sampling points whose time deviation is greater than a preset synchronization threshold. Determine the degree of outlier between the temperature measurement point values ​​and the mean in the temperature field distribution data of the shell surface, and mark the temperature measurement points with an outlier degree greater than the preset jump threshold as outlier points; After removing outliers, the data are uniformly converted into dimensionless proportional values ​​and normalized to the same order of magnitude to obtain a standardized monitoring dataset.

[0007] Furthermore, based on a standardized monitoring dataset, the current capacitance retention rate and the current internal resistance growth rate of the supercapacitor under test are calculated, and a weighted sum is obtained to obtain a health baseline value, including: Determine the ratio between the measured charge of the supercapacitor under full charge and the factory-specified rated charge, and multiply the ratio by a preset capacity weighting coefficient to obtain the current capacity retention rate. Determine the ratio between the current equivalent series resistance value and the factory-nominated reference resistance value, and multiply the reciprocal of the ratio by a preset internal resistance weighting coefficient to obtain the current internal resistance growth rate.

[0008] Furthermore, based on the duration of voltage decay exceeding the rated voltage limit in the terminal voltage decay sequence, the integrated heat value above the warning temperature in the casing surface temperature field distribution data, and the extreme value of a single charge-discharge depth, a stress accumulation index is constructed, including: Determine the cumulative duration for which the voltage value at each sampling point in the terminal voltage decay sequence exceeds the upper limit of the rated voltage, and use a preset voltage stress coefficient to multiply the cumulative duration to obtain the voltage stress component; The integral of the difference between the temperature value at each temperature measurement point and the warning temperature in the temperature field distribution data of the shell surface is determined over time, and the temperature stress component is obtained by multiplying the integral result using a preset temperature stress coefficient. Determine the ratio of the charge change amplitude to the rated charge during a single charge and discharge cycle, and multiply the ratio using a preset depth stress coefficient to obtain the depth stress component; The stress accumulation index is obtained by summing the voltage stress component, temperature stress component, and depth stress component.

[0009] Furthermore, a decay trend line of the health baseline value as a function of the equivalent cyclic equivalence is constructed using historical monitoring data. A double exponential smoothing algorithm is then used to extrapolate this decay trend line to obtain the predicted remaining lifespan of the supercapacitor, including: The equivalent cycle equivalent sequence is obtained by multiplying the actual number of charge-discharge cycles by the normalization coefficient of the depth of charge-discharge per cycle. After sorting the health baseline values ​​according to the equivalent cyclic equivalent sequence, linear interpolation is performed to obtain a decay scatter sequence with equal intervals. The first smoothed sequence is obtained by performing a first exponential smoothing on the decaying scatter sequence, and the second smoothed sequence is obtained by performing a second exponential smoothing on the first smoothed sequence. The trend component is calculated based on the difference between the first smoothed sequence and the second smoothed sequence. The trend component is extrapolated to a preset prediction step size to obtain the remaining life prediction value.

[0010] Furthermore, after constructing a decay trend line of the health baseline value as a function of the equivalent cyclic equivalence using historical monitoring data, and extrapolating the decay trend line using a double exponential smoothing algorithm to obtain the predicted remaining lifespan of the supercapacitor, the process also includes: The prediction confidence factor is calculated based on the data missing rate, the number of abnormal jump points, and the variance of the fitting residual of the decay trend line in the standardized monitoring dataset. A request to review the prediction results is triggered when the confidence factor is lower than the preset confidence threshold.

[0011] Furthermore, based on the data missing rate, the number of anomalous jump points, and the variance of the fitting residuals of the decay trend line in the standardized monitoring dataset, a prediction confidence factor is calculated, including: The missing data rate is obtained by determining the proportion of missing values ​​in the standardized monitoring dataset to the total data volume. The missing data confidence component is obtained by multiplying the missing data rate based on the preset missing penalty coefficient. The number of abnormal jump points is obtained by determining the number of times the difference between adjacent sampling points in the terminal voltage decay sequence exceeds the preset jump limit. The jump confidence component is obtained by multiplying the number of abnormal jump points according to the preset jump penalty coefficient. Determine the variance of the fitting residuals of the decay trend line, and multiply the variance values ​​based on the preset residual penalty coefficient to obtain the residual confidence component; The sum of the components of the missing confidence component, the jump confidence component, and the residual confidence component is determined, and the difference between the baseline confidence constant and the sum of the components is used as the prediction confidence factor.

[0012] To achieve the above objectives, the present invention also provides a supercapacitor lifetime intelligent prediction system, comprising: The data acquisition module is used to acquire monitoring data of the supercapacitor under test within a preset monitoring window. The monitoring data includes terminal voltage decay sequence, ripple current RMS value sequence, shell surface temperature field distribution data, and actual charge and discharge cycles. The monitoring data is processed by time alignment and outlier removal to obtain a standardized monitoring dataset. The benchmark calculation module is used to calculate the current capacity retention rate and the current internal resistance growth rate of the supercapacitor under test based on the standardized monitoring dataset, and then sum them by weight to obtain the health benchmark value. The index calculation module is used to construct the stress accumulation index based on the duration of the voltage decay sequence exceeding the rated voltage limit, the integral heat value above the warning temperature in the temperature field distribution data of the casing surface, and the extreme value of the single charge and discharge depth. The lifespan prediction module is used to construct a decay trend line of the health baseline value as a function of the equivalent cyclic equivalence using historical monitoring data. The decay trend line is extrapolated using a double exponential smoothing algorithm to obtain the predicted remaining lifespan value of the supercapacitor.

[0013] Furthermore, it also includes: The reliable prediction module is used to calculate the prediction reliability factor based on the data missing rate, the number of abnormal jump points, and the variance of the fitting residual of the decay trend line in the standardized monitoring dataset. The review request module is used to trigger a review request for the prediction results when the confidence factor is lower than the preset confidence threshold.

[0014] Furthermore, the reliable prediction module is used for: The missing data rate is obtained by determining the proportion of missing values ​​in the standardized monitoring dataset to the total data volume. The missing data confidence component is obtained by multiplying the missing data rate based on the preset missing penalty coefficient. The number of abnormal jump points is obtained by determining the number of times the difference between adjacent sampling points in the terminal voltage decay sequence exceeds the preset jump limit. The jump confidence component is obtained by multiplying the number of abnormal jump points according to the preset jump penalty coefficient. Determine the variance of the fitting residuals of the decay trend line, and multiply the variance values ​​based on the preset residual penalty coefficient to obtain the residual confidence component; The sum of the components of the missing confidence component, the jump confidence component, and the residual confidence component is determined, and the difference between the baseline confidence constant and the sum of the components is used as the prediction confidence factor.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention discloses a method and system for intelligent prediction of supercapacitor lifetime. The method acquires monitoring data of the supercapacitor under test within a preset monitoring window to obtain a standardized monitoring dataset. Based on the standardized monitoring dataset, the current capacity retention rate and the current internal resistance growth rate are calculated to obtain a health baseline value. A stress accumulation index is constructed based on the duration of voltage decay exceeding the rated voltage limit in the terminal voltage decay sequence, the integral heat value above the warning temperature in the shell surface temperature field distribution data, and the extreme value of a single charge-discharge depth. Using the decay trend line of the health baseline value changing with the equivalent cyclic equivalence, a double exponential smoothing algorithm is used to extrapolate the decay trend line to obtain the predicted remaining lifetime value of the supercapacitor. By combining the health baseline value and the stress accumulation index, along with the double exponential smoothing time-series extrapolation and the online correction mechanism for prediction errors, a comprehensive improvement in lifetime prediction accuracy, reliability, and adaptability is achieved. Attached Figure Description

[0016] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A flowchart illustrating an intelligent prediction method for supercapacitor lifetime in an embodiment of the present invention is shown. Figure 2 A schematic diagram of the structure of a supercapacitor lifetime intelligent prediction system according to an embodiment of the present invention is shown. Detailed Implementation

[0017] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0018] In the description of this application, it should be understood that the terms "center", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0019] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0020] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0021] The following is a description of preferred embodiments of the present invention in conjunction with the accompanying drawings.

[0022] like Figure 1 As shown, an embodiment of the present invention discloses a smart prediction method for supercapacitor lifetime, comprising: S110: Acquire monitoring data of the supercapacitor under test within a preset monitoring window. The monitoring data includes terminal voltage decay sequence, ripple current RMS value sequence, shell surface temperature field distribution data, and actual number of charge-discharge cycles. Perform time-series alignment and outlier removal on the monitoring data to obtain a standardized monitoring dataset. S120: Calculate the current capacity retention rate and the current internal resistance growth rate of the supercapacitor under test based on the standardized monitoring dataset, and obtain the health baseline value by weighted summation; S130: Based on the duration of the voltage drop sequence exceeding the upper limit of the rated voltage, the integral heat value above the warning temperature in the temperature field distribution data of the casing surface, and the extreme value of the depth of charge and discharge in a single charge and discharge cycle, a stress accumulation index is constructed. S140: Construct a decay trend line of the health baseline value as a function of the equivalent cyclic equivalent using historical monitoring data, and extrapolate the decay trend line using a double exponential smoothing algorithm to obtain the predicted remaining lifespan of the supercapacitor.

[0023] In some embodiments of this application, the monitoring data is subjected to time-series alignment and outlier removal to obtain a standardized monitoring dataset, including: Determine the time deviation between the sampling timestamp of the terminal voltage decay sequence and the sampling timestamp of the ripple current RMS value sequence, and perform interpolation resampling on sampling points whose time deviation is greater than a preset synchronization threshold. Determine the degree of outlier between the temperature measurement point values ​​and the mean in the temperature field distribution data of the shell surface, and mark the temperature measurement points with an outlier degree greater than the preset jump threshold as outlier points; After removing outliers, the data are uniformly converted into dimensionless proportional values ​​and normalized to the same order of magnitude to obtain a standardized monitoring dataset.

[0024] In this embodiment, the sampling timestamp is uniformly generated by the BMS system, and the time deviation is obtained by comparing the time tag difference between voltage and current data packets. The preset synchronization threshold is set to 1 second. When the deviation is greater than 1 second, linear interpolation is used for resampling to align the data to the whole minute. The temperature field distribution data includes 6 temperature measurement points. The outlier is calculated by dividing the absolute difference between the temperature of each point and the mean of the 6 points by the mean. For example, if the temperature of a point is 65 degrees Celsius and the mean is 55 degrees Celsius, the outlier is (65-55)÷55≈0.182. The preset jump threshold is set to 0.15. Points exceeding this threshold are marked as outliers. After removal, the mean of the nearest points is used to fill the gap. The dimensionless proportional value conversion is performed according to the characteristics of different parameters: the terminal voltage is calculated as proportional voltage / rated upper limit voltage, the ripple current is calculated as effective value / rated ripple current, the temperature is calculated as (measured value - ambient temperature) / (upper limit temperature - ambient temperature), and the number of cycles is calculated as the current number of cycles / rated cycle life. Normalization to the same order of magnitude is achieved using maximum and minimum value normalization, mapping all data to the [0, 1] interval for easier subsequent fusion calculations. Interpolation and resampling ensure strict synchronization of voltage and current, outlier determination effectively identifies abnormal temperature rises caused by sensor malfunctions or transient thermal shocks, and normalization eliminates differences in different physical dimensions.

[0025] The beneficial effects of the above technical solutions are: time deviation detection and interpolation resampling ensure strict synchronization of multi-source data, outlier degree quantification identifies temperature outliers, elimination and filling maintain data integrity, dimensionless scaling preserves the physical meaning of parameters, and normalization achieves comparability of multiple parameters, laying a high-quality data foundation for subsequent feature extraction.

[0026] In some embodiments of this application, the current capacitance retention rate and the current internal resistance growth rate of the supercapacitor under test are calculated based on a standardized monitoring dataset, and a weighted sum is obtained to obtain a health baseline value, including: Determine the ratio between the measured charge of the supercapacitor under full charge and the factory-specified rated charge, and multiply the ratio by a preset capacity weighting coefficient to obtain the current capacity retention rate. Determine the ratio between the current equivalent series resistance value and the factory-nominated reference resistance value, and multiply the reciprocal of the ratio by a preset internal resistance weighting coefficient to obtain the current internal resistance growth rate.

[0027] In this embodiment, a fully charged state refers to a terminal voltage reaching the rated voltage of 2.7 volts. The measured charge is obtained by the coulomb integral of discharging at a constant current of 100 amperes to 1.35 volts. For example, if the measured charge is 2750 coulombs and the rated charge is 3000 coulombs, the ratio is 2750 ÷ 3000 ≈ 0.917. The preset capacity weighting coefficient is set to 0.6, and the current capacity retention rate is 0.917 × 0.6 = 0.550. The equivalent series resistance is measured using the 10 milliohm AC injection method. The current value is, for example, 1.8 milliohms, and the factory standard value is 1.5 milliohms. The ratio is 1.8 ÷ 1.5 = 1.2, and the reciprocal is 0.833. The preset internal resistance weighting coefficient is 0.4, and the current internal resistance growth rate is 0.833 × 0.4 = 0.333. The health baseline value is 0.550 + 0.333 = 0.883, which is 88.3 out of 100. The sum of the weighting coefficients is 1 to ensure that the health baseline value is within the range of 0-1.

[0028] The beneficial effects of the above technical solution are: full-load measured charge accurately reflects the available capacity; ratio calculation quantifies the degree of capacity decay; the inverse of the equivalent series resistance is processed to convert the increase in internal resistance into a decrease in health; weight allocation reflects the dominance of capacity; weighted integration integrates the two dimensions of comprehensive electrical performance; and the health benchmark value provides a unified quantitative evaluation standard, providing a reliable basis for trend analysis.

[0029] In some embodiments of this application, a stress accumulation index is constructed based on the duration of the voltage drop sequence exceeding the rated voltage upper limit, the integral heat value above the warning temperature in the casing surface temperature field distribution data, and the extreme value of a single charge-discharge depth, including: Determine the cumulative duration for which the voltage value at each sampling point in the terminal voltage decay sequence exceeds the upper limit of the rated voltage, and use a preset voltage stress coefficient to multiply the cumulative duration to obtain the voltage stress component; The integral of the difference between the temperature value at each temperature measurement point and the warning temperature in the temperature field distribution data of the shell surface is determined over time, and the temperature stress component is obtained by multiplying the integral result using a preset temperature stress coefficient. Determine the ratio of the charge change amplitude to the rated charge during a single charge and discharge cycle, and multiply the ratio using a preset depth stress coefficient to obtain the depth stress component; The stress accumulation index is obtained by summing the voltage stress component, temperature stress component, and depth stress component.

[0030] In this embodiment, the rated voltage upper limit is 2.7 volts, and the cumulative duration is recorded by a timer for the number of seconds the voltage exceeds 2.7 volts. The preset voltage stress coefficient is set to 0.001 based on electrochemical breakdown experimental data, and the voltage stress component is 1800 × 0.001 = 1.8. The warning temperature is 55 degrees Celsius, and the integrated heat value is calculated as ∑(T-55) × Δt, where T is the temperature value at the measurement point, and Δt is the warning temperature. For example, if the temperature at a certain point is 60 degrees Celsius for 10 minutes, the temperature difference is 5 degrees Celsius, and the integral is 5 × 600 = 3000 degrees Celsius·second. The preset temperature stress coefficient is set to 0.0005 based on the capacitor life temperature acceleration model, and the temperature stress component is 3000 × 0.0005 = 1.5. The extreme value of a single charge-discharge cycle is determined by monitoring the change in charge during a single cycle. For example, if the charge in a certain cycle is 2400 coulombs and the rated charge is 3000 coulombs, the ratio is 2400 ÷ 3000 = 0.8. The preset depth stress coefficient is set to 2.0 based on the deep cycle accelerated aging test, and the depth stress component is 0.8 × 2.0 = 1.6. The stress accumulation index is 1.8 + 1.5 + 1.6 = 4.9. This index has no unit; the larger the value, the more severe the cumulative damage. The voltage stress coefficient of 0.001 reflects the accelerating effect of overvoltage on the decomposition of the internal dielectric, the temperature stress coefficient of 0.0005 reflects the lifetime-temperature relationship described by the Arrhenius equation, and the depth stress coefficient of 2.0 reflects the mechanical stress damage to the electrode material caused by deep cycling.

[0031] The beneficial effects of the above technical solution are: the overpressure stress is accurately quantified by the over-limit duration, the temperature accumulation effect is comprehensively evaluated by the integrated heat, the depth ratio reflects the cyclic intensity, the preset stress coefficient is scientifically calibrated based on accelerated life test, the three components are added to construct a multi-stress linear accumulation model, and the stress accumulation index intuitively represents the comprehensive damage degree, providing a physical basis for the prediction of attenuation trend.

[0032] In some embodiments of this application, a decay trend line of the health baseline value as a function of the equivalent cyclic equivalence is constructed using historical monitoring data. A double exponential smoothing algorithm is then used to extrapolate the decay trend line to obtain the predicted remaining lifetime value of the supercapacitor, including: The equivalent cycle equivalent sequence is obtained by multiplying the actual number of charge-discharge cycles by the normalization coefficient of the depth of charge-discharge per cycle. After sorting the health baseline values ​​according to the equivalent cyclic equivalent sequence, linear interpolation is performed to obtain a decay scatter sequence with equal intervals. The first smoothed sequence is obtained by performing a first exponential smoothing on the decaying scatter sequence, and the second smoothed sequence is obtained by performing a second exponential smoothing on the first smoothed sequence. The trend component is calculated based on the difference between the first smoothed sequence and the second smoothed sequence. The trend component is extrapolated to a preset prediction step size to obtain the remaining life prediction value.

[0033] In this embodiment, the actual number of charge-discharge cycles is recorded by the BMS cycle counter, for example, currently accumulating to 1200 cycles. The normalization coefficient for single charge-discharge depth is calculated based on the ratio of single charge change to rated charge, for example, an average depth of 0.85, resulting in an equivalent cycle count of 1200 × 0.85 = 1020 cycles. The equivalent cycle count sequence is extracted from historical data, for example, starting from cycle 0 and taking a point every 100 cycles up to cycle 1500. The health baseline value corresponds to the value taken at each equivalent point, with slow early decay corresponding to 98 points, mid-term to 88 points, and rapid decay in the later stages. Linear interpolation uses the nearest neighbor interpolation method to convert non-equal interval data into an equal interval sequence with 100 equivalent intervals, facilitating smoothing calculations. The first exponential smoothing uses a smoothing coefficient α = 0.3, and the smoothing formula is S1(t) = α·Y(t) + (1-α)·S1(t-1), resulting in the first smoothed sequence. The second exponential smoothing uses the same α to smooth the S1 sequence again, resulting in the S2 sequence. The trend component b(t) = [S1(t) - S2(t)]·α / (1-α) reflects the decay slope. Where, S1(t): the value of the first smoothed sequence at the current time t, representing the result after exponential smoothing of the original data, reflecting the overall level component of the data. Y(t): the actual value of the original observed sequence at time t, i.e., the measured data of the health baseline value corresponding to the equivalent cyclic equivalent. S1(t-1): the value of the first smoothed sequence at the previous time t-1, used to recursively calculate the smoothed value at the current time. S2(t): the value of the second smoothed sequence at time t, representing the result after exponential smoothing the first smoothed sequence S1(t) again, further extracting deeper trend information. b(t): the value of the trend component at time t, reflecting the decay slope or changing trend. It is calculated by the difference between the two smoothing results; the larger the difference, the stronger the trend, used for extrapolating and predicting the direction and magnitude of future value changes. The preset prediction step size is set to 200 equivalent cycles, extrapolating to calculate the predicted health value 200 cycles later. To make it easier to understand, here is an example: Assuming the current equivalent cycle count is t = 1000 cycles, the historical data is as follows (α = 0.3): The original observation value Y(t): the baseline health value at t=900, 950, and 1000 times is 0.78, 0.75, and 0.72 respectively (showing a decreasing trend). First smooth sequence S1(t): S1(900) = 0.80 (initialization) S1(950)=0.3×0.75+0.7×0.80=0.785 S1(1000)=0.3×0.72+0.7×0.785=0.7655 The second smoothed sequence S2(t) S2(950)=0.3×0.785+0.7×0.80=0.7955 S2(1000)=0.3×0.7655+0.7×0.7955=0.7865 Trend component b(t) b(1000)=[S1(1000)-S2(1000)]×0.3 / (1-0.3)=(0.7655-0.7865)×0.4286=-0.0090 (The negative value indicates that the decay slope decreases by 0.009 every 100 cycles) Extrapolation prediction: Assuming the health failure threshold is 0.60, and the prediction step size is 200 iterations, after 1000 iterations, the predicted health = S1(1000) + b(1000) × 2 = 0.7655 + (-0.0090 × 2) = 0.7475. Continuing to extrapolate based on the trend, it will take approximately (0.7655 - 0.60) / 0.0090 ≈ 18.4 iterations to reduce the health to 0.60.

[0034] The remaining lifetime is predicted to be 18.4 equivalent cycles.

[0035] The beneficial effects of the above technical solutions are: equivalent cyclic equivalent conversion enables comparability of cycles at different depths; equal-interval interpolation facilitates time series processing; double exponential smoothing effectively tracks nonlinear decay trends; trend component extrapolation quantifies future decay rates; and prediction step size setting provides clear lifetime end criteria, significantly improving long-term prediction accuracy.

[0036] In some embodiments of this application, after constructing a decay trend line of the health baseline value as a function of the equivalent cyclic equivalence using historical monitoring data, and extrapolating the decay trend line using a double exponential smoothing algorithm to obtain the predicted remaining lifetime value of the supercapacitor, the method further includes: The prediction confidence factor is calculated based on the data missing rate, the number of abnormal jump points, and the variance of the fitting residual of the decay trend line in the standardized monitoring dataset. A request to review the prediction results is triggered when the confidence factor is lower than the preset confidence threshold.

[0037] In some embodiments of this application, a prediction confidence factor is calculated based on the data missing rate, the number of anomalous jump points, and the variance of the fitting residuals of the decay trend line in the standardized monitoring dataset, including: The missing data rate is obtained by determining the proportion of missing values ​​in the standardized monitoring dataset to the total data volume. The missing data confidence component is obtained by multiplying the missing data rate based on the preset missing penalty coefficient. The number of abnormal jump points is obtained by determining the number of times the difference between adjacent sampling points in the terminal voltage decay sequence exceeds the preset jump limit. The jump confidence component is obtained by multiplying the number of abnormal jump points according to the preset jump penalty coefficient. Determine the variance of the fitting residuals of the decay trend line, and multiply the variance values ​​based on the preset residual penalty coefficient to obtain the residual confidence component; The sum of the components of the missing confidence component, the jump confidence component, and the residual confidence component is determined, and the difference between the baseline confidence constant and the sum of the components is used as the prediction confidence factor.

[0038] In this embodiment, each monitoring window of the standardized monitoring dataset should have 1440 data points (24 hours × 60 minutes). If there are 72 missing values, the data missing rate is 72 ÷ 1440 = 0.05. The preset missing value penalty coefficient is set to 0.5, and the missing value confidence component is 0.05 × 0.5 = 0.025. The preset jump limit is set to 0.05 volts based on the voltage sampling noise level. For example, the number of times the difference between adjacent points in the terminal voltage sequence exceeds 0.05 volts is 15, and the number of abnormal jump points is 15. The preset jump limit penalty coefficient is set to 0.01, and the jump confidence component is 15 × 0.01 = 0.15. The variance of the residual fitting of the decay trend line is calculated as the squared average of the difference between the predicted health and the actual health, for example, a variance of 0.0036. The preset residual penalty coefficient is set to 10, and the residual confidence component is 0.0036 × 10 = 0.036. The baseline confidence constant is set to 1.0, and the prediction confidence factor is calculated as 1.0 - 0.025 - 0.15 - 0.036 = 0.789. The preset reliability threshold is set to 0.75. When the confidence factor (0.789) > 0.75, the prediction result is considered reliable. If the factor falls below 0.75, a review is triggered, and the system will re-collect data and re-execute the prediction. Each penalty coefficient is determined based on historical prediction error analysis to ensure sensitivity to data quality.

[0039] The beneficial effects of the above technical solution are: data missing rate quantifies data integrity, jump limit identifies sensor faults or interference, fitting residual variance evaluates model matching degree, three components systematically reflect the source of prediction uncertainty, benchmark constant minus penalty to obtain credibility, threshold-triggered verification ensures prediction reliability, and significantly improves the scientific nature of prediction result credibility assessment.

[0040] To further illustrate the technical concept of this invention, the technical solution of this invention will now be described in conjunction with specific application scenarios.

[0041] Correspondingly, such as Figure 2 As shown, this application also provides a supercapacitor lifetime intelligent prediction system, comprising: The data acquisition module is used to acquire monitoring data of the supercapacitor under test within a preset monitoring window. The monitoring data includes terminal voltage decay sequence, ripple current RMS value sequence, shell surface temperature field distribution data, and actual charge and discharge cycles. The monitoring data is processed by time alignment and outlier removal to obtain a standardized monitoring dataset. The benchmark calculation module is used to calculate the current capacity retention rate and the current internal resistance growth rate of the supercapacitor under test based on the standardized monitoring dataset, and then sum them by weight to obtain the health benchmark value. The index calculation module is used to construct the stress accumulation index based on the duration of the voltage decay sequence exceeding the rated voltage limit, the integral heat value above the warning temperature in the temperature field distribution data of the casing surface, and the extreme value of the single charge and discharge depth. The lifespan prediction module is used to construct a decay trend line of the health baseline value as a function of the equivalent cyclic equivalence using historical monitoring data. The decay trend line is extrapolated using a double exponential smoothing algorithm to obtain the predicted remaining lifespan value of the supercapacitor.

[0042] In some embodiments of this application, it also includes: The reliable prediction module is used to calculate the prediction reliability factor based on the data missing rate, the number of abnormal jump points, and the variance of the fitting residual of the decay trend line in the standardized monitoring dataset. The review request module is used to trigger a review request for the prediction results when the confidence factor is lower than the preset confidence threshold.

[0043] In some embodiments of this application, the reliable prediction module is used for: The missing data rate is obtained by determining the proportion of missing values ​​in the standardized monitoring dataset to the total data volume. The missing data confidence component is obtained by multiplying the missing data rate based on the preset missing penalty coefficient. The number of abnormal jump points is obtained by determining the number of times the difference between adjacent sampling points in the terminal voltage decay sequence exceeds the preset jump limit. The jump confidence component is obtained by multiplying the number of abnormal jump points according to the preset jump penalty coefficient. Determine the variance of the fitting residuals of the decay trend line, and multiply the variance values ​​based on the preset residual penalty coefficient to obtain the residual confidence component; The sum of the components of the missing confidence component, the jump confidence component, and the residual confidence component is determined, and the difference between the baseline confidence constant and the sum of the components is used as the prediction confidence factor.

[0044] In the description of the above embodiments, specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0045] Although the invention has been described above with reference to embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of the invention. In particular, as long as there is no structural conflict, the features in the embodiments disclosed in this invention can be combined with each other in any way. The fact that not all of these combinations are described in this specification is merely for the sake of brevity and resource conservation.

[0046] It will be understood by those skilled in the art that the above are merely preferred embodiments of the present invention and are not intended to limit the present invention. 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 intelligent prediction of supercapacitor lifetime, characterized in that, include: Acquire monitoring data of the supercapacitor under test within a preset monitoring window. The monitoring data includes terminal voltage decay sequence, ripple current RMS value sequence, shell surface temperature field distribution data, and actual number of charge-discharge cycles. Perform time-series alignment and outlier removal on the monitoring data to obtain a standardized monitoring dataset. The current capacity retention rate and the current internal resistance growth rate of the supercapacitor under test are calculated based on the standardized monitoring dataset, and the weighted sum is used to obtain the health baseline value. The stress accumulation index is constructed based on the duration of the voltage decay sequence exceeding the rated voltage limit, the integral heat value above the warning temperature in the temperature field distribution data of the casing surface, and the extreme value of the single charge and discharge depth. By using historical monitoring data, a decay trend line is constructed to show the change of the health baseline value with the equivalent cyclic equivalence. The decay trend line is extrapolated using a double exponential smoothing algorithm to obtain the predicted remaining lifespan of the supercapacitor.

2. The intelligent prediction method for supercapacitor lifetime according to claim 1, characterized in that, The monitoring data is subjected to time-series alignment and outlier removal to obtain a standardized monitoring dataset, including: Determine the time deviation between the sampling timestamp of the terminal voltage decay sequence and the sampling timestamp of the ripple current RMS value sequence, and perform interpolation resampling on sampling points whose time deviation is greater than a preset synchronization threshold. Determine the degree of outlier between the temperature measurement point values ​​and the mean in the temperature field distribution data of the shell surface, and mark the temperature measurement points with an outlier degree greater than the preset jump threshold as outlier points; After removing outliers, the data are uniformly converted into dimensionless proportional values ​​and normalized to the same order of magnitude to obtain a standardized monitoring dataset.

3. The intelligent prediction method for supercapacitor lifetime according to claim 1, characterized in that, Based on a standardized monitoring dataset, the current capacitance retention rate and current internal resistance growth rate of the supercapacitor under test are calculated, and a weighted sum is obtained to obtain a health baseline value, including: Determine the ratio between the measured charge of the supercapacitor under full charge and the factory-specified rated charge, and multiply the ratio by a preset capacity weighting coefficient to obtain the current capacity retention rate. Determine the ratio between the current equivalent series resistance value and the factory-nominated reference resistance value, and multiply the reciprocal of the ratio by a preset internal resistance weighting coefficient to obtain the current internal resistance growth rate.

4. The intelligent prediction method for supercapacitor lifetime according to claim 1, characterized in that, Based on the duration of voltage decay exceeding the rated voltage limit in the terminal voltage decay sequence, the integrated heat value above the warning temperature in the casing surface temperature field distribution data, and the extreme value of a single charge-discharge cycle, a stress accumulation index is constructed, including: Determine the cumulative duration for which the voltage value at each sampling point in the terminal voltage decay sequence exceeds the upper limit of the rated voltage, and use a preset voltage stress coefficient to multiply the cumulative duration to obtain the voltage stress component; The integral of the difference between the temperature value at each temperature measurement point and the warning temperature in the temperature field distribution data of the shell surface is determined over time, and the temperature stress component is obtained by multiplying the integral result using a preset temperature stress coefficient. Determine the ratio of the charge change amplitude to the rated charge during a single charge and discharge cycle, and multiply the ratio using a preset depth stress coefficient to obtain the depth stress component; The stress accumulation index is obtained by summing the voltage stress component, temperature stress component, and depth stress component.

5. The intelligent prediction method for supercapacitor lifetime according to claim 1, characterized in that, A decay trend line of the health baseline value as a function of equivalent cyclic equivalence is constructed using historical monitoring data. A double exponential smoothing algorithm is then used to extrapolate this decay trend line to obtain the predicted remaining lifespan of the supercapacitor, including: The equivalent cycle equivalent sequence is obtained by multiplying the actual number of charge-discharge cycles by the normalization coefficient of the depth of charge-discharge per cycle. After sorting the health baseline values ​​according to the equivalent cyclic equivalent sequence, linear interpolation is performed to obtain a decay scatter sequence with equal intervals. The first smoothed sequence is obtained by performing a first exponential smoothing on the decaying scatter sequence, and the second smoothed sequence is obtained by performing a second exponential smoothing on the first smoothed sequence. The trend component is calculated based on the difference between the first smoothed sequence and the second smoothed sequence. The trend component is extrapolated to a preset prediction step size to obtain the remaining life prediction value.

6. The intelligent prediction method for supercapacitor lifetime according to claim 1, characterized in that, After constructing a decay trend line of the health baseline value as a function of equivalent cyclic equivalence using historical monitoring data, and extrapolating the decay trend line using a double exponential smoothing algorithm to obtain the predicted remaining lifespan of the supercapacitor, the process also includes: The prediction confidence factor is calculated based on the data missing rate, the number of abnormal jump points, and the variance of the fitting residual of the decay trend line in the standardized monitoring dataset. A request to review the prediction results is triggered when the confidence factor is lower than the preset confidence threshold.

7. The intelligent prediction method for supercapacitor lifetime according to claim 6, characterized in that, Based on the data missing rate, the number of anomalous abrupt change points, and the variance of the fitting residuals of the decay trend line in the standardized monitoring dataset, a prediction confidence factor is calculated, including: The missing data rate is obtained by determining the proportion of missing values ​​in the standardized monitoring dataset to the total data volume. The missing data confidence component is obtained by multiplying the missing data rate based on the preset missing penalty coefficient. The number of abnormal jump points is obtained by determining the number of times the difference between adjacent sampling points in the terminal voltage decay sequence exceeds the preset jump limit. The jump confidence component is obtained by multiplying the number of abnormal jump points according to the preset jump penalty coefficient. Determine the variance of the fitting residuals of the decay trend line, and multiply the variance values ​​based on the preset residual penalty coefficient to obtain the residual confidence component; The sum of the components of the missing confidence component, the jump confidence component, and the residual confidence component is determined, and the difference between the baseline confidence constant and the sum of the components is used as the prediction confidence factor.

8. A supercapacitor lifetime intelligent prediction system, applied to the supercapacitor lifetime intelligent prediction method as described in any one of claims 1-7, characterized in that, include: The data acquisition module is used to acquire monitoring data of the supercapacitor under test within a preset monitoring window. The monitoring data includes terminal voltage decay sequence, ripple current RMS value sequence, shell surface temperature field distribution data, and actual charge and discharge cycles. The monitoring data is processed by time alignment and outlier removal to obtain a standardized monitoring dataset. The benchmark calculation module is used to calculate the current capacity retention rate and the current internal resistance growth rate of the supercapacitor under test based on the standardized monitoring dataset, and then sum them by weight to obtain the health benchmark value. The index calculation module is used to construct the stress accumulation index based on the duration of the voltage decay sequence exceeding the rated voltage limit, the integral heat value above the warning temperature in the temperature field distribution data of the casing surface, and the extreme value of the single charge and discharge depth. The lifespan prediction module is used to construct a decay trend line of the health baseline value as a function of the equivalent cyclic equivalence using historical monitoring data. The decay trend line is extrapolated using a double exponential smoothing algorithm to obtain the predicted remaining lifespan value of the supercapacitor.

9. The intelligent prediction system for supercapacitor lifespan according to claim 8, characterized in that, Also includes: The reliable prediction module is used to calculate the prediction reliability factor based on the data missing rate, the number of abnormal jump points, and the variance of the fitting residual of the decay trend line in the standardized monitoring dataset. The review request module is used to trigger a review request for the prediction results when the confidence factor is lower than the preset confidence threshold.

10. The intelligent prediction system for supercapacitor lifespan according to claim 9, characterized in that, The reliable prediction module is used for: The missing data rate is obtained by determining the proportion of missing values ​​in the standardized monitoring dataset to the total data volume. The missing data confidence component is obtained by multiplying the missing data rate based on the preset missing penalty coefficient. The number of abnormal jump points is obtained by determining the number of times the difference between adjacent sampling points in the terminal voltage decay sequence exceeds the preset jump limit. The jump confidence component is obtained by multiplying the number of abnormal jump points according to the preset jump penalty coefficient. Determine the variance of the fitting residuals of the decay trend line, and multiply the variance values ​​based on the preset residual penalty coefficient to obtain the residual confidence component; The sum of the components of the missing confidence component, the jump confidence component, and the residual confidence component is determined, and the difference between the baseline confidence constant and the sum of the components is used as the prediction confidence factor.

Images