Diode Package Lifetime Prediction Method

Abstract

The present invention relates to the technical field of packaged diode prediction and analysis, and specifically to a diode package life prediction method, comprising the following steps: collecting historical operating data of diodes under different working conditions; regularly monitoring microstructural changes of diode packaging materials through non-invasive microscopic imaging technology, and calculating degradation rates; combining historical operating data and degradation rates to construct a comprehensive data model containing multiple key parameters affecting diode life; developing a life prediction model to predict the remaining effective life of diodes under specific working conditions; verifying the accuracy of the life prediction model through experiments; and applying the optimized life prediction model to the production and maintenance process of diodes. The present invention takes into account the changes in various operating currents, temperatures and other environmental factors that a diode may encounter during its life cycle, thereby making a more accurate prediction of its remaining effective life.

Description

Technical Field

[0001] The invention relates to the technical field of packaged diode prediction and analysis, and in particular to a diode package life prediction method. Background Art

[0002] Diode packaging refers to the integration of packaged diodes. In the manufacture and maintenance of modern electronic equipment, packaged diodes are key components, and their stability and reliability are crucial to the performance of the entire system. Packaged diodes not only need to withstand normal operating loads of current and voltage, but also have to deal with potential degradation caused by environmental factors such as temperature changes, mechanical stress, and humidity. Although packaging technology is designed to protect the semiconductor materials inside the diode and provide the necessary mechanical support and thermal management, in actual use, these factors may still cause the performance of the packaged diode to degrade and ultimately lead to failure.

[0003] Current lifetime prediction methods mainly rely on simple degradation models and statistical analysis. These models often assume that the diode operates under certain constant operating conditions, without taking into account the dynamic changes and complexity of the working environment in actual applications. Especially for packaged diodes, due to the complexity of the packaging structure, their internal temperature and stress distribution may change with time and working conditions, making lifetime prediction more difficult.

[0004] Therefore, an advanced lifetime prediction method is needed that can take into account the actual behavior of packaged diodes under varying operating conditions and comprehensively analyze the nonlinear degradation process caused by temperature fluctuations, current overload, and packaging material aging. Such a prediction method can provide more accurate lifetime estimates, enabling manufacturers to more effectively plan maintenance cycles, optimize designs, and prevent potential failure modes during the product development stage. Summary of the invention

[0005] Based on the above purpose, the present invention provides a diode package life prediction method.

[0006] The diode package life prediction method comprises the following steps:

[0007] S1: Collect historical operating data of the diode under different working conditions, including operating current, operating temperature and working time;

[0008] S2: Regularly monitor the microstructural changes of the diode packaging materials by non-invasive microscopic imaging techniques (such as scanning electron microscopy or atomic force microscopy) to detect degradation caused by long-term current and temperature stress and calculate the degradation rate;

[0009] S3: Combine the historical operating data and degradation rates obtained in S1 and S2 to build a comprehensive data model that includes multiple key parameters that affect the diode life;

[0010] S4: Based on the key factors, develop a lifetime prediction model to predict the remaining useful lifetime of the diode under specific operating conditions;

[0011] S5: Verify the accuracy of the life prediction model through experiments, and adjust and optimize the life prediction model until the prediction result and the actual life meet the predetermined standard;

[0012] S6: Apply the optimized lifetime prediction model to the production and maintenance process of diodes to achieve real-time monitoring and prediction of diode lifetime.

[0013] Furthermore, the S1 specifically includes:

[0014] S11: Integrate sensors in the diode circuit to monitor and record the working current and working temperature of the diode under working and stress test conditions in real time;

[0015] S12: Use the timestamp function to assign a time tag to each data collection and track the specific working time corresponding to each data point;

[0016] S13: By deploying environmental monitoring equipment in the use environment of the diode, data on environmental factors that affect the performance of the diode, including ambient temperature and humidity, are collected;

[0017] S14: Establish a data acquisition unit responsible for receiving data from the diode body and the environmental monitoring equipment, and storing the data in a database;

[0018] S15: Set a data collection strategy, including presetting the frequency of data collection, the format of data storage, and the length of time for data retention.

[0019] Furthermore, the S2 specifically includes:

[0020] S21: Select a diode sample and place it in a test environment simulating actual working conditions, including a predetermined current load and temperature range;

[0021] S22: Regularly perform microstructural observations on selected diode samples using scanning electron microscopy, transmission electron microscopy, or atomic force microscopy;

[0022] S23: During each imaging process, pay attention to the degradation phenomena of microcracks, material delamination, and interface degradation in the diode packaging material, and record the location, size, and morphology of the degradation phenomena;

[0023] S24: Analyze the trends of microstructural changes by comparing consecutive imaging results to identify and quantify the degradation rates caused by long-term current and temperature stresses.

[0024] Furthermore, the degradation rate calculation in S24 specifically includes:

[0025] The degradation parameter D is defined as a quantitative indicator of microstructural changes. The degradation parameter D includes the average length of microcracks, the area ratio of material delamination, and the depth of interface degradation.

[0026] After each microscopic imaging, the imaging results were processed using image analysis software to quantify the value of the degradation parameter D;

[0027] The time point t of each imaging is recorded and paired with the corresponding degradation parameter D to form a series of time-degradation data points: {(t1, D1), (t2, D2), ..., (t n ,D n )};

[0028] Use the following formula to calculate the degradation rate Wherein, ΔD is the change of the degradation parameter between adjacent imaging time points, and Δt is the time difference between adjacent imaging time points;

[0029] Considering the influence of working conditions on the degradation rate, a correction factor C is introduced. The correction factor C is adjusted based on the working current and temperature data of the diode. The corrected degradation rate R′

[0030] The calculation formula is: R′=C×R, where C is a function based on empirical data or physical model: C=α×I β ×T γ , where I represents the operating current, T represents the operating temperature, and α, β, and γ are coefficients obtained by fitting the experimental data to describe the influence of current and temperature on the degradation rate;

[0031] The calculated modified degradation rate R′ is taken as one of the input parameters and incorporated into the diode lifetime prediction model.

[0032] Furthermore, the S3 specifically includes:

[0033] S31: Obtain historical operating data of the diode, including operating current I, operating temperature T, and operating time t;

[0034] S32: using the calculated modified degradation rate R′ as a key parameter in the diode degradation process;

[0035] S33: combining the historical operation data with the corrected degradation rate R′ to form a multidimensional data set, where each data point consists of {I, T, t, R′}, where each parameter represents an aspect that affects the diode life;

[0036] S34: A comprehensive data model was constructed using multivariate linear regression to describe the relationship between diode lifetime and key parameters.

[0037] Furthermore, the multivariate linear regression method for constructing a comprehensive data model specifically includes:

[0038] Define the diode life L as the dependent variable, and the operating current I, operating temperature T, operating time t and modified degradation rate R′ as independent variables;

[0039] The form of constructing a multiple linear regression model is: L = β0 + β1I + β2T + β3t +

[0040] β4R′+∈, where β0 is the intercept term, β1, β2, β3 and β4 are model parameters, representing the influence of each independent variable on the dependent variable L, and ∈ is the error term, representing the random variation that the model cannot explain;

[0041] Use a historical data set containing multiple observations, each of which includes I, T, t, R′ and the corresponding L value;

[0042] The least squares method is used to estimate the model parameters β0, β1, β2, β3 and β4. The goal of the least squares method is to minimize the sum of squares of the error term ∈. The specific calculation formula is: Where n is the total number of data points, L i ,I i 、T i ,t i and R i ′ are the values ​​of life, operating current, operating temperature, operating time and corrected degradation rate of the i-th data point respectively;

[0043] Use the multivariate linear regression function in Python to solve the optimal values ​​of the model parameters;

[0044] Perform statistical tests on the model, including R 2 The F test and t test were used to evaluate the goodness of fit of the model and the significance of the parameters.

[0045] Furthermore, the life prediction model is based on an improved Cox proportional hazard model, specifically including:

[0046] Time-dependent covariate: In order to describe the dynamic changes of diode working conditions and degradation rate, a time-dependent covariate X is introduced i(t), the value of each covariate changes with time t, reflecting the impact of changes over time: X i (t) = f i (t,α i ), where f i (t,α i ) is a linear function of time t, describing the covariate X i Changes with time t, where α i is the parameter set of the linear function. Considering the nonlinear law followed by the degradation process of the diode, a nonlinear term is introduced to capture the nonlinear relationship. The risk function based on time dependence is expressed as:

[0047] Among them, f j (·) represents the introduced nonlinear function, h(t|X(t)) represents the instantaneous hazard rate (or risk function) at time t given the covariate X(t), which represents the instantaneous probability of diode failure at time t, and h0(t) is the baseline risk function, which represents the probability of diode failure when all covariates X(t) are equal. i When (t) is 0, the instantaneous risk rate at time t, exp is a natural exponential function, which is used to ensure that the risk rate is positive, β i is the covariate X i The coefficient of (t) reflects the influence weight of the covariate in the risk rate, X i (t) is the function of the value of the i-th covariate at time t, which changes with time and represents the i-th key factor affecting the diode life, such as operating current, operating temperature, etc. n is the covariate X i The total number of (t), γ j is a nonlinear function f j The coefficient reflects the influence weight of the nonlinear function in the risk rate, f j (X j (t)) represents the jth nonlinear function, which is used to capture the covariate X j (t) has a nonlinear relationship with the hazard rate, for example, X j (t) 2 、log(X j (t)) or other functional forms based on physical mechanisms, m is a nonlinear function f j Total number of;

[0048] Piecewise baseline risk function: A piecewise function is used to represent the baseline risk function h0(t) to adapt to the characteristics of diode failure data. For the time period [t k-1 ,t k ]:h0(t)=h 0k (t),t k-1 ≤t <t k , where tk-1 ,t k represents the start and end time of the kth time period, and each piecewise function h 0k (t) Function forms include constant, linear, and exponential forms to adapt to risk characteristics in different time periods;

[0049] Risk prediction and remaining life estimation: Based on the improved Cox proportional hazard model, the instantaneous risk rate h(t|X(t)) and the remaining life prediction of each diode under given working conditions are calculated. The remaining life prediction is based on the survival function S(t|X(t)), which takes into account time-dependent covariates and nonlinear relationships:

[0050] in, It is the integral of the instantaneous hazard rate h(u|X(u)) from time 0 to t. The remaining life is estimated by solving S(T|X(T))=p, where is used to calculate the survival function, p is the given survival probability threshold, and T is the corresponding time point.

[0051] Further, the solution S(T|X(T))=p is used to estimate the remaining life based on the survival function S(t|X(t)). In the Coxt proportional hazard model, the survival function S(t|X(t)) is obtained by the cumulative hazard function H(t), which is the integral of the instantaneous hazard function. The relationship between the survival function S(t|X(t)) and the cumulative hazard function H(t) is:

[0052] S(t|X(t))=exp(-H(t));

[0053] The cumulative risk function H(t) is defined as:

[0054]

[0055] For the Coxt proportional hazards model, the cumulative hazard function is further expressed as:

[0056]

[0057] Given a survival probability threshold p, find the time point T such that S(T) = p, that is, solve the following equation:

[0058]

[0059] Taking the logarithm gives:

[0060]

[0061] Solving the equations uses numerical integration and numerical methods, which is done by:

[0062] For each time point u, calculate the value of h(u|X(u));

[0063] Numerically integrate these values ​​from u = 0 to u = T;

[0064] Adjust the value of T until you find a T that satisfies the equation.

[0065] Furthermore, the S6 specifically includes:

[0066] S61: Select sample diodes, and randomly select multiple diodes from a production batch as test samples;

[0067] S62: Record key factors, record key factor data of the sample diode under actual working conditions;

[0068] S63: Accelerated life test: Accelerated life test is performed on sample diodes to simulate the degradation process under long-term operating conditions in a short time;

[0069] S64: data collection, collecting the failure data of the diode during the test, including the failure time and the corresponding working conditions;

[0070] S65: Model prediction, using the collected data to calculate the predicted lifetime of the diode through the lifetime prediction model;

[0071] S66: Result comparison, comparing the predicted lifespan with the actual lifespan observed in the experiment to assess the degree of agreement between the predicted results and the actual results;

[0072] S67: Model optimization, adjusting the parameters of the life prediction model according to the deviation between the predicted results and the actual results, including re-estimating the coefficients β and γ in the Cox model and using the maximum likelihood estimation method to estimate the coefficients in the model;

[0073] S68: Accuracy evaluation, using statistical methods to assess the predictive accuracy of the model;

[0074] Iterative verification: Repeat S63 to S68 until the accuracy of the prediction result meets the predetermined standard.

[0075] Beneficial effects of the present invention:

[0076] The present invention provides a life prediction model that can accurately reflect the degradation process of the diode under actual working conditions by introducing time-dependent covariates and considering the nonlinear relationship of the covariates. This method can take into account the changes in various operating currents, temperatures and other environmental factors that the diode may encounter during its life cycle, thereby making a more accurate prediction of its remaining effective life.

[0077] The present invention enables manufacturers and maintenance engineers to formulate more scientific maintenance and replacement plans based on the model prediction results, verify the accuracy of the model through experiments and optimize the model according to the experimental results. The present invention ensures a high degree of consistency between the prediction results and the actual lifespan, reduces equipment downtime caused by diode failures, and reduces maintenance costs.

[0078] The present invention uses an improved Cox proportional hazard model to predict the diode life, which can accurately capture and quantify the complex factors affecting the diode life. By introducing time-dependent covariates and nonlinear relationships, the model can not only accurately describe the behavior of the diode under changing working conditions, but also adapt to the different degradation characteristics that may be exhibited by the diode at different stages, significantly improving the accuracy of life prediction, and providing a strong scientific basis for the use, maintenance and reliability management of the diode, thereby helping to extend the service life of the diode, reduce unexpected failures, reduce costs, and improve the stability and safety of the overall system. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0080] Figure 1 A schematic diagram of a prediction method flow chart of an embodiment of the present invention;

[0081] Figure 2 It is a schematic diagram of an improved Cox proportional hazard model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0082] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0083] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0084] like Figure 1-2 As shown, the diode package life prediction method includes the following steps:

[0085] S1: Collect historical operating data of the diode under different working conditions, including operating current, operating temperature and working time;

[0086] S2: Regularly monitor the microstructural changes of the diode packaging materials by non-invasive microscopic imaging techniques (such as scanning electron microscopy or atomic force microscopy) to detect degradation caused by long-term current and temperature stress and calculate the degradation rate;

[0087] S3: Combine the historical operating data and degradation rates obtained in S1 and S2 to build a comprehensive data model that includes multiple key parameters that affect the diode life;

[0088] S4: Based on the key factors, develop a lifetime prediction model to predict the remaining useful lifetime of the diode under specific operating conditions;

[0089] S5: Verify the accuracy of the life prediction model through experiments, and adjust and optimize the life prediction model until the prediction result and the actual life meet the predetermined standard;

[0090] S6: Apply the optimized lifetime prediction model to the production and maintenance process of diodes to achieve real-time monitoring and prediction of diode lifetime.

[0091] S1 specifically includes:

[0092] S11: Integrate sensors in the diode circuit to monitor and record the working current and working temperature of the diode under working and stress test conditions in real time;

[0093] S12: Use the timestamp function to assign a time tag to each data collection and track the specific working time corresponding to each data point;

[0094] S13: By deploying environmental monitoring equipment in the use environment of the diode, data on environmental factors that affect the performance of the diode, including ambient temperature and humidity, are collected;

[0095] S14: Establish a data acquisition unit responsible for receiving data from the diode body and the environmental monitoring equipment, and storing the data in a database;

[0096] S15: Set a data collection strategy, including presetting the frequency of data collection, the format of data storage, and the length of time for data retention.

[0097] S2 specifically includes:

[0098] S21: Select a diode sample and place it in a test environment simulating actual working conditions, including a predetermined current load and temperature range;

[0099] S22: Regularly perform microstructural observations on selected diode samples using scanning electron microscopy, transmission electron microscopy, or atomic force microscopy;

[0100] S23: During each imaging process, pay attention to the degradation phenomena of microcracks, material delamination, and interface degradation in the diode packaging material, and record the location, size, and morphology of the degradation phenomena;

[0101] S24: Analyze the trends of microstructural changes by comparing consecutive imaging results to identify and quantify the degradation rates caused by long-term current and temperature stresses.

[0102] The degradation rate calculation in S24 specifically includes:

[0103] The degradation parameter D is defined as a quantitative indicator of microstructural changes. The degradation parameter D includes the average length of microcracks, the area ratio of material delamination, and the depth of interface degradation.

[0104] After each microscopic imaging, the imaging results were processed using image analysis software to quantify the value of the degradation parameter D;

[0105] The time point t of each imaging is recorded and paired with the corresponding degradation parameter D to form a series of time-degradation data points: {(t1, D1), (t2, D2), ..., (t n ,D n )};

[0106] Use the following formula to calculate the degradation rate Wherein, ΔD is the change of the degradation parameter between adjacent imaging time points, and Δt is the time difference between adjacent imaging time points;

[0107] Considering the influence of working conditions on the degradation rate, a correction factor C is introduced. The correction factor C is adjusted based on the working current and temperature data of the diode. The corrected degradation rate R′

[0108] The calculation formula is: R′=C×R, where C is a function based on empirical data or physical model: C=α×I β ×T γ , where I represents the operating current, T represents the operating temperature, and α, β, and γ are coefficients obtained by fitting the experimental data to describe the influence of current and temperature on the degradation rate;

[0109] The calculated modified degradation rate R′ is taken as one of the input parameters and incorporated into the diode lifetime prediction model to improve the accuracy and reliability of the prediction.

[0110] S3 specifically includes:

[0111] S31: Obtain historical operating data of the diode, including operating current I, operating temperature T, and operating time t;

[0112] S32: taking the calculated corrected degradation rate R′ as a key parameter in the diode degradation process; the operating current I, the operating temperature T, the operating time t, and the corrected degradation rate R′ are the key parameters;

[0113] S33: combining the historical operation data with the corrected degradation rate R′ to form a multidimensional data set, where each data point consists of {I, T, t, R′}, where each parameter represents an aspect that affects the diode life;

[0114] S34: A comprehensive data model was constructed using multivariate linear regression to describe the relationship between diode lifetime and key parameters.

[0115] The comprehensive data model constructed by multivariate linear regression specifically includes:

[0116] Define the diode life L as the dependent variable, and the operating current I, operating temperature T, operating time t and modified degradation rate R′ as independent variables;

[0117] The form of constructing a multiple linear regression model is: L = β0 + β1I + β2T + β3t +

[0118] β4R′+∈, where β0 is the intercept term, β1, β2, β3 and β4 are model parameters, representing the influence of each independent variable on the dependent variable L, and ∈ is the error term, representing the random variation that the model cannot explain;

[0119] Use a historical data set containing multiple observations, each of which includes I, T, t, R′ and the corresponding L value;

[0120] The least squares method is used to estimate the model parameters β0, β1, β2, β3 and β4. The goal of the least squares method is to minimize the sum of squares of the error term ∈. The specific calculation formula is: Where n is the total number of data points, L i ,I i 、T i ,t i and R i ′ are the values ​​of life, operating current, operating temperature, operating time and corrected degradation rate of the i-th data point respectively;

[0121] Use the multivariate linear regression function in Python to solve the optimal values ​​of the model parameters;

[0122] Perform statistical tests on the model, including R 2 The F test and t test were used to evaluate the goodness of fit of the model and the significance of the parameters.

[0123] The life expectancy prediction model is based on the improved Cox proportional hazard model, which includes:

[0124] Time-dependent covariate: In order to describe the dynamic changes of diode working conditions and degradation rate, a time-dependent covariate X is introduced i (t), the value of each covariate changes with time t, reflecting the impact of changes over time: X i (t) = f i (t,α i ), where f i (t,α i ) is a linear function of time t, describing the covariate X i Changes with time t, where α i is the parameter set of the linear function. Considering the nonlinear law followed by the degradation process of the diode, a nonlinear term is introduced to capture the nonlinear relationship. The risk function based on time dependence is expressed as:

[0125] Among them, f j (·) represents the introduced nonlinear function, h(t|X(t)) represents the instantaneous hazard rate (or risk function) at time t given the covariate X(t), which represents the instantaneous probability of diode failure at time t, and h0(t) is the baseline risk function, which represents the probability of diode failure when all covariates X(t) are equal. i When (t) is 0, the instantaneous risk rate at time t, exp is a natural exponential function, which is used to ensure that the risk rate is positive, βi is the covariate X i The coefficient of (t) reflects the influence weight of the covariate in the risk rate, X i (t) is the function of the value of the i-th covariate at time t, which changes with time and represents the i-th key factor affecting the diode life, such as operating current, operating temperature, etc. n is the covariate X i The total number of (t), γ j is a nonlinear function f j The coefficient reflects the influence weight of the nonlinear function in the risk rate, f j (X j (t)) represents the jth nonlinear function, which is used to capture the covariate X j (t) has a nonlinear relationship with the hazard rate, for example, X j (t) 2 、log(X j (t)) or other functional forms based on physical mechanisms, m is a nonlinear function f j Total number of;

[0126] Piecewise baseline risk function: A piecewise function is used to represent the baseline risk function h0(t) to adapt to the characteristics of diode failure data. For the time period [t k-1 ,t k ]:h0(t)=h 0k (t),t k-1 ≤t <t k , where t k-1 ,t k represents the start and end time of the kth time period, and each piecewise function h 0k (t) Function forms include constant, linear, and exponential forms to adapt to risk characteristics in different time periods;

[0127] Risk prediction and remaining life estimation: Based on the improved Cox proportional hazard model, the instantaneous risk rate h(t|X(t)) and the remaining life prediction of each diode under given working conditions are calculated. The remaining life prediction is based on the survival function S(t|X(t)), which takes into account time-dependent covariates and nonlinear relationships:

[0128] in, It is the integral of the instantaneous hazard rate h(u|X(u)) from time 0 to t. The remaining life is estimated by solving S(T|X(T))=p, where is used to calculate the survival function, p is the given survival probability threshold, and T is the corresponding time point.

[0129] Solve S(T|X(T))=p to estimate the remaining life based on the survival function S(t|X(t)). In the Coxt proportional hazard model, the survival function S(t|X(t)) is obtained by the cumulative hazard function H(t). The cumulative hazard function H(t) is the integral of the instantaneous hazard function. The relationship between the survival function S(t|X(t)) and the cumulative hazard function H(t) is:

[0130] S(t|X(t))=exp(-H(t));

[0131] The cumulative risk function H(t) is defined as:

[0132]

[0133] For the Coxt proportional hazards model, the cumulative hazard function is further expressed as:

[0134]

[0135] Given a survival probability threshold p, find the time point T such that S(T) = p, that is, solve the following equation:

[0136]

[0137] Taking the logarithm gives:

[0138]

[0139] Solving the equations uses numerical integration and numerical methods, which is done by:

[0140] For each time point u, calculate the value of h(u|X(u));

[0141] Numerically integrate these values ​​from u = 0 to u = T;

[0142] Adjust the value of T until you find a T that satisfies the equation.

[0143] In practice, it is implemented in a software package through programming and numerical methods such as the Newton-Raphson iteration method are used to solve T.

[0144] In the present invention, it is assumed that the goal is to predict the remaining useful life of a diode under specific operating conditions, and a set of historical data has been acquired, the data including time-dependent records of operating current I, operating temperature T, operating time t and corrected degradation rate R′, as well as the failure time of each diode.

[0145] The modified Coxt proportional hazards model was used to analyze these data, and a set of estimated model parameters β i and γ j,Now, we want to predict the remaining useful lifetime of a particular diode under the current or predicted ,operating conditions.

[0146] The application steps are as follows:

[0147] 1. Parameter estimation: Use historical data to estimate the parameters of the Cox model, including β i , γ j and h0(t).

[0148] 2. Risk function calculation: For a new diode, or a diode operating under specific working conditions, calculate its instantaneous risk function h(t|X(t)) based on the current operating current, temperature and other factors.

[0149] 3. Calculation of survival probability: Calculate the cumulative risk function H(t) by integrating the instantaneous risk function, and then calculate the survival function S(t|X(t))=exp(-H(t)).

[0150] 4. Remaining lifetime estimation: Determine a survival probability threshold p, which represents the probability that the diode will continue to operate without failure before a certain point in the future. Solve the equation S(T|X(T))=p to find the corresponding time point T, which can be regarded as the predicted remaining effective lifetime.

[0151] For example, suppose a diode currently operates at 50 mA and 70°C and has been working for 1000 hours. We know from historical data that the corrected degradation rate R′ of the diode under similar conditions is 0.02h^-1. We want to know how long this diode will continue to work without failure with a 95% probability while maintaining the current conditions.

[0152] We set a survival probability threshold p = 0.95 and then use our model and estimated parameters to solve the equation S(T|X(T)) = 0.95. Using numerical methods, we find a time point T at which the survival probability equals 0.95. This time point T represents the predicted time at which the diode has a 95% probability of continuing to work without failure under given working conditions. In this way, manufacturers or maintenance engineers can use this prediction to plan maintenance schedules, decide whether to replace diodes, or adjust working conditions to extend the service life of diodes. This process can be automated through programming so that real-time data can be used to continuously update life predictions, thereby achieving more accurate predictions and more efficient maintenance strategies.

[0153] S6 specifically includes:

[0154] S61: Select sample diodes, and randomly select multiple diodes from a production batch as test samples;

[0155] S62: Record key factors, record key factor data of the sample diode under actual working conditions;

[0156] S63: Accelerated life test: Accelerated life test is performed on sample diodes to simulate the degradation process under long-term operating conditions in a short time;

[0157] S64: data collection, collecting the failure data of the diode during the test, including the failure time and the corresponding working conditions;

[0158] S65: Model prediction, using the collected data to calculate the predicted lifetime of the diode through the lifetime prediction model;

[0159] S66: Result comparison, comparing the predicted lifespan with the actual lifespan observed in the experiment to assess the degree of agreement between the predicted results and the actual results;

[0160] S67: Model optimization, adjusting the parameters of the life prediction model according to the deviation between the predicted results and the actual results, including re-estimating the coefficients β and γ in the Cox model and using the maximum likelihood estimation method to estimate the coefficients in the model;

[0161] S68: Accuracy assessment, using statistical methods (such as mean square error, correlation coefficient or determination coefficient, etc.) to evaluate the predictive accuracy of the model;

[0162] Iterative verification: Repeat S63 to S68 until the accuracy of the prediction result meets the predetermined standard.

[0163] The predetermined standard can be set according to the needs of the actual application, for example, requiring that the difference between the predicted lifespan and the actual lifespan does not exceed 5% or the correlation coefficient is not less than 0.9.

[0164] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present invention is limited to these examples. Under the concept of the present invention, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the present invention as described above, which are not provided in detail for the sake of simplicity.

[0165] The present invention is intended to cover all such substitutions, modifications and variations that fall within the broad scope of the claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A diode package life prediction method, characterized in that: The following steps are involved: S1: Collect historical operating data of the diode under different working conditions, including operating current, operating temperature and working time; S2: Regularly monitor the microstructural changes of diode packaging materials through non-invasive microscopic imaging techniques to detect degradation caused by long-term current and temperature stress and calculate the degradation rate, including: S21: Select a diode sample and place it in a test environment simulating actual working conditions, including a predetermined current load and temperature range; S22: Regularly perform microstructural observations on selected diode samples using scanning electron microscopy, transmission electron microscopy, or atomic force microscopy; S23: During each imaging process, pay attention to the degradation phenomena of microcracks, material delamination, and interface degradation in the diode packaging material, and record the location, size, and morphology of the degradation phenomena; S24: Analyze the development trend of microstructural changes by comparing continuous imaging results to identify and quantify the degradation rate caused by long-term current and temperature stress. The degradation rate calculation specifically includes: The degradation parameter D is defined as a quantitative indicator of microstructural changes. The degradation parameter D includes the average length of microcracks, the area ratio of material delamination, and the depth of interface degradation. After each microscopic imaging, the imaging results were processed using image analysis software to quantify the value of the degradation parameter D; The time point t of each imaging is recorded and paired with the corresponding degradation parameter D to form a series of time-degradation data points: {(t1, D1), (t2, D2), ..., (t n , D n )}; The degradation rate R is calculated using the following formula: Wherein, ΔD is the change of the degradation parameter between adjacent imaging time points, and Δt is the time difference between adjacent imaging time points; Considering the influence of working conditions on the degradation rate, a correction factor C is introduced. The correction factor C is adjusted based on the working current and temperature data of the diode. The corrected degradation rate R′ The calculation formula is: R′=C×R, where C is a function based on empirical data or physical model: C=α×I β ×T γ , where I represents the operating current, T represents the operating temperature, and α, β, and γ are coefficients obtained by fitting the experimental data to describe the influence of current and temperature on the degradation rate; The calculated modified degradation rate R′ is used as one of the input parameters and incorporated into the diode life prediction model; S3: Combine the historical operating data and degradation rates obtained in S1 and S2 to build a comprehensive data model that includes multiple key parameters that affect the diode life; S4: Based on the key factors, develop a lifetime prediction model to predict the remaining useful lifetime of the diode under specific operating conditions; S5: Verify the accuracy of the life prediction model through experiments, and adjust and optimize the life prediction model until the prediction result and the actual life meet the predetermined standard; S6: Apply the optimized lifetime prediction model to the production and maintenance process of diodes to achieve real-time monitoring and prediction of diode lifetime.

2. The diode package life prediction method according to claim 1, characterized in that: The S1 specifically includes: S11: Integrate sensors in the diode circuit to monitor and record the working current and working temperature of the diode under working and stress test conditions in real time; S12: Use the timestamp function to assign a time tag to each data collection and track the specific working time corresponding to each data point; S13: By deploying environmental monitoring equipment in the use environment of the diode, data on environmental factors that affect the performance of the diode, including ambient temperature and humidity, are collected; S14: Establish a data acquisition unit responsible for receiving data from the diode body and the environmental monitoring equipment, and storing the data in a database; S15: Set a data collection strategy, including presetting the frequency of data collection, the format of data storage, and the length of time for data retention.

3. The diode package life prediction method according to claim 1, characterized in that: The S3 specifically includes: S31: Obtain historical operating data of the diode, including operating current I, operating temperature T, and operating time t; S32: using the calculated modified degradation rate R′ as a key parameter in the diode degradation process; S33: combining the historical operation data with the corrected degradation rate R′ to form a multidimensional data set, where each data point consists of {I, T, t, R′}, where each parameter represents an aspect that affects the diode life; S34: A comprehensive data model was constructed using multivariate linear regression to describe the relationship between diode lifetime and key parameters.

4. The diode package life prediction method according to claim 3, characterized in that: The multivariate linear regression method for constructing a comprehensive data model specifically includes: Define the diode life L as the dependent variable, and the operating current I, operating temperature T, operating time t and modified degradation rate R′ as independent variables; The form of the multivariate linear regression model is: L = β0 + β1I + β2T + β3t + β4R′ + ∈, where β0 is the intercept term, β1, β2, β3 and β4 are model parameters, representing the influence of each independent variable on the dependent variable L, and ∈ is the error term, representing the random variation that the model cannot explain; Use a historical data set containing multiple observations, each of which contains I, T, t, R′ and the corresponding L value; The least squares method is used to estimate the model parameters β0, β1, β2, β3 and β4. The goal of the least squares method is to minimize the sum of squares of the error term ∈. The specific calculation formula is: Where n is the total number of data points, L i ,I i , T i ,t i and R i ′ are the values ​​of life, operating current, operating temperature, operating time and corrected degradation rate of the i-th data point respectively; Use the multivariate linear regression function in Python to solve the optimal values ​​of the model parameters; Statistical tests of the model, including R 2 The F test and t test were used to evaluate the goodness of fit of the model and the significance of the parameters.

5. The diode package life prediction method according to claim 4, characterized in that: The life prediction model is based on an improved Cox proportional hazard model, and specifically includes: Time-dependent covariate: In order to describe the dynamic changes of diode working conditions and degradation rate, a time-dependent covariate X is introduced i (t), the value of each covariate changes with time t, reflecting the impact of changes over time: X i (t) = f i (t,α i ), where f i (t,α i ) is a linear function of time t, describing the covariate X i Changes with time t, where α i is the parameter set of the linear function. Considering the nonlinear law followed by the degradation process of the diode, a nonlinear term is introduced to capture the nonlinear relationship. The risk function based on time dependence is expressed as: Among them, f j (·) represents the introduced nonlinear function, h(t|X(t)) represents the instantaneous hazard rate at time t given the covariate X(t), represents the instantaneous probability of diode failure at time t, and h0(t) is the baseline hazard function, which represents the probability of diode failure when all covariates X i When (t) is 0, the instantaneous risk rate at time t, exp is a natural exponential function, which is used to ensure that the risk rate is positive, β i is the covariate X i The coefficient of (t) reflects the influence weight of the covariate in the risk rate, X i (t) is the function of the value of the i-th covariate at time t, and n is the covariate X i The total number of (t), γ j is the coefficient of the nonlinear function f, reflecting the influence weight of the nonlinear function in the risk rate, f j (X j (t)) represents the jth nonlinear function, which is used to capture the covariate X j (t) is a nonlinear relationship between the risk rate and m is a nonlinear function f j Total number of; Piecewise baseline risk function: A piecewise function is used to represent the baseline risk function h0(t) to adapt to the characteristics of diode failure data. For the time period [t k-1 , t k ]:h0(t)=h 0k (t), t k-1 ≤t<t k , where t k-1 , t k represents the start and end time of the kth time period, and each piecewise function h 0k (t) Function forms include constant, linear, and exponential forms to adapt to risk characteristics in different time periods; Risk prediction and remaining life estimation: Based on the improved Cox proportional hazard model, the instantaneous risk rate h(t|X(t)) and the remaining life prediction of each diode under given working conditions are calculated. The remaining life prediction is based on the survival function S(t|X(t)), which takes into account time-dependent covariates and nonlinear relationships: in, It is the integral of the instantaneous hazard rate h(u|X(u)) from time 0 to t. The remaining life is estimated by solving S(T|X(T))=p, where is used to calculate the survival function, p is the given survival probability threshold, and T is the corresponding time point.

6. The diode package life prediction method according to claim 5, characterized in that: The solution S(T|X(T))=p is used to estimate the remaining life based on the survival function S(t|X(t)). In the Coxt proportional hazard model, the survival function S(t|X(t)) is obtained by the cumulative hazard function H(t). The cumulative hazard function H(t) is the integral of the instantaneous hazard function. The relationship between the survival function S(t|X(t)) and the cumulative hazard function H(t) is: S(t|X(t))=exp(-H(t)); The cumulative risk function H(t) is defined as: For the Coxt proportional hazards model, the cumulative hazard function is further expressed as: Given a survival probability threshold p, find the time point T such that S(T) = p, that is, solve the following equation: Taking the logarithm gives: Solving the equations uses numerical integration and numerical methods, which is done by: For each time point u, calculate the value of h(u|X(u)); Numerically integrate these values ​​from u = 0 to u = T; Adjust the value of T until you find a T that satisfies the equation.

7. The diode package life prediction method according to claim 6, characterized in that: The S6 specifically includes: S61: Select sample diodes, and randomly select multiple diodes from a production batch as test samples; S62: Record key factors, record key factor data of the sample diode under actual working conditions; S63: Accelerated life test: Accelerated life test is performed on sample diodes to simulate the degradation process under long-term operating conditions in a short time; S64: data collection, collecting the failure data of the diode during the test, including the failure time and the corresponding working conditions; S65: Model prediction, using the collected data to calculate the predicted lifetime of the diode through the lifetime prediction model; S66: Result comparison, comparing the predicted lifespan with the actual lifespan observed in the experiment to assess the degree of agreement between the predicted results and the actual results; S67: Model optimization, adjusting the parameters of the life prediction model according to the deviation between the predicted results and the actual results, including re-estimating the coefficients β and γ in the Cox model and using the maximum likelihood estimation method to estimate the coefficients in the model; S68: Accuracy evaluation, using statistical methods to assess the predictive accuracy of the model; Iterative verification: Repeat S63 to S68 until the accuracy of the prediction result meets the predetermined standard.

Images