CN113987932A - MOSFET service life prediction method based on time series model - Google Patents
MOSFET service life prediction method based on time series model Download PDFInfo
- Publication number
- CN113987932A CN113987932A CN202111251625.1A CN202111251625A CN113987932A CN 113987932 A CN113987932 A CN 113987932A CN 202111251625 A CN202111251625 A CN 202111251625A CN 113987932 A CN113987932 A CN 113987932A
- Authority
- CN
- China
- Prior art keywords
- data
- model
- aging
- mosfet
- autocorrelation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/04—Ageing analysis or optimisation against ageing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Testing Of Individual Semiconductor Devices (AREA)
Abstract
The invention discloses a MOSFET service life prediction method based on a time series model, which comprises the following steps: s1, acquiring aging data of the MOSFET device through experiments; s2, preprocessing data and establishing a data model; s3, selecting parameters of an autoregressive moving average model and training the model; and S4, life prediction based on the time series model. The method uses the time sequence model to predict the service life of the MOSFET device, is simple and convenient to operate, has high accuracy, and can improve the use reliability of the power MOSFET device.
Description
Technical Field
The invention relates to the field of semiconductors, in particular to a MOSFET service life prediction method based on a time series model.
Background
With the rapid development of the semiconductor industry, power electronic devices are applied to a large number of large-scale fields, wherein metal-oxide semiconductor field effect transistors (abbreviated as MOSFETs) are core devices in the power electronic systems, but the power electronic devices often fail due to the complex working environment. Therefore, the reliability assessment of the operation condition of the MOSFET device is of great significance, if the time of the failure of the MOSFET device, namely the service life prediction, can be predicted by a certain technical means, and corresponding early warning schemes are prepared in advance, the occurrence of some accidents can be avoided, and the safety of a power electronic system is improved, so that the research on the service life prediction direction of the MOSFET device is of great practical significance.
Disclosure of Invention
The invention provides a MOSFET service life prediction method based on a time series model. The method predicts the residual service life of the MOSFET through the time model, and has the advantages of simple structure, high accuracy and strong practicability.
In order to achieve the purpose of the invention, the invention adopts the technical scheme that:
provided is a MOSFET life prediction method, which comprises the following steps:
s1, acquiring aging data of the MOSFET device through experiments;
s2, preprocessing data and establishing a data model;
s3, selecting parameters of an autoregressive moving average model and training the model;
and S4, life prediction based on the time series model.
Further, the specific method of step S1 is:
the method comprises the steps of carrying out a power cycle accelerated aging experiment on an MOSFET device, actively heating a chip through loss generated by the device, increasing the internal temperature of the device, accelerating aging of the device, acquiring switching voltage and switching current of the device through a sensor in the experimental aging process, and obtaining on-resistance Rds and on through calculating the switching voltage and dividing the switching current by a digital signal processing module, so as to obtain the change condition of the on-resistance of the MOSFET along with aging of the device and establish an aging data set.
Further, the specific method of step S2 includes the following steps:
s2-1, carrying out error judgment on the data in the aging data set, wherein the judgment criterion on the n data in a single cycle is as follows:
if it is not
X is theniFor measuring error data
If it is not
X is theniFor normal measurement data
Wherein x isiIs the ith data recorded in a single cycle;
is the arithmetic mean of all data in a single cycle;
s2-2, after error data in a single cycle are all removed, an arithmetic mean value of the remaining data is obtained to serve as an aging characteristic value of the current cycle, and the aging data set after processing only corresponds to one aging characteristic value in each cycle;
s2-3, performing curve fitting on the processed aging data to obtain the on-resistance Rds,onAs the device ages (i.e., the number of cycles N), the fitted data model is as follows:
Rds,on＝aexp(bN)+R0
wherein a and b are parameters of a fitting model; r0Is the initial value of the on-resistance.
Further, the specific method of step S3 includes the following steps:
s3-1, selecting an autoregressive moving average model (ARIMA (p, d, q) model) as a time sequence model, wherein the three corresponding main model parameters are the order p of the autoregressive model, and the order q of the moving average model and the difference order d when the time sequence presents stable autocorrelation;
s3-2, initializing a model parameter d to be 0, and then calculating and drawing an autocorrelation graph of the aging data, wherein the autocorrelation graph comprises an autocorrelation function graph and a partial autocorrelation function graph;
s3-3, checking the stationarity of the autocorrelation result, namely judging whether the autocorrelation and partial autocorrelation function graph of the data fluctuate within the range of 0.1 above and below the abscissa axis, if the autocorrelation result is not stable, carrying out once differential processing on the data, then checking the stationarity of the autocorrelation result, and so on until the autocorrelation result of the data is stable, wherein the differential order of the data is the value of d; in order to avoid loss of important information in data caused by excessive differential operation, d is set to be within a range of 0-2;
s3-4, determining the values of model parameters p and q by using Bayesian Information Criterion (BIC) so as to determine the final optimal time series model, wherein the BIC has the following formula:
BIC＝kln(n)-2ln(L)
wherein k is the number of parameters of the model, n is the number of samples, and L is a likelihood function;
setting a range for p and q, carrying out permutation and combination from a low order to a high order, and finally determining parameters according to a BIC criterion, namely, the minimum BIC value corresponds to an optimal model.
Further, the specific method of step S4 is:
and (4) predicting data by using the trained time series model in the step (S3), setting an early warning threshold, and estimating the residual service life of the MOSFET device according to the experimental cycle number of the current device, wherein the corresponding cycle number is the end point of the service life of the device when the predicted value of a certain time is greater than or equal to the threshold.
The invention has the beneficial effects that: the method calculates the on-resistance by obtaining the switching voltage and the switching current of the MOSFET in the working process, then processes data and fits a curve, and trains and determines a time sequence model to predict the service life. Compared with the prior art, the method provided by the invention has the advantages of simple structure, simplicity and convenience in operation, strong practicability and capability of improving the use reliability of devices.
Drawings
FIG. 1 is a schematic flow diagram of the process.
Detailed Description
The following description of the embodiments of the present invention is provided to facilitate the understanding of the present invention by those skilled in the art, but it should be understood that the present invention is not limited to the scope of the embodiments, and it will be apparent to those skilled in the art that various changes may be made without departing from the spirit and scope of the invention as defined and defined in the appended claims, and all matters produced by the invention using the inventive concept are protected.
As shown in fig. 1, the MOSFET lifetime prediction method based on the time series model includes the following steps:
s1, acquiring aging data of the MOSFET device through experiments;
s2, preprocessing data and establishing a data model;
s3, selecting parameters of an autoregressive moving average model and training the model;
and S4, life prediction based on the time series model.
The specific method of step S1 is:
the method comprises the steps of carrying out a power cycle accelerated aging experiment on an MOSFET device, actively heating a chip through loss generated by the device, increasing the internal temperature of the device, accelerating aging of the device, acquiring switching voltage and switching current of the device through a sensor in the experimental aging process, and obtaining on-resistance Rds and on through calculating the switching voltage and dividing the switching current by a digital signal processing module, so as to obtain the change condition of the on-resistance of the MOSFET along with aging of the device and establish an aging data set.
The specific method of step S2 includes the steps of:
s2-1, carrying out error judgment on the data in the aging data set, wherein the judgment criterion on the n data in a single cycle is as follows:
if it is not
X is theniFor measuring error data
If it is not
X is theniFor normal measurement data
Wherein x isiIs the ith data recorded in a single cycle;
is the arithmetic mean of all data in a single cycle;
s2-2, after error data in a single cycle are all removed, an arithmetic mean value of the remaining data is obtained to serve as an aging characteristic value of the current cycle, and the aging data set after processing only corresponds to one aging characteristic value in each cycle;
s2-3, performing curve fitting on the processed aging data to obtain the on-resistance Rds,onAs the device ages (i.e., the number of cycles N), the fitted data model is as follows:
Rds,on＝aexp(bN)+R0
wherein a and b are parameters of a fitting model; r0Is the initial value of the on-resistance.
The specific method of step S3 includes the steps of:
s3-1, selecting an autoregressive moving average model (ARIMA (p, d, q) model) as a time sequence model, wherein the three corresponding main model parameters are the order p of the autoregressive model, and the order q of the moving average model and the difference order d when the time sequence presents stable autocorrelation;
s3-2, initializing a model parameter d to be 0, and then calculating and drawing an autocorrelation graph of the aging data, wherein the autocorrelation graph comprises an autocorrelation function graph and a partial autocorrelation function graph;
s3-3, checking the stationarity of the autocorrelation result, namely judging whether the autocorrelation and partial autocorrelation function graph of the data fluctuate within the range of 0.1 above and below the abscissa axis, if the autocorrelation result is not stable, carrying out once differential processing on the data, then checking the stationarity of the autocorrelation result, and so on until the autocorrelation result of the data is stable, wherein the differential order of the data is the value of d; in order to avoid loss of important information in data caused by excessive differential operation, d is set to be within a range of 0-2;
s3-4, determining the values of model parameters p and q by using Bayesian Information Criterion (BIC) so as to determine the final optimal time series model, wherein the BIC has the following formula:
BIC＝kln(n)-2ln(L)
wherein k is the number of parameters of the model, n is the number of samples, and L is a likelihood function;
setting a range for p and q, carrying out permutation and combination from a low order to a high order, and finally determining parameters according to a BIC criterion, namely, the minimum BIC value corresponds to an optimal model.
The specific method of step S4 is:
and (4) predicting data by using the trained time series model in the step (S3), setting an early warning threshold, and estimating the residual service life of the MOSFET device according to the experimental cycle number of the current device, wherein the corresponding cycle number is the end point of the service life of the device when the predicted value of a certain time is greater than or equal to the threshold.
In summary, the invention provides a MOSFET lifetime prediction method based on a time series model. The method mainly adopts a time series model to predict the service life, and selects an autoregressive moving average model as the time series model.
Claims (5)
1. A MOSFET service life prediction method based on a time series model is characterized by comprising the following steps:
s1, acquiring aging data of the MOSFET device through experiments;
s2, preprocessing data and establishing a data model;
s3, selecting parameters of an autoregressive moving average model and training the model;
and S4, life prediction based on the time series model.
2. The method for predicting the lifetime of a MOSFET according to claim 1, wherein the step S1 is performed by:
performing power cycle accelerated aging experiment on the MOSFET device, actively heating the chip through the loss generated by the device, increasing the internal temperature of the device, accelerating the aging of the device, and performing power cycle accelerated aging experiment on the MOSFET device in the experimental aging processThe sensor obtains the switching voltage and the switching current of the device, and the digital signal processing module calculates the switching voltage divided by the switching current to obtain the on-resistance Rds,onTherefore, the change condition of the on-resistance of the MOSFET along with the aging of the device is obtained, and an aging data set is established.
3. The method for predicting the lifetime of a MOSFET according to claim 1, wherein the step S2 comprises the following steps:
s2-1, carrying out error judgment on the data in the aging data set, wherein the judgment criterion on the n data in a single cycle is as follows:
if it is not
X is theniFor measuring error data
If it is not
X is theniFor normal measurement data
Wherein x isiIs the ith data recorded in a single cycle;
is the arithmetic mean of all data in a single cycle;
s2-2, after error data in a single cycle are all removed, an arithmetic mean value of the remaining data is obtained to serve as an aging characteristic value of the current cycle, and the aging data set after processing only corresponds to one aging characteristic value in each cycle;
s2-3, performing curve fitting on the processed aging data to obtain the on-resistance Rds,onAs the device ages (i.e., the number of cycles N), the fitted data model is as follows:
Rds,on＝aexp(bN)+R0
wherein a and b are parameters of a fitting model; r0Is the initial value of the on-resistance.
4. The method for predicting the lifetime of a MOSFET according to claim 1, wherein the step S3 comprises the following steps:
s3-1, selecting an autoregressive moving average model (ARIMA (p, d, q) model) as a time sequence model, wherein the three corresponding main model parameters are the order p of the autoregressive model, and the order q of the moving average model and the difference order d when the time sequence presents stable autocorrelation;
s3-2, initializing a model parameter d to be 0, and then calculating and drawing an autocorrelation graph of the aging data, wherein the autocorrelation graph comprises an autocorrelation function graph and a partial autocorrelation function graph;
s3-3, checking the stationarity of the autocorrelation result, namely judging whether the autocorrelation and partial autocorrelation function graph of the data fluctuate within the range of 0.1 above and below the abscissa axis, if the autocorrelation result is not stable, carrying out once differential processing on the data, then checking the stationarity of the autocorrelation result, and so on until the autocorrelation result of the data is stable, wherein the differential order of the data is the value of d; in order to avoid loss of important information in data caused by excessive differential operation, d is set to be within a range of 0-2;
s3-4, determining the values of model parameters p and q by using Bayesian Information Criterion (BIC) so as to determine the final optimal time series model, wherein the BIC has the following formula:
BIC＝kln(n)-2ln(L)
wherein k is the number of parameters of the model, n is the number of samples, and L is a likelihood function;
setting a range for p and q, carrying out permutation and combination from a low order to a high order, and finally determining parameters according to a BIC criterion, namely, the minimum BIC value corresponds to an optimal model.
5. The method for predicting the lifetime of a MOSFET according to claim 1, wherein the step S4 is performed by:
and (4) predicting data by using the trained time series model in the step (S3), setting an early warning threshold, and estimating the residual service life of the MOSFET device according to the experimental cycle number of the current device, wherein the corresponding cycle number is the end point of the service life of the device when the predicted value of a certain time is greater than or equal to the threshold.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111251625.1A CN113987932A (en) | 2021-10-26 | 2021-10-26 | MOSFET service life prediction method based on time series model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111251625.1A CN113987932A (en) | 2021-10-26 | 2021-10-26 | MOSFET service life prediction method based on time series model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113987932A true CN113987932A (en) | 2022-01-28 |
Family
ID=79742091
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111251625.1A Pending CN113987932A (en) | 2021-10-26 | 2021-10-26 | MOSFET service life prediction method based on time series model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113987932A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114818393A (en) * | 2022-06-28 | 2022-07-29 | 北京芯可鉴科技有限公司 | Semiconductor device failure time prediction method, device, equipment and medium |
| CN115308558A (en) * | 2022-08-29 | 2022-11-08 | 北京智芯微电子科技有限公司 | Method and device for predicting service life of CMOS (complementary Metal oxide semiconductor) device, electronic equipment and medium |
| CN115495924A (en) * | 2022-10-10 | 2022-12-20 | 国营芜湖机械厂 | MOSFET service life prediction method based on ARIMA model |
-
2021
- 2021-10-26 CN CN202111251625.1A patent/CN113987932A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114818393A (en) * | 2022-06-28 | 2022-07-29 | 北京芯可鉴科技有限公司 | Semiconductor device failure time prediction method, device, equipment and medium |
| CN115308558A (en) * | 2022-08-29 | 2022-11-08 | 北京智芯微电子科技有限公司 | Method and device for predicting service life of CMOS (complementary Metal oxide semiconductor) device, electronic equipment and medium |
| CN115495924A (en) * | 2022-10-10 | 2022-12-20 | 国营芜湖机械厂 | MOSFET service life prediction method based on ARIMA model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113987932A (en) | MOSFET service life prediction method based on time series model | |
| CN112149860A (en) | Automatic anomaly detection method and system | |
| CN111465866B (en) | Sensor fault detection using paired sample correlation | |
| CN116628616A (en) | Data processing method and system for high-power charging energy | |
| CN116957120A (en) | Device state history trend anomaly prediction method based on data analysis | |
| CN117150244B (en) | Intelligent power distribution cabinet state monitoring method and system based on electrical parameter analysis | |
| KR101348635B1 (en) | Diagnosis apparatus and methof for broken rotor bar in induction motors | |
| CN116990683B (en) | Driving motor locked rotor detection system and detection method based on electric variable | |
| CN107544008B (en) | Vehicle-mounted IGBT state monitoring method and device | |
| CN113987900A (en) | IGBT service life prediction method based on extended Kalman particle filter | |
| CN114186492A (en) | Service life prediction method of SiC MOSFET device based on gated cyclic unit neural network | |
| CN113670428B (en) | Transformer vibration online abnormality detection method | |
| CN113945818A (en) | MOSFET service life prediction method | |
| Zheng et al. | Reliability analysis of multi-stage degradation with stage-varying noises based on the nonlinear Wiener process | |
| CN114076882B (en) | MMC submodule IGBT open-circuit fault diagnosis method based on model prediction | |
| CN115219842A (en) | Electromechanical device fault location and alarm protection device | |
| CN111722060B (en) | Distribution line early fault severity evaluation method based on waveform characteristics | |
| CN113919116A (en) | IGBT remaining service life prediction method based on GARCH model | |
| CN112834942A (en) | Battery management system service life testing method and device based on temperature alternation test | |
| CN113987899A (en) | IGBT service life prediction method based on unscented Kalman particle filter | |
| CN112485629B (en) | IGBT converter health assessment method based on harmonic analysis | |
| Chen et al. | Generalized CCA with applications for fault detection and estimation | |
| CN113794413B (en) | Permanent magnet motor driving system current sensor fault type identification method and device | |
| Wu et al. | A remaining useful life prediction method of SiC MOSFET considering failure threshold uncertainty | |
| CN116773889A (en) | Health management system and method for mining power electronic equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |