A chip life prediction system and method using artificial intelligence technology

Abstract

The present invention discloses a chip life prediction system and method using artificial intelligence technology, including step S100: respectively obtaining simulated aging trajectory curves of chips of different specifications and models when life data is obtained before leaving the factory; respectively performing use time stage processing on the simulated aging trajectory curves of different chips; step S200: performing curve fitting based on the aging characteristic values ​​of each chip at different use time stages to obtain the aging characteristic value trajectory curve of each chip; step S300: performing initial matching judgment to obtain a matching judgment result, the matching judgment result includes the need to adjust the life prediction time of the chip to be tested, the need not to adjust the life prediction time of the chip to be tested, and the need to re-match the simulated aging trajectory curve of the chip to be tested; step S400: verifying the matching judgment result; step S500: matching and predicting the life of the chip to be tested according to the verified matching judgment result.

Description

Technical Field

[0001] The present invention relates to the technical field of artificial intelligence equipment life prediction, and specifically to a chip life prediction system and method using artificial intelligence technology. Background Art

[0002] In the prior art, when predicting the life of a chip, a corresponding life aging trajectory curve is often obtained based on the historical chip usage, and the data of the same specifications and models newly put into use are also consistently predicted by referring to the previously obtained life aging trajectory curve; however, in actual use, the usage scenarios of chips are often complex and diverse, because the actual working environments required for different chips are also different, and the impact of different external environments or different human operating habits on the service life of the chip is also different; some chips age prematurely compared to other chips of the same specifications and models due to long-term harsh environments or long-term improper human operations; some chips may also have their performance fundamentally destroyed due to long-term harsh environments or long-term improper human operations, so at these times, if these chips are predicted according to the original aging curve trajectory, the results obtained are often inaccurate and have no reference value. Summary of the invention

[0003] The purpose of the present invention is to provide a chip life prediction system and method using artificial intelligence technology to solve the problems raised in the above background technology.

[0004] In order to solve the above technical problems, the present invention provides the following technical solutions: a chip life prediction method using artificial intelligence technology, characterized in that the prediction method includes:

[0005] Step S100: The prediction system obtains the simulated aging trajectory curves of different specifications and models when the life data is obtained before leaving the factory; the simulated aging trajectory curve refers to the trajectory curve of the chip performance parameter value obtained under the standard supply condition with the decay change of the use time; the simulated aging trajectory curves of different chips are respectively processed by the use time to obtain the segmented periods of the use time of different chips and the aging characteristic values ​​corresponding to different chips in each segmented period;

[0006] Step S200: performing curve fitting based on the aging characteristic values ​​of each chip at different usage durations to obtain an aging characteristic value trajectory curve of each chip;

[0007] Step S300: The prediction system performs initial matching judgment based on the information of the chip under test in combination with the corresponding simulation aging trajectory curve and the aging characteristic value trajectory curve to obtain a matching judgment result, wherein the matching judgment result includes whether the life prediction duration of the chip under test needs to be adjusted, whether the life prediction duration of the chip under test needs not to be adjusted, and whether the simulation aging trajectory curve of the chip under test needs to be re-matched;

[0008] Step S400: verifying the matching judgment result that requires adjustment of the life prediction duration of the chip under test and the matching judgment result that requires re-matching of the simulated aging trajectory curve of the chip under test;

[0009] Step S500: Match and predict the life of the chip to be tested according to the verified matching judgment result.

[0010] Furthermore, step S100 includes:

[0011] Step S101: Assume that the service life of the chip of the i-th specification model is T i , T i The unit is A; based on T i Randomly set several division time periods t i , a i The setting method of the value corresponds to a method of dividing the usage time, t i The unit is A -1 ; Based on t i The service life T of the chip of the i-th specification model i Divide into N usage duration stages; calculate each interval A in k usage duration stages -2 The performance change rate p obtained by the unit; where A -1 Indicates the next level unit of A; A -2 Indicates the next two units of A, A -1 The next level unit;

[0012] Step S102: In the x-th usage duration division method, the number of performance change rates in the k-th usage duration stage is Q K , the jth performance change rate in the kth usage duration period is The average value of all performance change rates during the k-th usage duration period is

[0013] Step S103: Calculate the square sum of deviations in each usage duration stage in the xth usage duration division method Calculate the sum of squares of deviations of all usage duration stages in the xth usage duration division method Calculate the variance fitting goodness of the x-th usage time division method

[0014] Step S104: According to steps S101 to S103, the variance fitting goodness corresponding to different usage time division methods for chips of different specifications and models is calculated respectively, and the usage time division method with the variance fitting goodness closest to 1 is selected as the best division method for each specification and model chip; the division time period corresponding to the best division method is used as the segmentation period for the usage time stage processing of the chip of this specification and model, and several usage time stages are obtained, and the average value of all performance change rates in each usage time stage is used as the aging characteristic value corresponding to each segmentation period;

[0015] The above-mentioned use time stage processing of the simulated aging trajectory curves of various chips is to obtain the average aging characteristics of various chips in different time periods; because the chip is a long-life component, but the aging speed of a brand new chip, a chip that has been used for a period of time, and a chip that has been used for more than half of the time is different. In the present application, the method of using time stage processing for different chips uses clustering thinking, so that in the final division method, the performance change rate within the same use time stage has the greatest similarity, while the performance change rate between different use time stages has the greatest difference; according to the performance change rate of each use time stage, the stage locking of the chip life prediction can be achieved, so that the final chip life prediction result is more accurate.

[0016] Further, step S200 includes:

[0017] Step S201: The final segment period is recorded as h, and the number of duration stages is recorded as E. h , the aging characteristic value of the i-th usage duration stage is R i ; The time range of the i-th usage duration stage is [(i-1)h, ih]; the time range of each usage duration stage is the horizontal axis, and the aging characteristic value of each usage duration stage is the vertical axis, and E is obtained. h coordinate points, each coordinate point is in the form of: [[(i-1)h, ih], R i ];

[0018] Step S202: E corresponding to each chip h The curve fitting is performed on the coordinate points to obtain the trajectory curve of the aging characteristic value of each chip in different use time stages;

[0019] The trajectory curves of the aging characteristic values of each of the above-obtained chips in different usage duration stages are exactly the curves obtained after performing a stage processing on the simulated aging trajectory curves of the chips, and these curves can better reflect the different life aging characteristics of different chips, and can also reflect the performance quality between different chips. Similar to a chip with better performance compared to a chip with poorer performance, perhaps their initial aging characteristic values are close at the beginning of the usage duration stage, but the time required for the chip with good performance to enter the next sub-aging characteristic value from the initial aging characteristic value will be longer than the time required for the chip with poor performance to enter the next sub-aging characteristic value from the initial aging characteristic value; that is to say, the chip with good performance can maintain the initial aging characteristic value at the beginning for a longer time period than the chip with poor performance maintains the initial aging characteristic value at the beginning; or perhaps the time required for them to enter the next sub-aging characteristic value from the initial aging characteristic value is close, but the difference between the initial aging characteristic values is large, that is, the initial aging characteristic value of the chip with good performance is much smaller than the initial aging characteristic value of the chip with poor performance; therefore, the trajectory curves of the aging characteristic values of chips of each specification model can be used as an intuitive characteristic curve for distinguishing chips of different performances and different specification models.

[0020] Further, step S300 includes:

[0021] Step S301: Assume that the simulated aging trajectory curve L1 and the aging characteristic value trajectory curve L2 are initially matched according to the specification model of the chip to be tested; obtain the actual usage duration Ta that the chip to be tested has been actually put into use until the current moment of life prediction, the actual performance parameter value Ya until the current moment of life prediction, the sub-period Ra to which Ta belongs in the curve L2, and the aging characteristic value Ua corresponding to the sub-period Ra; obtain the standard performance parameter value Yb according to Ta in the curve L1;

[0022] Step S302: Obtain the usage duration Tb according to Ya in the curve L1, obtain the sub-period Rb to which Tb belongs in the curve L2, and the aging characteristic value Ub corresponding to the sub-period Rb; obtain the standard performance parameter value Yb according to Ta in the curve L1;

[0023] Step S302: If Ya > Yb, determine that the life prediction duration of the chip to be tested needs to be adjusted; if Ya = Yb, determine that the life prediction duration of the chip to be tested does not need to be adjusted; if Ya < Yb, and Ra = Rb or Ra ≠ Rb, that is the value of is included in (Ua, Ub), determine that the life prediction duration of the chip to be tested needs to be adjusted; if Ya < Yb, and Ra ≠ Rb, that is the value of is not included in (Ua, Ub), determine that the simulated aging trajectory curve of the chip to be tested needs to be rematched;

[0024] Ya > Yb means that the actual performance parameter value Ya is greater than the performance parameter value Yb obtained according to the standard simulated aging trajectory curve. That is, the current chip is in a healthy state, and it may be due to less usage frequency and proper storage, and its service life will be longer than the life predicted according to the actual usage duration. Ya = Yb means that the actual performance parameter value Ya is the same as the performance parameter value Yb obtained according to the standard simulated aging trajectory curve. That is, the current chip is in a healthy state, and the chip's performance parameter values change completely according to the change law of the standard simulated aging trajectory curve. Ya < Yb means that the actual performance parameter value Ya is less than the performance parameter value Yb obtained according to the standard simulated aging trajectory curve. That is, the current chip may be in an unhealthy state, and it may be due to high usage frequency and improper storage, and its service life will be shorter than the life predicted according to the actual usage duration. That is, for chips of the same specification model, although they have been used for a short time, their aging characteristic values are equal to those that appear after a long time of use. If Ra = Rb or Ra ≠ Rb, it means that the chip has indeed experienced premature aging, but it still changes its performance parameter values according to the change law of the standard simulated aging trajectory curve, only showing stage-jumping aging. Ya < Yb and Ra ≠ Rb means that the chip has indeed experienced premature aging, but it does not change its performance parameter values according to the change law of the standard simulated aging trajectory curve. It may be due to high usage frequency and improper storage, resulting in a performance transformation, and its aging law becomes the same as that of another specification model with poorer performance. This means that the life prediction based on the original standard simulated aging trajectory curve matched according to the specification model is not very accurate, and the simulated aging trajectory curve of the chip to be tested needs to be rematched.

[0025] Further, step S400 includes:

[0026] Step S401: Based on artificial intelligence, capture several attention operation matters that will damage the chip performance during the actual use of chips with the same specification model as the chip to be tested.

[0027] Step S402: If a matching judgment result that requires adjustment of the life prediction duration appears during the life prediction of a certain chip to be tested, retrieve the historical operation records of the certain chip to be tested before the life prediction. If the historical operation records show that several attention operation matters have been implemented, the verification result is passed.

[0028] Step S403: If a matching judgment result that requires re - matching the simulation aging trajectory curve appears during the life prediction of a certain chip to be tested, retrieve the historical operation records of the chip to be tested before life prediction. If there are several attention operation items implemented in the historical operation records, and the implementation times are greater than the implementation times of the several attention operation items in step S402, or the duration of any one implementation record of the attention operation item is greater than the duration of the same attention operation item implemented in step S402, then the verification result is passed.

[0029] Further, the process of step S500 for matching and predicting the life of the chip to be tested includes:

[0030] When Ya > Yb, obtain the usage duration Tb from curve L1 according to Ya, and use the usage duration Tb as the actual starting duration for life prediction of the chip to be tested. That is, adjust the usage duration Ta of the chip to be tested along curve L1 to start life prediction from Tb to obtain the life prediction result;

[0031] The above means that the current chip is in a healthy state, and it may be due to less usage frequency and proper storage, so its service life is longer than the life estimated according to the actual usage duration. Its accurate starting life time needs to be moved forward;

[0032] When Ya = Yb, perform life prediction on the chip to be tested along curve L1 starting from Ta to obtain the life prediction result;

[0033] The above means that the current chip is in a healthy state, and the chip completely follows the change law of the standard simulation aging trajectory curve for the change of performance parameter values;

[0034] When Ya < Yb, and Ra = Rb or Ra ≠ Rb, At this time, use the usage duration Tb as the actual starting duration for life prediction of the chip to be tested. That is, adjust the usage duration Ta of the chip to be tested along curve L1 to start life prediction from Tb to obtain the life prediction result;

[0035] The above means that the chip has an early aging phenomenon, but it still follows the change law of the standard simulation aging trajectory curve for the change of performance parameter values, only there is a jump - stage aging. Its accurate starting life time needs to be moved backward;

[0036] When Ya < Yb, and Ra ≠ Rb, At this time, substitute the actual usage duration Ta until the current moment of life prediction and the actual performance parameter value Ya until the current moment of life prediction of the chip to be tested into the simulation aging trajectory curve L1 of the i - th specification and model of the chip iAnd the simulated aging trajectory curve L1 i The corresponding aging characteristic value trajectory curve L2 i According to Ta, respectively, in the curve L1 i The corresponding standard performance parameter value Yb is obtained i ; Filter out those that meet Yb i -Ya, the deviation value obtained between them is less than the deviation threshold, and the simulation aging trajectory curve set S1 is obtained; the total performance parameter value Yq of the chip to be tested when it leaves the factory is obtained, and the calculation is performed The corresponding aging characteristic value trajectory curve L2 in the simulation aging trajectory curve set S1 i Filter out the aging characteristic values ​​and The same simulation aging trajectory curves are obtained to obtain the simulation aging trajectory curve set S2, and the For the segmented periods to which the aging characteristic values ​​are equal, the simulated aging trajectory curve with the farthest interval between the segmented period range and Ta is determined as the simulated aging trajectory curve that is re-matched with the life prediction of the chip to be tested;

[0037] The above means that the chip has aged prematurely, but its performance parameter values ​​have not changed according to the changing rules of the standard simulation aging trajectory curve. The performance has changed due to high frequency of use and improper storage. To find a simulation aging trajectory curve that matches the life prediction for it, first find a simulation aging trajectory curve with a higher curve matching degree based on the actually measured and obtained Ya and Ta, and then further match the aging characteristic values ​​obtained from the aging characteristic value trajectory curves corresponding to these simulation aging trajectory curves, that is, find another specification model chip for the chip to be tested with the highest matching degree between the simulation aging trajectory curve and the aging characteristic value trajectory curve, the performance of this other specification model chip is consistent with the actual performance of the chip to be tested after the performance change, and the life prediction result of the chip to be tested according to the standard simulation aging trajectory curve of the chip with this specification model is more accurate.

[0038] In order to better implement the above method, a chip life prediction system using artificial intelligence technology is also proposed. The prediction system includes: an aging data acquisition module, an aging cycle stage processing module, a characteristic value processing module, an initial matching judgment module, a verification module, and a matching prediction module;

[0039] Aging data acquisition module, used to obtain the simulated aging trajectory curve of chips of different specifications and models before they leave the factory;

[0040] The aging cycle phase processing module is used to receive the data from the aging data acquisition module and perform phase processing on the simulated aging trajectory curves of different chips according to the usage time;

[0041] The characteristic value processing module is used to receive the data in the aging cycle stage processing module, obtain the aging characteristic value of each chip at different use time stages, and perform curve fitting to obtain the aging characteristic value trajectory curve of each chip;

[0042] An initial matching judgment module is used to perform initial matching judgment based on the information of the chip under test in combination with the corresponding simulation aging trajectory curve and the aging characteristic value trajectory curve to obtain a matching judgment result;

[0043] A verification module, used for receiving data from the initial matching judgment module and performing data verification on the data;

[0044] The matching prediction module is used to receive data from the verification module and complete the life matching and prediction of the chip to be tested.

[0045] Furthermore, the aging cycle stage processing module includes: a time period setting unit, a calculation unit, and a optimization unit;

[0046] A division time period setting unit, used for randomly setting a number of division time periods based on the lifespans of chips of different specifications and models;

[0047] The calculation unit is used to complete the calculation of the performance change rate, the sum of squares of deviations, and the goodness of fit of variance involved in the aging cycle stage processing;

[0048] The optimization unit is used to receive the data in the calculation unit and respectively complete the selection of the optimal division time period for chips of different specifications and models.

[0049] Further, the initial matching judgment module includes: a data capturing unit, a judgment unit, and a matching analysis unit;

[0050] A data capture unit, used to capture all judgment data required in the life prediction process of the chip under test;

[0051] A judgment unit, used for receiving data from the data capture unit and performing matching judgment based on the data;

[0052] The matching analysis unit is used to receive the data in the judgment unit, and based on the data, obtain the matching analysis results of the life prediction time adjustment required for the chip under test and the matching analysis results of the re-matching of the simulated aging trajectory curve of the chip under test.

[0053] Compared with the prior art, the beneficial effect achieved by the present invention is that: in the process of predicting the chip life, the present invention takes into account the actual use of the chip, the influence of the actual working environment of the chip or human operating habits on the chip's service life, which may cause premature aging or fundamental destruction of performance due to the complexity and diversity of the chip's usage scenarios. In the process of predicting the life of the chip to be tested, the precise life start time is located or the appropriate life prediction-related curve trajectory is re-matched, thereby improving the accuracy and credibility of the final chip life prediction result. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0055] Figure 1 It is a flow chart of a chip life prediction method using artificial intelligence technology of the present invention;

[0056] Figure 2 It is a schematic diagram of the structure of a chip life prediction system using artificial intelligence technology in the present invention;

[0057] Figure 3 This is an embodiment of a chip life prediction method using artificial intelligence technology of the present invention Figure 1 ;

[0058] Figure 4 This is an embodiment of a chip life prediction method using artificial intelligence technology of the present invention Figure 2 . DETAILED DESCRIPTION

[0059] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0060] See also Figure 1-Figure 4 The present invention provides a technical solution: a chip life prediction method using artificial intelligence technology, characterized in that the prediction method includes:

[0061] Step S100: The prediction system obtains the simulated aging trajectory curves of different specifications and models when the life data is obtained before leaving the factory; the simulated aging trajectory curve refers to the trajectory curve of the chip performance parameter value obtained under the standard supply condition with the decay change of the use time; the simulated aging trajectory curves of different chips are respectively processed by the use time to obtain the segmented periods of the use time of different chips and the aging characteristic values ​​corresponding to different chips in each segmented period;

[0062] Wherein, step S100 includes:

[0063] Step S101: Assume that the service life of the chip of the i-th specification model is T i , T i The unit is A; based on T i Randomly set several division time periods t i , a i The setting method of the value corresponds to a method of dividing the usage time, t i The unit is A -1 ; Based on t i The service life T of the chip of the i-th specification model i Divide into N usage duration stages; calculate each interval A in k usage duration stages -2 The performance change rate p obtained by the unit; where A -1 Indicates the next level unit of A; A -2 Indicates the next two units of A, A -1 The next level unit;

[0064] For example, in this embodiment, T i The unit is year, then t i Unit A -1 For the moon, A -2 For God; also because of T i The unit is month, then t i Unit A -1 For God, A -2 for hours;

[0065] Step S102: In the x-th usage duration division method, the number of performance change rates in the k-th usage duration stage is Q K , the jth performance change rate in the kth usage duration period is The average value of all performance change rates during the k-th usage duration period is

[0066] Step S103: Calculate the square sum of deviations in each usage duration stage in the xth usage duration division method Calculate the sum of squares of deviations of all usage duration stages in the xth usage duration division method Calculate the variance fitting goodness of the x-th usage time division method

[0067] Step S104: According to steps S101 to S103, the variance fitting goodness corresponding to different usage time division methods for chips of different specifications and models is calculated respectively, and the usage time division method with the variance fitting goodness closest to 1 is selected as the best division method for each specification and model chip; the division time period corresponding to the best division method is used as the segmentation period for the usage time stage processing of the chip of this specification and model, and several usage time stages are obtained, and the average value of all performance change rates in each usage time stage is used as the aging characteristic value corresponding to each segmentation period;

[0068] Step S200: performing curve fitting based on the aging characteristic values ​​of each chip at different usage durations to obtain an aging characteristic value trajectory curve of each chip;

[0069] Wherein, step S200 includes:

[0070] Step S201: The final segment period is recorded as h, and the number of duration stages is recorded as E. h , the aging characteristic value of the i-th usage duration stage is R i ; The time range of the i-th usage duration stage is [(i-1)h, ih]; the time range of each usage duration stage is the horizontal axis, and the aging characteristic value of each usage duration stage is the vertical axis, and E is obtained. h coordinate points, each coordinate point is in the form of: [[(i-1)h, ih], R i ];

[0071] Step S202: E corresponding to each chip h The curve fitting is performed on the coordinate points to obtain the trajectory curve of the aging characteristic value of each chip in different use time stages;

[0072] For example, for a chip of a certain model specification, the simulation aging trajectory curve when it obtains the life data before leaving the factory shows that its total life duration is 15 years. The best division method is to take every 3 years as a usage duration stage. Therefore, the number of usage duration stages is 5, which are (0, 3), (3, 6), (6, 9), (9, 12), and (12, 15) respectively. In the first usage duration stage (0, 3), its aging characteristic value is 0.1%. In the second usage duration stage (3, 6), its aging characteristic value is 0.14%. In the third usage duration stage (6, 9), its aging characteristic value is 0.19%. In the fourth usage duration stage (9, 12), its aging characteristic value is 0.21%. In the fifth usage duration stage (12, 15), its aging characteristic value is 0.25%. Then the trajectory curve of its aging characteristic value Figure 3 is shown as

[0073] Step S300: The prediction system makes an initial matching judgment based on the information of the chip to be tested, combined with the corresponding simulation aging trajectory curve and the aging characteristic value trajectory curve, to obtain a matching judgment result. The matching judgment result includes that the life prediction duration of the chip to be tested needs to be adjusted, the life prediction duration of the chip to be tested does not need to be adjusted, and the simulation aging trajectory curve of the chip to be tested needs to be rematched;

[0074] Among them, step S300 includes:

[0075] Step S301: Assume that the simulation aging trajectory curve L1 and the aging characteristic value trajectory curve L2 are initially matched according to the specification model of the chip to be tested. Obtain the actual usage duration Ta of the chip to be tested until the current moment of life prediction, the actual performance parameter value Ya until the current moment of life prediction, the sub-period Ra to which Ta belongs in curve L2, and the aging characteristic value Ua corresponding to the sub-period Ra. Obtain the standard performance parameter value Yb according to Ta in curve L1;

[0076] Step S302: Obtain the usage duration Tb according to Ya in curve L1, obtain the sub-period Rb to which Tb belongs in curve L2, and the aging characteristic value Ub corresponding to the sub-period Rb. Obtain the standard performance parameter value Yb according to Ta in curve L1;

[0077] Step S302: If Ya > Yb, it is determined that the life prediction duration of the chip to be tested needs to be adjusted; if Ya = Yb, the life prediction duration of the chip to be tested does not need to be adjusted; if Ya < Yb, and Ra = Rb or Ra ≠ Rb, that is the value of is included in (Ua, Ub), it is determined that the life prediction duration of the chip to be tested needs to be adjusted; if Ya < Yb, and Ra ≠ Rb, that is The value of is not included in (Ua, Ub), and it is determined that the simulated aging trajectory curve of the chip under test needs to be re-matched;

[0078] Step S400: verifying the matching judgment result that requires adjustment of the life prediction duration of the chip under test and the matching judgment result that requires re-matching of the simulated aging trajectory curve of the chip under test;

[0079] Wherein, step S400 includes:

[0080] Step S401: capturing, based on artificial intelligence, several precautions for operation that may damage chip performance during actual use of a chip with the same specifications and model as the chip to be tested;

[0081] Step S402: If a matching judgment result that requires adjustment of the life prediction duration is found when performing life prediction on a chip under test, the historical operation record of the chip under test during the life prediction is retrieved. If the historical operation record shows that a number of precautionary operations have been implemented, the verification result is passed.

[0082] Step S403: If a matching judgment result that requires rematching of the simulated aging trajectory curve is found when predicting the life of a chip to be tested, retrieve the historical operation records of the chip to be tested before the life prediction. If the historical operation records show that a number of attention operations have been implemented, and the number of implementations is greater than the number of implementations of a number of attention operations in step S402, or the duration of the attention operations in any one implementation record is greater than the duration of the same attention operations in step S402, then the verification result is passed;

[0083] Step S500: matching and predicting the life of the chip to be tested according to the verified matching judgment result;

[0084] The process of matching and predicting the life of the chip to be tested in step S500 includes:

[0085] When Ya>Yb, the usage time Tb is obtained according to Ya in the curve L1, and the usage time Tb is used as the actual starting time for the life prediction of the chip to be tested, that is, the usage time Ta of the chip to be tested is adjusted along the curve L1 to start the life prediction from Tb to obtain the life prediction result;

[0086] For example, if Yb is 76 and Ta is 3 years, if the Ya of the chip to be tested is 78, then according to Figure 4As shown, corresponding to the performance parameter value of 78 is t1, and t1 is two years and three months; adjust the usage duration of 3 years for the chip under test along curve L1 to start from two years and three months for life prediction to obtain the life prediction result. If the simulation aging trajectory curve shows that its total life duration is 15 years when the chip under test obtains life data before leaving the factory, then the remaining life of the chip under test is 12 years and nine months;

[0087] When Ya = Yb, perform life prediction on the chip under test along curve L1 starting from Ta to obtain the life prediction result;

[0088] When Ya < Yb, and Ra = Rb or Ra ≠ Rb, use the usage duration Tb as the actual starting duration for life prediction of the chip under test, that is, adjust the usage duration Ta of the chip under test along curve L1 to start from Tb for life prediction to obtain the life prediction result;

[0089] For example, Yb is 76, Ta is 3 years. If the Ya of the chip under test is 73, then according to Figure 4 as shown, corresponding to the performance parameter value of 73 is t2, and t2 is three years and three months; Ra = Rb or Ra ≠ Rb, adjust the usage duration of 3 years for the chip under test along curve L1 to start from three years and three months for life prediction to obtain the life prediction result. If the simulation aging trajectory curve shows that its total life duration is 15 years when the chip under test obtains life data before leaving the factory, then the remaining life of the chip under test is 11 years and nine months;

[0090] When Ya < Yb, and Ra ≠ Rb, substitute the actual usage duration Ta of the chip under test until the current moment of life prediction and the actual performance parameter value Ya until the current moment of life prediction into the simulation aging trajectory curve L1 of the i-th specification and model chip i and the aging characteristic value trajectory curve L2 i corresponding to the simulation aging trajectory curve L1 i respectively; obtain the corresponding standard performance parameter value Yb in curve L1 i according to Ta i ; screen out several simulation aging trajectory curves that satisfy the deviation value obtained between Yb i - Ya less than the deviation threshold to obtain the simulation aging trajectory curve set S1; obtain the total performance parameter value Yq when the chip under test leaves the factory, and calculate screen out the aging characteristic value trajectory curve L2 i corresponding to each in the simulation aging trajectory curve set S1 The same simulation aging trajectory curves are obtained to obtain the simulation aging trajectory curve set S2, and the For the segmented periods to which the aging characteristic values ​​are equal, the simulated aging trajectory curve with the farthest interval between the segmented period range and Ta is determined as the simulated aging trajectory curve that is re-matched with the life prediction of the chip to be tested.

[0091] In order to better implement the above method, a chip life prediction system using artificial intelligence technology is also proposed. The prediction system includes: an aging data acquisition module, an aging cycle stage processing module, a characteristic value processing module, an initial matching judgment module, a verification module, and a matching prediction module;

[0092] Aging data acquisition module, used to obtain the simulated aging trajectory curve of chips of different specifications and models before they leave the factory;

[0093] The aging cycle phase processing module is used to receive the data from the aging data acquisition module and perform phase processing on the simulated aging trajectory curves of different chips according to the usage time;

[0094] The aging cycle stage processing module includes: a time period setting unit, a calculation unit, and a selection unit;

[0095] A division time period setting unit, used for randomly setting a number of division time periods based on the lifespans of chips of different specifications and models;

[0096] The calculation unit is used to complete the calculation of the performance change rate, the sum of squares of deviations, and the goodness of fit of variance involved in the aging cycle stage processing;

[0097] The characteristic value processing module is used to receive the data in the aging cycle stage processing module, obtain the aging characteristic value of each chip at different use time stages, and perform curve fitting to obtain the aging characteristic value trajectory curve of each chip;

[0098] An initial matching judgment module is used to perform initial matching judgment based on the information of the chip under test in combination with the corresponding simulation aging trajectory curve and the aging characteristic value trajectory curve to obtain a matching judgment result;

[0099] A verification module, used for receiving data from the initial matching judgment module and performing data verification on the data;

[0100] The matching prediction module is used to receive the data in the verification module and complete the life matching and prediction of the chip to be tested;

[0101] The optimization unit is used to receive the data in the calculation unit and select the optimal division time period for chips of different specifications and models;

[0102] The initial matching judgment module includes: a data capturing unit, a judgment unit, and a matching analysis unit;

[0103] A data capture unit, used to capture all judgment data required in the life prediction process of the chip under test;

[0104] A judgment unit, used for receiving data from the data capture unit and performing matching judgment based on the data;

[0105] The matching analysis unit is used to receive the data in the judgment unit, and based on the data, obtain the matching analysis results of the life prediction time adjustment required for the chip under test and the matching analysis results of the re-matching of the simulated aging trajectory curve of the chip under test.

[0106] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0107] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A chip life prediction method using artificial intelligence technology, characterized in that: The prediction method comprises: Step S100: The prediction system obtains simulated aging trajectory curves of chips of different specifications and models when obtaining life data before leaving the factory; the simulated aging trajectory curve refers to a trajectory curve of the chip performance parameter value obtained under the standard supply condition and the decay change of the chip performance parameter value with the use time; the simulated aging trajectory curves of different chips are respectively processed by using time to obtain the segmented periods of the different chips and the aging characteristic values ​​corresponding to the different chips in each segmented period; Step S200: performing curve fitting based on the aging characteristic values ​​of each chip at different usage durations to obtain an aging characteristic value trajectory curve of each chip; Step S300: The prediction system performs initial matching judgment based on the information of the chip under test in combination with the corresponding simulation aging trajectory curve and the aging characteristic value trajectory curve to obtain a matching judgment result, wherein the matching judgment result includes whether the life prediction duration of the chip under test needs to be adjusted, whether the life prediction duration of the chip under test needs not to be adjusted, and whether the simulation aging trajectory curve of the chip under test needs to be re-matched; Step S400: verifying the matching judgment result that requires adjustment of the life prediction duration of the chip under test and the matching judgment result that requires re-matching of the simulated aging trajectory curve of the chip under test; Step S500: matching and predicting the life of the chip to be tested according to the verified matching judgment result; The step S100 includes: Step S101: Assume that the service life of the chip of the i-th specification model is T i , T i The unit is A; based on T i Randomly set several division time periods t i , a i The setting method of the value corresponds to a method of dividing the usage time, t i The unit is A -1 ; Based on t i The service life T of the chip of the i-th specification model i Divide into N usage duration stages; calculate each interval A in k usage duration stages -2 The performance change rate p obtained by the unit; where A -1 Indicates the next level unit of A; A -2 Indicates the next two units of A, A -1 The next level unit; Step S102: In the x-th usage duration division method, the number of performance change rates in the k-th usage duration stage is Q k , the jth performance change rate in the kth usage duration period is The average value of all performance change rates during the k-th usage duration period is Step S103: Calculate the square sum of deviations in each usage duration stage in the x-th usage duration division method: Calculate the sum of squares of deviations of all usage duration stages in the xth usage duration division method Calculate the variance fitting goodness of the x-th usage time division method Step S104: According to steps S101 to S103, respectively calculate the variance fitting goodness of each method for dividing the usage time of chips of different specifications and models, and select the usage time division method with the variance fitting goodness of closest to 1 as the optimal division method for each specification and model of chip; use the division time period corresponding to the optimal division method as the segmentation period for dividing the usage time of the chip of this specification and model into stages, and obtain several usage time stages, and use the average value of all performance change rates in each usage time stage as the aging characteristic value corresponding to each segmentation period.

2. According to claim 1, a chip life prediction method using artificial intelligence technology is characterized in that: The step S200 includes: Step S201: The final segment period is recorded as h, and the number of duration stages is recorded as E. h , the aging characteristic value of the i-th usage duration stage is R i ; The time range of the i-th usage duration stage is [(i-1)h, ih]; the time range of each usage duration stage is the horizontal axis, and the aging characteristic value of each usage duration stage is the vertical axis, and E is obtained. h coordinate points, each coordinate point is in the form of: [[(i-1)h, ih], R i ]; Step S202: E corresponding to each chip h The curve fitting is performed on the coordinate points to obtain the trajectory curve of the aging characteristic value of each chip in different usage time stages.

3. The chip life prediction method using artificial intelligence technology according to claim 1 is characterized in that: The step S300 includes: Step S301: Assume that the specification model of the chip to be tested is initially matched to the simulation aging trajectory curve L1 and the aging characteristic value trajectory curve L2; obtain the actual use time Ta of the chip to be tested until the current time of life prediction, the actual performance parameter value Ya until the current time of life prediction, the segment period Ra to which Ta belongs in the curve L2, and the aging characteristic value Ua corresponding to the segment period Ra; obtain the standard performance parameter value Yb in the curve L1 according to Ta; Step S302: Obtain the usage time Tb in the curve L1 according to Ya, obtain the segment period Rb to which Tb belongs in the curve L2, and the aging characteristic value Ub corresponding to the segment period Rb; Step S303: If Ya > Yb, it is determined that the life prediction duration of the chip under test needs to be adjusted; if Ya = Yb, it is determined that the life prediction duration of the chip under test does not need to be adjusted; if Ya < Yb and Ra = Rb, that is the value of is included in (Ua, Ub), it is determined that the life prediction duration of the chip under test needs to be adjusted; If Ya < Yb and Ra ≠ Rb, that is the value of is not included in (Ua, Ub), it is determined that the simulation aging trajectory curve of the chip to be tested needs to be rematched.

4. The chip life prediction method using artificial intelligence technology according to claim 1 is characterized in that: The step S400 includes: Step S401: capturing, based on artificial intelligence, several precautions for operation that may damage chip performance during actual use of a chip with the same specifications and model as the chip to be tested; Step S402: If a matching judgment result that requires adjustment of the life prediction duration is found when performing life prediction on a chip under test, retrieve the historical operation record of the chip under test during the life prediction. If the historical operation record shows that a number of precautionary operations have been implemented, the verification result is passed. Step S403: If a matching judgment result that requires re-matching of the simulated aging trajectory curve is found during life prediction for a chip to be tested, retrieve the historical operation records of the chip to be tested during the life prediction. If the historical operation records show that a number of precautionary operations have been implemented, and the number of implementations is greater than the number of implementations of a number of precautionary operations in step S402, or the duration of the precautionary operations in any implementation record is greater than the duration of the same precautionary operations in step S402, then the verification result is passed.

5. The chip life prediction method using artificial intelligence technology according to claim 3 is characterized in that: The process of matching and predicting the life of the chip to be tested in step S500 includes: When Ya>Yb, the usage time Tb is obtained according to Ya in the curve L1, and the usage time Tb is used as the actual starting time for the life prediction of the chip under test, that is, the usage time Ta of the chip under test is adjusted along the curve L1 to start the life prediction from Tb to obtain the life prediction result; When Ya=Yb, the life prediction of the chip to be tested is performed along the curve L1 starting from Ta to obtain a life prediction result; When Ya < Yb and Ra = Rb, the usage duration Tb is used as the actual starting duration for predicting the life of the chip to be measured. That is, the usage duration Ta of the chip to be measured is adjusted along the curve L1 to start from Tb for life prediction to obtain a life prediction result. When Ya < Yb and Ra ≠ Rb, the actually used duration Ta of the chip under test until the current moment of life prediction and the actual performance parameter value Ya until the current moment of life prediction are respectively substituted into the simulation aging trajectory curve L1 of the i-th specification and model of the chip i and the aging characteristic value trajectory curve L2 i corresponding to the simulation aging trajectory curve L1 i ; according to Ta, the corresponding standard performance parameter value Yb is obtained in the curve L1 i ; several simulation aging trajectory curves that satisfy the deviation value obtained between Yb i -Ya being less than the deviation threshold are selected to obtain the simulation aging trajectory curve set S1; the total performance parameter value Yq of the chip under test at the time of factory is obtained, and the aging characteristic value trajectory curves L2 corresponding to each in the simulation aging trajectory curve set S1 i are screened to select the simulation aging trajectory curves with the aging characteristic value equal to to obtain the simulation aging trajectory curve set S2, and the segmented periods to which the aging characteristic values equal to belong are respectively obtained, and the simulation aging trajectory curve with the largest distance between the range of the segmented period and Ta is determined as the simulation aging trajectory curve to be rematched with the chip under test for life prediction.​​ 6. A chip life prediction system using artificial intelligence technology applied to the chip life prediction method using artificial intelligence technology as claimed in any one of claims 1 to 5, characterized in that: The prediction system includes: an aging data acquisition module, an aging cycle stage processing module, a characteristic value processing module, an initial matching judgment module, a verification module, and a matching prediction module; The aging data acquisition module is used to obtain the simulated aging trajectory curve of chips of different specifications and models when obtaining life data before leaving the factory; The aging cycle stage processing module is used to receive the data in the aging data acquisition module and perform stage processing on the simulated aging trajectory curves of different chips according to the length of use; The characteristic value processing module is used to receive the data in the aging cycle stage processing module, obtain the aging characteristic value of each chip at different usage time stages, and perform curve fitting to obtain the aging characteristic value trajectory curve of each chip; The initial matching judgment module is used to perform initial matching judgment by combining the information of the chip to be tested with the corresponding simulation aging trajectory curve and the aging characteristic value trajectory curve to obtain a matching judgment result; The verification module is used to receive the data in the initial matching judgment module and perform data verification on the data; The matching prediction module is used to receive the data in the verification module and complete the life matching and prediction of the chip to be tested.

7. The chip life prediction system using artificial intelligence technology according to claim 6 is characterized in that: The aging cycle stage processing module includes: a time period setting unit, a calculation unit, and a optimization unit; The division time period setting unit is used to randomly set a number of division time periods based on the lifespans of chips of different specifications and models; The calculation unit is used to complete the calculation of the performance change rate, the calculation of the sum of squares of deviations, and the calculation of the variance goodness of fit involved in the aging cycle stage processing; The optimization unit is used to receive the data in the calculation unit and respectively complete the selection of the optimal division time period for chips of different specifications and models.

8. The chip life prediction system using artificial intelligence technology according to claim 6 is characterized in that: The initial matching judgment module includes: a data capturing unit, a judging unit, and a matching analysis unit; The data capture unit is used to capture all judgment data required in the life prediction process of the chip to be tested; The judgment unit is used to receive the data in the data capture unit and perform matching judgment based on the data; The matching analysis unit is used to receive the data in the judgment unit, and obtain the matching analysis results of the life prediction duration adjustment required for the chip under test and the matching analysis results of the re-matching of the simulated aging trajectory curve of the chip under test based on the data.

Images