CN111553611A - Method and device for obtaining design yield - Google Patents
Method and device for obtaining design yield Download PDFInfo
- Publication number
- CN111553611A CN111553611A CN202010381928.4A CN202010381928A CN111553611A CN 111553611 A CN111553611 A CN 111553611A CN 202010381928 A CN202010381928 A CN 202010381928A CN 111553611 A CN111553611 A CN 111553611A
- Authority
- CN
- China
- Prior art keywords
- input
- class
- parameter
- sampling points
- sampling
- 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.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Educational Administration (AREA)
- General Physics & Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- Economics (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- Computer Hardware Design (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Complex Calculations (AREA)
Abstract
The invention discloses a method and a device for obtaining design yield, which can quickly position a first class input parameter space which may not meet performance indexes by amplifying standard deviation, obtain a first class function based on sampling points corresponding to the first class input parameter space and corresponding performance measurement values, randomly sample based on the standard deviation of the original distribution of each input random parameter of a design object to obtain a second class sampling point, obtaining a performance estimation value by utilizing the first class function for the second class sampling points falling in the first class input parameter space, and determining the third class sampling points which do not accord with the performance index, and calculating the design yield of the design object which accords with the performance index in the corresponding overall random parameter space according to the number of the third type sampling points and the number of the second type sampling points, thereby realizing the improvement of the efficiency of obtaining the design yield which accords with the performance index.
Description
Technical Field
The invention relates to the technical field of computer simulation, in particular to a method and a device for acquiring design yield.
Background
For a design object with a high dimension of input random parameter space and a flattened input random parameter distribution, for example, a chip design is performed with the support of a modern integrated circuit manufacturing process, since the design object pursues an extremely high yield, such as 99.999999%, a yield of a Monte Carlo analysis (Monte Carlo Method) design object with a high sigma, or referred to as a design yield, is required at this time.
The conventional low-sigma Monte Carlo analysis method faces technical challenges in speed/time and accuracy when used for high-sigma Monte Carlo analysis, resulting in extremely inefficient analysis of design yield.
Disclosure of Invention
In view of this, the present invention discloses a method and an apparatus for obtaining a design yield, so as to achieve the purposes of improving the efficiency of obtaining the design yield meeting performance indexes and accelerating the speed of monte carlo analysis.
In order to achieve the purpose, the technical scheme disclosed by the invention is as follows:
in a first aspect, the present invention discloses a method for obtaining a design yield, the method comprising:
amplifying the standard deviation of each input random parameter of the design object, and randomly sampling each input random parameter based on the amplified standard deviation to obtain a plurality of first-class sampling points;
acquiring a performance measurement value of each first-type sampling point;
determining a first type of input parameter space which possibly does not accord with design indexes based on the performance measurement values of the first type of sampling points and corresponding performance index values;
obtaining a first class function representing the corresponding relation between parameter values possibly not conforming to the design index and performance measurement values based on sampling points corresponding to the first class input parameter space corresponding to each input random parameter and the corresponding performance measurement values;
based on the standard deviation of the original distribution of each input random parameter of the design object, randomly sampling to obtain a second class of sampling points, calculating the second class of sampling points falling in the first class of input parameter space by using the first class function to obtain a performance estimation value, and determining a third class of sampling points which do not accord with the design index according to the performance estimation value;
and calculating the design yield of the design object which meets the design index in the corresponding overall random parameter space according to the number of the third type of sampling points and the number of the second type of sampling points.
Optionally, the determining, based on the performance measurement value of the first type of sampling point and the corresponding performance index value, a first type of input parameter space that may not meet a design index includes:
selecting a sampling point set of which the performance measurement value is within a preset range from the first type of sampling points, wherein the preset range is obtained according to a performance index value corresponding to an input random parameter;
for any input random parameter in the design object, the value ranges of the input random parameters corresponding to all sampling points in the selected sampling point set;
and determining the value ranges corresponding to all the input random parameters in the design object as the first-class input parameter space.
Optionally, the obtaining a first-class function representing a correspondence between a parameter value that may not meet a design index and a performance measurement value based on a sampling point corresponding to the first-class input parameter space and the corresponding performance measurement value corresponding to each input random parameter includes:
subdividing the first-class input parameter space corresponding to each input random parameter to obtain a plurality of subdivided first-class input parameter spaces;
and performing function extraction on the sampling points corresponding to the plurality of subdivided first-class input parameter spaces and the corresponding performance measurement values to obtain a first-class function representing the corresponding relation between the parameter values possibly not conforming to the design index and the performance measurement values.
Optionally, before the amplifying the standard deviation of each input random parameter of the design object and randomly sampling each input random parameter based on the amplified standard deviation to obtain a plurality of first-class sampling points, the method further includes:
dividing the whole random parameter space corresponding to the design object to obtain a plurality of subdivided spaces so as to obtain the second type of sampling points and the third type of sampling points corresponding to each subdivided space;
calculating the design yield in the input random parameter space according to the number of the third type of sampling points and the number of the second type of sampling points, and the method comprises the following steps:
and calculating to obtain the design yield corresponding to the whole random parameter space based on the number of the second type of sampling points, the number of the third type of sampling points and the ratio of each subdivided space to the whole random parameter space corresponding to each subdivided space.
Optionally, the dividing the random parameter space to obtain each subdivision space includes:
and dividing the random parameter space by taking the parameter intermediate value point as a center and respectively taking the upper value boundary direction and the lower value boundary direction according to the sequence of the space from large to small to obtain a plurality of subdivided spaces.
Optionally, the step of amplifying the standard deviation of each input random parameter of the design object, and randomly sampling each input random parameter based on the amplified standard deviation to obtain a plurality of first-class sampling points includes:
amplifying the standard deviation of each input random parameter of the design object by using a plurality of different amplification factors respectively to obtain a plurality of amplified standard deviations;
and randomly sampling each input random parameter based on each amplified standard deviation to obtain a plurality of first-class sampling points.
Optionally, the process of determining the overall random parameter space corresponding to the design object is as follows:
acquiring the sensitivity of each input random parameter corresponding to the design object, and selecting the input random parameter with the sensitivity greater than or equal to a preset threshold value;
and determining the area range of the selected input random parameter as an integral random parameter space corresponding to the design object.
Optionally, the randomly sampling to obtain a second type of sampling points based on the standard deviation of the original distribution of each input random parameter of the design object includes:
randomly sampling any input random parameter of the design object, and judging whether the value of the input random parameter is in a first-class input parameter space corresponding to the input random parameter;
if the value of the input random parameter is in the corresponding first type of input parameter space, continuing to randomly sample the next input random parameter;
if the value of the input random parameter is not in the corresponding first-class input parameter space, marking the current sampling point outside the first-class input parameter space;
if the values of all input random parameters of the design object are in the corresponding first-class input parameter space, marking the current sampling point in the first-class input parameter space;
and all the sampling points marked in the first type of input parameter space and all the sampling points marked outside the first type of output parameter space form the second type of sampling points.
In a second aspect, the present invention discloses an apparatus for obtaining design yield, the apparatus comprising:
the amplifying unit is used for amplifying the standard deviation of each input random parameter of the design object and randomly sampling each input random parameter based on the amplified standard deviation to obtain a plurality of first-class sampling points;
the first acquisition unit is used for acquiring the performance measurement value of each first-class sampling point;
the determining unit is used for determining a first type of input parameter space which possibly does not accord with the design index based on the performance measurement value of the first type of sampling point and the corresponding performance index value;
a second obtaining unit, configured to obtain a first class function representing a correspondence between a parameter value that may not meet a design index and a performance measurement value based on a sampling point corresponding to the first class input parameter space and the corresponding performance measurement value corresponding to each input random parameter;
the sampling determining unit is used for performing random sampling to obtain second-class sampling points based on the standard deviation of the original distribution of each input random parameter of the design object, calculating the second-class sampling points in the first-class input parameter space by using the first-class function to obtain performance estimation values, and determining third-class sampling points which do not accord with performance indexes according to the performance estimation values;
and the calculating unit is used for calculating the design yield in the input random parameter space according to the number of the third type of sampling points and the number of the second type of sampling points.
Optionally, the second obtaining unit includes:
the subdivision module is used for subdividing the first-class input parameter space corresponding to each input random parameter to obtain a plurality of subdivided first-class input parameter spaces;
and the function extraction module is used for performing function extraction on the sampling points corresponding to the plurality of subdivided first-class input parameter spaces and the corresponding performance measurement values to obtain a first-class function representing the corresponding relation between the parameter values possibly not meeting the design index and the performance measurement values.
According to the scheme, a first class input parameter space which possibly does not accord with the performance index is quickly positioned by amplifying the standard deviation, a first class function is obtained based on sampling points corresponding to the first class input parameter space and corresponding performance measurement values, random sampling is carried out based on the standard deviation of the original distribution of each input random parameter of a design object to obtain a second class sampling point, a performance estimation value is obtained for the second class sampling point falling in the first class input parameter space by using the first class function, a third class sampling point which does not accord with the performance index is determined, the design yield which accords with the performance index of the design object in the corresponding overall random parameter space is calculated according to the number of the third class sampling points and the number of the second class sampling points, and the efficiency of obtaining the design yield which accords with the performance index is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
FIG. 1 is a schematic flow chart illustrating a method for obtaining a design yield according to an embodiment of the present disclosure;
FIG. 2 is a schematic flow chart illustrating a process for obtaining a corresponding first type function of a parameter value and a performance measurement value in each subdivision space in a first type input parameter space according to an embodiment of the present application;
fig. 3 is a structural diagram of a monte carlo analysis result in the entire input parameter space calculated according to the size of each subdivision space and the monte carlo analysis result disclosed in the embodiment of the present application;
FIG. 4 is a schematic diagram of an input random parameter space partition disclosed in an embodiment of the present application;
FIG. 5 is a schematic diagram of another input random parameter space partition disclosed in an embodiment of the present application;
FIG. 6 is a schematic diagram illustrating an apparatus for obtaining design yield according to an embodiment of the present disclosure;
fig. 7 is a schematic structural diagram of another apparatus for obtaining design yield according to an embodiment of the present disclosure.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1, a flow chart of a method for obtaining a design yield according to an embodiment of the present disclosure is shown, as shown in fig. 1, the method mainly includes the following steps:
s110: and amplifying the standard deviation of each input random parameter of the design object, and randomly sampling each input random parameter based on the amplified standard deviation to obtain a plurality of first-class sampling points.
The process of amplifying the standard deviation of each input random parameter of the design object is as follows:
by multiplying the standard deviation of each input random parameter of the design object by the magnification factor. Wherein the amplification factor may be a real number greater than 1. The preferred amplification factor of this scheme is not less than 3.
The purpose of amplifying the standard deviation of each input random parameter of the design object is to change the distribution probability of each input random parameter, so that the sampling with extremely low probability can be obviously improved, and the precision and the speed of the high-sigma Monte Carlo analysis are improved.
The standard deviation of each input random parameter of the design object is amplified by respectively using a plurality of different amplification factors to obtain a plurality of amplified standard deviations, and each input random parameter is randomly sampled based on each amplified standard deviation to obtain a plurality of first-class sampling point sets. Then, the processes of S120 to S160 are performed for the sets of the plurality of first-type sampling points, respectively.
S120: and acquiring the performance measurement value of each first-class sampling point.
And carrying out computer simulation on each first-type sampling point to obtain a corresponding performance measurement value.
Wherein, the computer simulation is to execute a simulation program on a computer to obtain a performance measurement value representing the design object.
S130: and determining a first type of input parameter space which possibly does not accord with the design index based on the performance measurement value of the first type of sampling point and the corresponding performance index value.
The parameter space is a value space formed by the value ranges of all the input random parameters; the first type of input parameter space refers to a value space in which the value of each input random parameter does not meet the design index.
Firstly, a sampling point set of which the performance measurement value is within a preset range is selected from first-class sampling points, then, for any input random parameter in a design object, the value ranges of the input random parameters corresponding to all the sampling points in the selected sampling point set are determined, and finally, the value ranges corresponding to all the input random parameters in the design object are determined to be a first-class input parameter space.
The preset range is obtained according to the performance index value of each input random parameter.
Further, the process of specifically determining the first type of input parameter space that may not meet the design criteria is as follows:
let the measured value of the property be VmiThe performance index value is Vspec，Vmi≤VspecThe samples of (A) are samples meeting design criteria, Vmi＞VspecThe samples of β are samples of the first type of sample points.
For any input random parameter PiAccording to the selected n sampling points Si，jTo obtain the random parameter PiDiscrete value of (V)i，j. Further, the input random parameter P in the selected sampling setiObtaining a value range [ V ] from each discrete value ofi，min,Vi，max]. Value range [ V ] of all input random parameters in design object under selected sampling seti，min,Vi，max]Together forming a first type of input parameter space that may not meet design criteria.
Wherein β is a positive number less than 1, and for high sigma analysis, β typically takes a value of 0.8.
j is 1, …, n. n is the selected number of samples.
Upper limit point Vi，min＝min(Vi，1，Vi，2，…，Vi，n) Lower limit point Vi，max＝max(Vi，1，Vi，2，…，Vi，n)。
It should be noted that, in practical application, the first-class input parameter space which may not meet the design index may be determined jointly according to the value range of each input random parameter in the selected sampling set and the theoretical value range of the input random parameter.
If the value range of the input random parameter under the selected sample is consistent with the theoretical value range of the input random parameter, the value range of the input random parameter under the selected sample can be ignored when the first type of input parameter space which possibly does not accord with the design index is determined, namely the theoretical value range of the input random parameter is obtained.
If the upper limit or the lower limit of the value range of the input random parameter under the selected sample is close to the upper limit or the lower limit of the theoretical value range of the input random parameter, the value between the upper limit of the value range of the input random parameter under the selected sample and the upper limit of the theoretical value range of the input random parameter, and the value between the lower limit of the value range of the input random parameter under the selected sample and the lower limit of the theoretical value range of the input random parameter cause the performance measurement value not to accord with the performance index value, the upper limit or the lower limit of the theoretical value range of the input random parameter is used as the upper limit or the lower limit of the value range of the input random parameter under the selected sample, and therefore the sample in the minimum probability value area is avoided from being missed.
S140: based on sampling points corresponding to the first-class input parameter space corresponding to each input random parameter and corresponding performance measurement values, a first-class function representing the corresponding relation between parameter values possibly not conforming to the design index and the performance measurement values is obtained.
The first-class input parameter space and the number of sampling points determined in the previous step are both small, so that the high-precision first-class function can be extracted conveniently from the function extraction point, and the efficiency of extracting the first-class function is improved.
In one embodiment of the present application, to further improve the accuracy of the monte carlo analysis, as shown in fig. 2, the samples in the first type of input parameter space that may not meet the design criteria may be further divided, thereby further subdividing the first type of input parameter space that may not meet the design criteria into a plurality of smaller first type of input parameter spaces that may not meet the design criteria. Then, performing function extraction on the sampling points corresponding to each subdivided first-class input parameter space to obtain a first-class function corresponding to the parameter value and the performance measurement value in each subdivided space in the first-class input parameter space, for example, f in fig. 21(P1，P2，…，Pn)、f2(P1，P2，…，Pn)、f3(P1，P2，…，Pn). Subdividing the first type of input parameter space and extracting functions according to the subdivided input parameter space improves the accuracy of extracting the functions and accelerates the efficiency of extracting the functions. In one embodiment of the present application, the sampling points of the first type of input parameter space may be divided by using a clustering algorithm, for example, a K-means clustering algorithm (K-means clustering algorithm).
S150: based on the standard deviation of the original distribution of each input random parameter of the design object, random sampling is carried out to obtain a second class of sampling points, the second class of sampling points falling in the first class of input parameter space are calculated by utilizing a first class function to obtain a performance estimation value, and a third class of sampling points which do not accord with the design index are determined according to the performance estimation value.
In an embodiment of the present application, when the random parameter space is randomly sampled to obtain the second type of sampling points, the parameter space other than the first type of input parameter space is incompletely sampled to increase the sampling speed.
Specifically, the process of obtaining the second type of sampling points is as follows: in the process of obtaining a complete sampling point, randomly sampling any input random parameter of a design object, and judging whether the value of the input random parameter is in a first type of input parameter space corresponding to the input random parameter; if the value of the input random parameter is in the corresponding first type of input parameter space, continuing to randomly sample the next input random parameter; if the value of the input random parameter is not in the corresponding first-class input parameter space, marking the current sampling point outside the first-class input parameter space; and if the values of all input random parameters of the design object are in the corresponding first-class input parameter space, marking the current sampling point in the first-class input parameter space.
And all the sampling points marked in the first-class input parameter space and all the sampling points marked outside the first-class output parameter space form second-class sampling points.
The second type of sampling points is a set of all sampling points corresponding to the design object.
Each sampling point corresponds to a design object, and each sampling point comprises a plurality of input random parameters.
If the sampling value of a certain input random parameter falls in [ V ]i，min,Vi，max]And further utilizing the extracted first class function to calculate the performance estimation value of the sampling point, and determining whether the sampling point does not accord with the design index according to the performance estimation value. If the sampling value of the input random parameter does not fall in [ V ]i，min,Vi，max]Indicating that the sampling point meets the design index.
And judging whether the sampling point corresponding to the design object is in the first-class input parameter space, if the sampling point corresponding to the design object is not in the first-class input parameter space, finishing the construction of the current sampling point, and not sampling other input parameters in the sampling, thereby saving the sampling time and improving the sampling efficiency.
Further, in the application scenario of subdividing the first-class input parameter space and then extracting the function in S140, referring to fig. 2, the performance estimation value of the sampling point is calculated for the class sampling point falling in the subdivided first-class input parameter space by using the first-class function corresponding to the subdivided first-class input parameter space, so as to more accurately obtain the performance estimation value of the second-class sampling point.
S160: and calculating the design yield of the design object which accords with the design index in the corresponding overall random parameter space according to the number of the third type of sampling points and the number of the second type of sampling points.
And calculating to obtain the design yield corresponding to the overall random parameter space based on the number of the second type sampling points and the number of the third type sampling points corresponding to each subdivided space and the ratio of each subdivided space to the overall random parameter space.
According to the method for obtaining the design yield, the standard deviation of the distribution of the input random parameters is amplified, so that the distribution probability of the input random parameters is changed, the sampling probability of the sampling points obtained by small-probability sampling is increased, and therefore the precision and the speed of high-sigma Monte Carlo analysis are improved. In addition, the first-class input parameter space which is determined in the scheme and possibly does not accord with the design index is smaller in number of the sampling points, so that the extraction time for extracting the function from the sampling points corresponding to the first-class input parameter space is shorter, the function with high precision is favorably extracted, and the calculation efficiency and precision of the design yield are finally improved.
Referring to fig. 3, a process of dividing the input random parameter space to obtain each subdivided space, and then executing S110 to S150 in each subdivided space in parallel is shown, and a monte carlo analysis result in the whole input parameter space is calculated according to the size of each subdivided space and the monte carlo analysis result, that is, the yield of the whole input parameter space meeting the design index is calculated.
Wherein, K1, K2, K3 and K4 are subspaces after the input random parameter space is divided. After the steps of S110 to S150 in the above embodiment are performed on the subspaces of K1, K2, K3, and K4, respectively, the numbers of the third type of sampling points corresponding to the subspaces of K1, K2, K3, and K4 are obtained, respectively.
And then, calculating the design yield of each subspace according to the number of the third type sampling points and the number of the second type sampling points corresponding to each subspace of K1, K2, K3 and K4. Specifically, assume that the size of the whole random parameter space is V, and the size of each subdivision space is V1、V2、…、VnThe sampling point (i.e. the second type of sampling point) of each subdivision space is S1、S2、…、SnThe sampling point (i.e. the sampling point of the third kind) of each subdivision space which does not meet the design criteria is F1、F2、…、Fn。
The yield rate of the whole input parameter space meeting the design index is
In one embodiment of the present invention, the process of partitioning the input random parameter space into smaller subdivision spaces may be as follows:
by the parameter mean value point Vi，centerAnd dividing the random parameter space to obtain a plurality of subdivided spaces by taking the random parameter space as a center and respectively taking the boundary directions up and down and according to the sequence of the spaces from large to small.
It should be noted that, in the subdivided space obtained by the division, the region division granularity close to the parameter intermediate value point is coarse, the obtained subdivided space is large, and the region division granularity far from the parameter intermediate value point is fine, so that the obtained subdivided space is small.
In the high-sigma Monte Carlo analysis, the subdivision space close to the parameter intermediate value point is not the region of interest, and the subdivision space far away from the parameter intermediate value point is the region of interest. Therefore, for the attention area, under the condition of the same sampling quantity, the divided small space can improve the sampling density, and samples which do not accord with the performance index can be obtained more generally.
Referring to fig. 4, which shows a schematic diagram of an input random parameter space division provided in an embodiment of the present application, subspace 1, subspace 2, and subspace 3 in fig. 4 are subdivided spaces after the input random parameter space is divided, where an input random parameter P isiHas an upper limit point of Vi，maxThe lower limit point is Vi，minThe parameter median is Vi，center。
Distance Vi，centerThe closer subspace 1 is relatively larger, the distance Vi，centerThe more distant subspaces 2 and 3 are relatively small.
Referring to fig. 5, a schematic diagram of another input random parameter space division provided in the embodiment of the present application is shown, and the subspace 11, the subspace 21, the subspace 31, the subspace 22, and the subspace 32 in fig. 5 are subdivided spaces after the input random parameter space division.
Wherein, the distance Vi，centerNearest subspace 11 is largest, distance Vi，centerThe closer subspaces 21 and 31 are relatively smaller, the distance Vi，centerThe more distant subspaces 22 and subspaces 32 are relatively smaller.
The space division mode provided by the embodiment can further improve the sampling probability of the range sampled by a smaller probability, thereby improving the analysis efficiency and the analysis precision.
In one embodiment of the present application, in order to reduce the dimension of the input random parameter space, the input random parameters with extremely low sensitivity are ignored, and the input random parameters with relatively high sensitivity are reserved, so that the number of the input random parameters is reduced, and the dimension of the input random parameter space is reduced.
The specific process of reducing the dimension of the input random parameter space is as follows:
calculating the sensitivity of each input random parameter corresponding to the performance index of the design object, selecting the input random parameters with the sensitivity greater than or equal to a preset threshold (namely, filtering the input random parameters with the sensitivity less than the preset threshold, wherein the input random parameters with the sensitivity less than the preset threshold have almost no influence on the performance index of the design object), and determining the area range of the selected input random parameters as an integral random parameter space corresponding to the design object, namely, subsequently calculating the design yield according to the reserved parameter space of each input random parameter.
The preset threshold may be set according to actual requirements, and the present invention is not particularly limited.
To facilitate understanding of the process of simplifying the dimension of the random parameter space, the following description is given by way of example, where the original input random parameter space is (P)1，P2，…，Pn) Wherein n is the original input random parameter number, i.e. the dimension of the original parameter space is n, and the input random parameter space after degeneration is (P)1，P2，…，Pm) The dimension of the new random parameter space is m, m<n。
In the embodiment of the application, the dimensionality of the input random parameter space is simplified, the low-dimensionality parameter space is divided, and the efficiency of dividing the random input parameter space is improved.
Based on the method embodiment for obtaining the design yield disclosed by the embodiment of the application, the embodiment of the application also correspondingly discloses an embodiment of a device for obtaining the design yield.
As shown in fig. 6, the apparatus for obtaining design yield mainly includes:
the amplifying unit 601 is configured to amplify a standard deviation of each input random parameter of the design object, and randomly sample each input random parameter based on the amplified standard deviation to obtain a plurality of first-class sample points.
Further, the amplifying unit 601 includes:
and the amplifying module is used for amplifying the standard deviation of each input random parameter of the design object by using a plurality of different amplifying factors respectively to obtain a plurality of amplified standard deviations.
And the sampling module is used for randomly sampling each input random parameter based on each amplified standard deviation to obtain a plurality of first-class sampling points.
A first obtaining unit 602, configured to obtain a performance measurement value of each first type of sampling point.
The determining unit 603 is configured to determine a first type of input parameter space that may not meet the design index based on the performance measurement value of the first type of sampling point and the corresponding performance index value.
Further, the determining unit 603 includes:
and the selecting module is used for selecting a sampling point set of which the performance measurement value is within a preset range from the first type of sampling points.
The preset range is obtained according to the performance index value corresponding to the input random parameter.
And the corresponding module is used for selecting the value range of the input random parameter corresponding to all the sampling points in the sampling point set for any input random parameter in the design object.
The first determining module is used for determining that the value ranges corresponding to all the input random parameters in the design object are the first type of input parameter space.
A second obtaining unit 604, configured to obtain a first class function representing a corresponding relationship between a parameter value that may not meet a design index and a performance measurement value based on a sampling point corresponding to the first class input parameter space corresponding to each input random parameter and the corresponding performance measurement value.
Further, the second obtaining unit 604 includes:
and the subdivision module is used for subdividing the first-class input parameter space corresponding to each input random parameter to obtain a plurality of subdivided first-class input parameter spaces.
And the function extraction module is used for performing function extraction on the sampling points corresponding to the plurality of subdivided first-class input parameter spaces and the corresponding performance measurement values to obtain a first-class function representing the corresponding relation between the parameter values possibly not meeting the design index and the performance measurement values.
The sampling determining unit 605 is configured to perform random sampling to obtain a second type of sampling points based on a standard deviation of an original distribution of each input random parameter of the design object, calculate a performance estimation value for the second type of sampling points falling in the first type of input parameter space by using a first type of function, and determine a third type of sampling points that do not meet the performance index according to the performance estimation value.
Further, the sampling determination unit 605 that performs random sampling to obtain the second type of sampling points based on the standard deviation of the original distribution of each input random parameter of the design object includes:
and the judging module is used for randomly sampling any input random parameter of the design object and judging whether the value of the input random parameter is in a first-class input parameter space corresponding to the input random parameter.
And the random sampling module is used for continuously performing random sampling on the next input random parameter if the value of the input random parameter is in the corresponding first-class input parameter space.
And the first marking module is used for marking the current sampling point out of the first-class input parameter space if the value of the input random parameter is not in the corresponding first-class input parameter space.
And the second marking module is used for marking the current sampling point in the first-class input parameter space if all the values of the input random parameters of the design object are in the corresponding first-class input parameter space.
And all the sampling points marked in the first-class input parameter space and all the sampling points marked outside the first-class output parameter space form second-class sampling points.
The calculating unit 606 is configured to calculate a design yield in the input random parameter space according to the number of the third type of sampling points and the number of the second type of sampling points.
Further, the calculating unit 606 for determining the overall random parameter space corresponding to the design object includes:
and the acquisition module is used for acquiring the sensitivity of each input random parameter corresponding to the design object and selecting the input random parameter with the sensitivity greater than or equal to a preset threshold value.
And the second determining module is used for determining the area range of the selected input random parameter as an integral random parameter space corresponding to the design object.
The embodiment of the application discloses a device for obtaining design yield, which can quickly position a first class of input parameter space which may not meet performance indexes by amplifying standard deviation, obtain a first class of functions based on sampling points corresponding to the first class of input parameter space and corresponding performance measurement values, randomly sample to obtain a second class of sampling points based on the standard deviation of the original distribution of each input random parameter of a design object, obtaining a performance estimation value by utilizing the first class function for the second class sampling points falling in the first class input parameter space, and determining the third class sampling points which do not accord with the performance index, and calculating the design yield of the design object which accords with the performance index in the corresponding overall random parameter space according to the number of the third type sampling points and the number of the second type sampling points, thereby realizing the improvement of the efficiency of obtaining the design yield which accords with the performance index.
As shown in fig. 7, a schematic structural diagram of another apparatus for obtaining a design yield according to the embodiment of the present application is provided, where the apparatus further includes, based on the embodiment shown in fig. 6: the unit 701 is divided.
The dividing unit 701 is configured to divide the overall random parameter space corresponding to the design object to obtain a plurality of subdivided spaces, so as to obtain a second-class sample point and a third-class sample point corresponding to each subdivided space.
Further, the dividing unit 701 is specifically configured to:
and dividing the random parameter space by taking the parameter intermediate value point as a center and respectively taking the upper value boundary direction and the lower value boundary direction according to the sequence of the space from large to small to obtain a plurality of subdivided spaces.
Correspondingly, the calculating unit 606 is specifically configured to calculate to obtain the design yield corresponding to the overall random parameter space based on the number of the second type of sampling points, the number of the third type of sampling points, and the ratio of each subdivided space to the overall random parameter space.
In the embodiment of the application, the random parameter space is divided to obtain a plurality of subdivided spaces, and the design yield corresponding to the overall random parameter space is calculated based on the number of the second type sampling points and the number of the third type sampling points corresponding to each subdivided space and the ratio of each subdivided space to the overall random parameter space. And the ratio of the number of the second type of sampling points and the number of the third type of sampling points corresponding to the subdivided space to the whole random parameter space is utilized, so that the design yield corresponding to the whole random parameter space is more accurately obtained.
While, for purposes of simplicity of explanation, the foregoing method embodiments have been described as a series of acts or combination of acts, it will be appreciated by those skilled in the art that the present invention is not limited by the illustrated ordering of acts, as some steps may occur in other orders or concurrently with other steps in accordance with the invention. Further, those skilled in the art should also appreciate that the embodiments described in the specification are preferred embodiments and that the acts and modules referred to are not necessarily required by the invention.
It should be noted that, in the present specification, the embodiments are all described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments may be referred to each other. For the device-like embodiment, since it is basically similar to the method embodiment, the description is simple, and for the relevant points, reference may be made to the partial description of the method embodiment.
The steps in the method of each embodiment of the invention can be sequentially adjusted, combined and deleted according to actual needs.
The device and the modules and sub-modules in the terminal in the embodiments of the present invention can be combined, divided and deleted according to actual needs.
In the embodiments provided in the present invention, it should be understood that the disclosed terminal, apparatus and method may be implemented in other ways. For example, the above-described terminal embodiments are merely illustrative, and for example, the division of a module or a sub-module is only one logical division, and there may be other divisions when the terminal is actually implemented, for example, a plurality of sub-modules or modules may be combined or integrated into another module, or some features may be omitted or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or modules, and may be in an electrical, mechanical or other form.
The modules or sub-modules described as separate parts may or may not be physically separate, and parts that are modules or sub-modules may or may not be physical modules or sub-modules, may be located in one place, or may be distributed over a plurality of network modules or sub-modules. Some or all of the modules or sub-modules can be selected according to actual needs to achieve the purpose of the solution of the present embodiment.
In addition, each functional module or sub-module in each embodiment of the present invention may be integrated into one processing module, or each module or sub-module may exist alone physically, or two or more modules or sub-modules may be integrated into one module. The integrated modules or sub-modules may be implemented in the form of hardware, or may be implemented in the form of software functional modules or sub-modules.
Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
The foregoing is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, various modifications and decorations can be made without departing from the principle of the present invention, and these modifications and decorations should also be regarded as the protection scope of the present invention.
Claims (10)
1. A method for obtaining design yield, the method comprising:
amplifying the standard deviation of each input random parameter of the design object, and randomly sampling each input random parameter based on the amplified standard deviation to obtain a plurality of first-class sampling points;
acquiring a performance measurement value of each first-type sampling point;
determining a first type of input parameter space which possibly does not accord with design indexes based on the performance measurement values of the first type of sampling points and corresponding performance index values;
obtaining a first class function representing the corresponding relation between parameter values possibly not conforming to the design index and performance measurement values based on sampling points corresponding to the first class input parameter space corresponding to each input random parameter and the corresponding performance measurement values;
based on the standard deviation of the original distribution of each input random parameter of the design object, randomly sampling to obtain a second class of sampling points, calculating the second class of sampling points falling in the first class of input parameter space by using the first class function to obtain a performance estimation value, and determining a third class of sampling points which do not accord with the design index according to the performance estimation value;
and calculating the design yield of the design object which meets the design index in the corresponding overall random parameter space according to the number of the third type of sampling points and the number of the second type of sampling points.
2. The method of claim 1, wherein determining a first type of input parameter space that may not meet design criteria based on the performance measurement values and corresponding performance metric values for the first type of sample points comprises:
selecting a sampling point set of which the performance measurement value is within a preset range from the first type of sampling points, wherein the preset range is obtained according to a performance index value corresponding to an input random parameter;
for any input random parameter in the design object, the value ranges of the input random parameters corresponding to all sampling points in the selected sampling point set;
and determining the value ranges corresponding to all the input random parameters in the design object as the first-class input parameter space.
3. The method according to claim 1, wherein the obtaining a first class function representing a correspondence between parameter values that may not meet design criteria and performance measurement values based on sampling points and corresponding performance measurement values corresponding to the first class input parameter space corresponding to each input random parameter comprises:
subdividing the first-class input parameter space corresponding to each input random parameter to obtain a plurality of subdivided first-class input parameter spaces;
and performing function extraction on the sampling points corresponding to the plurality of subdivided first-class input parameter spaces and the corresponding performance measurement values to obtain a first-class function representing the corresponding relation between the parameter values possibly not conforming to the design index and the performance measurement values.
4. The method according to claim 1, further comprising, before amplifying the standard deviation of each input random parameter of the design object and randomly sampling each input random parameter based on the amplified standard deviation to obtain a plurality of first type sampling points:
dividing the whole random parameter space corresponding to the design object to obtain a plurality of subdivided spaces so as to obtain the second type of sampling points and the third type of sampling points corresponding to each subdivided space;
calculating the design yield in the input random parameter space according to the number of the third type of sampling points and the number of the second type of sampling points, and the method comprises the following steps:
and calculating to obtain the design yield corresponding to the whole random parameter space based on the number of the second type of sampling points, the number of the third type of sampling points and the ratio of each subdivided space to the whole random parameter space corresponding to each subdivided space.
5. The method of claim 4, wherein the dividing the random parameter space into the subdivided spaces comprises:
and dividing the random parameter space by taking the parameter intermediate value point as a center and respectively taking the upper value boundary direction and the lower value boundary direction according to the sequence of the space from large to small to obtain a plurality of subdivided spaces.
6. The method of claim 1, wherein amplifying the standard deviation of each input random parameter of the design object, and randomly sampling each input random parameter based on the amplified standard deviation to obtain a plurality of first type sampling points, comprises:
amplifying the standard deviation of each input random parameter of the design object by using a plurality of different amplification factors respectively to obtain a plurality of amplified standard deviations;
and randomly sampling each input random parameter based on each amplified standard deviation to obtain a plurality of first-class sampling points.
7. The method of claim 1, wherein determining the overall stochastic parameter space for the design object is performed as follows:
acquiring the sensitivity of each input random parameter corresponding to the design object, and selecting the input random parameter with the sensitivity greater than or equal to a preset threshold value;
and determining the area range of the selected input random parameter as an integral random parameter space corresponding to the design object.
8. The method of claim 1, wherein randomly sampling the second type of sample points based on a standard deviation of an original distribution of each input random parameter of the design object comprises:
randomly sampling any input random parameter of the design object, and judging whether the value of the input random parameter is in a first-class input parameter space corresponding to the input random parameter;
if the value of the input random parameter is in the corresponding first type of input parameter space, continuing to randomly sample the next input random parameter;
if the value of the input random parameter is not in the corresponding first-class input parameter space, marking the current sampling point outside the first-class input parameter space;
if the values of all input random parameters of the design object are in the corresponding first-class input parameter space, marking the current sampling point in the first-class input parameter space;
and all the sampling points marked in the first type of input parameter space and all the sampling points marked outside the first type of output parameter space form the second type of sampling points.
9. An apparatus for obtaining design yield, the apparatus comprising:
the amplifying unit is used for amplifying the standard deviation of each input random parameter of the design object and randomly sampling each input random parameter based on the amplified standard deviation to obtain a plurality of first-class sampling points;
the first acquisition unit is used for acquiring the performance measurement value of each first-class sampling point;
the determining unit is used for determining a first type of input parameter space which possibly does not accord with the design index based on the performance measurement value of the first type of sampling point and the corresponding performance index value;
a second obtaining unit, configured to obtain a first class function representing a correspondence between a parameter value that may not meet a design index and a performance measurement value based on a sampling point corresponding to the first class input parameter space and the corresponding performance measurement value corresponding to each input random parameter;
the sampling determining unit is used for performing random sampling to obtain second-class sampling points based on the standard deviation of the original distribution of each input random parameter of the design object, calculating the second-class sampling points in the first-class input parameter space by using the first-class function to obtain performance estimation values, and determining third-class sampling points which do not accord with performance indexes according to the performance estimation values;
and the calculating unit is used for calculating the design yield in the input random parameter space according to the number of the third type of sampling points and the number of the second type of sampling points.
10. The apparatus of claim 9, wherein the second obtaining unit comprises:
the subdivision module is used for subdividing the first-class input parameter space corresponding to each input random parameter to obtain a plurality of subdivided first-class input parameter spaces;
and the function extraction module is used for performing function extraction on the sampling points corresponding to the plurality of subdivided first-class input parameter spaces and the corresponding performance measurement values to obtain a first-class function representing the corresponding relation between the parameter values possibly not meeting the design index and the performance measurement values.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010381928.4A CN111553611B (en) | 2020-05-08 | 2020-05-08 | Method and device for obtaining design yield |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010381928.4A CN111553611B (en) | 2020-05-08 | 2020-05-08 | Method and device for obtaining design yield |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111553611A true CN111553611A (en) | 2020-08-18 |
| CN111553611B CN111553611B (en) | 2023-05-23 |
Family
ID=72001151
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010381928.4A Active CN111553611B (en) | 2020-05-08 | 2020-05-08 | Method and device for obtaining design yield |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111553611B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TW201237647A (en) * | 2010-10-27 | 2012-09-16 | Solido Design Automation Inc | Method and system for identifying rare-event failure rates |
| US9836564B1 (en) * | 2015-04-09 | 2017-12-05 | Cadence Design Systems, Inc. | Efficient extraction of the worst sample in Monte Carlo simulation |
| CN110610009A (en) * | 2018-06-14 | 2019-12-24 | 复旦大学 | SRAM circuit yield analysis method based on Bayesian model |
-
2020
- 2020-05-08 CN CN202010381928.4A patent/CN111553611B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TW201237647A (en) * | 2010-10-27 | 2012-09-16 | Solido Design Automation Inc | Method and system for identifying rare-event failure rates |
| US9836564B1 (en) * | 2015-04-09 | 2017-12-05 | Cadence Design Systems, Inc. | Efficient extraction of the worst sample in Monte Carlo simulation |
| CN110610009A (en) * | 2018-06-14 | 2019-12-24 | 复旦大学 | SRAM circuit yield analysis method based on Bayesian model |
Non-Patent Citations (3)
| Title |
|---|
| 侯本伟;杜修力: "考虑节点失效网络可靠性计算的重要度抽样随机模拟", 系统工程理论与实践 * |
| 刘佩;姚谦峰;: "采用重要抽样法的结构动力可靠度计算", 计算力学学报 * |
| 骆勇鹏;黄方林;伍彦斌;鲁四平;苏泽平: "基于逐步回归分析和Bootstrap重抽样的铁路钢桁桥不确定参数识别", 公路交通科技 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111553611B (en) | 2023-05-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109685092B (en) | Clustering method, equipment, storage medium and device based on big data | |
| CN107341220B (en) | Multi-source data fusion method and device | |
| CN107979431B (en) | Method, device and equipment for spectrum sensing based on Riemann median | |
| CN107071788B (en) | Spectrum sensing method and device in cognitive wireless network | |
| CN110825826A (en) | Clustering calculation method, device, terminal and storage medium | |
| CN109949176A (en) | It is a kind of based on figure insertion social networks in abnormal user detection method | |
| CN102799616B (en) | Outlier point detection method in large-scale social network | |
| CN113205225A (en) | Method, system and data platform for identifying key factors of carbon emission peak | |
| CN115309753B (en) | Data rapid reading method of efficient environment-friendly intelligent sample research and development system | |
| CN112131220B (en) | Data report processing method and device | |
| CN110348717B (en) | Base station value scoring method and device based on grid granularity | |
| CN117407828B (en) | Data analysis method applied to sponge city rainwater collection system | |
| CN111553611A (en) | Method and device for obtaining design yield | |
| CN110895533A (en) | Form mapping method and device, computer equipment and storage medium | |
| CN115934699A (en) | Abnormal data screening method and device, electronic equipment and storage medium | |
| CN104635206B (en) | A kind of method and device of wireless location | |
| CN110597807A (en) | Data expansion method, device, terminal and medium based on data analysis | |
| WO2020258951A1 (en) | Method and device for acquiring user residence location, and computer-readable storage medium | |
| Liu et al. | Maximum likelihood abundance estimation from capture-recapture data when covariates are missing at random | |
| CN113946717A (en) | Sub-map index feature obtaining method, device, equipment and storage medium | |
| CN115809400A (en) | Method and device for evaluating lithium content in brine, storage medium and electronic equipment | |
| CN114077865A (en) | IDmapping method based on factor analysis and graph clustering | |
| CN111385116A (en) | Multi-dimensional correlation characteristic analysis method and device for high-interference cell | |
| CN108509967A (en) | A kind of clustering method and device, server | |
| CN115563193B (en) | Big data analysis processing method for digital information |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |