CN105843990A - Probability circuit simulation method based on steepest descent method and bisection method - Google Patents
Probability circuit simulation method based on steepest descent method and bisection method Download PDFInfo
- Publication number
- CN105843990A CN105843990A CN201610148150.6A CN201610148150A CN105843990A CN 105843990 A CN105843990 A CN 105843990A CN 201610148150 A CN201610148150 A CN 201610148150A CN 105843990 A CN105843990 A CN 105843990A
- Authority
- CN
- China
- Prior art keywords
- probability
- value
- unit
- output signal
- groups
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/30—Circuit design
- G06F30/39—Circuit design at the physical level
- G06F30/398—Design verification or optimisation, e.g. using design rule check [DRC], layout versus schematics [LVS] or finite element methods [FEM]
Abstract
The invention discloses a probability circuit simulation method based on a steepest descent method and a bisection method. The probability circuit simulation method comprises following steps: (1) determining a corresponding relation between a probability unit and an output signal and determining a "gradient" relation therebetween at the same time; (2) acquiring a probability unit combination matrix with respect to each individual output signal, and meeting required precision requirements by using the bisection method to adjust error probabilities; and (3) carrying out overall optimization so as to obtain a final result, allowing a value of each probability unit to be a maximum value and a minimum value in each row in step 2 to obtain two groups of probability combinations and carry out simulation, finding out an output signal, whose error probabilities are at two sides of a target value and error probability difference value is the biggest in two groups of tests, using the bisection method to optimize the final result again and repeating the above steps until all the output error probability values reach a required precision. The probability circuit simulation method is advantageous in that a probability unit needing to be adjusted can be chosen in a quite precise manner during adjustment of a probability value; an error probability value of the probability unit is adjusted in a quite precise manner; and more energy is acquired and saved.
Description
Technical field
The present invention relates to, based on probability circuit emulation technology, relate to the parameter probability valuing quickly adjusting probability unit, be specifically related to
A kind of probability circuit emulation mode based on steepest descent method Yu two way classification.
Background technology
Along with the development of IC industry, while chip processing speed quickly improves, energy consumption the most constantly increases, and with
Complementary metal oxide semiconductors (CMOS) cmos device size and enter nanometer era, all kinds of internal interference factors are to CMOS
The impact of device electrology characteristic is the most obvious.Meanwhile, along with the acceleration of each electronic product is popularized, people obtain in expectation
While obtaining higher hardware performance, it is also desirable to equipment power dissipation can be reduced.For the problems referred to above, researchers propose
The concept of PCMOS (Probabilistic complementary metal-oxide semiconductor) probability device, the most defeated
Go out and non-determined, but there is the cmos device of probability mistake.And in order to PCMOS probability circuit is applied to
In real system it is necessary to solve how it is carried out simulation modeling problem.Based on on-site programmable gate array FPGA
It is fast that simulation hardware method has simulation velocity, and the advantages such as simulation process is highly controllable, are a kind of splendid emulation modes.
Three step matrixes build method and can apply to the mistake adjustment of probabilistic module in probability circuit simulation process, and algorithm thinking
Simpler, it is easy to realize.But this algorithm maximum problematically, its to adjust process the most loaded down with trivial details, and lack certain
Regularity, when circuit under test is complex, should be accomplished by trial the most repeatedly in this way and adjust and emulation, because of
And tackle large-scale circuit, this process needs to take a substantial amount of time, and has had a strong impact on its range of application.Steepest decline and
Two way classification is then able to well solve.
Three step matrixes build the adjustment process of method to be needed in the face of two problems: one is that target output is all had by multiple probability unit
When affecting, how to determine and which is preferentially revised;And after determining that probability unit to be revised, the most effectively adjust its probability
It is worth to export close to target the soonest.Steepest descent method and two way classification are then able to well solve the two problem.
The core concept that two way classification combines with steepest descent method is, checking same the most relevant with each output of probability unit
Time, calculate the ratio that its probability gradient, i.e. output error probability difference are poor with output probability, then according to probability gradient for each
Output signal is adjusted, and obtains probability matrix faster, finally utilizes probability gradient to carry out probability tune for entirety output
Whole, to obtain final result.Utilize this concept of probability gradient to find and currently output is affected maximum of probability unit, utilize
Its error probability is adjusted by two way classification, thus adjusts error probability more accurately so that optimum results is restrained faster.
Summary of the invention
It is an object of the invention to provide a kind of probability circuit emulation mode based on steepest descent method Yu two way classification, in this situation
Under can adjust error probability the most accurately, to some extent solve for want of regular and cause adjusting with imitative
The biggest problem.
To achieve these goals, the technical scheme is that
A kind of probability circuit emulation mode based on steepest descent method Yu two way classification, it is characterised in that comprise the steps:
(1) probability unit and output signal " gradient " relation between the two are determined;Each probability unit error probability depends on
Secondary it is set to two groups of different values, sees whether that mistake occurs, calculate twice error probability poor, i.e. gradient;
(2) steepest descent method and two way classification is utilized to obtain error probability value matrix, successively for each output signal,
Probability unit error probability is spread a little at random, obtains two groups of probability that distance objective value both sides are nearest;Find two groups of probits
The probability unit that middle probability difference is maximum with gradient product, is set in two groups corresponding probability by maximum probability unit error probability
Intermediate value, emulates again, finds two groups of probability that distance objective value both sides are nearest in three class values;Repeat step (2) straight
Required probability required precision is reached to output error probability;
(3) value making each probability unit takes maximum and the minima of every string in step (2) respectively, obtains
Two groups of probabilistic combination emulate respectively;Error probability in two groups of tests is found to be positioned at maximum one of target value on both sides and difference defeated
Go out signal;Finding in the probability unit relevant with this output, the probability unit that probability difference is maximum with gradient product, by it
Error probability is set in two groups corresponding probability intermediate value, again emulates, finds distance objective value both sides nearest in three class values
Two groups of probability, repeat above step (3) until output error probability reaches required probability required precision.
In described step (1), determine probability unit and output signal " gradient " relation between the two, wherein gradient Gmn
For
Wherein, m and n has been expressed as m output signal and n probability unit;pemn1And pemn2For each generally
Two different error probability values that rate unit is respectively provided with, after emulation, pomn1For by pemn1The output signal of impact
Error probability, pomn2For by pemn2The error probability of the output signal of impact.
In described step (1), according to calculating GmnMethod, after testing all of probability unit, just obtain probability " ladder
Degree " matrix G:
Wherein, the corresponding output signal of every a line in matrix G, the corresponding probability unit of every string, the value in matrix
Represent the output signal " gradient " in this probability unit dimension, i.e. error probability downward gradient.
Described step (2) particularly as follows:
(21) to some output signal k in circuit, from the matrix obtained, the probability relevant to output signal k is obtained
Unit, spreads a little at random to the error probability value of this probability unit, i.e. randomly generates several probabilistic combination, emulation
After test, it is thus achieved that corresponding output error probability;
(22) from which more than in the combination presetting output error probability, choose minimum one group and be designated as
PA(pk1a, pk2a…pkna), it is designated as from less than choose maximum the combination of preset value one group
PB(pk1b, pk2b…pknb);
(23) according to the gradient relation obtained in step (1), probability difference and corresponding ladder in two groups of probabilistic combination are found
Numbering x of one probability unit of the product maximum absolute value of degree, corresponding calculating formula is with following formula (1-3);
(24) numbering x obtained according to formula (1-3), the error probability value of order numbering x is PAAnd PBMiddle correspondence takes
The intermediate value of value, i.e. utilizes two way classification to be adjusted to formula (1-4);
(25) completing the once amendment to parameter probability valuing, amended two groups of probabilistic combination are respectively
PA’(pk1a, pk2a…pkxc…pkna) and PB’(pk1b, pk2b…pkxc…pknb);
(26) P is being calculated respectivelyA, PB, PA’,PBIn ' four kind probabilistic combination, output error probability is respectively from positive and negative two
Step (22)-(25), closest to two groups of preset value, are repeated until the error probability obtained reaches required in individual direction
Probability required precision;
(27) again for next output signal, step (22)-(26) are repeated, until all of output signal is whole
It is disposed, has the most just obtained error probability value matrix E:
Wherein, the corresponding output signal of every a line, the corresponding probability unit of every string.
In described step (3), concretely comprise the following steps:
(31) value making each probability unit takes the error probability value matrix E formula obtained in step (2) respectively
(1-5) in the middle of, the maximum of every string and minima, obtain two groups of probabilistic combination, be denoted as P respectivelyMAXAnd PMIN
(32) by PMAXAnd PMINIn substitution formula (1-8), obtain YMAX={ YMAX1,YMAX2…YMAXmAnd
YMIN={ YMIN1,YMIN2…YMINm}
WhereinFor system transfer matrix;
(33) output signal that in latter event, two error probability value difference values are maximum is found, then for this output
Signal utilizes the method for formula (1-3), finds the probability unit that corresponding probability difference is maximum with gradient product, then profit
By formula (1-4), its probit is set in two groups of probabilistic combination the intermediate value of corresponding probit, again carries out emulation testing.
The present invention utilizes this concept of probability gradient to find currently affects maximum of probability unit to output, utilizes two way classification to it
Error probability is adjusted, thus adjusts error probability more accurately so that optimum results is restrained faster.Compare original
The mode complexity of three step matrixes to reduce, the time of consuming and simulation times also to reduce.Export for each,
Can accurately find associated probability unit, also be able to the value quickly adjusting probability unit to reach simultaneously
To required probability output precision, improve simulation efficiency.
Accompanying drawing explanation
Fig. 1 is present invention probability circuit substantially emulation mode flow chart based on steepest descent method Yu two way classification.
Fig. 2 is the present invention detailed emulation mode flow chart of probability circuit based on steepest descent method Yu two way classification.
Detailed description of the invention
By description below and combine accompanying drawing, the present invention will become more fully apparent, and accompanying drawing is for explaining the enforcement of the present invention
Example.
It is probability circuit emulation mode based on FPGA that steepest of the present invention declines with two way classification, allows at certain at output probability
In the range of one, by adjusting the state that the parameter probability valuing of probability unit reaches the most energy-conservation.Use steepest to decline to carry with two way classification
The efficiency of high emulation, reduces simulation times.
As it is shown in figure 1, the present invention's comprises the following steps that (1) determines probability unit and output signal " ladder between the two
Degree " relation;Each probability unit error probability is set to two groups of different values successively, sees whether mistake occur, calculates two
Secondary error probability is poor, i.e. gradient;
(2) steepest descent method and two way classification is utilized to obtain error probability value matrix, successively for each output signal,
Probability unit error probability is spread a little at random, obtains two groups of probability that distance objective value both sides are nearest;Find two groups of probits
The probability unit that middle probability difference is maximum with gradient product, is set in two groups corresponding probability by maximum probability unit error probability
Intermediate value, emulates again, finds two groups of probability that distance objective value both sides are nearest in three class values;Repeat step (2) straight
Required probability required precision is reached to output error probability;
(3) value making each probability unit takes maximum and the minima of every string in step (2) respectively, obtains
Two groups of probabilistic combination emulate respectively;Error probability in two groups of tests is found to be positioned at maximum one of target value on both sides and difference defeated
Go out signal;Finding in the probability unit relevant with this output, the probability unit that probability difference is maximum with gradient product, by it
Error probability is set in two groups corresponding probability intermediate value, again emulates, finds distance objective value both sides nearest in three class values
Two groups of probability, repeat above step (3) until output error probability reaches required probability required precision.
As in figure 2 it is shown, the present invention specifically comprises the following steps that
Step S1, determines the corresponding relation between probability unit and output signal and " gradient " relation, to each probability
Unit is respectively provided with different error probabilities, is denoted as pemn1And pemn2, emulating a period of time respectively, record is affected by
Twice different error probability of output signal, be denoted as po respectivelymn1And pomn2.Wherein n represents the numbering of probability unit,
M represents the numbering of output signal, thus can calculate gradient Gmn;
Step S2, repeats step S1 until all of probability unit is all completed, writes down all of difference and just can obtain
To probability gradient matrix G
The corresponding output signal of every a line of matrix obtained by step S2, the corresponding probability unit of every string, matrix
In value represent the output signal " gradient " in this probability unit dimension, i.e. error probability decrease speed, so every a line
In the probability unit corresponding to maximum be exactly that this output error probability declines the fastest dimension.
Step S3, utilizes steepest to decline and obtains probability matrix with two way classification, for some output signal in circuit,
As exported k, from the matrix that the first step obtains, obtain associated probability unit, its error probability value is carried out
Spread a little at random, i.e. randomly generate several probabilistic combination, several clock cycle of emulation testing, it is thus achieved that export mistake accordingly
Probability by mistake
Step S4, chooses a group minimum more than output error probability in the combination that step S3 obtains and is designated as
PA(pk1a, pk2a…pkna), it is designated as from less than choose maximum the combination of preset value one group
PB(pk1b, pk2b…pknb).Further according to the gradient relation obtained in the first step, find probability difference in two groups of probabilistic combination
Numbering x of one probability unit of the product maximum absolute value of value and its gradient.
Step S5, the numbering obtained according to step S4, making its error probability value is PAAnd PBThe centre of middle corresponding value
Value, i.e. utilizes two way classification to adjust;
This completes the once amendment to parameter probability valuing, amended two groups of probabilistic combination are respectively
PA’(pk1a, pk2a…pkxc…pkna) and PB’(pk1b, pk2b…pkxc…pknb).Calculating P respectivelyA, PB,
PA', PBIn ' four kind probabilistic combination, output error probability respectively from positive and negative both direction closest to two groups of preset value
Step S6, repeats above step S3-S5 for next output signal, until all of output signal is whole
It is disposed, has the most just obtained error probability value matrix E;
Step S7, utilizes steepest to decline and obtains result with two way classification optimization, make the value of each probability unit take respectively
In second step, the maximum of every string and minima, obtain two groups of probabilistic combination, be denoted as P respectivelyMAXAnd PMIN;
Step S8, PMAXAnd PMINTwo groups of probits carry out emulation testing respectively, obtain its output probability, will PMAXWith
PMINSubstitute into following formula;
Calculating through above formula can obtain YMAX={ YMAX1,YMAX2…YMAXmAnd YMIN={ YMIN1,YMIN2…YMINm}
Step S9, finds the output signal that two error probability differences are maximum, and it is right to find according to the method for step S4
Answering a probability unit of probability difference and the maximum of gradient product, the two way classification that recycle step S5 is used is by its probability
Value is set in two groups of probabilistic combination the intermediate value of corresponding probit, again carries out emulation testing;
Repeat step S7-S9, until the error probability value of all outputs has all reached required precision;Terminate.
Above in association with flow chart, the present invention is described in detail.The advantage of two points of prompt drop method maximums is to adjust
Whole probit can select the probability unit that need to adjust more accurately, adjusts its error probability value more accurately, it is to avoid
The invalid simulation problems caused because lacking directivity during random adjustment.And the improvement that causes because lacking regularity is too small
Problem, thus the simulation times that need to carry out can be greatly reduced, improve simulation velocity, and then expand and utilize FPGA pair
PCMOS circuit carries out the range of application emulated.
Claims (5)
1. a probability circuit emulation mode based on steepest descent method Yu two way classification, it is characterised in that comprise the steps:
(1) probability unit and output signal " gradient " relation between the two are determined;Each probability unit error probability depends on
Secondary it is set to two groups of different values, sees whether that mistake occurs, calculate twice error probability poor, i.e. gradient;
(2) steepest descent method and two way classification is utilized to obtain error probability value matrix, successively for each output signal,
Probability unit error probability is spread a little at random, obtains two groups of probability that distance objective value both sides are nearest;Find two groups of probits
The probability unit that middle probability difference is maximum with gradient product, is set in two groups corresponding probability by maximum probability unit error probability
Intermediate value, emulates again, finds two groups of probability that distance objective value both sides are nearest in three class values;Repeat step (2) straight
Required probability required precision is reached to output error probability;
(3) value making each probability unit takes maximum and the minima of every string in step (2) respectively, obtains
Two groups of probabilistic combination emulate respectively;Error probability in two groups of tests is found to be positioned at maximum one of target value on both sides and difference defeated
Go out signal;Finding in the probability unit relevant with this output, the probability unit that probability difference is maximum with gradient product, by it
Error probability is set in two groups corresponding probability intermediate value, again emulates, finds distance objective value both sides nearest in three class values
Two groups of probability, repeat above step (3) until output error probability reaches required probability required precision.
2. probability circuit emulation mode based on steepest descent method Yu two way classification as claimed in claim 1, it is characterised in that
In described step (1), determine probability unit and output signal " gradient " relation between the two, wherein gradient GmnFor
Wherein, m and n has been expressed as m output signal and n probability unit;pemn1And pemn2For each generally
Two different error probability values that rate unit is respectively provided with, after emulation, pomn1For by pemn1The output signal of impact
Error probability, pomn2For by pemn2The error probability of the output signal of impact.
3. probability circuit emulation mode based on steepest descent method Yu two way classification as claimed in claim 2, it is characterised in that
In described step (1), according to calculating GmnMethod, after testing all of probability unit, just obtain probability " ladder
Degree " matrix G:
Wherein, the corresponding output signal of every a line in matrix G, the corresponding probability unit of every string, the value in matrix
Represent the output signal " gradient " in this probability unit dimension, i.e. error probability downward gradient.
4. probability circuit emulation mode based on steepest descent method Yu two way classification as claimed in claim 3, it is characterised in that
Described step (2) particularly as follows:
(21) to some output signal k in circuit, from the matrix obtained, the probability relevant to output signal k is obtained
Unit, spreads a little at random to the error probability value of this probability unit, i.e. randomly generates several probabilistic combination, emulation
After test, it is thus achieved that corresponding output error probability;
(22) from which more than in the combination presetting output error probability, choose minimum one group and be designated as
PA(pk1a, pk2a...pkna), it is designated as from less than choose maximum the combination of preset value one group
PB(pk1b, pk2b...pknb);
(23) according to the gradient relation obtained in step (1), probability difference and corresponding ladder in two groups of probabilistic combination are found
Numbering x of one probability unit of the product maximum absolute value of degree, corresponding calculating formula is with following formula (1-3);
(24) numbering x obtained according to formula (1-3), the error probability value of order numbering x is PAAnd PBMiddle correspondence takes
The intermediate value of value, i.e. utilizes two way classification to be adjusted to formula (1-4);
(25) completing the once amendment to parameter probability valuing, amended two groups of probabilistic combination are respectively
PA’(pk1a, pk2a...pkxc...pkna) and PB’(pk1b, pk2b...pkxc...pknb);
(26) P is being calculated respectivelyA, PB, PA’,PBIn ' four kind probabilistic combination, output error probability is respectively from positive and negative two
Step (22)-(25), closest to two groups of preset value, are repeated until the error probability obtained reaches required in individual direction
Probability required precision;
(27) again for next output signal, step (22)-(26) are repeated, until all of output signal is whole
It is disposed, has the most just obtained error probability value matrix E:
Wherein, the corresponding output signal of every a line, the corresponding probability unit of every string.
5. probability circuit emulation mode based on steepest descent method Yu two way classification as claimed in claim 4, it is characterised in that
In described step (3), concretely comprise the following steps:
(31) value making each probability unit takes the error probability value matrix E formula obtained in step (2) respectively
(1-5) in the middle of, the maximum of every string and minima, obtain two groups of probabilistic combination, be denoted as P respectivelyMAXAnd PMIN
(32) by PMAXAnd PMINIn substitution formula (1-8), obtain YMAX={ YMAX1,YMAX2...YMAXmAnd
YMIN={ YMIN1,YMIN2...YMINm}
WhereinFor system transfer matrix;
(33) output signal that in latter event, two error probability value difference values are maximum is found, then for this output
Signal utilizes the method for formula (1-3), finds the probability unit that corresponding probability difference is maximum with gradient product, then profit
By formula (1-4), its probit is set in two groups of probabilistic combination the intermediate value of corresponding probit, again carries out emulation testing.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610148150.6A CN105843990B (en) | 2016-03-15 | 2016-03-15 | A kind of probability circuit emulation mode based on steepest descent method and dichotomy |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610148150.6A CN105843990B (en) | 2016-03-15 | 2016-03-15 | A kind of probability circuit emulation mode based on steepest descent method and dichotomy |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105843990A true CN105843990A (en) | 2016-08-10 |
CN105843990B CN105843990B (en) | 2018-10-12 |
Family
ID=56588195
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610148150.6A Expired - Fee Related CN105843990B (en) | 2016-03-15 | 2016-03-15 | A kind of probability circuit emulation mode based on steepest descent method and dichotomy |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105843990B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110970084A (en) * | 2018-09-29 | 2020-04-07 | 杭州广立微电子有限公司 | Method for quickly correcting SRAM test voltage and SRAM test circuit |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103117750A (en) * | 2013-01-14 | 2013-05-22 | 天津大学 | Error probability estimation method of low-density parity check codes |
US8504974B1 (en) * | 2011-09-14 | 2013-08-06 | Xilinx, Inc. | Analysis of circuit designs |
-
2016
- 2016-03-15 CN CN201610148150.6A patent/CN105843990B/en not_active Expired - Fee Related
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8504974B1 (en) * | 2011-09-14 | 2013-08-06 | Xilinx, Inc. | Analysis of circuit designs |
CN103117750A (en) * | 2013-01-14 | 2013-05-22 | 天津大学 | Error probability estimation method of low-density parity check codes |
Non-Patent Citations (1)
Title |
---|
何光进: "多信号模型在故障诊断中的应用", 《计算机测量与控制》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110970084A (en) * | 2018-09-29 | 2020-04-07 | 杭州广立微电子有限公司 | Method for quickly correcting SRAM test voltage and SRAM test circuit |
CN110970084B (en) * | 2018-09-29 | 2021-09-07 | 杭州广立微电子股份有限公司 | Method for quickly correcting SRAM test voltage and SRAM test circuit |
Also Published As
Publication number | Publication date |
---|---|
CN105843990B (en) | 2018-10-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103676881B (en) | A kind of dynamic bottleneck analytical method of semiconductor production line | |
CN110348574B (en) | ZYNQ-based universal convolutional neural network acceleration structure and design method | |
CN103761368B (en) | A kind of cable mesh reflector antenna surface accuracy and method for designing while tension state | |
CN102682345B (en) | Based on the traffic flow forecasting method of two variable optimal learning rate Fast Learning neural network | |
CN111008502A (en) | Fault prediction method for complex equipment driven by digital twin | |
CN111125947B (en) | Modeling method and related device for crimping IGBT thermal network model | |
CN105116733B (en) | Modified form population optimizing neutral net supersonic motor control system and its method | |
CN111224404B (en) | Power flow rapid control method for electric power system with controllable phase shifter | |
CN105184027A (en) | Power load modeling method based on interactive multi-model algorithm | |
CN110110434A (en) | A kind of initial method that Probabilistic Load Flow deep neural network calculates | |
CN104408325A (en) | Optimization method of micro-electronic encapsulation thermal expansion coefficient reliability matching | |
CN114166318A (en) | Ultrasonic water meter flow data calibration method based on deep learning | |
CN109444547A (en) | RFID chip impedance measurement method and device based on two-port network | |
CN110008914A (en) | A kind of pattern recognition system neural network based and recognition methods | |
CN105843990A (en) | Probability circuit simulation method based on steepest descent method and bisection method | |
CN112861429A (en) | Wind turbine engine room transfer function calculation method | |
CN103559541A (en) | Back propagation method for out-of-order data stream in big data | |
CN101477582B (en) | Model modification method for a semiconductor device | |
CN110348157A (en) | The noise simulation method and system of dynamic comparer | |
CN106777621A (en) | A kind of neutral net space reflection modeling method for packaged transistor | |
Tai et al. | Measuring the manufacturing yield for processes with multiple manufacturing lines | |
CN104991447A (en) | Thermal power generating unit variable-load speed predicting method based on wavelet neural network | |
CN108833156A (en) | A kind of appraisal procedure and system of the simulation performance index for power telecom network | |
CN116248202A (en) | Method for realizing radio frequency channel calibration based on deep learning | |
CN108829948A (en) | The method for building up in bad point library and establish system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20181012 Termination date: 20200315 |
|
CF01 | Termination of patent right due to non-payment of annual fee |