CN103885867A - Online evaluation method of performance of analog circuit - Google Patents
Online evaluation method of performance of analog circuit Download PDFInfo
- Publication number
- CN103885867A CN103885867A CN201410123824.8A CN201410123824A CN103885867A CN 103885867 A CN103885867 A CN 103885867A CN 201410123824 A CN201410123824 A CN 201410123824A CN 103885867 A CN103885867 A CN 103885867A
- Authority
- CN
- China
- Prior art keywords
- sample
- fuzzy
- support vector
- vector regression
- samples
- 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
Landscapes
- Complex Calculations (AREA)
Abstract
An online evaluation method of performance of an analog circuit includes the steps that first, types of kernel functions and parameters of fuzzy clustering support vector regression machine models are determined, a set cutting fuzzy C-mean clustering algorithm is adopted for clustering data samples, c classes are generated, the fuzzy membership degree of each class of data samples is calculated in combination with k neighboring thought, and c fuzzy clustering support vector regression machine models are generated; the generated c fuzzy clustering support vector regression machine models are used for training training samples in the data samples, the MSE of the training samples is calculated, and when the MSE is larger than the algorithm termination condition (please see the formula in the specifications), circulation is ended, and the final fuzzy clustering support vector regression machine models are determined; if not, after new samples are added in, the size relationship between the total number of the current samples and the sample capacity is judged, and incremental updating and decrement updating are performed. The method has the advantages that wrong values and interference parameters are effectively processed, evaluation accuracy is guaranteed, and online operation accuracy is improved.
Description
Technical field
The present invention relates to a kind of on-line evaluation method of mimic channel performance, be specifically related to a kind of on-line evaluation method of the mimic channel performance based on fuzzy clustering support vector regression.
Background technology
Along with the develop rapidly of electronic technology, the ever-increasing while of the widespread use of mimic channel, complexity and closeness, its reliability of operation is required also day by day improving.And mimic channel is because of diversity, the complicacy of himself and fault thereof, makes existing Performance Evaluation Technique slower development.Existing mimic channel performance evaluation and detection method adopt the evaluation method of instrument and equipment more, and this evaluation method shortcoming is that cost is high, portable poor.
The development of the intellectual technologies such as neural network, fuzzy logic, genetic algorithm, for mimic channel performance evaluation provides the useful space, particularly support vector machine is subject to more wide research and application.Though at present all equipment evaluation methods of realizing take support vector machine as gordian technique have the advantages such as evaluation precision is excellent, portable, low cost, but factory's real current situation is not considered in its existing design, the data that gathered, because being subject to the existing wrong value of factory's actual environment and disturbing, produce wrong evaluation and test therefrom.
Summary of the invention
The technical problem to be solved in the present invention is to provide that a kind of evaluation precision is excellent, portable, low cost, fast, precisely, can realize the effectively on-line evaluation method of the mimic channel performance of processing of wrong value, interference value.
The present invention adopts following technical scheme:
An on-line evaluation method for mimic channel performance, its concrete steps are as follows:
Step 1: determine the kernel function type of fuzzy clustering support vector regression model, fuzzy clustering support vector regression model parameter has initialization clusters number c, carries out the sample size of decrement process
, nearest neighbor point number k, algorithm end condition
;
Step 2: adopt cut set Fuzzy C-Means Clustering Algorithm to carry out cluster to data sample, generate c class, the data sample in every class, in conjunction with k neighbour thought, is calculated to its fuzzy membership, generate c fuzzy clustering support vector regression submodel;
Step 3: use c the fuzzy clustering support vector regression submodel generating to train the training sample in data sample, calculate its root-mean-square error MSE, judge root-mean-square error MSE and algorithm end condition
magnitude relationship, when root-mean-square error MSE> algorithm end condition
, stop circulation, determine final fuzzy clustering support vector regression model; Otherwise, enter step 4;
Step 4: utilize in the data-optimized modeling process in rolling time window, model, along with online updating is carried out in the rolling of time window, adds after new samples, judges the magnitude relationship of current total sample number and sample size, current total sample number M≤sample size
, carry out incremental update, return to step 2; Otherwise adopt decrement to upgrade, return to step 2.
The on-line evaluation method of above-mentioned mimic channel performance, in step 1, the kernel function of fuzzy clustering support vector regression submodel adopts gaussian kernel function, the nuclear parameter penalty factor of gaussian kernel function
with core width
optimizing process as follows:
1.2, by optimizing
collection and
grid search method optimizing parameter pair for centralized procurement
carry out cross validation;
The sample set D of data sample is divided into S group
, arbitrarily
group as training set is
, remain one group as checking collection
, can repeat
inferior, its Generalization Capability P can evaluate by through type (1),
In formula:
be
group checking collection,
for the input of checking collection sample,
for the output of checking collection sample,
for
the parameter vector obtaining during as training sample, by regression equation group
Obtain regression parameter
, wherein,
for gaussian kernel function,
for Lagrange multiplier vector,
for bias vector; With the input of checking collection sample
, obtain the output of system
;
1.3, circulation select parameter to and carry out cross validation, calculate every pair of parameter Generalization Capability P until grid search stop, and then the parameter pair of P minimum in assurance formula (1)
for optimum.
The on-line evaluation method of above-mentioned mimic channel performance, cut set Fuzzy C-Means Clustering in step 2, utilize the data-optimized modeling in rolling time window, model is along with online updating is carried out in the rolling of time window, realize online location to newly adding sample, and upgrade support vector regression submodel, when the mutual clustering algorithm of increase and decrease that adopts is incremental clustering algorithm:
If current sampling instant fuzzy clustering support vector regression model data sample number is M, wherein
, N is off-line sample number;
1.1, initialization exponential factor m, generates the threshold value of new class
, carry out the initialization clusters number d of incremental process;
1.2, calculate the cut set factor
with the newly-increased sample of current sampling instant
to a upper cluster centre that sampling instant is definite
inner product norm
and have
;
1.3, order
, return
corresponding
value, be denoted as
; If simultaneously
, calculate
for
class be subordinate to index
, right
be handled as follows: judgement
with the threshold value that generates new class
magnitude relationship;
If
, and
, the fuzzy membership based on clustering algorithm so
, known by maximum membership grade principle
affiliated class, and by all sample substitutions in such
Wherein,
represent to add and produce after new samples
the new cluster centre of class, z is the subscript of all samples in such,
for the fuzzy membership of all samples of being obtained by cut set Fuzzy C-Means Clustering Algorithm;
When
, so for new samples M sets up a new class, and using this sample as the cluster centre that increases class newly; Increase a fuzzy clustering support vector regression submodel simultaneously, and with the newly-increased model of this sample training;
1.4, compute classes centre deviation is
,
for the error information row that obtain after application model prediction,
for all classes center median, availability deciding is carried out at Dui Lei center, when
,
for median deviation, such,, for disturbing, deletes such; When
, such is wrong value, deletes such; When
, such is normal sample, wherein
;
1.5, along with step 1.2 is returned in the arrival of newly-increased sample, until sample number reaches Sample Maximal capacity
.
The on-line evaluation method of above-mentioned mimic channel performance, cut set Fuzzy C-Means Clustering in step 2, model is along with online updating is carried out in the rolling of time window, utilize the data-optimized modeling in rolling time window, realize online location to newly adding sample, and upgrade support vector regression submodel, when the mutual clustering algorithm of increase and decrease that adopts is decrement clustering algorithm:
If present Fuzzy cluster support vector regression model is M because incremental learning is updated to data sample number,
,
for Sample Maximal capacity;
1.1, initialization exponential factor m, generates the threshold value of new class
, carry out the initialization clusters number of decrement process
;
1.2, find out sample
in corresponding class, be subordinate to minimum element, and delete this element;
1.3, will delete after all sample substitution formulas (3) in such
, obtain such new cluster centre, wherein,
represent to add and produce after new samples
the new cluster centre of class, z is the subscript of all samples in such,
for the fuzzy membership of all samples of being obtained by cut set Fuzzy C-Means Clustering Algorithm;
1.4, calculate the cut set factor
with the newly-increased sample of current sampling instant
cluster centre after upgrading to step 1.3
inner product norm
:
1.5, order
, return
corresponding
value, be denoted as S; If simultaneously
, calculate
for
class be subordinate to index
, right
be handled as follows: judgement
with the threshold value that generates new class
magnitude relationship,
If
and
, the fuzzy membership based on clustering algorithm so
, known by maximum membership grade principle
affiliated class, and by all sample substitutions in such
Wherein,
represent to add and produce after new samples
the new cluster centre of class, z is the subscript of all samples in such,
for the fuzzy membership of all samples of being obtained by cut set Fuzzy C-Means Clustering Algorithm;
2. work as
, so for new samples M sets up a new class, and using this sample as newly-increased Lei Lei center; Increase a fuzzy clustering support vector regression submodel simultaneously, and with the newly-increased model of this sample training;
1.6, compute classes centre deviation is
,
for the error information row that obtain after application model prediction,
for all classes center median, availability deciding is carried out at Dui Lei center, when
,
for median deviation, such,, for disturbing, deletes such; When
, such is wrong value, deletes such; When
, such is normal sample, wherein
;
1.7, along with step 1.2 is returned in the arrival of newly-increased sample, until evaluation procedure finishes.
The on-line evaluation method of above-mentioned mimic channel performance, cut set Fuzzy C-Means Clustering in step 2, utilizes the data-optimized modeling in rolling time window, to newly adding sample to realize online location, and upgrades support vector regression submodel;
In step 2, the fuzzy membership of Fuzzy Support Vector Regression machine is
(6)
In formula,
for the fuzzy membership of all samples of being obtained by cut set Fuzzy C-Means Clustering Algorithm,
for the fuzzy membership based on tight ness rating;
In formula: k is nearest neighbor point number.
The on-line evaluation method of above-mentioned mimic channel performance, in the increment self refresh of the support vector regression model of step 4 is:
Formula (9) represents that current sampling instant fuzzy clustering support vector regression model is existing
individual sample, after next sampling instant data message obtains, the fuzzy clustering support vector regression model after self refresh, in formula:
for the nuclear matrix of current sampling instant,
,
for vector of unit length,
,
,
for model parameter after increase sample,
for Lagrange multiplier after increment self refresh,
for increment self refresh model amount of bias,
for adding the output of model after new samples,
for penalty factor;
The decrement self refresh model parameter of the support vector regression model of step 4 is
In formula,
for Lagrange multiplier after decrement self refresh,
for the output of model after decrement self refresh,
,
for
the element of the capable k row of k, and
The present invention compared with prior art has following beneficial effect:
1) mimic channel performance online evaluation method proposed by the invention, compared with tradition employing exact instrument evaluation method, system cost of the present invention is lower, and real-time is more excellent, good portability, more easily popularization.
2) mimic channel performance online evaluation method proposed by the invention only adopts simple cut set Fuzzy C-cluster to combine with support vector regression, the theory of considering the self refresh of increase and decrease amount has designed mimic channel performance online evaluation method, with utilize robust, the mimic channel evaluation method of the designs such as filtering is compared, the present invention formerly evaluation method is simple in structure, data calculated amount is little, can effectively process wrong value, interference parameter etc., more can effectively promote precision and the speed of mimic channel on-line evaluation, the cost that has solved traditional instrument evaluation method is high, real-time is poor and related algorithm is complicated, calculated amount is large, overabundance of data and overflow problem of producing etc., be applicable to the on-line evaluation of the electronic product performance of mimic channel and the evaluation and test of available oscillograph.
Accompanying drawing explanation
Fig. 1 is that mimic channel performance online of the present invention is evaluated process flow diagram;
Fig. 2 is mimic channel performance online evaluation system structural drawing of the present invention;
The analog circuit test object adopting in Fig. 3 the present invention;
Fig. 4 is the actual curve figure of the mimic channel tested of the present invention;
Fig. 5 is the training sample of the mimic channel tested of the present invention and the regression curve drawing by training sample.
Embodiment
As shown in Figure 1, on-line evaluation concrete steps are as follows:
Step 1, learning training sample.
Adopt cut set Fuzzy C-Means Clustering Algorithm to carry out cluster to data sample, generate c class;
Fuzzy clustering support vector regression model parameter has initialization clusters number c, carries out the sample size of decrement process
, nearest neighbor point number k, algorithm end condition
;
The kernel function of fuzzy clustering support vector regression model adopts gaussian kernel function, the nuclear parameter penalty factor of gaussian kernel function
with core width
optimizing process as follows:
1.2, by optimizing
collection and
grid search method optimizing parameter pair for centralized procurement
carry out cross validation;
The sample set D of data sample is divided into S group
, arbitrarily
group as training set is
, remain one group as checking collection
, can repeat
inferior, its Generalization Capability P can evaluate by through type (1),
In formula:
be
group checking collection,
for the input of checking collection sample,
for the output of checking collection sample,
for
the parameter vector obtaining during as training sample, by regression equation group
Obtain regression parameter
, wherein,
for gaussian kernel function,
for Lagrange multiplier vector,
for bias vector; With the input of checking collection sample
, obtain the output of system
;
1.3, circulation select parameter to and carry out cross validation, calculate every pair of parameter Generalization Capability P until grid search stop, and then the parameter pair of P minimum in assurance formula (1)
for optimum.
Step 3: in conjunction with increase and decrease amount self refresh study, calculate fuzzy membership
To generating c the data sample in class in step 2 in conjunction with k neighbour thought, calculate its fuzzy membership, generate c fuzzy clustering support vector regression submodel;
Wherein, the fuzzy membership of Fuzzy Support Vector Regression machine is
(6)
In formula,
for the fuzzy membership of all samples of being obtained by cut set Fuzzy C-Means Clustering Algorithm,
for the fuzzy membership based on tight ness rating;
step 4: forecast sample is predicted, calculated its root-mean-square error;
Use c the fuzzy clustering support vector regression submodel generating to train the training sample in data sample, calculate its root-mean-square error MSE, judge root-mean-square error MSE and algorithm end condition
magnitude relationship, when root-mean-square error MSE> algorithm end condition
, stop circulation, determine final fuzzy clustering support vector regression model; Otherwise, enter step 5.
Step 5: utilize in the data-optimized modeling process in rolling time window, model, along with online updating is carried out in the rolling of time window, adds after new samples, judges the magnitude relationship of current total sample number and sample size, current total sample number M≤sample size
, carry out incremental update, return to step 2; Otherwise adopt decrement to upgrade, return to step 2.
While further illustrating cut set Fuzzy C-Means Clustering, utilize the data-optimized modeling in rolling time window, model is along with online updating is carried out in the rolling of time window, realize online location to newly adding sample, and upgrade support vector regression submodel, when the mutual clustering algorithm of increase and decrease that adopts is incremental clustering algorithm:
If current sampling instant fuzzy clustering support vector regression model data sample number is M, wherein
, N is off-line sample number;
1.1, initialization exponential factor m=2, generates the threshold value of new class
=0.1, carry out the initialization clusters number d of incremental process;
1.2, calculate the cut set factor
with the newly-increased sample of current sampling instant
to a upper cluster centre that sampling instant is definite
inner product norm
and have
;
1.3, order
, return
corresponding
value, be denoted as
; If simultaneously
, calculate
for
class be subordinate to index
, right
be handled as follows: judgement
with the threshold value that generates new class
magnitude relationship;
If
, and
, the fuzzy membership based on clustering algorithm so
, known by maximum membership grade principle
affiliated class, and by all sample substitutions in such
Wherein,
represent to add and produce after new samples
the new cluster centre of class, z is the subscript of all samples in such,
for the fuzzy membership of all samples of being obtained by cut set Fuzzy C-Means Clustering Algorithm;
When
, so for new samples M sets up a new class, and using this sample as the cluster centre that increases class newly; Increase a fuzzy clustering support vector regression submodel simultaneously, and with the newly-increased model of this sample training;
1.4, compute classes centre deviation is
,
for the error information row that obtain after application model prediction,
for all classes center median, availability deciding is carried out at Dui Lei center, when
,
for median deviation, such,, for disturbing, deletes such; When
, such is wrong value, deletes such; When
, such is normal sample, wherein
;
1.5, along with step 1.2 is returned in the arrival of newly-increased sample, until sample number reaches Sample Maximal capacity
.
When the mutual clustering algorithm of increase and decrease that adopts is decrement clustering algorithm:
If present Fuzzy cluster support vector regression model is M because incremental learning is updated to data sample number,
,
for Sample Maximal capacity;
1.1, initialization exponential factor m=2, generates the threshold value of new class
=0.1, carry out the initialization clusters number of decrement process
;
1.2, find out sample
in corresponding class, be subordinate to minimum element, and delete this element;
1.3, will delete after all sample substitution formulas (3) in such
, obtain such new cluster centre, wherein,
represent to add and produce after new samples
the new cluster centre of class, z is the subscript of all samples in such,
for the fuzzy membership of all samples of being obtained by cut set Fuzzy C-Means Clustering Algorithm;
1.4, calculate the cut set factor
with the newly-increased sample of current sampling instant
cluster centre after upgrading to step 1.3
inner product norm
:
1.5, order
, return
corresponding
value, be denoted as S; If simultaneously
, calculate
for
class be subordinate to index
, right
be handled as follows: judgement
with the threshold value that generates new class
magnitude relationship,
1. work as
,
If
and
, the fuzzy membership based on clustering algorithm so
, known by maximum membership grade principle
affiliated class, and by all sample substitutions in such
Wherein,
represent to add and produce after new samples
the new cluster centre of class, z is the subscript of all samples in such,
for the fuzzy membership of all samples of being obtained by cut set Fuzzy C-Means Clustering Algorithm;
2. work as
, so for new samples M sets up a new class, and using this sample as newly-increased Lei Lei center; Increase a fuzzy clustering support vector regression submodel simultaneously, and with the newly-increased model of this sample training;
1.6, compute classes centre deviation is
,
for the error information row that obtain after application model prediction,
for all classes center median, availability deciding is carried out at Dui Lei center, when
,
for median deviation, such,, for disturbing, deletes such; When
, such is wrong value, deletes such; When
, such is normal sample, wherein
;
1.7, along with step 1.2 is returned in the arrival of newly-increased sample, until evaluation procedure finishes.
The increment self refresh of support vector regression model is:
Formula (9) represents that current sampling instant fuzzy clustering support vector regression model is existing
individual sample, after next sampling instant data message obtains, the fuzzy clustering support vector regression model after self refresh, in formula:
for the nuclear matrix of current sampling instant,
,
for vector of unit length,
,
,
for model parameter after increase sample,
for Lagrange multiplier after increment self refresh,
for increment self refresh model amount of bias,
for adding the output of model after new samples,
for penalty factor;
The decrement self refresh model parameter of support vector regression model is
In formula,
for Lagrange multiplier after decrement self refresh,
for the output of model after decrement self refresh,
,
for
the element of the capable k row of k, and
。
Take the mimic channel of Fig. 3 as tested object, its actual performance curve as shown in Figure 4.Mimic channel by the on-line evaluation method of mimic channel performance of the present invention for Fig. 3, as shown in Figure 2, first measurand is carried out to data acquisition, after input computer system, the Performance Appraisal System forming by the on-line evaluation method of mimic channel performance of the present invention is tested.The training sample of the mimic channel of test and pick recurrence performance curve that the training sample after default, interference parameter draws as Fig. 5.Can be found out by Fig. 4 and Fig. 5, the actual performance curve of mimic channel and the recurrence performance curve with on-line evaluation method acquisition of the present invention match, and mimic channel method of evaluating performance of the present invention is described, the performance state of energy precisetest mimic channel.
Evaluation method of the present invention can be extended in the performance online evaluation of the electronic product using oscillograph as evaluating tool.
Claims (6)
1. an on-line evaluation method for mimic channel performance, is characterized in that: its concrete steps are as follows:
Step 1: determine the kernel function type of fuzzy clustering support vector regression model, fuzzy clustering support vector regression model parameter has initialization clusters number c, carries out the sample size of decrement process
, nearest neighbor point number k, algorithm end condition
;
Step 2: adopt cut set Fuzzy C-Means Clustering Algorithm to carry out cluster to data sample, generate c class, the data sample in every class, in conjunction with k neighbour thought, is calculated to its fuzzy membership, generate c fuzzy clustering support vector regression submodel;
Step 3: use c the fuzzy clustering support vector regression submodel generating to train the training sample in data sample, calculate its root-mean-square error MSE, judge root-mean-square error MSE and algorithm end condition
magnitude relationship, when root-mean-square error MSE> algorithm end condition
, stop circulation, determine final fuzzy clustering support vector regression model; Otherwise, enter step 4;
Step 4: utilize the data-optimized modeling in rolling time window, model, along with online updating is carried out in the rolling of time window, adds after new samples, judges the magnitude relationship of current total sample number and sample size, current total sample number M≤sample size
, carry out incremental update, return to step 2; Otherwise adopt decrement to upgrade, return to step 2.
2. the on-line evaluation method of mimic channel performance according to claim 1, is characterized in that: in step 1, the kernel function of fuzzy clustering support vector regression submodel adopts gaussian kernel function, the nuclear parameter penalty factor of gaussian kernel function
with core width
optimizing process as follows:
1.2, by optimizing
collection and
grid search method optimizing parameter pair for centralized procurement
carry out cross validation;
The sample set D of data sample is divided into S group
, arbitrarily
group as training set is
, remain one group as checking collection
, can repeat
inferior, its Generalization Capability P can evaluate by through type (1),
(1)
In formula:
be
group checking collection,
for the input of checking collection sample,
for the output of checking collection sample,
for
the parameter vector obtaining during as training sample, by regression equation group
(2)
Obtain regression parameter
, wherein,
for gaussian kernel function,
for Lagrange multiplier vector,
for bias vector; With the input of checking collection sample
, obtain the output of system
;
1.3, circulation select parameter to and carry out cross validation, calculate every pair of parameter Generalization Capability P until grid search stop, and then the parameter pair of P minimum in assurance formula (1)
for optimum.
3. the on-line evaluation method of mimic channel performance according to claim 1, it is characterized in that: cut set Fuzzy C-Means Clustering in step 2, utilize the data-optimized modeling in rolling time window, model is along with online updating is carried out in the rolling of time window, realize online location to newly adding sample, and upgrade support vector regression submodel, when the mutual clustering algorithm of increase and decrease that adopts is incremental clustering algorithm:
If current sampling instant fuzzy clustering support vector regression model data sample number is M, wherein
, N is off-line sample number;
1.1, initialization exponential factor m, generates the threshold value of new class
, carry out the initialization clusters number d of incremental process;
1.2, calculate the cut set factor
with the newly-increased sample of current sampling instant
to a upper cluster centre that sampling instant is definite
inner product norm
and have
;
1.3, order
, return
corresponding
value, be denoted as
; If simultaneously
, calculate
for
class be subordinate to index
, right
be handled as follows: judgement
with the threshold value that generates new class
magnitude relationship;
If
, and
, the fuzzy membership based on clustering algorithm so
, known by maximum membership grade principle
affiliated class, and by all sample substitutions in such
Wherein,
represent to add and produce after new samples
the new cluster centre of class, z is the subscript of all samples in such,
for the fuzzy membership of all samples of being obtained by cut set Fuzzy C-Means Clustering Algorithm;
When
, so for new samples M sets up a new class, and using this sample as the cluster centre that increases class newly; Increase a fuzzy clustering support vector regression submodel simultaneously, and with the newly-increased model of this sample training;
1.4, compute classes centre deviation is
,
for the error information row that obtain after application model prediction,
for all classes center median, availability deciding is carried out at Dui Lei center, when
,
for median deviation, such,, for disturbing, deletes such; When
, such is wrong value, deletes such; When
, such is normal sample, wherein
;
1.5, along with step 1.2 is returned in the arrival of newly-increased sample, until sample number reaches Sample Maximal capacity
.
4. the on-line evaluation method of mimic channel performance according to claim 1, it is characterized in that: cut set Fuzzy C-Means Clustering in step 2, model is along with online updating is carried out in the rolling of time window, utilize the data-optimized modeling in rolling time window, realize online location to newly adding sample, and upgrade support vector regression submodel, when the mutual clustering algorithm of increase and decrease that adopts is decrement clustering algorithm:
If present Fuzzy cluster support vector regression model is M because incremental learning is updated to data sample number,
,
for Sample Maximal capacity;
1.1, initialization exponential factor m, generates the threshold value of new class
, carry out the initialization clusters number of decrement process
;
1.2, find out sample
in corresponding class, be subordinate to minimum element, and delete this element;
1.3, will delete after all sample substitution formulas (3) in such
, obtain such new cluster centre, wherein,
represent to add and produce after new samples
the new cluster centre of class, z is the subscript of all samples in such,
for the fuzzy membership of all samples of being obtained by cut set Fuzzy C-Means Clustering Algorithm;
1.4, calculate the cut set factor
with the newly-increased sample of current sampling instant
cluster centre after upgrading to step 1.3
inner product norm
:
1.5, order
, return
corresponding
value, be denoted as S; If simultaneously
, calculate
for
class be subordinate to index
, right
be handled as follows: judgement
with the threshold value that generates new class
magnitude relationship,
If
and
, the fuzzy membership based on clustering algorithm so
, known by maximum membership grade principle
affiliated class, and by all sample substitutions in such
Wherein,
represent to add and produce after new samples
the new cluster centre of class, z is the subscript of all samples in such,
for the fuzzy membership of all samples of being obtained by cut set Fuzzy C-Means Clustering Algorithm;
2. work as
, so for new samples M sets up a new class, and using this sample as newly-increased Lei Lei center; Increase a fuzzy clustering support vector regression submodel simultaneously, and with the newly-increased model of this sample training;
1.6, compute classes centre deviation is
,
for the error information row that obtain after application model prediction,
for all classes center median, availability deciding is carried out at Dui Lei center, when
,
for median deviation, such,, for disturbing, deletes such; When
, such is wrong value, deletes such; When
, such is normal sample, wherein
;
1.7, along with step 1.2 is returned in the arrival of newly-increased sample, until evaluation procedure finishes.
5. the on-line evaluation method of mimic channel performance according to claim 1, it is characterized in that: cut set Fuzzy C-Means Clustering in step 2, utilize the data-optimized modeling in rolling time window, to newly adding sample to realize online location, and upgrade support vector regression submodel;
In step 2, the fuzzy membership of Fuzzy Support Vector Regression machine is
In formula,
for the fuzzy membership of all samples of being obtained by cut set Fuzzy C-Means Clustering Algorithm,
for the fuzzy membership based on tight ness rating;
(7)
(8)
In formula: k is nearest neighbor point number.
6. the on-line evaluation method of mimic channel performance according to claim 1, is characterized in that: the increment self refresh at the support vector regression model of step 4 is:
(9)
Formula (9) represents that current sampling instant fuzzy clustering support vector regression model is existing
individual sample, after next sampling instant data message obtains, the fuzzy clustering support vector regression model after self refresh, in formula:
for the nuclear matrix of current sampling instant,
,
for vector of unit length,
,
,
for model parameter after increase sample,
for Lagrange multiplier after increment self refresh,
for increment self refresh model amount of bias,
for adding the output of model after new samples,
for penalty factor;
The decrement self refresh model parameter of the support vector regression model of step 4 is
In formula,
for Lagrange multiplier after decrement self refresh,
for the output of model after decrement self refresh,
,
for
the element of the capable k row of k, and
。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410123824.8A CN103885867B (en) | 2014-03-28 | 2014-03-28 | Online evaluation method of performance of analog circuit |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410123824.8A CN103885867B (en) | 2014-03-28 | 2014-03-28 | Online evaluation method of performance of analog circuit |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103885867A true CN103885867A (en) | 2014-06-25 |
CN103885867B CN103885867B (en) | 2015-04-01 |
Family
ID=50954775
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410123824.8A Expired - Fee Related CN103885867B (en) | 2014-03-28 | 2014-03-28 | Online evaluation method of performance of analog circuit |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103885867B (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107218964A (en) * | 2017-05-23 | 2017-09-29 | 中国人民解放军国防科学技术大学 | A kind of decision method of test sample capacity character |
CN109147875A (en) * | 2018-08-08 | 2019-01-04 | 合肥学院 | The dissolution of contaminated water oxygen concentration prediction technique of support vector regression algorithm based on fuzzy clustering |
CN109377110A (en) * | 2018-12-13 | 2019-02-22 | 洛阳博得天策网络科技有限公司 | Evaluation method and system for brand content assets |
CN112485650A (en) * | 2020-11-30 | 2021-03-12 | 电子科技大学 | Analog circuit fault parameter range identification method based on PBI |
CN113076693A (en) * | 2021-04-02 | 2021-07-06 | 东南大学 | Road surface compaction quality evaluation method based on support vector machine and hidden horse model |
CN115358178A (en) * | 2022-08-11 | 2022-11-18 | 山东大学 | Circuit yield analysis method based on fusion neural network |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102087337A (en) * | 2009-12-04 | 2011-06-08 | 哈尔滨理工大学 | Annealing genetic optimization method for diagnosing excitation of nonlinear analog circuit |
CN103245907A (en) * | 2013-01-30 | 2013-08-14 | 中国人民解放军海军航空工程学院 | Artificial circuit fault diagnosis pattern sorting algorithm |
-
2014
- 2014-03-28 CN CN201410123824.8A patent/CN103885867B/en not_active Expired - Fee Related
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102087337A (en) * | 2009-12-04 | 2011-06-08 | 哈尔滨理工大学 | Annealing genetic optimization method for diagnosing excitation of nonlinear analog circuit |
CN103245907A (en) * | 2013-01-30 | 2013-08-14 | 中国人民解放军海军航空工程学院 | Artificial circuit fault diagnosis pattern sorting algorithm |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107218964A (en) * | 2017-05-23 | 2017-09-29 | 中国人民解放军国防科学技术大学 | A kind of decision method of test sample capacity character |
CN107218964B (en) * | 2017-05-23 | 2020-01-24 | 中国人民解放军国防科学技术大学 | Method for judging capacity character of test subsample |
CN109147875A (en) * | 2018-08-08 | 2019-01-04 | 合肥学院 | The dissolution of contaminated water oxygen concentration prediction technique of support vector regression algorithm based on fuzzy clustering |
CN109377110A (en) * | 2018-12-13 | 2019-02-22 | 洛阳博得天策网络科技有限公司 | Evaluation method and system for brand content assets |
CN112485650A (en) * | 2020-11-30 | 2021-03-12 | 电子科技大学 | Analog circuit fault parameter range identification method based on PBI |
CN112485650B (en) * | 2020-11-30 | 2021-09-14 | 电子科技大学 | Analog circuit fault parameter range identification method based on PBI |
CN113076693A (en) * | 2021-04-02 | 2021-07-06 | 东南大学 | Road surface compaction quality evaluation method based on support vector machine and hidden horse model |
CN115358178A (en) * | 2022-08-11 | 2022-11-18 | 山东大学 | Circuit yield analysis method based on fusion neural network |
CN115358178B (en) * | 2022-08-11 | 2023-04-07 | 山东大学 | Circuit yield analysis method based on fusion neural network |
Also Published As
Publication number | Publication date |
---|---|
CN103885867B (en) | 2015-04-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Sun et al. | Using Bayesian deep learning to capture uncertainty for residential net load forecasting | |
CN103885867B (en) | Online evaluation method of performance of analog circuit | |
Corizzo et al. | Anomaly detection and repair for accurate predictions in geo-distributed big data | |
CN112686464A (en) | Short-term wind power prediction method and device | |
CN109242041A (en) | A kind of electric energy meter abnormal deviation data examination method, device, equipment and storage medium | |
Yang et al. | Computational optimization, modelling and simulation: Recent trends and challenges | |
Nichiforov et al. | Deep learning techniques for load forecasting in large commercial buildings | |
WO2015085308A1 (en) | Computer-implemented data analysis methods and systems for wind energy assessments | |
EP3739494A2 (en) | Method, apparatus, system, and program for optimizing solid electrolytes for li-ion batteries using bayesian optimization | |
CN103488869A (en) | Wind power generation short-term load forecast method of least squares support vector machine | |
De Wilde et al. | Building simulation approaches for the training of automated data analysis tools in building energy management | |
CN104899135A (en) | Software defect prediction method and system | |
CN107730097B (en) | Bus load prediction method and device and computing equipment | |
JP2019537079A (en) | How to build stochastic models for large-scale renewable energy data | |
CN103559303A (en) | Evaluation and selection method for data mining algorithm | |
CN109981749A (en) | A kind of cloud workflow task running time prediction method promoted based on limit gradient | |
CN117454492A (en) | Electromechanical construction method and system based on BIM technology | |
Lu et al. | Neural network interpretability for forecasting of aggregated renewable generation | |
Xiyun et al. | Wind power probability interval prediction based on bootstrap quantile regression method | |
CN108205713A (en) | A kind of region wind power prediction error distribution determination method and device | |
CN103679484A (en) | Novel method for analyzing E-commerce consistency based on behavior Petri network | |
Li et al. | Improved LSTM-based prediction method for highly variable workload and resources in clouds | |
CN108694472B (en) | Prediction error extreme value analysis method, device, computer equipment and storage medium | |
CN113408808A (en) | Training method, data generation method, device, electronic device and storage medium | |
Zhang et al. | RSVRs based on feature extraction: a novel method for prediction of construction projects’ costs |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20150401 Termination date: 20160328 |