CN106897774B - Multiple soft measurement algorithm cluster modeling methods based on Monte Carlo cross validation - Google Patents
Multiple soft measurement algorithm cluster modeling methods based on Monte Carlo cross validation Download PDFInfo
- Publication number
- CN106897774B CN106897774B CN201710056647.XA CN201710056647A CN106897774B CN 106897774 B CN106897774 B CN 106897774B CN 201710056647 A CN201710056647 A CN 201710056647A CN 106897774 B CN106897774 B CN 106897774B
- Authority
- CN
- China
- Prior art keywords
- algorithm
- sample
- monte carlo
- submodel
- outlier
- 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.)
- Expired - Fee Related
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N7/00—Computing arrangements based on specific mathematical models
- G06N7/01—Probabilistic graphical models, e.g. probabilistic networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Abstract
Multiple soft measurement algorithm cluster modeling methods based on Monte Carlo cross validation that the invention discloses a kind of, aspect is searched to the model evaluation aspect and outlier of Models Sets group analysis method and has carried out systematic difference, attempts integrated leaming system of the construction one under Models Sets group analysis method frame.In terms of outlier lookup, a variety of soft measurement algorithms are selected to model industrial process after rejecting outlier using the Monte Carlo cross validation algorithm based on Models Sets group analysis method, then integrate to the prediction result of each soft measurement algorithm.In terms of model evaluation, by generating big quantity training cluster, influence of the selection of training set for model evaluation result is eliminated, the diversity of data is improved.Compared to current other methods, the present invention improves the accuracy of prediction, from the variation of whole prediction effect of each algorithm of statistical angle analysis before and after rejecting outlier by rejecting outlier.
Description
Technical field
The invention belongs to industrial process control field more particularly to a kind of multiple soft surveys based on Monte Carlo cross validation
Quantity algorithm cluster modeling method.
Background technique
In recent years, due to industrial benefit and quality requirement, hard measurement has become an important field of research.
In the process industrials such as chemical industry, fermentation, biology, metallurgy, petroleum, food, to realize bounder control, running on process units
Optimal working condition, the more quality products of production, it is necessary to many important process variables of strict control.However, often
It is difficult to use in line sensor directly these important process variables to be measured to come, therefore there have been the methods of hard measurement.
It is necessary to carry out data prediction, such as the rejecting of data normalization, outlier etc. before soft sensor modeling.Normalizing
Change processing is the number selected data become according to given method between [0,1].Normalized purpose is that data is made to exist as far as possible
On the same order of magnitude, avoid keeping prediction result inaccurate because of each differing greatly for variable.Outlier refers in data
There are one or several numerical value biggish data of diversity ratio compared with other numerical value, is also singular value, extremum.Its Producing reason
It is the contingency due to experiment condition and method, or fault or true and normal data when observation, record, calculating,
Only showed in current experiment some are extreme.The presence of outlier will affect the estimated performance of entire model, therefore having must
It is rejected.
The present invention searches aspect to the model evaluation aspect and outlier of Models Sets group analysis method and has carried out answering for system
With integrated leaming system of the trial construction one under Models Sets group analysis method frame.In terms of outlier lookup, using base
A variety of soft measurement algorithms are selected after rejecting outlier in the Monte Carlo cross validation algorithm MC of Models Sets group analysis method
Industrial process is modeled, then the prediction result of each soft measurement algorithm is integrated.In terms of model evaluation, pass through generation
Big quantity training cluster, eliminates influence of the selection of training set for model evaluation result, improves the diversity of data.With it is previous
The method that model evaluation is carried out from a single output valve it is different, the output of Models Sets group analysis method is one point
Cloth, so as to obtain more conclusions from statistical angle.
Summary of the invention
It is an object of the invention to the hypothesis limitations for existing method, provide a kind of based on Monte Carlo cross validation
Multiple soft measurement algorithm cluster modeling methods.
The purpose of the present invention is achieved through the following technical solutions: multiple soft surveys based on Monte Carlo cross validation
Quantity algorithm cluster modeling method, which comprises the following steps:
(1) industrial process data sample is inputted to system, all samples is normalized, so that each variable
Mean value is 0, variance 1.These data are stored in historical data base.
(2) the Monte Carlo cross validation algorithm MC based on Models Sets group analysis method is used, outlier is rejected.Models Sets
Group analysis method step is:
A) random generation N (N is greater than 0 natural number) a Sub Data Set is concentrated from initial data with monte carlo method;
B) submodel (classification or recurrence) is established to each Sub Data Set;
C) for statistical analysis to the output of N number of submodel.
The step of Monte Carlo cross validation algorithm, is formulated according to the frame of Models Sets group analysis method, its step
Suddenly it is:
A) N number of training subset and N number of test subset are generated at random from initial data concentration with monte carlo method;
B) N number of model is established, is predicted on N number of test set;
C) each of working as sample for raw data set has a group to predict error, unites to this group of prediction errors
Meter analysis.
The mean value and variance that the prediction error of each sample can be calculated, the mean value for predicting error and variance are made respectively
For abscissa and ordinate, so that it may obtain piece image, by segmenting the image into four regions, normal sample can be distinguished
And outlier.It predicts error mean and the smallest region of variance, is exactly normal sample local area.The sample in other regions is rejected, that
Outlier removal just completes.
N (n is greater than 0 natural number) a normal sample after rejecting outlier is stored in historical data base.According still further to mould
Type cluster analysis method and step carries out following step (3) to (5).
(3) it is taken out in the data set using the binary matrix sampling method BMS based on Monte carlo algorithm after rejecting outlier
Take the sample of a% (50≤a≤80) as training sample, it is remaining to be used as test sample, it comprises the concrete steps that: in normal sample collection
The sample mark 1 of middle random selection a%, remaining mark 0, marking 1 position, to represent the sample selected as training set sample, mark 0
Be used as test set sample.BMS duplicate sampling n times, symbiosis is at N number of training set and N number of test set.These data are stored in history
Database.
(4) a soft measurement algorithm conduct of the m natural number of 3≤m≤8 (m for) suitable for different industrial process characteristics is selected
The subalgorithm of integrated study.N number of training set is called from database, and these training sets are distinguished with hard measurement submodel algorithm
N number of submodel is established, corresponding N number of test set in database is then called to be predicted.If the prediction that each submodel obtains
As a result are as follows:
yi,j(t), i=1,2 ..., m, j=1,2 ..., N, t=1,2 ..., n × (1-a%)
If the legitimate reading of test set are as follows:
Yi(t), i=1,2 ..., N, t=1,2 ..., n × (1-a%)
The respective N number of root-mean-square error RMSEP of each submodel algorithm is acquired by following formulai,j:
It will be spare in modeling data and each hard measurement submodel algorithm parameter deposit historical data base.
(5) modeling data is called from database, and m hard measurement submodel algorithm is carried out using Bayes's Integrated Algorithm
It is integrated, the weight of each subalgorithm distribution is obtained, is stored in spare in historical data base, the specific steps are as follows:
A) the coefficient Z of Bayes's Integrated Algorithm is acquired by following formulai:
If MiI-th of subalgorithm model is represented, bayesian prior probability is set as:
If S represents test data set, Bayesian likelihood probability are as follows:
Bayes posterior probability are as follows:
B) summation finally is weighted to the predicted value of each submodel again, obtains final predicted value:
Off-line modeling is completed.Calculate N number of root-mean-square error of Bayes's integrated model.By each submodel and Bayes
Root-mean-square error before and after the rejecting outlier of integrated model is for statistical analysis, evaluation model superiority and inferiority.
(6) by online process data input system, after m × N number of soft measurement algorithm model prediction, according to obtaining before
Weight, pass through the integrated final prediction result distribution for acquiring process data to be predicted of Bayes.
The beneficial effects of the present invention are: the present invention is directed to outlier problems, propose based on Monte Carlo cross validation
Multiple soft measurement algorithm cluster modeling methods.Compared to other current flexible measurement methods, the present invention is improved by rejecting outlier
The accuracy of prediction, and model cluster analysis is used to carry out model evaluation, from statistical angle carry out evaluation model
Quality, it can thus be seen that the variation of whole prediction effect of each algorithm before and after rejecting outlier.
Detailed description of the invention
Fig. 1 is the flow chart of Models Sets group analysis method;
Fig. 2 is normal sample and outlier region segmentation schematic diagram;
Fig. 3 is the schematic diagram of binary matrix method of sampling BMS;
Fig. 4 is the flow chart of multiple soft measurement algorithm cluster modeling methods based on Monte Carlo cross validation;
Fig. 5 is the mean value and variance distribution map of the prediction error of each sample;
Fig. 6 is the root-mean-square error distribution map of tetra- sample points of A, B, C, D;
Fig. 7 is the statistical chart of root-mean-square error distribution of each algorithm before and after outlier rejecting, wherein (a) is PCR calculation
The statistical chart of root-mean-square error distribution of the method before and after outlier rejecting is (b) that PLS algorithm is square before and after outlier rejecting
The statistical chart of root error distribution, (c) statistical chart being distributed for root-mean-square error of the ICR algorithm before and after outlier rejecting (d) are
The statistical chart of root-mean-square error distribution of the KPLS algorithm before and after outlier rejecting is (e) BP algorithm before and after outlier rejecting
Root-mean-square error distribution statistical chart.
Specific embodiment
The present invention searches the model evaluation aspect and outlier of Models Sets group analysis method to overcome outlier problems
Aspect has carried out systematic difference, attempts integrated leaming system of the construction one under Models Sets group analysis method frame.From
Group's value searches aspect, using the Monte Carlo cross validation algorithm MC based on Models Sets group analysis method, after rejecting outlier,
It selects a variety of soft measurement algorithms to model industrial process, then the prediction result of each soft measurement algorithm is integrated.In mould
Type evaluation aspect is eliminated influence of the selection of training set for model evaluation result, is improved by generating big quantity training cluster
The diversity of data.
Illustrate effectiveness of the invention below in conjunction with the example of a specific industrial process.In ammonia synthesis process process
In, methane decarburization unit can generate hydrogen, and carbon is but still with CO and CO2Form exist.The work of high-low temperature degree converting unit
With being exactly that CO is converted into CO2, and CO2It can be by CO2Absorption tower absorbs, and re-uses in urea synthesizing unit as raw material.
CO transformation is successively to carry out in the process by following reaction formula:
Gas contains only CO 0.29% (butt volume) after transformation.Conversion reaction increases H2, while generating and CO equivalent
CO2.The reaction carries out in the presence of catalyst, and low temperature and high vapour concentration are conducive to balance, and high temperature is conducive to reaction speed,
But high vapour concentration will be such that reaction speed is substantially reduced, because atm number will lead to shortening (with catalyst) time of contact.High-low temperature
Converting unit totally 27 variables are spent, as shown in table 1, including 26 conventional process variables and 1 quality variable, CO is exported and contains
Amount.
With reference to Fig. 4, according to the flow chart of multiple soft measurement algorithm cluster modeling methods based on Monte Carlo cross validation,
The key step difference of the technical solution adopted by the present invention is as follows:
The first step inputs 3000 industrial process data samples to system, and all samples are normalized, so that
The mean value of each variable is 0, variance 1.These data are stored in historical data base.
Monte Carlo cross validation algorithm MC of the second step based on Models Sets group analysis method rejects outlier.Model
Cluster analysis method as shown in Figure 1, step is:
A) random generation N (N is greater than 0 natural number) a Sub Data Set is concentrated from initial data with monte carlo method;
B) submodel (classification or recurrence) is established to each Sub Data Set;
C) for statistical analysis to the output of N number of submodel.
The step of Monte Carlo cross validation algorithm, is formulated according to the frame of Models Sets group analysis method, its step
Suddenly it is:
A) N (N=1000) a training is generated at random from 3000 industrial process data sample sets with monte carlo method
Subset and N number of test subset;
B) N number of model is established, is predicted on N number of test set;
C) each of working as sample for raw data set has a group to predict error, unites to this group of prediction errors
Meter analysis.
The mean value and variance that the prediction error of each sample can be calculated, the mean value for predicting error and variance are made respectively
For abscissa and ordinate, image as shown in Figure 5 is obtained.Pass through average value MEAN=0.01's and variance STD=0.0004
Straight line segments the image into four regions, and the region of error mean and the smallest lower left corner of variance is predicted in Fig. 5, is exactly normal sample
Local area.The sample in other regions is rejected, n (n=2733) a normal sample is left.
The distribution of the prediction error of point A, B, C, the D respectively in four regions is drawn in Fig. 6.It will be appreciated from fig. 6 that being located at just
The D point prediction error of normal sample area is minimum, and it is to peel off that the variance or mean value of three point prediction errors in other regions are bigger
Value.For 3000 samples, Monte Carlo cross validation method can make each sample point have a group prediction error original
Because being: 1000 wheels are repeated in Monte Carlo, in each round, choose 70% sample as training set, remaining 30% sample
As test set.For a sample, it is drawn in epicycle as training set sample, in next round it is possible that by drawing
As test set sample, it is drawn the probability as test set sample also close to 30%, that is, 1000 wheels in 1000 wheels
After end, a sample has more than 300 prediction error, therefore can acquire more than respective 300 prediction to each sample
The mean value and variance of error, draw Fig. 6, judge outlier.As shown in the frequency histogram of Fig. 6, each sample of A, B, C, D is pre-
The frequency for surveying error is all more than 300.
2733 normal samples after rejecting outlier are stored in historical data base.It is walked according still further to Models Sets group analysis method
The rapid following step three that carries out is to step 5.
Third step is using the binary matrix sampling method BMS based on Monte carlo algorithm from the data set after rejecting outlier
Extract a% (a=70) sample be used as training sample, it is remaining be used as test sample, comprise the concrete steps that: normal sample concentration
Randomly choose the sample mark 1 of a%, remaining mark 0, marking 1 position, to represent the sample selected as training set sample, marks 0
As test set sample.BMS duplicate sampling N (N=1000) is secondary, and symbiosis is at N number of training set and N number of test set.As shown in figure 3,
Every a line represents a wheel sampled result, and each column represent the situation that the sample is selected in each wheel sampling.These data are deposited
Enter historical data base.
4th step selects son of m (m=5) a soft measurement algorithm suitable for different industrial process characteristics as integrated study
Algorithm.Selection is suitable for principle component analysis PCR, the Partial Least Squares PLS of linear process, the BP mind suitable for non-linear process
Independent component analysis method ICR through network, KPLS core least square method, and suitable for nongausian process is as integrated study
Submodel algorithm.By testing repeatedly, the pivot number of PCR, PLS are selected as 14, the independent pivot number of ICR is selected as 11,
The nuclear parameter of KPLS is selected as 35, and the number of plies of BP neural network selects single layer, number of nodes selection 3.N number of training is called from database
Collection, and N number of submodel is established respectively to these training sets with hard measurement submodel algorithm, then call corresponding N in database
A test set is predicted.If the prediction result that each submodel obtains are as follows:
yi,j(t), i=1,2 ..., m, j=1,2 ..., N, t=1,2 ..., n × (1-a%)
If the legitimate reading of test set are as follows:
Yi(t), i=1,2 ..., N, t=1,2 ..., n × (1-a%)
The respective N number of root-mean-square error RMSEP of each submodel algorithm is acquired by following formulai,j:
It will be spare in modeling data and each hard measurement submodel algorithm parameter deposit historical data base.
5th step calls modeling data from database, using Bayes's Integrated Algorithm to m hard measurement submodel algorithm
It is integrated, obtains the weight of each subalgorithm distribution, be stored in spare in historical data base, the specific steps are as follows:
A) the coefficient Z of Bayes's Integrated Algorithm is acquired by following formulai:
If MiI-th of subalgorithm model is represented, bayesian prior probability is set as:
If S represents test data set, Bayesian likelihood probability are as follows:
Bayes posterior probability are as follows:
B) summation finally is weighted to the predicted value of each submodel again, obtains final predicted value:
Off-line modeling is completed.Calculate N number of root-mean-square error of Bayes's integrated model.As shown in fig. 7, by each submodule
Root-mean-square error before and after the rejecting outlier of type and Bayes's integrated model is for statistical analysis, evaluation model superiority and inferiority.
6th step is by online process data input system, after m × N number of soft measurement algorithm model prediction, according to before
Obtained weight passes through the integrated final prediction result distribution for acquiring process data to be predicted of Bayes.
Fig. 7 is the statistical chart of root-mean-square error distribution of each algorithm before and after outlier rejecting, and (a) is that PCR algorithm exists
Outlier rejects the statistical chart of the root-mean-square error distribution of front and back, (b) misses for root mean square of the PLS algorithm before and after outlier rejecting
The statistical chart of difference cloth, (c) statistical chart being distributed for root-mean-square error of the ICR algorithm before and after outlier rejecting, (d) is KPLS
The statistical chart of root-mean-square error distribution of the algorithm before and after outlier rejecting, it is (e) equal before and after outlier rejecting for BP algorithm
The statistical chart of square error distribution.As shown in Figure 7 for Bayes's integrated model and any one subalgorithm model, with illiteracy
The distribution that special Carlow cross validation method rejects the root-mean-square error after outlier more keeps left, and root-mean-square error RMSEP's is averaged
Value is smaller, and the standard deviation of distribution also has reduction substantially, that is, the precision of prediction rejected after outlier improves.
It may finally obtain following conclusion:
Conclusion one, in terms of model evaluation, the index of model evaluation is logical in hard measurement prediction, and the most commonly used is root mean square mistakes
Difference, but in Models Sets group analysis method, the evaluation index of prediction result quality is the distribution of root-mean-square error.It is special using covering
The method of Carlow stochastical sampling, this can reduce shadow of the selection for model evaluation result of training set sample to the maximum extent
It rings, reduces the dependence for being selected the subsample as training set;And carry out the good of evaluation model from statistical angle
It is bad, it may be seen that the whole prediction effect variation before and after rejecting outlier, can obtain more information.
Conclusion two, in terms of outlier rejecting, the Monte Carlo cross validation algorithm based on model cluster analysis is to each
A sample point all obtains a group prediction error, according to the mean value and variance available sample point of the prediction error of each sample
Distribution map, by can clearly select normal sample local area in figure according to region division.
According to an embodiment of the invention, multiple soft measurement algorithm cluster modeling methods based on Monte Carlo cross validation exist
After eliminating outlier, prediction accuracy is improved, carries out model evaluation using model cluster analysis, from statistical angle
Degree carrys out the quality of evaluation model, it can thus be seen that the whole prediction effect variation before and after rejecting outlier, is more believed
Breath.Aspect fully is searched to the model evaluation aspect and outlier of Models Sets group analysis method and has carried out systematic difference.
Table 1: input/output variable explanation
Above-described embodiment is used to illustrate the present invention, rather than limits the invention, in spirit of the invention and
In scope of protection of the claims, to any modifications and changes that the present invention makes, protection scope of the present invention is both fallen within.
Claims (4)
1. a kind of multiple soft measurement algorithm cluster modeling methods based on Monte Carlo cross validation, which is characterized in that including with
Lower step:
(1) industrial process data sample is inputted to system, all samples is normalized, so that the mean value of each variable
It is 0, variance 1;These data are stored in historical data base;
(2) the Monte Carlo cross validation algorithm MC based on Models Sets group analysis method is used, outlier is rejected;Rejecting is peeled off
N normal sample after value is stored in historical data base, and n is the natural number greater than 0;According still further to Models Sets group analysis method step into
Row following step (3)-(5);
(3) the binary matrix sampling method BMS based on Monte carlo algorithm is used, is extracted in the data set after rejecting outlier
The sample of a% is remaining to be used as test sample, 50≤a≤80 as training sample;BMS duplicate sampling n times, symbiosis is at N number of instruction
Practice collection and N number of test set, N is the natural number greater than 0;These data are stored in historical data base;
(4) subalgorithm of the m soft measurement algorithm suitable for different industrial process characteristics as integrated study is selected, m is 3≤m
≤ 8 natural number;N number of training set is called from database, and N is established respectively to these training sets with hard measurement submodel algorithm
Then a submodel calls corresponding N number of test set in database to be predicted;If the prediction result that each submodel obtains
Are as follows:
yi,j(t), i=1,2, m, j=1,2, N, t=1,2, n × (1-a%)
If the legitimate reading of test set are as follows:
Yj(t), j=1,2, N, t=1,2, n × (1-a%)
The respective N number of root-mean-square error RMSEP of each submodel algorithm is acquired by following formulai,j:
It will be spare in modeling data and each hard measurement submodel algorithm parameter deposit historical data base;
(5) modeling data is called from database, and m hard measurement submodel algorithm is collected using Bayes's Integrated Algorithm
At obtaining the weight of each subalgorithm distribution, be stored in spare in historical data base;Off-line modeling is completed;
(6) by online process data input system, after m × N number of soft measurement algorithm model prediction, according to the power obtained before
Weight passes through the integrated final prediction result distribution for acquiring process data to be predicted of Bayes.
2. multiple soft measurement algorithm cluster modeling methods based on Monte Carlo cross validation according to claim 1, special
Sign is, in the step (2), the Models Sets group analysis method specifically:
A) N number of Sub Data Set is generated at random from initial data concentration with monte carlo method;
B) a classification submodel is established to each Sub Data Set or returns submodel;
C) for statistical analysis to the output of N number of submodel;
The step of Monte Carlo cross validation algorithm, is formulated according to the frame of Models Sets group analysis method, its step
Suddenly it is:
A) N number of training subset and N number of test subset are generated at random from initial data concentration with monte carlo method;
B) N number of model is established, is predicted on N number of test set;
C) each of working as sample for raw data set has a group to predict error, predicts that error carries out statistical to this group
Analysis;The mean value and variance that the prediction error of each sample can be calculated, using the mean value and variance of predicting error as cross
Coordinate and ordinate, so that it may obtain piece image, by segmenting the image into four regions, can distinguish normal sample and from
Group's value;N normal sample after rejecting outlier is stored in historical data base.
3. multiple soft measurement algorithm cluster modeling methods based on Monte Carlo cross validation according to claim 1, special
Sign is, the step (3) specifically: using the binary matrix sampling method BMS based on Monte carlo algorithm from rejecting outlier
The sample that a% is extracted in data set afterwards is used as training sample, remaining as test sample, comprises the concrete steps that: in normal sample
This concentration randomly chooses the sample mark 1 of a%, remaining mark 0, and marking 1 position, to represent the sample selected as training set sample,
Mark 0 is used as test set sample;BMS duplicate sampling n times, symbiosis is at N number of training set and N number of test set;These data are stored in
Historical data base.
4. multiple soft measurement algorithm cluster modeling methods based on Monte Carlo cross validation according to claim 1, special
Sign is, Bayes's Integrated Algorithm specific steps in the step (5) are as follows:
A) the coefficient Z of Bayes's Integrated Algorithm is acquired by following formulai:
If MiI-th of subalgorithm model is represented, bayesian prior probability is set as:
If S represents test data set, Bayesian likelihood probability are as follows:
Bayes posterior probability are as follows:
B) summation finally is weighted to the predicted value of each submodel again, obtains final predicted value:
Calculate N number of root-mean-square error of Bayes's integrated model;By the rejecting of each submodel and Bayes's integrated model from
The root-mean-square error of group's value front and back is for statistical analysis, evaluation model superiority and inferiority.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710056647.XA CN106897774B (en) | 2017-01-25 | 2017-01-25 | Multiple soft measurement algorithm cluster modeling methods based on Monte Carlo cross validation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710056647.XA CN106897774B (en) | 2017-01-25 | 2017-01-25 | Multiple soft measurement algorithm cluster modeling methods based on Monte Carlo cross validation |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106897774A CN106897774A (en) | 2017-06-27 |
CN106897774B true CN106897774B (en) | 2019-01-18 |
Family
ID=59199231
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710056647.XA Expired - Fee Related CN106897774B (en) | 2017-01-25 | 2017-01-25 | Multiple soft measurement algorithm cluster modeling methods based on Monte Carlo cross validation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106897774B (en) |
Families Citing this family (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108492026B (en) * | 2018-03-06 | 2021-05-11 | 宁波大学 | Soft measurement method based on integrated orthogonal component optimization regression analysis |
CN109272508B (en) * | 2018-08-02 | 2021-10-26 | 哈尔滨工程大学 | Petri network image segmentation method based on rough set and rough entropy |
CN109146076A (en) * | 2018-08-13 | 2019-01-04 | 东软集团股份有限公司 | model generating method and device, data processing method and device |
CN109859807B (en) * | 2019-03-05 | 2021-07-30 | 复旦大学 | Quantum Monte Carlo method for realizing full-topology sampling of quantum dimer model |
CN111105628A (en) * | 2019-12-23 | 2020-05-05 | 北京首钢自动化信息技术有限公司 | Parking lot portrait construction method and device |
US11769426B2 (en) | 2020-02-18 | 2023-09-26 | BlueOwl, LLC | Systems and methods for creating driving challenges |
CN111613266A (en) * | 2020-05-20 | 2020-09-01 | 中南大学 | Outlier detection method based on quantitative structure-activity relationship |
CN114297054B (en) * | 2021-12-17 | 2023-06-30 | 北京交通大学 | Software defect number prediction method based on subspace mixed sampling |
CN114970698A (en) * | 2022-05-17 | 2022-08-30 | 河海大学 | Metering equipment operation performance prediction method based on improved LWPLS |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104914723A (en) * | 2015-05-22 | 2015-09-16 | 浙江大学 | Industrial process soft measurement modeling method based on cooperative training partial least squares model |
CN105868164A (en) * | 2016-03-19 | 2016-08-17 | 浙江大学 | Soft measurement modeling method based on monitored linear dynamic system model |
-
2017
- 2017-01-25 CN CN201710056647.XA patent/CN106897774B/en not_active Expired - Fee Related
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104914723A (en) * | 2015-05-22 | 2015-09-16 | 浙江大学 | Industrial process soft measurement modeling method based on cooperative training partial least squares model |
CN105868164A (en) * | 2016-03-19 | 2016-08-17 | 浙江大学 | Soft measurement modeling method based on monitored linear dynamic system model |
Non-Patent Citations (4)
Title |
---|
Artificial Intelligence techniques applied as estimator in chemical process systems - A literature survey;Ali J M etal.;《Expert Systems with Applications》;20151231;第42卷(第14期);第5915-5931页 |
Data-driven Soft Sensors in the process industry;Petr Kadlec B G etal.;《Computers & Chemical Engineering》;20091231;第33卷(第4期);第795-814页 |
基于蒙特卡洛模型集群分析方法的气相色谱保留指数的QSPR研究_景琳;景琳;《中国优秀硕士学位论文全文数据库 医药卫生科技辑》;20131015;第7-35页 |
基于蒙特卡罗法的锅炉煤质在线软测量特性模拟;刘福国;《煤炭学报》;20091130;第34卷(第11期);第1540-1545页 |
Also Published As
Publication number | Publication date |
---|---|
CN106897774A (en) | 2017-06-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106897774B (en) | Multiple soft measurement algorithm cluster modeling methods based on Monte Carlo cross validation | |
CN106897775B (en) | Soft-measuring modeling method based on Bayes's integrated study | |
JP6817426B2 (en) | Yield prediction system and method for machine learning-based semiconductor manufacturing | |
CN111079836B (en) | Process data fault classification method based on pseudo label method and weak supervised learning | |
CN107463993B (en) | Medium-and-long-term runoff forecasting method based on mutual information-kernel principal component analysis-Elman network | |
CN103914064A (en) | Industrial process fault diagnosis method based on multiple classifiers and D-S evidence fusion | |
CN111191726B (en) | Fault classification method based on weak supervision learning multilayer perceptron | |
CN108803520A (en) | A kind of dynamic process monitoring method rejected based on the non-linear autocorrelation of variable | |
CN113642666B (en) | Active enhancement soft measurement method based on sample expansion and screening | |
CN108334943A (en) | The semi-supervised soft-measuring modeling method of industrial process based on Active Learning neural network model | |
CN109389314A (en) | A kind of quality hard measurement and monitoring method based on optimal neighbour's constituent analysis | |
CN110569566A (en) | Method for predicting mechanical property of plate strip | |
CN105868164A (en) | Soft measurement modeling method based on monitored linear dynamic system model | |
CN112414715A (en) | Bearing fault diagnosis method based on mixed feature and improved gray level co-occurrence algorithm | |
CN115965110A (en) | Accurate measurement and calculation method for enterprise energy consumption image and carbon emission facing industrial park | |
CN115495991A (en) | Rainfall interval prediction method based on time convolution network | |
CN111898673A (en) | Dissolved oxygen content prediction method based on EMD and LSTM | |
CN110084301B (en) | Hidden Markov model-based multi-working-condition process working condition identification method | |
CN113283288B (en) | Nuclear power station evaporator eddy current signal type identification method based on LSTM-CNN | |
CN111062539A (en) | Total electric quantity prediction method based on secondary electric quantity characteristic clustering analysis | |
CN110196456A (en) | A kind of medium-term and long-term rainfall runoff forecasting method based on analog year grey correlation analysis | |
Zhang et al. | A novel hybrid model based on EMD-BPNN for forecasting US and UK stock indices | |
CN112949203B (en) | Board laser cutting quality judgment method based on electrical parameters and XGBOOST-NN algorithm | |
CN105069214A (en) | Process reliability evaluation method based on nonlinear correlation analysis | |
CN115034005A (en) | Model analysis visualization method for component residual service life prediction |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
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: 20190118 Termination date: 20200125 |
|
CF01 | Termination of patent right due to non-payment of annual fee |