CN109884893A - Dynamic lag estimation method between a kind of multi-process variable - Google Patents
Dynamic lag estimation method between a kind of multi-process variable Download PDFInfo
- Publication number
- CN109884893A CN109884893A CN201910152331.XA CN201910152331A CN109884893A CN 109884893 A CN109884893 A CN 109884893A CN 201910152331 A CN201910152331 A CN 201910152331A CN 109884893 A CN109884893 A CN 109884893A
- Authority
- CN
- China
- Prior art keywords
- data
- time
- delay
- relative
- data set
- 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
- 238000000034 method Methods 0.000 title claims abstract description 81
- 238000012549 training Methods 0.000 claims abstract description 8
- 238000013480 data collection Methods 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 6
- 238000004519 manufacturing process Methods 0.000 claims description 5
- 241001269238 Data Species 0.000 claims description 3
- 238000012544 monitoring process Methods 0.000 abstract description 6
- 238000005259 measurement Methods 0.000 abstract 1
- 230000000694 effects Effects 0.000 description 8
- GRYLNZFGIOXLOG-UHFFFAOYSA-N Nitric acid Chemical compound O[N+]([O-])=O GRYLNZFGIOXLOG-UHFFFAOYSA-N 0.000 description 5
- 229910017604 nitric acid Inorganic materials 0.000 description 5
- 238000007637 random forest analysis Methods 0.000 description 5
- 238000007796 conventional method Methods 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- 230000000052 comparative effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Abstract
The invention discloses dynamic lag estimation methods between a kind of multi-process variable, data in database are divided into two class of input data and output data first, wherein input data, that is, auxiliary variable, output data, that is, leading variable；Then Delay Parameters collection and input data Delay Parameters collection relative to output data, intermediate variable relative to output data of the input data relative to intermediate variable is acquired to training dataset in database；It is finally input relative to the Delay Parameters collection of intermediate variable with input data, it is exported relative to output data, intermediate variable relative to the Delay Parameters collection of output data with input data, obtain dynamic lag parametric prediction model between multi-process variable, it seeks inputting the Delay Parameters relative to output online, and substitute into the prediction model, the Delay Parameters of input and output are obtained in real time, and the invention enables the result of hard measurement is relatively reliable, be conducive to factory's monitoring data, keep factory's stable operation.
Description
Technical field
The invention belongs to chemical industry technical field of automatic control, and in particular to dynamic lag between a kind of multi-process variable
Estimation method.
Background technique
It is all to be input into as made of the cascade of multiple processes from raw material in most of technological process of productions in the factory
Product output usually requires one or even several hours, and the time lag information between each process is with external environment, front and back process
The factors dynamic change such as state, be not unalterable.The existing process variable prediction technique using Delay Parameters
The document method that is all based on fixed time lag parameter, since the Delay Parameters in practical factory between the process of front and back are with factory
The dynamic value of operating status variation, and it is not fixed value, if not considering the time lag information between input and output, or simple
Time lag information is thought of as definite value, the data training soft-sensing model after reusing reconstruct, then the causality of input and output
It will likely change, model training can be made not accurate enough, and then soft-sensing model estimated performance is caused to decline.
Summary of the invention
The object of the present invention is to provide dynamic lag estimation methods between a kind of multi-process variable, solve and deposit in the prior art
The problem of online soft sensor is not accurate enough, estimated performance decline.
The technical scheme adopted by the invention is that dynamic lag estimation method between a kind of multi-process variable, specifically according to
Lower step is implemented:
Data in database are divided into two class of input data and output data by step 1, wherein input data assists becoming
Amount, output data, that is, leading variable；
Step 2, training dataset in database is acquired input data relative to the Delay Parameters collection of intermediate variable and
Delay Parameters collection of the input data relative to output data, intermediate variable relative to output data；
Step 3, with input data relative to intermediate variable Delay Parameters collection be input, with input data relative to output
Data, intermediate variable are output relative to the Delay Parameters collection of output data, and dynamic lag parameter is pre- between generating multi-process variable
Survey model；
Step 4, using the input data in Pearson correlation coefficient method line solver actual production data set relative in
Between variable Delay Parameters, which is substituted into real time in trained Delay Parameters prediction model, prediction inputted
Delay Parameters of the data relative to output data, intermediate variable relative to output data.
The features of the present invention also characterized in that
The method that Delay Parameters are solved in step 2 is Pearson correlation coefficient method, fuzzy curve method, Generalized Quadratic correlation method
In any one.
Step 2 is specifically implemented according to the following steps:
Step 2.1 assumes the total n group data of database dataAll, and defining timer time initial value is 1；
Step 2.2 sets the Delay Parameters upper limit as delay_max, takes time to time+ from database dataAll
M-1 group data definition is data set D_time, wherein when m > delay_max, time are 1, takes the 1st from database dataAll
Into m group data deposit data set D_1, when D_time=D_1 at this time, time are h, taken from database dataAll h to
The total m group data of h+m-1 are stored in data set D_h, D_time=D_h；D_time (i, j) is the i-th row jth column in data set D_time
Data, wherein i=1,2,3m, j=1,2,3,1,2,3 column datas of data set D_time respectively correspond to input number
According to, intermediate variable, output data；
Data set D_time in step 2.2 is extended to data set D1, D2, D3: data set D1 as input by step 2.3
Data and intermediate variable, data set D2 are input data and output data, and data set D3 is intermediate variable and output data；
Step 2.4, the optimal time-delay parameter that data set D1, D2, D3 are sought using Pearson correlation coefficient method are simultaneously deposited respectively
Enter D_delayIn (time), D_delayOut1 (time), in D_delayOut2 (time)；
Step 2.5 enables time=time+1, is updated to data set D_time, if time+m-1=n, holds downwards
Row, return step 2.2 if time+m-1 < n repeat；
Step 2.6 obtains optimal time-delay parameter set D_delayIn of the input data relative to intermediate variable at this time, input
Optimal time-delay parameter set D_delayOut1 of the data relative to output data, intermediate variable relative to output data it is optimal when
Stagnant parameter set D_delayOut2.
Delay_max=30 in step 2.2.
Step 2.4 is specifically implemented according to the following steps:
Step 2.4.1, data set D1 is extended to delay_max+1 group data set: when delay time is w, w is data set
Delay time value of the second column data of D1 relative to the first column data, the group data set are named as D1_w, the first row of D1_w
Data are the 1st row of the first column data of D1 to m-w row data, and the second column data of D1_w is the of the second column data of D1
W+1 row is to m row data；
Step 2.4.2, the correlation of the corresponding data set of each delay time value w is acquired using Pearson correlation coefficient method
Coefficient value P (w), using the maximum value corresponding delay time in obtained correlation coefficient value collection P as required data set D_
The optimal time-delay parameter of data set D1 after the extension of time, and be stored in D_delayIn (time), data set D_time extension
The optimal time-delay same D1 of parameter acquiring method of data set D2, D3 afterwards, and it is stored in D_delayOut1 (time), D_ respectively
In delayOut2 (time).
Pearson correlation coefficient formula is as follows in step 2.4.2:
Wherein, P (w) be obtained delay time be w data set D1_w correlation coefficient value, N be data set D1_w
Length, x indicate after D1 extension required for seek the 1st column data in the corresponding data collection D1_w of corresponding correlation coefficient value, y
Demand takes the 2nd column data in the corresponding data collection D1_w of corresponding correlation coefficient value after indicating D1 extension.
Step 4 is specifically implemented according to the following steps:
Step 4.1, when input data and intermediate variable data reach m group by factory's online acquisition, establish data set
D_timeTest, D_timeTest take the currently collect the current 1st to m group data, and D_timeTest is pressed step 2.3
It is extended to data set D1, acquires optimal time-delay of the input variable relative to output variable with step 2.4 using correlation coefficient process
Parameter D_delayIn_test (1)；
Step 4.2, hereafter one group of data of every update, D_timeTest update primary, it is assumed that currently collect u group number
According to u >=m, then D_timeTest takes the current u-m+1 currently collected to u group data, asks current with step 2.3
Optimal time-delay parameter D_delayIn_test (u-m+1) of the moment input variable relative to output variable；
Step 4.3, the optimal time-delay parameter by the input variable acquired online relative to intermediate variable, substitute into step in real time
In the 3 Delay Parameters prediction models established, on-line prediction input data is relative to output data, intermediate variable relative to output
The Delay Parameters of data.
The invention has the advantages that dynamic lag estimation method between a kind of multi-process variable, inside input data
Delay time relationship and input data and output data between delay time relationship, train and only rely on input data
The dynamic lag parametric prediction model of Delay Parameters between being output and input, then by input data and required input and
Delay time between output substitutes into soft-sensing model, carries out On-line Estimation to output data.Compared to traditional use from
The soft-sensing model of the trained definite value Delay Parameters of line uses the dynamic lag parametric prediction model on-line prediction in this patent
The output result of the soft-sensing model of Delay Parameters is relatively reliable accurate.
Detailed description of the invention
Fig. 1 is the demonstration graph for taking 40 groups of data to constitute data set D_time in the present invention from database；
Fig. 2 is that data set D_time is extended to data set D1, the demonstration graph of D2, D3 by the present invention；
Fig. 3 is the present invention by taking data set D1 as an example, seeks the demonstration graph of the optimal time-delay parameter of data set D1；
Fig. 4 is that the present invention uses the random forest based on fixed time lag parameter to Shan drum corporation nitric acid process monitoring data
The effect picture that model is predicted；
Fig. 5 is that the present invention uses based on seeking the random of Delay Parameters online Shan drum corporation nitric acid process monitoring data
The effect picture that forest model is predicted；
Fig. 6 is flow chart of the invention.
Specific embodiment
The following describes the present invention in detail with reference to the accompanying drawings and specific embodiments.
The present invention is dynamic lag estimation method between a kind of multi-process variable, and flow chart is as shown in fig. 6, specifically according to following
Step is implemented:
Data in database are divided into two class of input data and output data by step 1, wherein input data assists becoming
Amount, output data, that is, leading variable；
Step 2, training dataset in database is acquired input data relative to the Delay Parameters collection of intermediate variable and
Delay Parameters collection of the input data relative to output data, intermediate variable relative to output data；
The method that Delay Parameters are solved in step 2 is Pearson correlation coefficient method, fuzzy curve method, Generalized Quadratic correlation method
In any one.
Step 2 is specifically implemented according to the following steps:
Step 2.1 assumes the total n group data of database dataAll, and defining timer time initial value is 1；
Step 2.2 sets the Delay Parameters upper limit as delay_max, takes time to time+ from database dataAll
M-1 group data definition is data set D_time, wherein m > delay_max, as shown in Figure 1, when time is 1, from database
The 1st is taken to be stored in data set D_1 in dataAll to m group data, at this time D_time=D_1, when time is h, from database
It is data set D_ that h to the h+m-1 total deposit of m group data data set D_h, D_time=D_h, D_time (i, j) are taken in dataAll
I-th row jth column data in time, wherein i=1,2,3m, j=1,2,3,1,2,3 column datas of data set D_time
It respectively corresponds as input data, intermediate variable, output data；
Data set D_time in step 2.2 is extended to data set D1, D2, D3 by step 2.3, and extension form is shown in Fig. 2:
Data set D1 is input data and intermediate variable, and data set D2 is input data and output data, and data set D3 is intermediate variable
With output data；
Step 2.4, the optimal time-delay parameter that data set D1, D2, D3 are sought using Pearson correlation coefficient method are simultaneously deposited respectively
Enter D_delayIn (time), D_delayOut1 (time), in D_delayOut2 (time)；
Step 2.5 enables time=time+1, is updated to data set D_time, if time+m-1=n, holds downwards
Row, return step 2.2 if time+m-1 < n repeat；
Step 2.6 obtains optimal time-delay parameter set D_delayIn of the input data relative to intermediate variable at this time, input
Optimal time-delay parameter set D_delayOut1 of the data relative to output data, intermediate variable relative to output data it is optimal when
Stagnant parameter set D_delayOut2.
Delay_max=30 in step 2.2.
Step 2.4 is specifically implemented according to the following steps:
Step 2.4.1, data set D1 is extended to delay_max+1 group data set: when delay time is w, w is data set
Delay time value of the second column data of D1 relative to the first column data, the group data set are named as D1_w, the first row of D1_w
Data are the 1st row of the first column data of D1 to m-w row data, and the second column data of D1_w is the of the second column data of D1
W+1 row is to m row data；
Step 2.4.2, the correlation of the corresponding data set of each delay time value w is acquired using Pearson correlation coefficient method
Coefficient value P (w), using the maximum value corresponding delay time in obtained correlation coefficient value collection P as required data set D_
The optimal time-delay parameter of data set D1 after the extension of time, and be stored in D_delayIn (time), as shown in figure 3, data set
The optimal time-delay same D1 of parameter acquiring method of data set D2, D3 after D_time extension, and it is stored in D_delayOut1 respectively
(time)、
In D_delayOut2 (time).
Pearson correlation coefficient formula is as follows in step 2.4.2:
Wherein, P (w) be obtained delay time be w data set D1_w correlation coefficient value, N be data set D1_w
Length, x indicate after D1 extension required for seek the 1st column data in the corresponding data collection D1_w of corresponding correlation coefficient value, y
Demand takes the 2nd column data in the corresponding data collection D1_w of corresponding correlation coefficient value after indicating D1 extension.
Step 3, with input data relative to intermediate variable, input data relative to output data, intermediate variable relative to
The Delay Parameters data set of this three of output data is training sample, has the l sample data of taking-up put back at random, carries out n1 altogether
Secondary sampling, generate n1 training set, with input data relative to intermediate variable Delay Parameters collection be input, with input data phase
It relative to the Delay Parameters collection of output data is output to get to dynamic between multi-process variable for output data, intermediate variable
Delay Parameters prediction model.
Wherein, BP neural network model, Random Forest model, support vector regression mould can be selected in delay time prediction model
Type etc..Suitable model can be selected according to the precision of prediction needs of oneself.Due to BP neural network model, Random Forest model,
Support vector regression model etc. is all well known prediction model, therefore careful description is not done to it.The present invention is predicted in Delay Parameters
Random Forest model is selected in model.
Step 4, using the input data in Pearson correlation coefficient method line solver actual production data set relative in
Between variable Delay Parameters, which is substituted into real time in trained Delay Parameters prediction model, prediction inputted
Delay Parameters of the data relative to output data, intermediate variable relative to output data, are specifically implemented according to the following steps:
Step 4.1, when input data and intermediate variable data reach m group by factory's online acquisition, establish data set
D_timeTest, D_timeTest take the currently collect the current 1st to m group data, and D_timeTest is pressed step 2.3
It is extended to data set D1, using Pearson correlation coefficient method, with step 2.4, acquires input variable relative to output variable most
Excellent Delay Parameters D_delayIn_test (1)；
Step 4.2, hereafter one group of data of every update, D_timeTest update primary, it is assumed that currently collect u group number
According to u >=m, then D_timeTest takes the current u-m+1 currently collected to u group data, asks current with step 2.3
Optimal time-delay parameter D_delayIn_test (u-m+1) of the moment input variable relative to output variable；
Step 4.3, the optimal time-delay parameter by the input variable acquired online relative to intermediate variable, substitute into step in real time
In the 3 Delay Parameters prediction models established, on-line prediction input data is relative to output data, intermediate variable relative to output
The Delay Parameters of data.
By taking the monitoring data during the drum corporation nitric acid of Shan as an example, established with traditional Delay Parameters estimation method is used
Soft-sensing model is compared, and has large increase on model prediction accuracy, and prediction result comparison diagram is shown in Fig. 4, Fig. 5, wherein Fig. 4 is
The effect that the present invention predicts Shan drum corporation nitric acid process monitoring data using the Random Forest model of fixed time lag parameter
Fruit figure, Fig. 5 are that the present invention uses based on seeking the random gloomy of Delay Parameters online Shan drum corporation nitric acid process monitoring data
Woods model, that is, online Delay Parameters estimation method uses the effect picture that this method is predicted, effect is as shown in table 1 below:
1 conventional method of table and context of methods Comparative result table
| Mean absolute error | Mean error | R2 | RMSE | |
| Conventional method | 19.59% | 4.615 | 0.08 | 6.7 |
| This patent method | 4.86% | 1.137 | 0.90 | 2.13 |
It is down to 4.86% by 19.59% as it can be seen from table 1 being compared with the traditional method average relative error, improvement effect
Up to 75.19%, mean error is down to 1.137 by 4.615, and improvement effect is up to 75.36%, that is, uses the method pair of this patent
The complex industrial process key variables prediction of having time lag has good effect.
Claims (7)
1. dynamic lag estimation method between a kind of multi-process variable, which is characterized in that be specifically implemented according to the following steps:
Data in database are divided into two class of input data and output data by step 1, wherein input data, that is, auxiliary variable, it is defeated
Data, that is, leading variable out；
Step 2 acquires input data to training dataset in database relative to the Delay Parameters collection of intermediate variable and input
Delay Parameters collection of the data relative to output data, intermediate variable relative to output data；
Step 3, with input data relative to intermediate variable Delay Parameters collection be input, with input data relative to output number
It relative to the Delay Parameters collection of output data is output to get pre- to dynamic lag parameter between multi-process variable according to, intermediate variable
Survey model；
Step 4 is become using the input data in Pearson correlation coefficient method line solver actual production data set relative to centre
The Delay Parameters of amount substitute into the Delay Parameters in trained Delay Parameters prediction model in real time, and prediction obtains input data
Delay Parameters relative to output data, intermediate variable relative to output data.
2. dynamic lag estimation method between a kind of multi-process variable according to claim 1, which is characterized in that the step
It is any one in Pearson correlation coefficient method, fuzzy curve method, Generalized Quadratic correlation method that the method for Delay Parameters is solved in 2
Kind.
3. dynamic lag estimation method between a kind of multi-process variable according to claim 2, which is characterized in that the step
2 are specifically implemented according to the following steps:
Step 2.1 assumes the total n group data of database dataAll, and defining timer time initial value is 1；
Step 2.2 sets the Delay Parameters upper limit as delay_max, takes time to time+m-1 group from database dataAll
Data definition is data set D_time, wherein when m > delay_max, time are 1, takes the 1st to m from database dataAll
In group data deposit data set D_1, D_time=D_1 takes h to h+m-1 from database dataAll when time is h at this time
Total m group data are stored in data set D_h, and D_time=D_h, D_time (i, j) are the i-th row jth column data in data set D_time,
Wherein, i=1,2,3m, j=1,2,3,1,2,3 column datas of data set D_time respectively correspond for input data, in
Between variable, output data；
Data set D_time in step 2.2 is extended to data set D1, D2, D3: data set D1 as input data by step 2.3
With intermediate variable, data set D2 is input data and output data, and data set D3 is intermediate variable and output data；
Step 2.4, the optimal time-delay parameter that data set D1, D2, D3 are sought using Pearson correlation coefficient method are simultaneously stored in D_ respectively
DelayIn (time), D_delayOut1 (time), in D_delayOut2 (time)；
Step 2.5 enables time=time+1, is updated to data set D_time, if time+m-1=n, executes downwards, if
Then return step 2.2 time+m-1 < n, repeat；
Step 2.6 obtains optimal time-delay parameter set D_delayIn of the input data relative to intermediate variable, input data at this time
Relative to the optimal time-delay parameter set D_delayOut1 of output data, intermediate variable is joined relative to the optimal time-delay of output data
Manifold D_delayOut2.
4. dynamic lag estimation method between a kind of multi-process variable according to claim 3, which is characterized in that the step
Delay_max=30 in 2.2, m=40.
5. dynamic lag estimation method between a kind of multi-process variable according to claim 3, which is characterized in that the step
2.4 are specifically implemented according to the following steps:
Step 2.4.1, data set D1 is extended to delay_max+1 group data set: when delay time is w, w is data set D1's
Delay time value of second column data relative to the first column data, the group data set are named as D1_w, the first column data of D1_w
For D1 the first column data the 1st row to m-w row data, the second column data of D1_w is the w+1 of the second column data of D1
It goes to m row data；
Step 2.4.2, the related coefficient of the corresponding data set of each delay time value w is acquired using Pearson correlation coefficient method
Value P (w), using the maximum value corresponding delay time in obtained correlation coefficient value collection P as required data set D_time's
The optimal time-delay parameter of data set D1 after extension, and be stored in D_delayIn (time), the number after data set D_time extension
According to the optimal time-delay same D1 of parameter acquiring method of collection D2, D3, and it is stored in D_delayOut1 (time), D_ respectively
In delayOut2 (time).
6. dynamic lag estimation method between a kind of multi-process variable according to claim 5, which is characterized in that the step
2.4.2 middle Pearson correlation coefficient formula is as follows:
Wherein, P (w) be obtained delay time be w data set D1_w correlation coefficient value, N be data set D1_w length
Degree, x indicate that required the 1st column data sought in the corresponding data collection D1_w of corresponding correlation coefficient value after D1 extension, y indicate
Demand takes the 2nd column data in the corresponding data collection D1_w of corresponding correlation coefficient value after D1 extension.
7. dynamic lag estimation method between a kind of multi-process variable according to claim 6, which is characterized in that the step
4 are specifically implemented according to the following steps:
Step 4.1, when input data and intermediate variable data reach m group by factory's online acquisition, establish data set D_
TimeTest, D_timeTest take the currently collect the current 1st to m group data, and D_timeTest is expanded by step 2.3
Exhibition is data set D1, and using correlation coefficient process, with step 2.4, the optimal time-delay for acquiring input variable relative to output variable is joined
Number D_delayIn_test (1)；
Step 4.2, hereafter one group of data of every update, D_timeTest update primary, it is assumed that currently collect u group data, u >
=m.Then D_timeTest takes the current u-m+1 currently collected to u group data, the current time asked with step 2.3
Optimal time-delay parameter D_delayIn_test (u-m+1) of the input variable relative to output variable；
Step 4.3, the optimal time-delay parameter by the input variable acquired online relative to intermediate variable substitute into step 3 institute in real time
In the Delay Parameters prediction model of foundation, on-line prediction input data is relative to output data, intermediate variable relative to output number
According to Delay Parameters.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910152331.XA CN109884893B (en) | 2019-02-28 | 2019-02-28 | Multi-process variable dynamic time lag estimation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910152331.XA CN109884893B (en) | 2019-02-28 | 2019-02-28 | Multi-process variable dynamic time lag estimation method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109884893A true CN109884893A (en) | 2019-06-14 |
| CN109884893B CN109884893B (en) | 2021-09-10 |
Family
ID=66930040
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910152331.XA Active CN109884893B (en) | 2019-02-28 | 2019-02-28 | Multi-process variable dynamic time lag estimation method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109884893B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112295255A (en) * | 2020-10-24 | 2021-02-02 | 四川泸天化创新研究院有限公司 | Intelligent control system and control method for methanol rectification device |
Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4612621A (en) * | 1983-03-17 | 1986-09-16 | The Babcock & Wilcox Company | Distributed system for optimizing the performance of a plurality of multi-stage steam turbines using function blocks |
| US5144549A (en) * | 1990-06-29 | 1992-09-01 | Massachusetts Institute Of Technology | Time delay controlled processes |
| JPH07114580A (en) * | 1993-10-18 | 1995-05-02 | Fujitsu Ltd | Delay time analysis system for logical device |
| CN103019094A (en) * | 2011-09-19 | 2013-04-03 | 费希尔-罗斯蒙特系统公司 | Inferential process modelling, quality prediction and fault detection using multi-stage data segregation |
| CN103279030A (en) * | 2013-03-07 | 2013-09-04 | 清华大学 | Bayesian framework-based dynamic soft measurement modeling method and device |
| WO2014011525A2 (en) * | 2012-07-12 | 2014-01-16 | Corning Incorporated | Multimode optical fiber systems with adjustable chromatic modal dispersion compensation |
| CN104503229A (en) * | 2014-11-24 | 2015-04-08 | 北京邮电大学 | Wave integral bilateral teleoperation control method based on LS-SVM (least square support vector machine) delay predication |
| CN105205224A (en) * | 2015-08-28 | 2015-12-30 | 江南大学 | Modeling method for soft measurement of time difference gaussian process regression based on fuzzy curve analysis |
| CN105429720A (en) * | 2015-11-25 | 2016-03-23 | 桂林航天工业学院 | Related delay estimation method based on EMD reconstruction |
| CN105846939A (en) * | 2016-03-24 | 2016-08-10 | 成都博思微科技有限公司 | System for method for accurately keeping synchronization of multiple modules |
| CN107273633A (en) * | 2017-06-29 | 2017-10-20 | 中南大学 | Varying delay method of estimation and flow time lag method of estimation is hydrocracked between multiple operation |
| US20180034470A1 (en) * | 2016-08-01 | 2018-02-01 | Kopin Corporation | Time delay in digitally oversampled sensor systems, apparatuses, and methods |
| CN107942667A (en) * | 2017-11-29 | 2018-04-20 | 辽宁石油化工大学 | Injection moulding process based on Time-varying time-delays and interference mixes 2D tracking and controlling methods |
| CN108614533A (en) * | 2018-05-28 | 2018-10-02 | 江南大学 | A kind of neural network modeling approach estimated based on NARX models and time lag |
| CN108647808A (en) * | 2018-04-11 | 2018-10-12 | 济南大学 | A kind of manufacturing parameter Optimization Prediction method, apparatus, equipment and storage medium |
-
2019
- 2019-02-28 CN CN201910152331.XA patent/CN109884893B/en active Active
Patent Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4612621A (en) * | 1983-03-17 | 1986-09-16 | The Babcock & Wilcox Company | Distributed system for optimizing the performance of a plurality of multi-stage steam turbines using function blocks |
| US5144549A (en) * | 1990-06-29 | 1992-09-01 | Massachusetts Institute Of Technology | Time delay controlled processes |
| JPH07114580A (en) * | 1993-10-18 | 1995-05-02 | Fujitsu Ltd | Delay time analysis system for logical device |
| CN103019094A (en) * | 2011-09-19 | 2013-04-03 | 费希尔-罗斯蒙特系统公司 | Inferential process modelling, quality prediction and fault detection using multi-stage data segregation |
| WO2014011525A2 (en) * | 2012-07-12 | 2014-01-16 | Corning Incorporated | Multimode optical fiber systems with adjustable chromatic modal dispersion compensation |
| CN103279030A (en) * | 2013-03-07 | 2013-09-04 | 清华大学 | Bayesian framework-based dynamic soft measurement modeling method and device |
| CN104503229A (en) * | 2014-11-24 | 2015-04-08 | 北京邮电大学 | Wave integral bilateral teleoperation control method based on LS-SVM (least square support vector machine) delay predication |
| CN105205224A (en) * | 2015-08-28 | 2015-12-30 | 江南大学 | Modeling method for soft measurement of time difference gaussian process regression based on fuzzy curve analysis |
| CN105429720A (en) * | 2015-11-25 | 2016-03-23 | 桂林航天工业学院 | Related delay estimation method based on EMD reconstruction |
| CN105846939A (en) * | 2016-03-24 | 2016-08-10 | 成都博思微科技有限公司 | System for method for accurately keeping synchronization of multiple modules |
| US20180034470A1 (en) * | 2016-08-01 | 2018-02-01 | Kopin Corporation | Time delay in digitally oversampled sensor systems, apparatuses, and methods |
| CN107273633A (en) * | 2017-06-29 | 2017-10-20 | 中南大学 | Varying delay method of estimation and flow time lag method of estimation is hydrocracked between multiple operation |
| CN107942667A (en) * | 2017-11-29 | 2018-04-20 | 辽宁石油化工大学 | Injection moulding process based on Time-varying time-delays and interference mixes 2D tracking and controlling methods |
| CN108647808A (en) * | 2018-04-11 | 2018-10-12 | 济南大学 | A kind of manufacturing parameter Optimization Prediction method, apparatus, equipment and storage medium |
| CN108614533A (en) * | 2018-05-28 | 2018-10-02 | 江南大学 | A kind of neural network modeling approach estimated based on NARX models and time lag |
Non-Patent Citations (16)
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112295255A (en) * | 2020-10-24 | 2021-02-02 | 四川泸天化创新研究院有限公司 | Intelligent control system and control method for methanol rectification device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109884893B (en) | 2021-09-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110674604B (en) | Transformer DGA data prediction method based on multi-dimensional time sequence frame convolution LSTM | |
| CN110024097A (en) | Semiconductors manufacture yield forecasting system and method based on machine learning | |
| CN107992976B (en) | Hot topic early development trend prediction system and prediction method | |
| CN109992921B (en) | On-line soft measurement method and system for thermal efficiency of boiler of coal-fired power plant | |
| WO2016101628A1 (en) | Data processing method and device in data modeling | |
| JP3637412B2 (en) | Time-series data learning / prediction device | |
| CN107146035B (en) | Method for calculating batch coefficients in large-goods production of knitted clothes | |
| CN107230349B (en) | A kind of online real-time short time traffic flow forecasting method | |
| CN108564326A (en) | Prediction technique and device, computer-readable medium, the logistics system of order | |
| CN112131782A (en) | Multi-loop intelligent factory edge side digital twin scene coupling device | |
| CN107403196B (en) | Method for predicting concentration of butane at bottom of debutanizer by instant learning modeling based on spectral clustering analysis | |
| CN111768000A (en) | Industrial process data modeling method for online adaptive fine-tuning deep learning | |
| CN112308298B (en) | Multi-scenario performance index prediction method and system for semiconductor production line | |
| CN113130014B (en) | Rare earth extraction simulation method and system based on multi-branch neural network | |
| CN106202377A (en) | A kind of online collaborative sort method based on stochastic gradient descent | |
| CN110019420A (en) | A kind of data sequence prediction technique and calculate equipment | |
| CN106997488A (en) | A kind of action knowledge extraction method of combination markov decision process | |
| CN106156844A (en) | For building Space-time Neural Network and utilizing its method and apparatus being predicted | |
| CN109754122A (en) | A kind of Numerical Predicting Method of the BP neural network based on random forest feature extraction | |
| CN110782083B (en) | Aero-engine standby demand prediction method based on deep Croston method | |
| CN109884893A (en) | Dynamic lag estimation method between a kind of multi-process variable | |
| CN109886316B (en) | Transformer state parameter combination prediction method based on cloud system similarity weight distribution | |
| CN109740221B (en) | Intelligent industrial design algorithm based on search tree | |
| CN116306832A (en) | Multimode generation antagonistic neural network modeling method and device for multidimensional sequence data | |
| CN110909492A (en) | Sewage treatment process soft measurement method based on extreme gradient lifting algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20240307 Address after: 1002, Building A, Zhiyun Industrial Park, No. 13 Huaxing Road, Henglang Community, Dalang Street, Longhua District, Shenzhen City, Guangdong Province, 518110 Patentee after: Shenzhen Wanzhida Technology Co.,Ltd. Country or region after: China Address before: 710048 No. 5 Jinhua South Road, Shaanxi, Xi'an Patentee before: XI'AN University OF TECHNOLOGY Country or region before: China |
|
| TR01 | Transfer of patent right |
Effective date of registration: 20240415 Address after: Room 216, 2nd Floor, Unit B, Building 4, Chang'an University Science and Technology Park, South Section 4555 Guangyuntan Avenue, Chanba Ecological Zone, Xi'an City, Shaanxi Province, 710000 Patentee after: Zhixiao Technology (Xi'an) Co.,Ltd. Country or region after: China Address before: 1002, Building A, Zhiyun Industrial Park, No. 13 Huaxing Road, Henglang Community, Dalang Street, Longhua District, Shenzhen City, Guangdong Province, 518110 Patentee before: Shenzhen Wanzhida Technology Co.,Ltd. Country or region before: China |