CN110909492A - Sewage treatment process soft measurement method based on extreme gradient lifting algorithm - Google Patents
Sewage treatment process soft measurement method based on extreme gradient lifting algorithm Download PDFInfo
- Publication number
- CN110909492A CN110909492A CN201911267973.0A CN201911267973A CN110909492A CN 110909492 A CN110909492 A CN 110909492A CN 201911267973 A CN201911267973 A CN 201911267973A CN 110909492 A CN110909492 A CN 110909492A
- Authority
- CN
- China
- Prior art keywords
- model
- data
- value
- soft measurement
- modeling
- 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.)
- Pending
Links
Images
Landscapes
- Activated Sludge Processes (AREA)
Abstract
The invention provides a sewage treatment process soft measurement method based on an extreme gradient lifting algorithm, and belongs to the field of industrial sewage treatment process soft measurement modeling and application. Firstly, aiming at the problem of missing values of data acquired in an industrial process, a neighboring algorithm model is adopted for filling the missing values; secondly, performing soft measurement modeling by using an extreme gradient lifting algorithm; and finally, adjusting 7 parameters of the model by using a grid searching method, thereby obtaining a more accurate soft measurement model. The method can improve the prediction precision of the ammonia nitrogen concentration of the effluent in the sewage treatment process.
Description
Technical Field
The invention belongs to the field of industrial sewage treatment process soft measurement modeling and application, and particularly relates to a soft measurement method for effluent quality parameter concentration in a sewage treatment process based on an extreme gradient lifting algorithm.
Background
Along with the progress and development of social economy, the pollution of human beings to the water environment is increasingly intensified, and the efficient treatment of sewage has more and more important significance to sustainable development. The concentration of ammonia nitrogen in water quality parameters is an important index for judging whether the water quality reaches the national discharge standard, however, the existing determination method for the concentration of ammonia nitrogen in water quality is complex in operation and low in real-time performance, and related instruments and meters are expensive in price and high in maintenance cost, so that a soft measurement method for the concentration of ammonia nitrogen in effluent water is generally required to be researched to meet the control requirement of an industrial process.
Aiming at the problems, an intelligent soft measurement method based on data driving is adopted, water quality parameters capable of being measured in real time are selected as auxiliary variables, a soft measurement model of the effluent ammonia nitrogen concentration is established, the effluent ammonia nitrogen concentration can be predicted on line, and the defect of time lag of the traditional method is effectively overcome.
The extreme gradient lifting algorithm has the advantages of good model interpretability, strong input data invariance, easier parameter adjustment and the like, and can quickly obtain an accurate, robust and strong explanatory model when being used for soft measurement of the concentration of the ammonia nitrogen in the effluent in the sewage treatment process.
Disclosure of Invention
Important indexes are obtained in the sewage treatment process: the invention provides a sewage treatment process soft measurement method based on an extreme gradient lifting algorithm. Firstly, aiming at the problem of missing values of data acquired in an industrial process, a neighboring algorithm model is adopted for filling the missing values; secondly, performing soft measurement modeling by using an extreme gradient lifting algorithm; and finally, adjusting 7 parameters of the model by using a grid searching method, thereby obtaining a more accurate soft measurement model.
The technical scheme adopted by the invention is as follows:
(1) collecting production data in a batch of sewage treatment processes, and establishing a proximity algorithm model for missing data filling;
(2) collecting variable values which can be measured on line in the sewage treatment process through a database, using the variable values as input quantity of a soft measurement modeling sample, using an effluent ammonia nitrogen concentration value obtained by off-line measurement as output quantity of the soft measurement modeling sample, and forming a soft measurement modeling sample set X ═ X1,x2,…,xi,…,xn]T,X∈Rn×m,xiIs a row vector of dimension 1 x m, representing the ith sample, i ═ 1,2, … n, n is the total number of samples, m is the total number of process variables, and R is the real number set;
(3) predicting and filling missing values of the soft measurement modeling sample set X by using the proximity algorithm model obtained in the step (1), dividing the processed soft measurement modeling sample set into two data sets,training data for modeling soft measurementsCollection, N1For the number of groups of the training data set, AkIs a row vector of dimension 1 x d, a set of input quantities for the soft-metric modeling samples,is AkTrue value of corresponding training sample data, k ═ 1,2, …, N1D is the dimension of each set of input quantities;test data set for modeling soft measurements, N2For testing the number of sets of data sets, BkIs a row vector of dimension 1 x d, a set of input quantities, y, for the soft-metric modeling sampleskIs BkTrue value of corresponding test data, k ═ 1,2, …, N2;d+1=m,N1+N2=n;
(4) Searching the optimal value of each parameter in the extreme gradient lifting algorithm by using a grid search method;
(5) after obtaining the optimal parameters, establishing a soft measurement model; acquiring a new data setN3Number of groups for new data set, CkIs a row vector of dimension 1 × d, k is 1,2, …, N3Will beInputting the concentration into a soft measurement model of the ammonia nitrogen concentration of the effluent in the sewage treatment process based on an extreme gradient lifting algorithm to obtain a real-time ammonia nitrogen concentration value of the effluent Is corresponding to CkThe soft measurement model output value of (1);
the specific operation steps of establishing the proximity algorithm model for missing data filling in the step (1) are as follows:
① collecting production data in a batch of sewage treatment process to form a data set for establishing an optimal proximity algorithm model, and performing data normalization processing on the numerical attribute column in the data set to meet the data format supported by the proximity model;
② dividing the normalized data into modeling data set and verification data set;
③ setting the interval of the proximity model parameters α, constructing a proximity model cluster Λ based on the modeling data set and different proximity model parameters α;
④, removing specific attribute column data from the verification data set, constructing missing value data and a missing value matrix, and bringing the missing value data and the missing value matrix into a model to obtain a prediction data set;
⑤ the model optimization objective function is used to screen the optimal proximity model, since the missing value type of the parameter data in the sewage treatment process is numerical data, the objective function S is
Where p denotes the number of data samples in the validation set, gfRepresenting the true value of each sample in the validation set in the missing value data column,is gfCorresponding model fill values, ε being a smoothing factor;
⑥ screening the adjacent model cluster according to the model optimization objective function to obtain the optimal adjacent model Lambda based on the original data and the predicted data of the verification data setbest;
The specific operation steps of searching for the optimal value of each parameter in the extreme gradient lifting algorithm by using the grid search method in the step (4) are as follows:
①, setting 7 initial search ranges of parameters eta belonging to [0.1,1], n _ estimators belonging to [50,800], max _ depth belonging to [1,15], min _ child _ weight belonging to [1,5], gamma belonging to [0,1], subsample [0,1], colsample _ byte belonging to [0,1], search steps respectively of eat: 0.1, n _ estimators: 10, max _ depth: 1, min _ child _ weight: 1, gamma: 0.1, subsample: 0.1, colsample _ byte: 0.1;
② selecting iterator type as gbtree, loss function type as linear, and regression rate as cross validation parameter, regression rate RsIs composed of
WhereinRepresenting the estimated values obtained by the soft measurement model,representing the mean of the first k true values, i.e.
③ A grid search method is used to search the combination Q of 7 parameters in the value ranger (7),Qr (7)Representing the value combination of 7 parameters, wherein the combination number has r groups; the number of initialization iterations s is 1, and the maximum value R of the regression ratem=0;
④ taking the s-th group parameter combination Qs (7)Establishing a soft measurement model based on an extreme gradient lifting algorithm;
⑤ calculation of RsWhen R iss>RmWhen it is established, then Rm=Rs,Qr+1 (7)=Qs (7)Otherwise go to ⑥;
⑥ when s < r is true, s is s +1, go to ④, otherwise go to ⑦;
⑦ taking the 7 parameters of the extreme gradient lifting algorithm as Qr+1 (7)Storing the data into a soft measurement database to obtain a soft measurement model;
the specific operation steps of the step ④ are as follows:
a is the parameter combination Q of the s groups (7)Based on training data setsGenerating a decision tree model of
F={f(x)=ωq(x)} (1)
Wherein F (x) represents the x-th regression tree, F represents the set space of the regression tree, q (x) represents the mapping relation between the sample and the leaf node in the tree model, and omegaq(x)Then representing the mapping relation between the weight omega of the leaf node and the tree structure q;
b, setting the maximum iteration number as K and the target function L (phi) of the extreme gradient lifting algorithm as
Wherein
L (phi) is composed of two parts, loss function and complexity, the loss functionRepresents the estimated value of the ith sampleAnd true valueTraining errors in between; f. ofkRepresents each tree model, Ω (f)k) The complexity of each tree is represented, T represents the number of leaf nodes, gamma and lambda are regular parameters for controlling the model structure, gamma is used for limiting the number of leaf nodes when a single tree is generated, and lambda is used for controlling the step length;
carrying out second-order Taylor expansion on the formula (2) to obtain a target function L of the t-th iteration(t)Is composed of
WhereinRepresenting the coefficients of the first-order expansion terms,representing the coefficients of the second-order expansion term,representing the estimated value of the ith sample after the t-1 iteration, and further using a regular term omega (f)t) Expanding and simplifying target function L of t-th iteration(t)Is written as
Wherein ω isjIs the weight value of the jth leaf node, IjSample represented at jth leaf node, TtRepresenting the number of leaf nodes in the t-th iteration;
derivation of equation (6) to obtain the objective function L(t)Minimum optimal weight value ωj *Is composed of
C, initializing the iteration time t as 1;
d, calculating a loss function of the t-th iterationCalculating L from equation (6)(t)From equation (7), ωj *(ii) a Computing the regression Tree f for the t-th iterationt(x) Corresponding optimal weight value is omegaj *(ii) a Calculate the ith sample A of the t iterationiIs estimated value ofi=1,2,…N1;
E, when the t is satisfied, turning to the step F; otherwise, if t is t +1, go to step D;
f, linearly combining the decision trees generated by each iteration to obtain a K decision tree-integrated soft measurement model based on the extreme gradient lifting algorithm
Wherein f ist(x) Representing a tree model obtained by the t iteration;
g will test the data setInputting the data into a soft measurement model to obtain an estimated valuek=1,2,…N2。
The invention has the beneficial effects that: the invention carries out soft measurement modeling based on the extreme gradient lifting algorithm on the nonlinear relation between the auxiliary variable which can be measured on line and the effluent ammonia nitrogen concentration in the sewage treatment process, carries out on-line soft measurement on the effluent ammonia nitrogen concentration which can not be measured on line in real time through the auxiliary variable which can be measured on line in real time in the sewage treatment process, and provides a method for the real-time on-line soft measurement of the effluent ammonia nitrogen concentration in the sewage treatment process.
Drawings
FIG. 1 is a general flow chart of a soft measurement method for a sewage treatment process based on an extreme gradient boost algorithm.
FIG. 2 is a flow chart for building a proximity algorithm model for missing data population.
FIG. 3 is a flow chart for soft measurement modeling based on an extreme gradient boosting algorithm.
FIG. 4 is a graph of the prediction result of the ammonia nitrogen concentration of the effluent after the soft measurement in the sewage treatment process based on the extreme gradient lifting algorithm, wherein: the ordinate NH4-N (mg/L) is the effluent ammonia nitrogen concentration (mg/L), the abscissa time (day) is the time (day), True Value represents the True Value of the effluent ammonia nitrogen concentration, and Predict Value represents the soft measurement Value of the effluent ammonia nitrogen concentration.
Detailed Description
The following further describes the embodiments of the present invention with reference to the drawings.
Referring to fig. 1,2 and 3, the extreme gradient lift-based soft measurement modeling of the sewage treatment process comprises the following steps:
step 1: collecting complete production data in a group of sewage treatment processes to perform proximity algorithm modeling;
step 2: acquiring production data of historical batches through a real-time database of a control system, wherein each auxiliary variable capable of being measured on line in real time is used as an input quantity of soft measurement modeling, and an ammonia nitrogen concentration value acquired through later-stage off-line measurement is used as an output quantity of the soft measurement modeling;
and step 3: carrying out missing value filling processing on the acquired data set by using a proximity algorithm;
and 4, step 4: setting a cross validation parameter regression rate, an iterator type gbtree and a loss function type linear;
and 5: setting the search step length of the value range of eta, n _ estimators, max _ depth, min _ child _ weight, gamma, subsample and colsample _ byte;
step 6: determining the optimal values of 7 parameters in the interval by using a grid searching method;
and 7: setting the optimal group of parameter values as final parameter values of an extreme gradient lifting algorithm, and carrying out soft measurement modeling;
and 8: and acquiring new values of auxiliary variables which can be measured on line in real time in the sewage treatment process, and directly inputting the new values into the soft measurement model to obtain corresponding predicted values of the ammonia nitrogen concentration of the effluent.
Example (b):
by adopting the extreme gradient lifting-based soft measurement modeling method for the sewage treatment process, 8 production batches are takenNext, each batch of data has a period of 14 days, a sampling interval of 15 minutes, and 1344 groups of data for each batch, and 10752 groups of data for the sewage treatment process, each batch representing a complete sewage treatment process, wherein 6 batches are used as a training data set for soft measurement modelingAiIs a row vector of dimensions 1 x 4, models a set of input quantities of a sample for soft measurements,is AiThe corresponding soft measurement models the output of the sample, i 1, 2.., 8064; 2 batches as a test data set for soft metrology modeling { (B)1,y1),(B2,y2),…,(B2688,y2688)},BiIs a row vector of 1 x 4 dimensions as a set of input quantities for the soft-metric modeling samples, yiIs BiA corresponding soft measurement sample output, i ═ 1, 2. The specific implementation mode is as follows:
setting cross validation parameters, selecting an iterator type and a loss function type, searching for optimal parameters of an effluent ammonia nitrogen concentration soft measurement model based on extreme gradient lifting by a grid search method, storing the determined parameters into a database, and then acquiring a new data setCiIs a row vector of dimensions 1 x 4,is CiCorresponding true value of ammonia nitrogen concentration, i ═ 1, 2.., 1344, will { C1,C2,…,C1344Inputting the data into a sewage treatment process soft measurement model based on an extreme gradient lifting algorithm to obtain a real-time ammonia nitrogen concentration value Is CiCorresponding soft measurement model output values, i ═ 1, 2.
As can be seen from FIG. 4, the soft measurement method for the concentration of the ammonia nitrogen in the effluent based on the extreme gradient lifting algorithm can accurately predict the concentration value of the ammonia nitrogen in the effluent in the sewage treatment process, and has high prediction precision.
The present invention is not intended to be limited to the particular embodiments shown above, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (1)
1. A sewage treatment process soft measurement method based on an extreme gradient lifting algorithm is characterized by comprising the following steps:
(1) collecting production data in a batch of sewage treatment processes, and establishing a proximity algorithm model for missing data filling;
(2) collecting variable values which can be measured on line in the sewage treatment process through a database, using the variable values as input quantity of a soft measurement modeling sample, using an effluent ammonia nitrogen concentration value obtained by off-line measurement as output quantity of the soft measurement modeling sample, and forming a soft measurement modeling sample set X ═ X1,x2,…,xi,…,xn]T,X∈Rn×m,xiIs a row vector of dimension 1 x m, representing the ith sample, i ═ 1,2, … n, n is the total number of samples, m is the total number of process variables, and R is the real number set;
(3) predicting and filling missing values of the soft measurement modeling sample set X by using the proximity algorithm model obtained in the step (1), dividing the processed soft measurement modeling sample set into two data sets,training data set for modeling soft measurements, N1For the number of groups of the training data set, AkIs a row vector of dimension 1 x d, a set of input quantities for the soft-metric modeling samples,is AkTrue value of corresponding training sample data, k ═ 1,2, …, N1D is the dimension of each set of input quantities;test data set for modeling soft measurements, N2For testing the number of sets of data sets, BkIs a row vector of dimension 1 x d, a set of input quantities, y, for the soft-metric modeling sampleskIs BkTrue value of corresponding test data, k ═ 1,2, …, N2;d+1=m,N1+N2=n;
(4) Searching the optimal value of each parameter in the extreme gradient lifting algorithm by using a grid search method;
(5) after obtaining the optimal parameters, establishing a soft measurement model; acquiring a new data setN3Number of groups for new data set, CkIs a row vector of dimension 1 × d, k is 1,2, …, N3Will beInputting the concentration into a soft measurement model of the ammonia nitrogen concentration of the effluent in the sewage treatment process based on an extreme gradient lifting algorithm to obtain a real-time ammonia nitrogen concentration value of the effluent Is corresponding to CkThe soft measurement model output value of (1);
the specific operation steps of establishing the proximity algorithm model for missing data filling in the step (1) are as follows:
① collecting production data in a batch of sewage treatment process to form a data set for establishing an optimal proximity algorithm model, and performing data normalization processing on the numerical attribute column in the data set to meet the data format supported by the proximity model;
② dividing the normalized data into modeling data set and verification data set;
③ setting the interval of the proximity model parameters α, constructing a proximity model cluster Λ based on the modeling data set and different proximity model parameters α;
④, removing specific attribute column data from the verification data set, constructing missing value data and a missing value matrix, and bringing the missing value data and the missing value matrix into a model to obtain a prediction data set;
⑤ the model optimization objective function is used to screen the optimal proximity model, since the missing value type of the parameter data in the sewage treatment process is numerical data, the objective function S is
Where p denotes the number of data samples in the validation set, gfRepresenting the true value of each sample in the validation set in the missing value data column,is gfCorresponding model fill values, ε being a smoothing factor;
⑥ screening the adjacent model cluster according to the model optimization objective function to obtain the optimal adjacent model Lambda based on the original data and the predicted data of the verification data setbest;
The specific operation steps of searching for the optimal value of each parameter in the extreme gradient lifting algorithm by using the grid search method in the step (4) are as follows:
①, setting 7 initial search ranges of parameters eta belonging to [0.1,1], n _ estimators belonging to [50,800], max _ depth belonging to [1,15], min _ child _ weight belonging to [1,5], gamma belonging to [0,1], subsample [0,1], colsample _ byte belonging to [0,1], search steps respectively of eat: 0.1, n _ estimators: 10, max _ depth: 1, min _ child _ weight: 1, gamma: 0.1, subsample: 0.1, colsample _ byte: 0.1;
② choosing the iterator type as gbtree, the loss function type is linear, the regression rate is selected as a cross validation parameter, and the regression rate R is selected assIs composed of
WhereinRepresenting the estimated values obtained by the soft measurement model,representing the mean of the first k true values, i.e.
③ A grid search method is used to search the combination Q of 7 parameters in the value ranger (7),Qr (7)Representing the value combination of 7 parameters, wherein the combination number has r groups; the number of initialization iterations s is 1, and the maximum value R of the regression ratem=0;
④ taking the s-th group parameter combination Qs (7)Establishing a soft measurement model based on an extreme gradient lifting algorithm;
⑤ calculation of RsWhen R iss>RmWhen it is established, then Rm=Rs,Qr+1 (7)=Qs (7)Otherwise go to ⑥;
⑥ when s < r is true, s is s +1, go to ④, otherwise go to ⑦;
⑦ taking the 7 parameters of the extreme gradient lifting algorithm as Qr+1 (7)Storing the data into a soft measurement database to obtain a soft measurement model;
the specific operation steps of the step ④ are as follows:
a is the parameter combination Q of the s groups (7)Based on training data setsGenerating a decision tree model of
F={f(x)=ωq(x)} (1)
Wherein F (x) represents the x-th regression tree, F represents the set space of the regression tree, q (x) represents the mapping relation between the sample and the leaf node in the tree model, and omegaq(x)Then representing the mapping relation between the weight omega of the leaf node and the tree structure q;
b, setting the maximum iteration number as K and the target function L (phi) of the extreme gradient lifting algorithm as
Wherein
L (phi) is composed of two parts, loss function and complexity, the loss functionRepresents the estimated value of the ith sampleAnd true valueTraining errors in between; f. ofkRepresents each tree model, Ω (f)k) The complexity of each tree is represented, T represents the number of leaf nodes, gamma and lambda are regular parameters for controlling the model structure, gamma is used for limiting the number of leaf nodes when a single tree is generated, and lambda is used for controlling the step length;
carrying out second-order Taylor expansion on the formula (2) to obtain a target function L of the t-th iteration(t)Is composed of
WhereinRepresenting the coefficients of the first-order expansion terms,representing the coefficients of the second-order expansion term,representing the estimated value of the ith sample after the t-1 iteration, and further using a regular term omega (f)t) Expanding and simplifying target function L of t-th iteration(t)Is written as
Wherein ω isjIs the weight value of the jth leaf node, IjSample represented at jth leaf node, TtRepresenting the number of leaf nodes in the t-th iteration;
derivation of equation (6) to obtain the objective function L(t)Minimum optimal weight value ωj *Is composed of
C, initializing the iteration time t as 1;
d, calculating a loss function of the t-th iterationCalculating L from equation (6)(t)From equation (7), ωj *(ii) a Computing the regression Tree f for the t-th iterationt(x) Corresponding optimal weight value is omegaj *(ii) a Calculate the t-th iterationThe ith sample AiIs estimated value of
E, when the t is satisfied, turning to the step F; otherwise, if t is t +1, go to step D;
f, linearly combining the decision trees generated by each iteration to obtain a K decision tree-integrated soft measurement model based on the extreme gradient lifting algorithm
Wherein f ist(x) Representing a tree model obtained by the t iteration;
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911267973.0A CN110909492A (en) | 2019-12-11 | 2019-12-11 | Sewage treatment process soft measurement method based on extreme gradient lifting algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911267973.0A CN110909492A (en) | 2019-12-11 | 2019-12-11 | Sewage treatment process soft measurement method based on extreme gradient lifting algorithm |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110909492A true CN110909492A (en) | 2020-03-24 |
Family
ID=69824715
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911267973.0A Pending CN110909492A (en) | 2019-12-11 | 2019-12-11 | Sewage treatment process soft measurement method based on extreme gradient lifting algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110909492A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112381221A (en) * | 2020-10-28 | 2021-02-19 | 华南理工大学 | Multi-output soft measurement method for sewage monitoring based on semi-supervised learning |
CN113295625A (en) * | 2021-04-30 | 2021-08-24 | 西安理工大学 | Machine vision dye concentration spectrum detection method based on extreme gradient promotion |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109858190A (en) * | 2019-03-13 | 2019-06-07 | 江南大学 | A kind of penicillin fermentation process soft measuring modeling method promoting regression tree based on drosophila algorithm optimization gradient |
CN110542748A (en) * | 2019-07-24 | 2019-12-06 | 北京工业大学 | knowledge-based robust effluent ammonia nitrogen soft measurement method |
-
2019
- 2019-12-11 CN CN201911267973.0A patent/CN110909492A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109858190A (en) * | 2019-03-13 | 2019-06-07 | 江南大学 | A kind of penicillin fermentation process soft measuring modeling method promoting regression tree based on drosophila algorithm optimization gradient |
CN110542748A (en) * | 2019-07-24 | 2019-12-06 | 北京工业大学 | knowledge-based robust effluent ammonia nitrogen soft measurement method |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112381221A (en) * | 2020-10-28 | 2021-02-19 | 华南理工大学 | Multi-output soft measurement method for sewage monitoring based on semi-supervised learning |
CN113295625A (en) * | 2021-04-30 | 2021-08-24 | 西安理工大学 | Machine vision dye concentration spectrum detection method based on extreme gradient promotion |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110084367B (en) | Soil moisture content prediction method based on LSTM deep learning model | |
CN110807554B (en) | Generation method and system based on wind power/photovoltaic classical scene set | |
CN109635245A (en) | A kind of robust width learning system | |
CN112182709B (en) | Method for rapidly predicting water drainage temperature of large reservoir stoplog gate layered water taking facility | |
CN101863088A (en) | Method for forecasting Mooney viscosity in rubber mixing process | |
CN112989711B (en) | Aureomycin fermentation process soft measurement modeling method based on semi-supervised ensemble learning | |
CN109616161B (en) | Fermentation process soft measurement method based on twin support vector regression machine | |
CN110442911B (en) | High-dimensional complex system uncertainty analysis method based on statistical machine learning | |
CN112949894B (en) | Output water BOD prediction method based on simplified long-short-term memory neural network | |
CN111754034A (en) | Time sequence prediction method based on chaos optimization neural network model | |
CN108595892A (en) | Soft-measuring modeling method based on time difference model | |
CN110909492A (en) | Sewage treatment process soft measurement method based on extreme gradient lifting algorithm | |
CN116757057A (en) | Air quality prediction method based on PSO-GA-LSTM model | |
CN111639111A (en) | Water transfer engineering-oriented multi-source monitoring data deep mining and intelligent analysis method | |
CN110908361B (en) | Fermentation process soft measurement method based on online twin support vector regression | |
CN117289652A (en) | Numerical control machine tool spindle thermal error modeling method based on multi-universe optimization | |
CN114971090A (en) | Electric heating load prediction method, system, equipment and medium | |
CN111832839A (en) | Energy consumption prediction method based on sufficient incremental learning | |
CN103279030B (en) | Dynamic soft measuring modeling method and device based on Bayesian frame | |
CN109408896B (en) | Multi-element intelligent real-time monitoring method for anaerobic sewage treatment gas production | |
CN113051806B (en) | Water quality BOD measurement method based on AQPSO-RBF neural network | |
CN114239397A (en) | Soft measurement modeling method based on dynamic feature extraction and local weighted deep learning | |
CN112381279B (en) | Wind power prediction method based on VMD and BLS combined model | |
CN110276478B (en) | Short-term wind power prediction method based on segmented ant colony algorithm optimization SVM | |
CN116316600A (en) | Wind power day-ahead prediction self-adaptive filtering method based on seasonal typing |
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 |