CN116596396A - Industrial polyethylene process quality prediction method based on K nearest neighbor interpolation and SLSTM - Google Patents
Industrial polyethylene process quality prediction method based on K nearest neighbor interpolation and SLSTM Download PDFInfo
- Publication number
- CN116596396A CN116596396A CN202310616717.8A CN202310616717A CN116596396A CN 116596396 A CN116596396 A CN 116596396A CN 202310616717 A CN202310616717 A CN 202310616717A CN 116596396 A CN116596396 A CN 116596396A
- Authority
- CN
- China
- Prior art keywords
- data
- slstm
- quality
- model
- prediction
- 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
- 238000000034 method Methods 0.000 title claims abstract description 102
- 230000008569 process Effects 0.000 title claims abstract description 82
- 239000004698 Polyethylene Substances 0.000 title claims abstract description 41
- -1 polyethylene Polymers 0.000 title claims abstract description 41
- 229920000573 polyethylene Polymers 0.000 title claims abstract description 41
- 238000012549 training Methods 0.000 claims abstract description 38
- 238000012795 verification Methods 0.000 claims abstract description 13
- 238000010606 normalization Methods 0.000 claims abstract description 9
- 238000012545 processing Methods 0.000 claims abstract description 7
- 239000013598 vector Substances 0.000 claims description 27
- 239000011159 matrix material Substances 0.000 claims description 12
- 230000006870 function Effects 0.000 claims description 10
- 230000015654 memory Effects 0.000 claims description 8
- 230000004913 activation Effects 0.000 claims description 6
- 238000004458 analytical method Methods 0.000 claims description 4
- 238000004519 manufacturing process Methods 0.000 description 5
- 238000012360 testing method Methods 0.000 description 3
- 230000003321 amplification Effects 0.000 description 2
- 239000000155 melt Substances 0.000 description 2
- 238000003199 nucleic acid amplification method Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 235000003642 hunger Nutrition 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 230000037351 starvation Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06395—Quality analysis or management
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24147—Distances to closest patterns, e.g. nearest neighbour classification
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Strategic Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Economics (AREA)
- Educational Administration (AREA)
- Development Economics (AREA)
- Artificial Intelligence (AREA)
- General Business, Economics & Management (AREA)
- Game Theory and Decision Science (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Marketing (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an industrial polyethylene process quality prediction method based on K neighbor interpolation and SLSTM, which comprises the steps of obtaining process data and quality data of industrial polyethylene, constructing a data set according to the process data and the quality data, expanding the data set, carrying out normalization processing, converting the normalized data set into a sequence data set, constructing an SLSTM model, training the SLSTM model according to a training set, inputting a verification set into the trained SLSTM model for prediction, calculating the prediction precision of the trained SLSTM model according to a prediction result, adjusting parameters of the trained SLSTM model when the prediction precision does not meet a threshold value, retraining according to the training set until the prediction precision of the trained SLSTM model meets the threshold value, obtaining process data and quality data with prediction, predicting according to the trained SLSTM model, and obtaining the quality data of industrial polyethylene at the current moment. And a hidden relation between the quality variable and the process variable is established, so that the accuracy of model prediction is improved.
Description
Technical Field
The invention relates to the field of industrial polyethylene process quality prediction, in particular to an industrial polyethylene process quality prediction method based on K nearest neighbor interpolation and SLSTM.
Background
Obtaining large amounts of high quality data can be difficult in certain industrial processes, and therefore how to utilize limited data to improve prediction accuracy is an important research task. The problem of low model prediction accuracy caused by insufficient data in the current industrial production process. The traditional cyclic neural network (RNN) and long-short-term memory network (LSTM) models can only capture the hidden relation between the process variables, neglect the influence of quality variables on the prediction result, and cause low precision on the prediction result.
Disclosure of Invention
The invention provides an industrial polyethylene process quality prediction method based on K nearest neighbor interpolation and SLSTM, which aims to overcome the technical problems.
An industrial polyethylene process quality prediction method based on K nearest neighbor interpolation and SLSTM comprises the following steps of
Step one, acquiring process data for producing industrial polyethylene by using a sensor, obtaining quality data by offline analysis of the produced industrial polyethylene, wherein the process data comprises five process variables, the quality data comprises one quality variable, constructing a data set according to the process data and the quality data,
expanding the data set based on the KNN algorithm to obtain an expanded data set,
step three, carrying out normalization processing on each piece of data in the expanded data set to obtain a normalized data set, converting the normalized data set into a sequence data set, dividing the sequence data set into a training set and a verification set,
step four, constructing an SLSTM model, training the SLSTM model according to a training set, obtaining a trained SLSTM model, inputting a verification set into the trained SLSTM model for prediction, obtaining a prediction result, calculating the prediction precision of the trained SLSTM model according to the prediction result, adjusting parameters of the trained SLSTM model when the prediction precision does not meet a threshold value, retraining according to the training set until the prediction precision of the trained SLSTM model meets the threshold value,
and fifthly, acquiring process data of industrial polyethylene produced at the current moment and quality data of the previous moment, normalizing the process data and the quality data according to a normalized data set, and inputting the normalized process data and the normalized quality data into a trained SLSTM model to predict, so as to acquire the quality data of industrial polyethylene produced at the current moment.
Preferably, the second step includes,
s21: acquiring a data set, representing process data in the data set as a data vector, and acquiring an nth data vector x in the data set n ，n＝1；
S22: respectively calculate x n Euclidean distance d from data vector of data set o-th data n，o ＝||x n -x o || 2 ，o＝n+1，n+2，……，n+m，||·|| 2 Represent Euclidean distance, n<m<L, L represents the number of samples in the dataset;
s23: will d n，o Ordering from small to large, acquiring such that d n，o The first k minimum vectors x p ，p∈o，k＜m；
S24: calculating the first k vectors x according to equation (1) p And as a new vector
wherein ,the sum of the first to five process variables, which represent the first k vectors, respectively, is based on +.>The values of the five process variables calculate the values of the quality variables, will +.>The five of the process variables and the quality variables are stored as new samples in the dataset,
s25: let n=n+1, if n+.l-m, return to S22, otherwise, end the loop.
Preferably, the normalizing each piece of data in the expanded dataset comprises normalizing each piece of data in the dataset according to formulas (2), (3),
wherein ,x′i，j and y′i Respectively representing the value of the ith sample, the jth process variable and the normalized value of the quality variable, x i，j Values, x, representing the jth process variable of the ith sample j，min Representing the minimum value, x, of the jth process variable in the dataset j，max Represents the maximum value, y, of the jth process variable in the dataset i Representing the quality data of the ith sample, y max and ymin Respectively, the maximum and minimum of quality data in the dataset.
Preferably, the updated formula of the SLSTM model is shown in formula (4),
in the formula ,ft Indicating forgetful door, i t Representing an input door,Representing a new candidate cell state, c t Representing a new cell state, c t-1 On the representationCell state at one instant o t Indicating the output gate, h t Represents the hidden state at time t, h t-1 Represents the hidden state, x, at time t-1 t Input vector, y representing time t t Output vector w representing time t f 、w i 、w c 、w o Respectively representing a forgetting gate weight parameter matrix, an input gate weight parameter matrix, a memory cell gate weight parameter matrix and an output gate weight parameter matrix; b f 、b i 、b c 、b o Respectively representing a forget gate bias term, an input gate bias term, a memory cell gate bias term and an output gate bias term; delta represents a sigmoid activation function, x represents an element-by-element multiplication operation, and tanh represents a hyperbolic tangent activation function.
Preferably, said calculating the prediction accuracy of the trained SLSTM model from the prediction results comprises calculating a root mean square error according to equation (5), taking the root mean square error as the prediction accuracy,
wherein n represents the number of samples, y i True value, y, representing the mass variable of the technical polyethylene i pre Representing the predicted result.
The invention provides an industrial polyethylene process quality prediction method based on K neighbor interpolation and SLSTM, which expands samples through K neighbor interpolation, solves the problem of insufficient samples, gradually reduces root mean square error on a verification set along with increase of amplified data, improves model prediction performance, establishes a hidden relation between quality variables and process variables by constructing an SLSTM model, and improves model prediction accuracy.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions of the prior art, the drawings that are needed in the embodiments or the description of the prior art will be briefly described below, it will be obvious that the drawings in the following description are some embodiments of the present invention, and that other drawings can be obtained according to these drawings without inventive effort to a person skilled in the art.
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 is a graph of the predicted results of different models of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
FIG. 1 is a flowchart of the method of the present invention, as shown in FIG. 1, the method of the present embodiment may include:
step one, acquiring process data for producing industrial polyethylene by using a sensor, obtaining quality data by offline analysis of the produced industrial polyethylene, wherein the process data comprises five process variables, the quality data comprises one quality variable, constructing a data set according to the process data and the quality data,
expanding the data set based on the KNN algorithm to obtain an expanded data set,
step three, carrying out normalization processing on each piece of data in the expanded data set to obtain a normalized data set, converting the normalized data set into a sequence data set, dividing the sequence data set into a training set and a verification set,
step four, constructing an SLSTM model, training the SLSTM model according to a training set, obtaining a trained SLSTM model, inputting a verification set into the trained SLSTM model for prediction, obtaining a prediction result, calculating the prediction precision of the trained SLSTM model according to the prediction result, adjusting parameters of the trained SLSTM model when the prediction precision does not meet a threshold value, retraining according to the training set until the prediction precision of the trained SLSTM model meets the threshold value,
and fifthly, acquiring process data of industrial polyethylene produced at the current moment and quality data of the previous moment, normalizing the process data and the quality data according to a normalized data set, and inputting the normalized process data and the normalized quality data into a trained SLSTM model to predict, so as to acquire the quality data of industrial polyethylene produced at the current moment.
Based on the scheme, the samples are expanded through K neighbor interpolation, the problem of insufficient samples is solved, along with the increase of amplified data, root mean square errors on a verification set are gradually reduced, model prediction performance is improved, a hidden relation between quality variables and process variables is established by constructing an SLSTM model, and model prediction accuracy is improved.
Step one, acquiring process data of industrial polyethylene production by using a sensor, and obtaining quality data by offline analysis of the industrial polyethylene production, wherein the process data comprises five process variables, the five process variables can be five sensors installed for different links in the production process, the data generated by the five sensors are respectively acquired, the quality data comprises a quality variable, and the quality variable can be a melt index. A dataset is constructed from the process data and the quality data. The sensor is used for collecting industrial polyethylene process data, the quality of the industrial polyethylene can be characterized by various indexes, the melt index is one index commonly used in production, and the data can be used for comprehensively evaluating the quality of the industrial polyethylene through statistical analysis.
And (3) data amplification, K neighbor interpolation, namely selecting a similar data set by using the basic idea of a KNN (K-Nearest Neighbors) algorithm, and calculating the average value of the selected data set to obtain virtual data. Step two, expanding the data set based on KNN algorithm to obtain an expanded data set,
the second step of the method comprises the steps of,
s21: acquiring a data set, representing process data in the data set as a data vector, and acquiring an nth data vector x in the data set n ，n＝1；
S22: respectively calculate x n Euclidean distance d from data vector of data set o-th data n，o ＝||x n -x o || 2 ，o＝n+1，n+2，……，n+m，||·|| 2 Represent Euclidean distance, n<m<L, L represents the number of samples in the dataset;
s23: will d n，o Ordering from small to large, acquiring such that d n，o The first k minimum vectors x p ，p∈o，k＜m；
S24: calculating the first k vectors x according to equation (1) p And as a new vector
wherein ,the sum of the first to five process variables, which represent the first k vectors, respectively, is based on +.>The values of the five process variables calculate the values of the quality variables, will +.>The five of the process variables and the quality variables are stored as new samples in the dataset,
s25: let n=n+1, if n+.l-m, return to S22, otherwise, end the loop.
Step three, carrying out normalization processing on each piece of data in the expanded data set, limiting the numerical value of all variables to be between 0 and 1 by adopting a maximum and minimum normalization processing method, carrying out normalization processing on each piece of data in the expanded data set, carrying out normalization processing on each piece of data in the data set according to formulas (2) and (3),
wherein ,x′i，j and y′i Respectively representing the value of the ith sample, the jth process variable and the normalized value of the quality variable, x i，j Values, x, representing the jth process variable of the ith sample j，min Representing the minimum value, x, of the jth process variable in the dataset j，max Represents the maximum value, y, of the jth process variable in the dataset i Representing the quality data of the ith sample, y max and ymin Respectively, the maximum and minimum of quality data in the dataset.
After normalization, a sequence data set is established, which is converted into supervised learning data, where the quality variables and the five other process variables are used as inputs to the model, i.e. the modeling data becomes [ x ]' i，1 ，x′ i，2 ，x′ i，4 ，x′ i，5 ，y′ i-1], wherein yi-1 Is the quality variable of the i-1 th sample. And then randomly disturbing modeling data according to rows, setting random seeds to ensure that the disturbed data of each training are consistent, and dividing the training set and the verification set according to the proportion of 3:1.
Obtaining a normalized data set, converting the normalized data set into a sequence data set, dividing the sequence data set into a training set and a verification set,
and fourthly, constructing an SLSTM model, wherein the SLSTM (Supervised Long Short-Term Memory) utilizes sample data containing quality variables and process variables in a training set to learn dynamic hidden states at the same time, and the hidden states contain dynamic information related to the quality in the quality variables, so that the quality regression prediction is facilitated. The updated formula of the SLSTM model is shown as formula (4),
in the formula ,ft Indicating forgetful door, i t Representing an input door,Representing a new candidate cell state, c t Representing a new cell state, c t-1 Indicating the state of the cell, o, at the previous time t Indicating the output gate, h t Represents the hidden state at time t, h t-1 Represents the hidden state, x, at time t-1 t Input vector, y representing time t t Output vector w representing time t f 、w i 、w c 、w o Respectively representing a forgetting gate weight parameter matrix, an input gate weight parameter matrix, a memory cell gate weight parameter matrix and an output gate weight parameter matrix; b f 、b i 、b c 、b o Respectively representing a forget gate bias term, an input gate bias term, a memory cell gate bias term and an output gate bias term; delta represents a sigmoid activation function, x represents an element-by-element multiplication operation, and tanh represents a hyperbolic tangent activation function.
Training the SLSTM model according to a training set, obtaining a trained SLSTM model, inputting a verification set into the trained SLSTM model for prediction, obtaining a prediction result, calculating the prediction accuracy of the trained SLSTM model according to the prediction result, wherein calculating the prediction accuracy of the trained SLSTM model according to the prediction result comprises calculating a root mean square error according to a formula (5), taking the root mean square error as the prediction accuracy,
wherein n represents the number of samples, y i True value, y, representing the mass variable of the technical polyethylene i pre Representing the predicted result.
When the prediction accuracy does not meet the threshold value, the parameters of the trained SLSTM model are adjusted and retrained according to the training set until the prediction accuracy of the trained SLSTM model meets the threshold value,
in performing SLSTM network training, there are two important parameters to adjust, epoch (epoch) and batch size.
Epoch refers to the number of single training iterations of all training data in the training process. The larger the epochs, the more iterated the data set will be by the network, but too large epochs may also cause over-fitting problems. In general, the value of the epoch can be gradually increased from small to large, and the value of the epoch is adjusted by comparing the curves of the loss functions on the training set and the test set, specifically, the training times when the curves of the loss functions reach the peak value in different training processes are respectively obtained, the average training times are calculated according to the plurality of training times, and the average training times are used as the value of the epoch.
Batch size refers to the number of samples used per training Batch. The larger the batch size, the more samples the model has per training, which increases the training speed of the model. However, too large a batch size may also cause memory starvation. In general, the value of the batch size can be gradually increased from small to large, and compared with the loss function curves on the training set and the test set to adjust the value of the batch size, specifically, the number of samples when the loss function curves reach the peak value in different training processes is respectively obtained, the average sample number is calculated according to the plurality of sample numbers, and the average sample number is used as the value of the batch size.
And fifthly, acquiring process data of industrial polyethylene produced at the current moment and quality data of the previous moment, normalizing the process data and the quality data according to a normalized data set, and inputting the normalized process data and the normalized quality data into a trained SLSTM model to predict, so as to acquire the quality data of industrial polyethylene produced at the current moment.
This example illustrates the effectiveness of the process taking as an example an industrial polyethylene quality prediction. For the amplified 162 samples, 160 samples were selected for comparative training in five groups: raw data, 40 samples, 80 data samples, 120 data samples and 160 data samples are amplified, and a prediction model is respectively built for the five groups of data by using an industrial polyethylene quality prediction method based on K neighbor interpolation and SLSTM. In order to verify the model prediction effect, the prediction results are analyzed by using root mean square error RMSE as an index, and the validity of the method is verified, wherein the prediction results of different models are shown in fig. 2.
TABLE 1 root mean square error values for different models
As shown in Table 1, it can be seen that the RMSE on the test set gradually decreased with increasing amplification data, and the model predictive performance improved. Experimental results show that the industrial polyethylene quality prediction method based on K nearest neighbor interpolation and SLSTM effectively improves the prediction performance of the system.
The whole beneficial effects are that:
the invention provides an industrial polyethylene process quality prediction method based on K neighbor interpolation and SLSTM, which expands samples through K neighbor interpolation, solves the problem of insufficient samples, gradually reduces root mean square error on a verification set along with increase of amplified data, improves model prediction performance, establishes a hidden relation between quality variables and process variables by constructing an SLSTM model, and improves model prediction accuracy.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the invention.
Claims (5)
1. The industrial polyethylene process quality prediction method based on K nearest neighbor interpolation and SLSTM is characterized by comprising the following steps of
Step one, acquiring process data for producing industrial polyethylene by using a sensor, obtaining quality data by offline analysis of the produced industrial polyethylene, wherein the process data comprises five process variables, the quality data comprises one quality variable, constructing a data set according to the process data and the quality data,
expanding the data set based on the KNN algorithm to obtain an expanded data set,
step three, carrying out normalization processing on each piece of data in the expanded data set to obtain a normalized data set, converting the normalized data set into a sequence data set, dividing the sequence data set into a training set and a verification set,
step four, constructing an SLSTM model, training the SLSTM model according to a training set, obtaining a trained SLSTM model, inputting a verification set into the trained SLSTM model for prediction, obtaining a prediction result, calculating the prediction precision of the trained SLSTM model according to the prediction result, adjusting parameters of the trained SLSTM model when the prediction precision does not meet a threshold value, retraining according to the training set until the prediction precision of the trained SLSTM model meets the threshold value,
and fifthly, acquiring process data of industrial polyethylene produced at the current moment and quality data of the previous moment, normalizing the process data and the quality data according to a normalized data set, and inputting the normalized process data and the normalized quality data into a trained SLSTM model to predict, so as to acquire the quality data of industrial polyethylene produced at the current moment.
2. The industrial polyethylene process quality prediction method based on K-nearest neighbor interpolation and SLSTM according to claim 1, wherein the second step comprises,
s21: acquiring a data set, representing process data in the data set as a data vector, and acquiring an nth data vector x in the data set n ，n＝1；
S22: respectively calculate x n Euclidean distance d from data vector of data set o-th data n，o ＝||x n -x o || 2 ，o＝n+1，n+2，……，n+m，||·|| 2 Representing Euclidean distance, n < m < L, L representing the number of samples in the data set;
s23: will d n，o Ordering from small to large, acquiring such that d n，o The first k minimum vectors x p ，p∈o，k<m；
S24: calculating the first k vectors x according to equation (1) p And as a new vector
wherein ,the sum of the first to five process variables, which represent the first k vectors, respectively, is based on +.>The values of the five process variables calculate the values of the quality variables, will +.>The five of the process variables and the quality variables are stored as new samples in the dataset,
s25: let n=n+1, if n is less than or equal to L-m, return to S22, otherwise, end the loop.
3. The industrial polyethylene process quality prediction method based on K-nearest neighbor interpolation and SLSTM according to claim 1, wherein normalizing each piece of data in the expanded dataset comprises normalizing each piece of data in the dataset according to equations (2), (3),
wherein ,x′i,j and y′i Respectively representing the value of the ith sample, the jth process variable and the normalized value of the quality variable, x i,j Values, x, representing the jth process variable of the ith sample j，min Representing the minimum value, x, of the jth process variable in the dataset j，max Represents the maximum value, y, of the jth process variable in the dataset i Representing the quality data of the ith sample, y max and ymim Respectively, the maximum and minimum of quality data in the dataset.
4. The industrial polyethylene process quality prediction method based on K nearest neighbor interpolation and SLSTM according to claim 1, wherein the update formula of the SLSTM model is shown in formula (4),
in the formula ,ft Indicating forgetful door, i t Representing an input door,Representing a new candidate cell state, c t Representing a new cell state, c t-1 Indicating the state of the cell, o, at the previous time t Indicating the output gate, h t Represents the hidden state at time t, h t-1 Represents the hidden state, x, at time t-1 t Input vector, y representing time t t Output vector w representing time t f 、w i 、w c 、w o Respectively represent forgetting gate weight parameter matrix, input gate weight parameter matrix and memory cell gate weight parameterOutputting a gate weight parameter matrix by the number matrix; b f 、b i 、b c 、b o Respectively representing a forget gate bias term, an input gate bias term, a memory cell gate bias term and an output gate bias term; delta represents a sigmoid activation function, x represents an element-by-element multiplication operation, and tanh represents a hyperbolic tangent activation function.
5. The industrial polyethylene process quality prediction method based on K-nearest neighbor interpolation and SLSTM according to claim 1, wherein calculating the prediction accuracy of the trained SLSTM model based on the prediction result comprises calculating a root mean square error according to equation (5), taking the root mean square error as the prediction accuracy,
wherein n represents the number of samples, yx represents the true value of the industrial polyethylene quality variable, y i pre Representing the predicted result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310616717.8A CN116596396A (en) | 2023-05-29 | 2023-05-29 | Industrial polyethylene process quality prediction method based on K nearest neighbor interpolation and SLSTM |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310616717.8A CN116596396A (en) | 2023-05-29 | 2023-05-29 | Industrial polyethylene process quality prediction method based on K nearest neighbor interpolation and SLSTM |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116596396A true CN116596396A (en) | 2023-08-15 |
Family
ID=87607987
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310616717.8A Pending CN116596396A (en) | 2023-05-29 | 2023-05-29 | Industrial polyethylene process quality prediction method based on K nearest neighbor interpolation and SLSTM |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116596396A (en) |
-
2023
- 2023-05-29 CN CN202310616717.8A patent/CN116596396A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Abonyi et al. | Modified Gath-Geva fuzzy clustering for identification of Takagi-Sugeno fuzzy models | |
| Amari et al. | Asymptotic statistical theory of overtraining and cross-validation | |
| CN111563706A (en) | Multivariable logistics freight volume prediction method based on LSTM network | |
| CN107992976B (en) | Hot topic early development trend prediction system and prediction method | |
| CN111144552B (en) | Multi-index grain quality prediction method and device | |
| CN111931983B (en) | Precipitation prediction method and system | |
| US20110060706A1 (en) | Information processing device, information processing method, and program | |
| Suryo et al. | Improved time series prediction using LSTM neural network for smart agriculture application | |
| CN114548592A (en) | Non-stationary time series data prediction method based on CEMD and LSTM | |
| US20110060707A1 (en) | Information processing device, information processing method, and program | |
| Susan et al. | Dynamic growth of hidden-layer neurons using the non-extensive entropy | |
| CN110738363A (en) | photovoltaic power generation power prediction model and construction method and application thereof | |
| Yang et al. | A neural network learning algorithm of chemical process modeling based on the extended Kalman filter | |
| CN113128666A (en) | Mo-S-LSTMs model-based time series multi-step prediction method | |
| CN117609905A (en) | Monitoring data prediction method, system, equipment and readable storage medium | |
| CN115062764B (en) | Intelligent illuminance adjustment and environmental parameter Internet of things big data system | |
| US20230092949A1 (en) | System and method for estimating model metrics without labels | |
| CN116596396A (en) | Industrial polyethylene process quality prediction method based on K nearest neighbor interpolation and SLSTM | |
| CN115062528A (en) | Prediction method for industrial process time sequence data | |
| CN110826184B (en) | Variational Bayesian identification method for NARX model structure and parameters under time-varying lag | |
| CN107220483B (en) | Earth temperature mode prediction method | |
| Dudukcu et al. | Comprehensive comparison of lstm variations for the prediction of chaotic time series | |
| Haque et al. | Improved adam-based feedforward deep neural network model for personalized asthma predictions | |
| CN113807606B (en) | Intermittent process quality online prediction method capable of explaining ensemble learning | |
| CN115719040B (en) | Penicillin fermentation process key variable soft measurement method and system |
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 |