CN111103420A - A method for predicting the quality of phenolic resin products under uncertainty of raw materials - Google Patents
A method for predicting the quality of phenolic resin products under uncertainty of raw materialsDownload PDFInfo
- Publication number
- CN111103420A CN111103420ACN201911140036.9ACN201911140036ACN111103420ACN 111103420 ACN111103420 ACN 111103420ACN 201911140036 ACN201911140036 ACN 201911140036ACN 111103420 ACN111103420 ACN 111103420A
- Authority
- CN
- China
- Prior art keywords
- phenolic resin
- product quality
- data
- model
- resin product
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
- G01N33/44—Resins; Plastics; Rubber; Leather
- G01N33/442—Resins; Plastics
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Chemical & Material Sciences (AREA)
- Food Science & Technology (AREA)
- Medicinal Chemistry (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
Description
Technical Field
The invention belongs to the field of online prediction of product quality in a chemical process, and particularly relates to a method for predicting the quality of a phenolic resin product under the condition of uncertain raw materials.
Background
Phenolic resin is a classical synthetic resin, which is a high molecular material formed by polymerizing phenols and aldehydes under the action of a catalyst. Different formulas are added during synthesis, modified phenolic resins with different functions can be obtained, and the modified phenolic resins have different excellent characteristics and are widely applied to anticorrosion engineering, adhesives, flame-retardant materials, grinding wheel manufacturing and rubber product industries. The phenolic resin is usually produced in batch production mode in various varieties and batches, the production process has time-varying, dynamic and nonlinear characteristics, and meanwhile, the phenolic resin is produced by taking a phenolic mixture produced at the upstream as a raw material. The uncertain composition of the mixture raw materials and the dynamic characteristics of the production process cause that the product quality cannot be kept stable, and the product quality is difficult to realize on-line measurement, so that the method is particularly important for predicting the product quality.
With the development of computers and technology, a great deal of process data and analytical data in the industrial phenolic resin production process is preserved. These data contain rich process information. The development of deep learning provides a new solution for solving the problem of product quality prediction in the phenolic resin production process with uncertain raw materials. Compared with the traditional artificial neural network, the cyclic neural network in deep learning has better capability of processing time series, and the process rule under uncertain raw materials can be mined by modeling the time series of the production process through the long-time memory network, so that the accurate prediction of the product quality is realized.
Disclosure of Invention
One of the purposes of the invention is to provide a method for predicting the quality of a phenolic resin product under the condition of uncertain raw materials, so that the quality of the produced phenolic resin product can be predicted in real time.
Based on the purpose, the invention provides a phenolic resin product quality prediction method based on a long-time and short-time memory network, which comprises the following steps:
(1) obtaining historical data of phenolic resin production process and corresponding batch product quality
(2) Preprocessing, feature selection and standardization are carried out on historical data;
(3) constructing a phenolic resin product quality prediction model based on a long-time and short-time memory network;
(4) training a phenolic resin product quality prediction model based on the data obtained in the step (2);
(5) and (3) acquiring real-time production data of the current batch, processing the data by using the pretreatment and standardization in the step (2), and predicting the product quality of the current batch through the phenolic resin product quality prediction model.
(6) And (3) when the real-time prediction accuracy in the step (5) does not meet the requirement, processing the data by using the pretreatment and standardization in the step (2) according to the real-time production data obtained in the step (5) as historical data, returning to the step (4) to train the model, and obtaining a new phenolic resin product quality prediction model.
The phenolic resin product prediction method based on the long-time and short-time memory network comprises the following steps:
the historical data is time series data of the production process of the phenolic resin and product quality data of corresponding batches.
The Long Short-Term Memory network (LSTM) is a recurrent neural network, and is suitable for processing and predicting regular information in time sequence data.
The phenolic resin product quality prediction method based on the long-time and short-time memory network is characterized in that a phenolic resin product quality prediction model based on the long-time and short-time memory network is constructed and trained on the basis of historical data of a phenolic resin production process and product quality data of corresponding batches, and the influence rule of uncertain raw materials on the product quality is obtained by extracting the internal trend rule of the historical data, so that the product quality of the phenolic resin production process of the current batch is effectively predicted.
Further, in the phenolic resin product quality prediction method based on the long-time and short-time memory network, historical data of the phenolic resin production process comprise alkylation liquid flow, liquid aldehyde flow, solid aldehyde flow, reaction temperature, pressure and reaction kettle weight, and product quality data of corresponding batches comprise softening point data obtained through laboratory analysis.
Furthermore, in the method for predicting the quality of the phenolic resin product based on the long-time and short-time memory network, in the step (2), the preprocessing of the historical data is to resample the time series data at equal time intervals, so that the dimension of the variable is reduced.
Furthermore, in the method for predicting the quality of the phenolic resin product based on the long-time and short-time memory network, in the step (2), the data is subjected to correlation analysis by using the maximum information coefficient for the feature selection, and the correlation analysis can be calculated by the following formula:
wherein x represents any variable in the historical data of the phenolic aldehyde production process, y represents the product quality index of the phenolic resin, and B is the 0.6 power of the total data. The maximum information coefficient is a decimal number between 0 and 1, and the larger the maximum information coefficient is, the stronger the correlation between the variable and the quality index is.
And selecting a variable with the maximum information coefficient of the product quality index larger than 0.3 as an input variable of the phenolic resin product quality prediction model.
Further, in the method for predicting the quality of a phenol resin product based on a long-term and short-term memory network, in the step (2), the data is normalized by normalizing the obtained data and the obtained product quality respectively, and the data is obtained by the following formula:
in formula (II), x'ikA k-dimensional variable x representing the ith lotikNormalized values in the range of [0,1 ]]The method has the advantages of no dimension,
and
respectively representing the minimum value and the maximum value in the k-dimension variable, and I represents the total number of batches.
Product quality was normalized and obtained by the following formula:
of formula (II) to (III)'iRepresents the final quality index y of the ith batchiNormalized numerical values in the range of [0,1 ]]The method has the advantages of no dimension,
and
the minimum and maximum values are indicated separately and I indicates the total number of batches.
Furthermore, in the method for predicting the quality of the phenolic resin product based on the long-term and short-term memory network, in the step (3), the process of constructing the phenolic resin product quality prediction model based on the long-term and short-term memory network includes establishing the long-term and short-term memory network composed of an LSTM layer and an output fully-connected layer, and the model parameters include network weights and offsets of the layers.
In the above scheme, the time t of the network (t e [1,2, …, m)]) The network cell state of (1) is the object, the input of the network is from the output h of the hidden layer at the time t-1t-1Time t-1 cell State Ct-1And input data x at time ttThe input data passes through a forgetting gate, an input gate and an output gate of the LSTM network to finally obtain the output h of the hidden layer at the time ttAnd t time cell status Ct. The specific operation steps of the door structure are as follows:
network input x at time ① ttAnd t-1 output h of the hidden layert-1Merging the input signals as the input of a forgetting gate, and finally outputting a result f of the forgetting gate by a sigmoid functiontNormalized to between 0 and 1, the specific formula is as follows:
ft＝σ(Wf*[ht-1,xt]+bf)
where σ denotes a Sigmoid activation function, WfWeight of forgetting gate, ht-1The output of the hidden layer is hidden for the previous moment,txas input at the current time, bfTo forget the biasing of the door.
② t time netThe channel input xtAnd the output h of the hidden layer at time t-1t-1The merged vector is subjected to sigmoid function to obtain a value i of the network cell needing to be updatedtMeanwhile, the merged vector also needs to be subjected to tanh function to obtain the candidate cell state
Specifically, the following formula:
it＝σ(Wi*[ht-1,xt]+bi)
Ct＝tanh(Wc*[ht-1,xt]+bc)
wherein, WiAs the weight of the input gate, biFor the bias of the input gate, tanh is the activation function, WcAs a weight of the candidate cell state, bcIs the bias of the candidate cell state.
③ use t-1 memory cell Ct-1And the output result f of the forgetting gatetMultiplication, which represents redundant information at a time before discarding; (ii) the candidate cell status
And output result i of input gatetMultiplication, which represents how much information in the candidate state cells is needed to update the memory cells; and adding the results of the two to complete the updating of the cells, wherein the specific formula is as follows:
Ct＝ft*Ct-1+it*Ct
④ LSTM network output value htIs the state C of the cell at time ttCurrent time input xtAnd the output h of the previous momentt-1Are jointly decided. Input x at time ttOutput h from time t-1t-1The merged vector of (1) is passed through a sigmoid function to indicate which information in the memory cells at the current moment needs to be output. State C of the cells at time ttOutput o of the tanh function and the sigmoid functiontMultiplying to obtain the output result h of the output gatetI.e. the output of the hidden layer at time t. Specifically, the following formula:
ot＝σ(Wo*[ht-1,xt]+bo)
ht＝ot*tanh(Ct)
wherein, WoAs weights of output gates, boIs the biasing of the output gate.
The weight and bias of each time network in the LSTM network are shared, and the output h of the hidden layer of the m time network is finally obtained through the transmission of m time datam. The output of the LSTM at the m moment is subjected to full connection layer to finally obtain the output value y of the prediction networkoutSpecifically, the following formula:
yout＝Wout×hm+bout
furthermore, in the method for predicting the quality of the phenolic resin product based on the long-time memory network, in the step (4), when the phenolic resin product quality prediction model is trained based on the production process and historical data of the quality of the corresponding batch of products, the data obtained in the step is sorted by adopting a sliding window, and the phenolic resin product quality prediction model is trained by adopting a time-based back propagation algorithm.
In the above scheme, the step (2) of sorting the data obtained by using the sliding window includes using m batches of data from the starting batch as the 1 st sample of the model, where the product quality corresponding to the sample is the product quality index of the mth batch, then using m batches of data from the 2 nd batch as the 2 nd sample of the model, where the product quality corresponding to the sample is the product quality index of the m +1 th batch, and so on. When the total batch number is I, the number of input samples of the model is I-m + 1. And keeping the front and back sequence of the samples, taking the first 80% of samples as a training sample set of the model, and taking the second 20% of samples as a testing sample set of the model for model training and testing.
Furthermore, in the method for predicting the quality of the phenolic resin product based on the long-time and short-time memory network, in the step (5), the data obtained after the real-time reading of the phenolic resin production process data is processed according to the step (2) is used as input, and the product quality of the phenolic resin of the current batch is predicted by using the phenolic resin product quality prediction model obtained in the step (4).
Furthermore, in the method for predicting the quality of the phenolic resin product based on the long-time and short-time memory network, in the step (6), the fact that the accuracy of the real-time prediction is not satisfactory means that the prediction of the quality of the product of the current batch is smaller than the factory-specified range, and the temperature value including but not limited to the softening point is outside the ± 2 ℃ interval of the final assay analysis value.
Drawings
FIG. 1 is a flow chart of the phenolic resin product quality prediction method based on a long-term and short-term memory network.
FIG. 2 is a block diagram of a phenolic resin product quality prediction model according to the present invention.
FIG. 3 is a diagram of the predicted effect of the implementation of the present invention.
Detailed Description
The technical scheme of the invention is further explained by combining the drawings and the embodiment of the specification.
As shown in fig. 1, the embodiment of the present invention and its specific implementation are as follows:
and (1) acquiring historical data of the production process of the phenolic resin and the quality of products of corresponding batches.
On a distributed control system configured in the production process, process variable data in the production process of the phenolic resin is collected through an OPC interface, wherein the process variable data comprises the flow of alkylation liquid, the flow of liquid aldehyde, the flow of solid aldehyde, reaction temperature, reaction pressure and reaction kettle weight, the sampling time is set to be 1 minute, and the phenolic resin process data is sampled to obtain a process data matrix of each batch. Collecting the latest 100 batches of phenolic resin production data, collecting the product softening point of the corresponding batch of phenolic resin analyzed by a laboratory as a quality index,
step (2) preprocessing, feature selection and standardization of the historical data obtained in the step (1)
In the pretreatment process of the step (2), 100 batches of normal phenolic resin production data collected in the step (1) are resampled at equal time intervals (the time intervals are set to be 10 minutes), and 28-dimensional data are obtained.
And (3) for 100 batches of data obtained by preprocessing in the step (2), according to a calculation formula of a maximum information coefficient:
wherein x represents any of the variables in the phenolic production process history data and y represents the softening point of the phenolic resin product. And (4) calculating to obtain the maximum information correlation coefficient of the 28-dimensional data, and selecting variables corresponding to the first 4 maximum values as independent variables of the model, namely the mass of the alkylation liquid, the mass of the liquid aldehyde, the temperature of the liquid aldehyde during addition and the temperature of the solid aldehyde after addition.
Normalizing the data after the feature selection in the step (2), wherein the normalization formula is as follows:
in formula (II), x'ikA k-dimensional variable x representing the ith lotikNormalized values in the range of [0,1 ]]The method has the advantages of no dimension,
and
respectively representing the minimum value and the maximum value in the k-dimension variable.
The softening point data, which is an index of the quality of phenolic resin products, is relatively concentrated and therefore is not standardized.
Step (3) constructing a phenolic resin product quality prediction model based on a long-time and short-time memory network;
the phenolic resin product quality prediction model is composed of 2 LSTM layers and 1 full-connection layer, wherein the dimension of input data is 4, the sequence length is 4, the number of neurons of the LSTM layers is 50, and the number of neurons of the full-connection layer is 1. The batch times of model training are set to be 30, the model training algebra is set to be 200, the loss function of the model is set to be the root mean square error, and the optimizer is adam.
Step (4) training a phenolic resin product quality prediction model based on the data obtained in the step (2);
and (4) processing the data obtained in the step (2) by adopting a sliding window method. And 4, setting a sliding window to be 4, starting from the initial batch of the phenolic resin production data normalized in the step 3, extracting 4 continuous batches of data as a first sample of the model, wherein a predicted value corresponding to the sample is the resin softening point of the 4 th batch of data, and the like. When the total number of batches is 100, the number of samples of the LSTM model is 97. The batch property of the sample is kept, and the first 80% of data is selected as training data, and the second 20% of data is selected as testing data.
And (3) taking the obtained data as input and output data, and training the network constructed in the step (3) by adopting adam as an optimizer to obtain a prediction model of the final quality index of the phenolic resin based on the LSTM.
And (5) acquiring real-time production data of the current batch, processing the data by using the pretreatment and standardization in the step (2), and predicting the product quality of the current batch by using the phenolic resin product quality prediction model.
And (3) reading the production process data of the phenolic resin in real time, calculating to obtain a standardized value of the process variable in the step (2), reading the combination of the latest previous 3 batches of historical data and the batch of data, taking the combination as the input of a model for predicting the softening point of the phenolic resin of the current batch, and predicting the softening point of the phenolic resin of the current batch of data by using the model trained in the step (4). The predicted effect is shown in fig. 3.
The above examples are only intended to illustrate the method and system of the present invention, but the present invention is not limited to the examples, and any simple modifications, equivalent changes and modifications made to the above examples according to the technical implementation of the present invention are within the scope of the present invention.
Claims (6)
Translated fromChinese
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911140036.9ACN111103420B (en) | 2019-11-20 | 2019-11-20 | A method for predicting the quality of phenolic resin products under uncertainty of raw materials |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911140036.9ACN111103420B (en) | 2019-11-20 | 2019-11-20 | A method for predicting the quality of phenolic resin products under uncertainty of raw materials |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111103420Atrue CN111103420A (en) | 2020-05-05 |
| CN111103420B CN111103420B (en) | 2022-05-10 |
Family
ID=70421604
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911140036.9AActiveCN111103420B (en) | 2019-11-20 | 2019-11-20 | A method for predicting the quality of phenolic resin products under uncertainty of raw materials |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111103420B (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111695743A (en)* | 2020-06-23 | 2020-09-22 | 信阳农林学院 | Industrial food safety production monitoring method, device, equipment and storage medium |
| CN112485260A (en)* | 2020-11-26 | 2021-03-12 | 常州微亿智造科技有限公司 | Workpiece defect detection method and device |
| CN114077950A (en)* | 2020-08-17 | 2022-02-22 | 株式会社日立制作所 | Quality correlation analysis method, quality prediction method, and quality control method |
| CN114139813A (en)* | 2021-12-03 | 2022-03-04 | 北京航空航天大学 | A product quality prediction method for process-type production equipment based on a self-updating model |
| JP2022167027A (en)* | 2021-04-22 | 2022-11-04 | 昭和電工マテリアルズ株式会社 | Model generation device, model generation method, model generation program, quality prediction model, and product quality prediction device |
| CN115438897A (en)* | 2022-05-18 | 2022-12-06 | 沈阳化工大学 | Industrial process product quality prediction method based on BLSTM neural network |
| CN115700597A (en)* | 2021-07-23 | 2023-02-07 | 北京航空航天大学 | A Method for Predicting Product Quality of Manufacturing Line Based on Federated Learning |
| CN116307528A (en)* | 2023-02-21 | 2023-06-23 | 云南中烟工业有限责任公司 | A Time Series Prediction Method of Process Quality Based on Graph Neural Network |
| CN119514808A (en)* | 2025-01-15 | 2025-02-25 | 陕西美力源乳业集团有限公司 | A method and system for detecting and analyzing the production quality of goat milk powder |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB1491730A (en)* | 1973-12-28 | 1977-11-16 | Charbonnages Ste Chimique | Production of phenol-formaldehyde resins |
| CN101598737A (en)* | 2009-06-26 | 2009-12-09 | 华东理工大学 | Soft measurement method for p-carboxybenzaldehyde content in purified terephthalic acid |
| CN103157772A (en)* | 2011-12-14 | 2013-06-19 | 鞍钢股份有限公司 | Forecasting method for internal quality of continuous casting billet |
| CN105956330A (en)* | 2016-05-31 | 2016-09-21 | 山东中烟工业有限责任公司 | Method and system for establishing product process quality prediction model based on site data |
| US20170293542A1 (en)* | 2016-04-06 | 2017-10-12 | Nec Laboratories America, Inc. | System failure prediction using long short-term memory neural networks |
| CN108764517A (en)* | 2018-04-08 | 2018-11-06 | 中南大学 | A kind of blast furnace molten iron silicon content trend method, equipment and storage medium |
| CN109523086A (en)* | 2018-11-26 | 2019-03-26 | 浙江蓝卓工业互联网信息技术有限公司 | The qualitative forecasting method and system of chemical products based on random forest |
| CN109711714A (en)* | 2018-12-24 | 2019-05-03 | 浙江大学 | A Quality Prediction Method for Manufacturing Assembly Products Based on Parallel Long Short-Term Memory Networks |
| CN110070217A (en)* | 2019-04-11 | 2019-07-30 | 武汉科技大学 | A kind of Forcasting Sinter Quality method of Kernel-based methods parameter |
| CN110400007A (en)* | 2019-07-05 | 2019-11-01 | 浙江大学 | Prediction method of molten iron quality based on improved gated recurrent neural network |
- 2019
- 2019-11-20CNCN201911140036.9Apatent/CN111103420B/enactiveActive
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB1491730A (en)* | 1973-12-28 | 1977-11-16 | Charbonnages Ste Chimique | Production of phenol-formaldehyde resins |
| CN101598737A (en)* | 2009-06-26 | 2009-12-09 | 华东理工大学 | Soft measurement method for p-carboxybenzaldehyde content in purified terephthalic acid |
| CN103157772A (en)* | 2011-12-14 | 2013-06-19 | 鞍钢股份有限公司 | Forecasting method for internal quality of continuous casting billet |
| US20170293542A1 (en)* | 2016-04-06 | 2017-10-12 | Nec Laboratories America, Inc. | System failure prediction using long short-term memory neural networks |
| CN105956330A (en)* | 2016-05-31 | 2016-09-21 | 山东中烟工业有限责任公司 | Method and system for establishing product process quality prediction model based on site data |
| CN108764517A (en)* | 2018-04-08 | 2018-11-06 | 中南大学 | A kind of blast furnace molten iron silicon content trend method, equipment and storage medium |
| CN109523086A (en)* | 2018-11-26 | 2019-03-26 | 浙江蓝卓工业互联网信息技术有限公司 | The qualitative forecasting method and system of chemical products based on random forest |
| CN109711714A (en)* | 2018-12-24 | 2019-05-03 | 浙江大学 | A Quality Prediction Method for Manufacturing Assembly Products Based on Parallel Long Short-Term Memory Networks |
| CN110070217A (en)* | 2019-04-11 | 2019-07-30 | 武汉科技大学 | A kind of Forcasting Sinter Quality method of Kernel-based methods parameter |
| CN110400007A (en)* | 2019-07-05 | 2019-11-01 | 浙江大学 | Prediction method of molten iron quality based on improved gated recurrent neural network |
Non-Patent Citations (4)
| Title |
|---|
| GUIXIANG XUE ETAL: "District Heating Load Prediction Algorithm Based on Feature Fusion LSTM Model", 《ENERGIES》* |
| 吕文洋: "基于神经网络的酚醛树脂纯度软测量方法研究", 《中国优秀硕士学位论文全文数据库》* |
| 李泽龙: "基于ARMAX-LSTM模型的高炉铁水质量预报研究", 《中国优秀硕士学位论文全文数据库》* |
| 罗娜等: "采用LSTM模型的Android应用行为一致性检测", 《浙江大学学报（工学版）》* |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111695743A (en)* | 2020-06-23 | 2020-09-22 | 信阳农林学院 | Industrial food safety production monitoring method, device, equipment and storage medium |
| CN111695743B (en)* | 2020-06-23 | 2024-04-05 | 信阳农林学院 | Industrial food safety production monitoring method, device, equipment and storage medium |
| CN114077950A (en)* | 2020-08-17 | 2022-02-22 | 株式会社日立制作所 | Quality correlation analysis method, quality prediction method, and quality control method |
| CN112485260B (en)* | 2020-11-26 | 2023-01-03 | 常州微亿智造科技有限公司 | Workpiece defect detection method and device |
| CN112485260A (en)* | 2020-11-26 | 2021-03-12 | 常州微亿智造科技有限公司 | Workpiece defect detection method and device |
| JP2022167027A (en)* | 2021-04-22 | 2022-11-04 | 昭和電工マテリアルズ株式会社 | Model generation device, model generation method, model generation program, quality prediction model, and product quality prediction device |
| JP7647288B2 (en) | 2021-04-22 | 2025-03-18 | 株式会社レゾナック | MODEL GENERATION DEVICE, MODEL GENERATION METHOD, MODEL GENERATION PROGRAM, QUALITY PREDICTION MODEL, AND PRODUCT QUALITY PREDICTION DEVICE |
| CN115700597A (en)* | 2021-07-23 | 2023-02-07 | 北京航空航天大学 | A Method for Predicting Product Quality of Manufacturing Line Based on Federated Learning |
| CN115700597B (en)* | 2021-07-23 | 2025-09-16 | 北京航空航天大学 | Product quality prediction method of manufacturing production line based on federal learning |
| CN114139813A (en)* | 2021-12-03 | 2022-03-04 | 北京航空航天大学 | A product quality prediction method for process-type production equipment based on a self-updating model |
| CN115438897A (en)* | 2022-05-18 | 2022-12-06 | 沈阳化工大学 | Industrial process product quality prediction method based on BLSTM neural network |
| CN116307528A (en)* | 2023-02-21 | 2023-06-23 | 云南中烟工业有限责任公司 | A Time Series Prediction Method of Process Quality Based on Graph Neural Network |
| CN119514808A (en)* | 2025-01-15 | 2025-02-25 | 陕西美力源乳业集团有限公司 | A method and system for detecting and analyzing the production quality of goat milk powder |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111103420B (en) | 2022-05-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111103420A (en) | A method for predicting the quality of phenolic resin products under uncertainty of raw materials | |
| CN110794093B (en) | Precision compensation method for discharged caustic alkali concentration measuring device in evaporation process | |
| CN110705692B (en) | A non-linear dynamic industrial process product prediction method based on spatiotemporal attention network | |
| CN112966891B (en) | A method for predicting river water environment quality | |
| CN110824914B (en) | An intelligent monitoring method for wastewater treatment based on PCA-LSTM network | |
| CN108920888B (en) | A method for identification of continuous stirred tank reaction process based on deep neural network | |
| CN111191726B (en) | A Fault Classification Method Based on Weakly Supervised Learning Multilayer Perceptron | |
| CN115963788B (en) | Online prediction method for key quality indicators of industrial processes with multiple sampling rates | |
| CN114239397A (en) | Soft sensor modeling method based on dynamic feature extraction and local weighted deep learning | |
| CN113569993A (en) | Method for constructing quality prediction model in polymerization reaction process | |
| CN108595892A (en) | Soft-measuring modeling method based on time difference model | |
| Liu et al. | Correntropy long short term memory soft sensor for quality prediction in industrial polyethylene process | |
| CN111079856B (en) | Multi-period intermittent process soft measurement modeling method based on CSJITL-RVM | |
| Graziani et al. | Design of a soft sensor for an industrial plant with unknown delay by using deep learning | |
| CN116930609A (en) | Electric energy metering error analysis method based on ResNet-LSTM model | |
| CN114239990A (en) | Time series data prediction method based on time series decomposition and LSTM | |
| CN103279030B (en) | Dynamic soft measuring modeling method and device based on Bayesian frame | |
| CN113609766A (en) | Soft measurement method based on depth probability latent model | |
| CN118031899A (en) | Multi-factor roadbed settlement prediction method based on deep learning | |
| CN110222825B (en) | Cement product specific surface area prediction method and system | |
| CN118839620B (en) | On-line soft measurement method and system for water quality parameters based on optimization algorithm | |
| CN107729988A (en) | Cyanobacterial bloom prediction method based on dynamic deep belief network | |
| CN103675005B (en) | The industrial melt index soft measurement instrument of optimum FUZZY NETWORK and method | |
| CN103675010B (en) | The industrial melt index soft measurement instrument of support vector machine and method | |
| CN119479884A (en) | Prediction method of sulfuric acid concentration at the outlet of electronic grade sulfuric acid absorption tower |
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 |