CN112464554A - Operating parameter optimization method of gasoline refining equipment - Google Patents

Operating parameter optimization method of gasoline refining equipment Download PDF

Info

Publication number
CN112464554A
CN112464554A CN202011212157.2A CN202011212157A CN112464554A CN 112464554 A CN112464554 A CN 112464554A CN 202011212157 A CN202011212157 A CN 202011212157A CN 112464554 A CN112464554 A CN 112464554A
Authority
CN
China
Prior art keywords
data
variables
parameters
values
octane number
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
Application number
CN202011212157.2A
Other languages
Chinese (zh)
Inventor
谢晓兰
翟青海
梁荣华
郑伟程
刘亚荣
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guilin University of Technology
Original Assignee
Guilin University of Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Guilin University of Technology filed Critical Guilin University of Technology
Priority to CN202011212157.2A priority Critical patent/CN112464554A/en
Publication of CN112464554A publication Critical patent/CN112464554A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • 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
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/12—Computing arrangements based on biological models using genetic models
    • G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming

Abstract

The invention discloses an operation parameter optimization method of gasoline refining equipment. Firstly, collecting a plurality of groups of original data and processing the original data, obtaining main characteristic vectors by using distance correlation coefficients, then establishing a prediction model of the final product octane number and sulfur content by using the obtained characteristic vectors through a BP neural network, finally obtaining the optimal operation parameter combination of the main characteristic vectors by using a genetic algorithm according to the established prediction model, and simultaneously adding the actually generated optimization results into a sample library for learning. The operating parameter optimization method of the gasoline refining equipment can accurately predict the final octane number and the sulfur content of a product, can obtain the optimized set value of the main operating variable, effectively reduces the octane number loss and realizes the intelligent optimization process of the gasoline refining process.

Description

Operating parameter optimization method of gasoline refining equipment
Technical Field
The invention relates to the technical field of industrial automatic control, in particular to an operation parameter optimization method of gasoline refining equipment based on a neural network and a genetic algorithm.
Background
At present, more than 70% of gasoline in China is produced by catalytic cracking, so that more than 95% of sulfur and olefin in finished gasoline come from catalytic cracking gasoline. Among them, olefin is an important chemical component that affects the octane number of gasoline (the higher the value is, the better the antiknock combustion ability of gasoline is), and sulfur causes adverse effects on automobile components and pollutes the environment. Therefore, it is necessary to refine catalytically cracked gasoline to meet gasoline quality requirements and to reduce octane number loss and sulfur content during processing. The catalytic gasoline adsorption desulfurization S-Zorb device has the characteristics of low hydrogen consumption, high desulfurization rate and small octane value loss, and can meet the increasingly-improved domestic environmental protection requirements. Therefore, the S-Zorb unit is one of the major domestic catalytic gasoline refining facilities.
However, the operating parameters of the S-Zorb plant include hydrogen-to-oil ratio, reaction filter differential pressure, reducer pressure, and the like, which exceed three hundred operating variables. Therefore, it is necessary to adjust the various operating parameters of the optimization device in order to obtain a high quality gasoline. The traditional optimization method generally determines the correlation between parameters and dependent variables for a data correlation method or adjusts partial parameters by a chemical mechanism method according to relevant important chemical reaction strength or chemical parameters, however, the required variables in a data correlation model are relatively few, the chemical mechanism has higher analysis requirements on raw materials and has not timely response to process optimization, and therefore, the two optimization methods have not ideal effects.
The method comprises the steps of collecting original data, cleaning the original data, processing abnormal values by adopting a maximum and minimum amplitude limiting method, a Lauda criterion and an averaging method, calculating correlation distance coefficients among all variables to generate a correlation distance coefficient matrix, further obtaining correlation scores, selecting main operation variables according to the scores, then establishing a prediction model of octane number and sulfur content through a neural network model, and finally obtaining the optimal operation parameter combination of the main operation variables by adopting a genetic algorithm.
Disclosure of Invention
The invention aims to provide an operation parameter optimization method of gasoline refining equipment, which is used for predicting the octane number and the sulfur content of gasoline under different operation variables; by optimizing the main operating variable parameters, the octane number loss in the gasoline refining process is reduced and the desulfurization effect is ensured.
In order to achieve the above purpose, the technical scheme adopted by the invention comprises the following steps:
s1, acquiring original data: recording the properties of raw materials, products and adsorbents and collecting data from refining equipment, wherein the data comprise adjustable operation variables such as hydrogen-oil ratio, pressure difference of a reaction filter, pressure of a reducer and the like on the equipment;
s2, data processing: processing the data obtained in step S1 to obtain normal data;
s3, feature selection: calculating the distance correlation coefficients of all the operation variables, then summing the distance correlation coefficients of the operation variables to obtain score coefficients, and sequencing according to the score coefficients to obtain 25 main operation variables;
s4, training a neural network model: sending the processed data sample into a preset neural network model for training;
s5, optimizing operation parameters: taking known better operation parameters as initialization variables, substituting the initialization variables into a genetic algorithm, and substituting the parameters into a prediction model to iterate to obtain optimal operation parameter variables;
step S6, relearning: and inputting the optimized main operation parameters into refining equipment to obtain the actual octane number and the sulfur content of the product. And adding the obtained parameters and the actual octane number and sulfur content of the product into the sample library again, so that the neural network model is more and more stable, and the prediction accuracy is higher and higher.
The step S2 data processing includes maximum minimum clipping, lazada criterion removal of outliers and mean replacement of outliers:
step S21, the maximum and minimum value amplitude limiting means that whether the actual measured value is abnormal is determined according to the theoretical value range of the operation variable, and if the actual measured value is abnormal, the actual measured value is removed;
in step S22, the Lauda criterion removing abnormal values means that a group of data collected is assumed to be (x)1,x2,...,xn) Calculating the arithmetic mean thereof as
Figure BDA0002759140920000021
And residual error
Figure BDA0002759140920000022
Then, the standard deviation sigma is calculated by using the formula (1), if the residual error | viI satisfies: | viIf | is greater than 3 σ, the data x is considerediIf the abnormal value is abnormal, the abnormal value needs to be removed;
Figure BDA0002759140920000023
step S23, the average method is that after removing the abnormal value, the average value of the characteristic of the abnormal value is calculated, and the abnormal value is replaced by the average value;
the step S3 feature selection includes the following steps:
s31, normalization processing is carried out on the data according to the formula (2):
Figure BDA0002759140920000024
wherein x is a value to be normalized, xmaxIs the largest value, x, in the collected data featuresminIs the smallest value among the collected data features;
s32, calculating distance correlation coefficients, wherein if X and Y are two groups of collected moduli which are n eigenvectors, the distance correlation coefficients of X and Y are calculated as follows:
first, the distances of all element pairs are calculated, the formula is as follows:
Figure BDA0002759140920000025
wherein (a)j,k) And (b)j,k) Respectively, the distance matrix between the respective elements of X and Y.
Then, a center distance matrix is calculated, and the calculation method is as follows:
Figure BDA0002759140920000031
wherein the content of the first and second substances,
Figure BDA0002759140920000032
and
Figure BDA0002759140920000033
respectively represent the average values of the j-th lines of a and b,
Figure BDA0002759140920000034
and
Figure BDA0002759140920000035
respectively represent the average values of the k-th columns of a and b,
Figure BDA0002759140920000036
and
Figure BDA0002759140920000037
respectively, the overall average of a, b.
Then, the distance variances of X and X, X and Y, and Y are respectively calculated, and the formula is as follows:
Figure BDA0002759140920000038
Figure BDA0002759140920000039
Figure BDA00027591409200000310
finally, the distance correlation coefficient is calculated, and the formula is as follows:
Figure BDA00027591409200000311
s33, after distance correlation coefficients among the features are calculated to generate a distance correlation coefficient matrix, summing the distance correlation coefficient matrix according to columns to obtain scoring coefficients, sorting according to the size of the scoring coefficients, and selecting 25 operation variables with highest scoring as main feature values;
the step S4 is that the neural network model adopts a BP neural network model, the number of layers of hidden layers is 9, data is divided into a training set and a test set according to the proportion of 8:2, the number of iterations is 500, and the learning rate is 0.01;
the step S5 parameter optimization refers to optimization of the operation variables by genetic algorithm. Firstly, taking initial operation parameters of a sample as a population initialization value, then coding and calculating population fitness, and then selecting individuals in a wheel disc mode and carrying out crossing and variation operation. And inputting the obtained operating variables into a neural network model to obtain the product sulfur content and octane number loss predicted value. And judging whether the sulfur content and octane number loss reach threshold values. If the threshold value is reached, the obtained operation variables are coded, and a better individual is continuously selected to obtain the operation variables meeting the requirements, otherwise, the operation variables are used as optimization parameters for subsequent actual adjustment operation.
The invention has the following beneficial effects and advantages:
(1) after the octane number and sulfur content prediction model is established, the optimal operation parameters can be quickly generated for the raw materials with different chemical properties through the parameter optimization prediction model based on the genetic algorithm, and compared with the traditional parameter optimization method, the method has the characteristics of timeliness and comprehensiveness;
(2) the octane number loss can be reduced by adopting the operation parameters optimized by the algorithm, so that the economic loss is reduced.
Drawings
FIG. 1 is a schematic view of an operation parameter optimization method of a gasoline refining apparatus according to the present invention.
FIG. 2 is a diagram of a BP neural network-based prediction model structure according to the present invention.
FIG. 3 is a flow chart of the present invention for optimizing operation parameters based on genetic algorithm.
Detailed Description
Example (b):
as shown in fig. 1, the technical solution of the present invention comprises six steps: the method comprises the steps of raw data acquisition, data processing, feature selection, neural network model training, operation parameter optimization and relearning.
The step S1 raw data acquisition: and (3) determining the property of the raw material and recording the operating parameters every three minutes, wherein the octane number and the sulfur content of the product are determined once every two hours due to the relatively complex determination of the octane number of the product, and the average value of the operating parameters within two hours is calculated, so that the octane number and the sulfur content of the product are considered as the influence results of the average value of the operating parameters. Uploading the data to a database after the sample data is recorded;
processing the data in the step S2, namely removing abnormal values of the acquired data according to a maximum and minimum amplitude limiting method and a Lauder criterion;
s3, selecting characteristics, namely normalizing the processed data and selecting 25 operation variables with strong independence according to the distance correlation coefficient;
the step S4 neural network model training: sending the data samples of the main operation variables subjected to data processing raw material properties, adsorbent property data and dimensionality reduction into a preset neural network model for training;
the step S5 optimizing operation parameters: and taking the existing main operating parameters for use as an initialization population, generating new operating parameters by using a genetic algorithm, substituting the new operating parameters into the prediction model, using the group of operating parameters as the use parameters if the octane number loss value and the sulfur content reach the standard, and continuing iteration if the octane number loss value and the sulfur content do not reach the standard.
The step S6 relearns: gradually adjusting the machine equipment parameters to the set parameters generated in the step S5, recording the actual octane number and the sulfur content, and adding the group of data as a new sample into the sample library.
The working process of the operating parameter optimization method of the gasoline refining equipment comprises the following steps:
firstly, recording the properties of raw materials and the variable values of operating parameters, and performing data processing;
because the operation variables are more and the variables have nonlinear and strong coupling relation, distance correlation coefficients are calculated pairwise by adopting the distance correlation coefficients, and thus a distance correlation coefficient matrix is generated. Summing the distance correlation coefficient matrixes according to columns to obtain the distance correlation coefficient score of each operation variable, and obtaining 25 main operation variables with weak correlation according to the score;
establishing a loss prediction model about octane number and sulfur content by adopting a BP neural network for raw material properties, adsorbent property data and selected main operation variables, wherein the proportion of a training set to a testing set is 8: 2;
and taking the existing superior main operating variables as initialization variables, obtaining new operating variables by adopting a genetic algorithm, substituting the new operating variables into a prediction model to judge whether the operating variables meet requirements, if so, operating the operating variables as actual operating variables, taking the obtained actual product octane number and sulfur content as new sample data for the neural network to relearn, and if not, continuing to iteratively generate new operating variables.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims (1)

1. A method for optimizing operating parameters of gasoline refining equipment is characterized by comprising the following steps:
s1, acquiring original data: recording the properties of raw materials, products and adsorbents and collecting a plurality of data from refining equipment, wherein the data comprise settable operation variables of hydrogen-oil ratio, differential pressure of a reaction filter and pressure of a reducer on the equipment;
s2, data processing: removing abnormal values from the data obtained in the step one to obtain normal data;
s3, feature selection: calculating distance correlation coefficients of all the operation variables, summing the operation variables to obtain scoring coefficients, and sequencing according to the scoring coefficients to obtain 25 main operation variables;
s4, training a neural network model: sending the processed data sample into a preset neural network model for training;
s5, parameter optimization: taking known better operation parameters as initialization variables, substituting the initialization variables into a genetic algorithm, and substituting the parameters into a prediction model to iterate to obtain optimal operation parameter variables;
s6, relearning, namely inputting the optimized main operating parameters into refining equipment to obtain the actual octane number and the sulfur content of the product; adding the obtained parameters and the actual octane number and sulfur content of the product into a sample library, so that a neural network model is more and more stable, and the prediction accuracy is more and more high;
the step S2 data processing comprises maximum and minimum value amplitude limiting and Lauda criterion removing abnormal values;
step S21, the maximum and minimum value amplitude limiting means that whether the actual measured value is abnormal is determined according to the theoretical value range of the operation variable, and if the actual measured value is abnormal, the actual measured value is removed;
in step S22, the Lauda criterion removing abnormal values means that a group of data collected is assumed to be (x)1,x2,...,xn) Calculating the arithmetic mean thereof as
Figure FDA0002759140910000014
And residual error
Figure FDA0002759140910000011
Then, the standard deviation sigma is calculated by using the formula (1), if the residual error | viI satisfies: | viIf | is greater than 3 σ, the data x is considerediIf the abnormal value is abnormal, the abnormal value needs to be removed;
Figure FDA0002759140910000012
the step S3 feature selection includes the following steps:
s31, normalization processing is carried out on the data according to the formula (2):
Figure FDA0002759140910000013
wherein x is a value to be normalized, xmaxIs the largest value, x, in the collected data featuresminIs the smallest value among the collected data features;
s32, calculating distance correlation coefficients, wherein if X and Y are two groups of collected moduli which are n eigenvectors, the distance correlation coefficients of X and Y are calculated as follows:
first, the distances of all element pairs are calculated, the formula is as follows:
Figure FDA0002759140910000021
(aj,k) And (b)j,k) Respectively are distance matrixes between respective elements of X and Y;
then, a center distance matrix is calculated, and the calculation method is as follows:
Figure FDA0002759140910000022
wherein the content of the first and second substances,
Figure FDA0002759140910000023
and
Figure FDA0002759140910000024
respectively represent the average values of the j-th lines of a and b,
Figure FDA0002759140910000025
and
Figure FDA0002759140910000026
respectively represent the average values of the k-th columns of a and b,
Figure FDA0002759140910000027
and
Figure FDA0002759140910000028
respectively representing the total mean values of a and b;
then, the distance variances of X and X, X and Y, and Y are respectively calculated, and the formula is as follows:
Figure FDA0002759140910000029
Figure FDA00027591409100000210
Figure FDA00027591409100000211
finally, the distance correlation coefficient is calculated, and the formula is as follows:
Figure FDA00027591409100000212
s33, after distance correlation coefficients among the features are calculated to generate a distance correlation coefficient matrix, summing the distance correlation coefficient matrix according to columns to obtain scoring coefficients, sorting according to the size of the scoring coefficients, and selecting 25 operation variables with highest scoring as main feature values;
the step S4 is that the neural network model adopts a BP neural network model, the number of layers of hidden layers is 9, data is divided into a training set and a test set according to the proportion of 8:2, the number of iterations is 500, and the learning rate is 0.01;
and step S5, optimizing parameters and selecting operation variables of a genetic algorithm, firstly taking initial operation parameters of a sample as population initialization values, then coding and calculating population fitness, then selecting individuals in a wheel disc mode and carrying out crossover and variation operations, inputting the obtained operation variables into a neural network model to obtain predicted values of sulfur content and octane number loss of products, judging whether the sulfur content and the octane number loss reach threshold values, if so, coding the obtained operation variables and continuously selecting more optimal individuals to obtain the operation variables meeting the requirements, otherwise, taking the operation variables as optimization parameters for subsequent actual adjustment operation.
CN202011212157.2A 2020-11-03 2020-11-03 Operating parameter optimization method of gasoline refining equipment Pending CN112464554A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011212157.2A CN112464554A (en) 2020-11-03 2020-11-03 Operating parameter optimization method of gasoline refining equipment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011212157.2A CN112464554A (en) 2020-11-03 2020-11-03 Operating parameter optimization method of gasoline refining equipment

Publications (1)

Publication Number Publication Date
CN112464554A true CN112464554A (en) 2021-03-09

Family

ID=74834999

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011212157.2A Pending CN112464554A (en) 2020-11-03 2020-11-03 Operating parameter optimization method of gasoline refining equipment

Country Status (1)

Country Link
CN (1) CN112464554A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113191496A (en) * 2021-05-17 2021-07-30 上海应用技术大学 Method for predicting and optimizing octane number loss of gasoline
CN114021463A (en) * 2021-11-09 2022-02-08 南通大学 Path optimization model in gasoline refining process

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104392098A (en) * 2014-10-27 2015-03-04 中国石油大学(北京) Method for predicting yield of catalytically cracked gasoline
CN107358021A (en) * 2017-06-01 2017-11-17 华南理工大学 DO prediction model establishment method based on BP neural network optimization
CN109034388A (en) * 2018-07-27 2018-12-18 湖北中烟工业有限责任公司 A kind of prediction model of cigarette material and mainstream smoke constituents based on Genetic Algorithm Optimized Neural Network
CN110046743A (en) * 2019-03-06 2019-07-23 上海交通大学 Energy Consumption of Public Buildings prediction technique and system based on GA-ANN
CN111027733A (en) * 2018-10-10 2020-04-17 中国石油化工股份有限公司 Petrochemical device product yield optimization method based on big data technology
CN111324989A (en) * 2020-03-19 2020-06-23 重庆大学 GA-BP neural network-based gear contact fatigue life prediction method

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104392098A (en) * 2014-10-27 2015-03-04 中国石油大学(北京) Method for predicting yield of catalytically cracked gasoline
CN107358021A (en) * 2017-06-01 2017-11-17 华南理工大学 DO prediction model establishment method based on BP neural network optimization
CN109034388A (en) * 2018-07-27 2018-12-18 湖北中烟工业有限责任公司 A kind of prediction model of cigarette material and mainstream smoke constituents based on Genetic Algorithm Optimized Neural Network
CN111027733A (en) * 2018-10-10 2020-04-17 中国石油化工股份有限公司 Petrochemical device product yield optimization method based on big data technology
CN110046743A (en) * 2019-03-06 2019-07-23 上海交通大学 Energy Consumption of Public Buildings prediction technique and system based on GA-ANN
CN111324989A (en) * 2020-03-19 2020-06-23 重庆大学 GA-BP neural network-based gear contact fatigue life prediction method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
张忠洋,李泽钦等: "GA辅助BP神经网络预测催化裂化装置汽油产率", 《石油炼制与化工》 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113191496A (en) * 2021-05-17 2021-07-30 上海应用技术大学 Method for predicting and optimizing octane number loss of gasoline
CN113191496B (en) * 2021-05-17 2024-02-02 上海应用技术大学 Gasoline octane number loss prediction and optimization method
CN114021463A (en) * 2021-11-09 2022-02-08 南通大学 Path optimization model in gasoline refining process

Similar Documents

Publication Publication Date Title
CN109814513B (en) Catalytic cracking unit optimization method based on data model
CN112464554A (en) Operating parameter optimization method of gasoline refining equipment
CN112668105B (en) Helicopter transmission shaft abnormity judgment method based on SAE and Mahalanobis distance
CN108647373A (en) A kind of industrial process flexible measurement method based on xgboost models
CN112508243B (en) Training method and device for multi-fault prediction network model of power information system
CN111007040A (en) Near infrared spectrum rapid evaluation method for rice taste quality
CN109299501B (en) Vibration spectrum analysis model optimization method based on workflow
CN110568483A (en) Automatic evaluation method for seismic linear noise suppression effect based on convolutional neural network
CN110991495B (en) Method, system, medium, and apparatus for predicting product quality in manufacturing process
CN109657411B (en) Modeling and optimizing method of solvent deasphalting device based on data driving
CN115188429A (en) Catalytic cracking unit key index modeling method integrating time sequence feature extraction
CN111949640A (en) Intelligent parameter adjusting method and system based on industrial big data
CN114757413A (en) Bad data identification method based on time sequence series analysis coupling neural network prediction
CN116431966A (en) Reactor core temperature anomaly detection method of incremental characteristic decoupling self-encoder
CN106281431B (en) A kind of real-time predicting method and device of hydrocracking unit reaction depth
CN106647285B (en) A kind of catalyst activity detection method based on soft-measuring technique
CN114202060A (en) Method for predicting methylene blue adsorption performance of biomass activated carbon based on deep neural network
CN114372693A (en) Transformer fault diagnosis method based on cloud model and improved DS evidence theory
CN114326639A (en) Industrial process performance evaluation method based on mixed attention convolutional neural network
CN113793652A (en) Spectrogram chemometrics analysis method based on segmented intelligent optimization
CN110705373A (en) Tank voltage signal classification method based on wavelet packet analysis and residual error network
CN111797574A (en) Integrated Gaussian process regression model method for polymer molecular weight distribution
CN111242350A (en) Continuous reforming device fractionation and extraction optimization method, system and medium
CN111178627A (en) Neural network hybrid optimization prediction method based on SPCA
CN110389948A (en) A kind of tail oil prediction technique of the hydrocracking unit based on data-driven

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
WD01 Invention patent application deemed withdrawn after publication
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20210309