CN110750858A - 4-NP reduction catalyst modeling prediction method based on ECSA Gaussian process regression - Google Patents
4-NP reduction catalyst modeling prediction method based on ECSA Gaussian process regression Download PDFInfo
- Publication number
- CN110750858A CN110750858A CN201910851343.1A CN201910851343A CN110750858A CN 110750858 A CN110750858 A CN 110750858A CN 201910851343 A CN201910851343 A CN 201910851343A CN 110750858 A CN110750858 A CN 110750858A
- Authority
- CN
- China
- Prior art keywords
- crow
- data
- gaussian process
- process regression
- reduction catalyst
- 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
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Monitoring And Testing Of Nuclear Reactors (AREA)
- Exhaust Gas After Treatment (AREA)
Abstract
The invention discloses a modeling and predicting method for a p-nitrophenol reduction catalyst based on ECSA Gaussian process regression, which comprises the following steps: adding a contraction and surrounding mechanism and simulated annealing to the crow search algorithm for improvement to obtain an enhanced crow search algorithm; preprocessing original input data by normalization, PCA dimensionality reduction and mean shift clustering; establishing a p-nitrophenol reduction catalyst prediction model based on a Gaussian process regression model by utilizing the processed data; optimizing hyper-parameters of the prediction model by utilizing an enhanced crow search algorithm; and screening out the optimal catalyst according to the optimized prediction model. The invention provides a new enhanced crow search algorithm which is used for optimizing model prediction of a p-nitrophenol reduction catalyst of Gaussian process regression, and improving model prediction precision and catalyst screening efficiency.
Description
Technical Field
The invention relates to a 4-NP reduction catalyst modeling prediction method based on ECSA Gaussian process regression, and belongs to the field of modeling and optimization of complex industrial processes.
Background
P-nitrophenol (4-NP) is generally used as an intermediate of fine chemicals such as pesticides, medicines, dyes and the like, and is widely applied in chemical production. However, 4-NP is also an organic substance with high toxicity and difficult biodegradation, so once the 4-NP enters the environment, the 4-NP has great harm and is listed as 68 pollutants for preferentially controlling the water environment in China and one of 129 pollutants for mainly controlling the water environment in the EPA in the United states. Therefore, the development of a high-efficiency p-nitrophenol reduction catalyst is of great significance.
At present, the screening of the catalyst is only obtained through human experience and blind experiments, and the experimental process is complicated and low-efficiency.
Disclosure of Invention
The invention aims to provide a 4-NP reduction catalyst modeling prediction method based on Gaussian process regression of ECSA (enhanced crow search algorithm), which not only can provide a method for optimizing a Gaussian process regression prediction model, but also can find out a functional relation between elements and catalytic activity by using the prediction model, and accurately predict the elements with the best catalytic performance in the elements by using the prediction model.
The invention is realized by adopting the following technical scheme: A4-NP reduction catalyst modeling prediction method based on Gaussian process regression of ECSA comprises the following steps:
the method comprises the following steps: obtaining original data to form a data set of a p-nitrophenol reduction catalyst prediction model based on an enhanced crow search algorithm optimized Gaussian process regression. The original data comprises 16 physicochemical properties of oxidation enthalpy of formation, first ionization energy, second ionization energy, electric dipole polarizability, boiling point, melting point, heat of dissolution, heat of vaporization, thermal conductivity, density, ionic radius, atomic weight, electronegativity, specific heat capacity, covalent radius and ionic valence of 63 chemical elements, wherein the 63 chemical elements are all elements except gaseous, radioactive and toxic elements in the periodic table of the elements;
step two: processing the original data: the method comprises the steps of firstly carrying out normalization processing on original data required by an experiment, then carrying out KPCA (kernel principal component analysis) dimension reduction on the normalized data, and reducing the dimension of the 16-dimensional data to the dimension containing 80-85% of main components of the 16-dimensional data. And then carrying out mean shift clustering on the elements by utilizing the similarity of the data types, and dividing 63 elements into m types. Selecting s representative elements in the m classes to obtain processed input data for establishing a prediction model, and performing a chemical experiment by using the s elements to obtain a reaction activity index (namely reaction time) of a corresponding catalyst to obtain required output data;
step three: establishing a p-nitrophenol reduction catalyst prediction model based on Gaussian process regression by using input and output data;
step four: optimizing the hyper-parameters of a p-nitrophenol reduction catalyst prediction model based on Gaussian process regression by adopting a crow search algorithm to obtain the p-nitrophenol reduction catalyst prediction model based on enhanced crow search algorithm optimized Gaussian process regression;
step five: and (3) predicting input data of other elements in the 63 elements except the s elements by adopting a well-established p-nitrophenol reduction catalyst prediction model based on an enhanced crow search algorithm optimized Gaussian process regression, completing the prediction of the catalytic performance of all the elements, and screening to obtain the element with the best catalytic performance.
The 4-NP reduction catalyst modeling prediction method based on the ECSA Gaussian process regression specifically realizes that the Gaussian process regression model is established by using the processed input data and output data, and comprises the following steps: the s input data and the output data are divided into training data (X, Y) and test data (X)*,Y*) Establishing a function model according to training input data and corresponding output data: y ═ f (x), given a priori to the gaussian process (x) is: (X) -GP (mu (X), k (X, X')), wherein mu is a mean function, k is a kernel function, and the kernel function adopts an RBF kernel function as follows:wherein X' is the kernel function center, sigma is the width parameter of the function, the radial action range of the function is controlled, and because the training data and the test data belong to the same distribution, the combined Gaussian distribution of the training data and the test data is obtained:according to training input and output data, utilizing kernel function and combining joint gaussDistribution, obtaining the input and output relation of the regression prediction model in the Gaussian process, and finally utilizing the test input data X*Input model to obtain predicted valueThe hyper-parameter x based on the enhanced crow search algorithm optimized Gaussian process regression model comprises a width parameter sigma of a kernel function and a regularization parameter C, namely x is { sigma, C }.
The modeling and predicting method for the 4-NP reduction catalyst based on the ECSA Gaussian process regression is characterized in that the crow search algorithm is improved to obtain an enhanced crow search algorithm, and comprises the following steps: tracking flight phase and disaggregation updating process of the crow;
in the crow tracking flight stage, the crow tracking flight is split into two stages according to probability, and the crow flight strategy in the first stage is as follows: x is the number ofi,iter+1=xi,iter+ri×fli,iter×(mj,iter-xi,iter) In the formula xi,iter+1Is the value of the post-location update hyper-parameter, xi,iterIs the value of the over-parameter, r, for the current iteration positioniIs a uniformly distributed random number, fl, within 0 to 1i,iterIs the flying radius of crow, mj,it erIs a memory location of crow;
the shrink wrap strategy in the second stage is formulated as follows:in the formulagbest is the best individual in the current population,a is linearly decreasing from 2 to 0 and r is [0,1 ]]Uniformly distributed random numbers within;
the solution set update process probability formula is as follows:where f (x) is a fitness function,xwhen the delta f is less than or equal to 0, the updated super parameter value is accepted as the next initial super parameter value, and when the delta f is more than or equal to 0, the probability r of accepting the updated super parameter value is calculated to generate [0,1 ] for]If r is less than or equal to s, the updated super parameter value is accepted as the next initial super parameter value, otherwise, the original super parameter value is still taken as the next initial super parameter value,
the method for carrying out hyper-parameter optimization on the established Gaussian process regression prediction model by using the enhanced crow search algorithm specifically comprises the following steps:
initializing parameter setting of an enhanced crow searching algorithm, the number of crow populations and a memory position of the crow;
by a fitness functionEvaluating fitness value of each individual crow, wherein Y represents real output corresponding to test input data X,representing test input data X*An expected output based on a gaussian process regression model;
updating the positions of the particles by using a flight strategy formula of the crow, evaluating a fitness function value of the new position, updating the super parameter value of the current position of the crow according to the simulated annealing probability, and updating the memory position of the crow;
and repeating the previous step until the maximum iteration number is reached or the fitness value meets the requirement of error precision, and stopping the optimization process.
The invention adopts a 4-NP reduction catalyst modeling prediction method based on ECSA Gaussian process regression, and optimizes a 4-NP reduction catalyst prediction model modeled by the Gaussian process regression by using an enhanced crow search algorithm. And selecting a catalyst prediction model with higher precision and better effect. And screening out the high-efficiency catalyst by utilizing a scientific catalyst prediction model.
Compared with the prior art, the invention has the advantages that:
1. the improved enhanced crow search algorithm has high convergence precision, can jump out of the local optimal solution, and has high-efficiency optimization performance.
2. A p-nitrophenol reduction catalyst prediction model based on Gaussian process regression is established, an enhanced crow search algorithm is used for optimizing the prediction model, and the performance of the prediction model is improved on the premise of ensuring the prediction accuracy. Meanwhile, a scientific and effective method is provided in the aspect of screening the p-nitrophenol reduction catalyst.
Drawings
FIG. 1 is a flow chart of an enhanced crow search algorithm.
FIG. 2 is a flow chart of a 4-NP reduction catalyst modeling prediction method based on Gaussian process regression of ECSA.
Detailed Description
The invention aims to provide a 4-NP reduction catalyst modeling prediction method based on Gaussian process regression of ECSA, which not only can provide a method for optimizing a Gaussian process regression prediction model, but also can find out a functional relation between elements and catalytic activity by using the prediction model, and accurately predict the elements with the best catalytic performance in the elements by using the prediction model.
A4-NP reduction catalyst modeling prediction method based on Gaussian process regression of ECSA comprises the following steps:
the method comprises the following steps: obtaining original data to form a data set of a p-nitrophenol reduction catalyst prediction model based on an enhanced crow search algorithm optimized Gaussian process regression. At present, 112 elements are totally contained in the periodic table of elements, and 63 elements are used as candidate elements of the catalyst besides gaseous, radioactive and toxic elements. The raw data includes 16 physicochemical properties of oxidation formation enthalpy, first ionization energy, second ionization energy, electric dipole polarizability, boiling point, melting point, heat of dissolution, heat of vaporization, thermal conductivity, density, ionic radius, atomic weight, electronegativity, specific heat capacity, covalent radius, and ionic valence of 63 chemical elements.
Step two: processing the original data: the method comprises the steps of firstly carrying out normalization processing on original data required by an experiment, then carrying out KPCA dimension reduction on the normalized data, reducing 16-dimensional data to a dimension containing 85% of main components of the 16-dimensional data, then carrying out mean shift clustering on elements by utilizing the similarity of data types, and classifying 63 elements into 7 classes according to a criterion that each class at least contains 2 elements. In the 7 types, 12 representative elements are selected according to laboratory conditions and experience so as to obtain processed input data for establishing a prediction model, and the 12 elements are utilized to carry out chemical experiments to obtain the reaction activity index of the corresponding catalyst so as to obtain required output data.
Step three: establishing a p-nitrophenol reduction catalyst prediction model based on Gaussian process regression by utilizing the processed input and output data; the specific implementation method comprises the following steps: dividing 12 input data and output data into training data (X, Y) and test data (X) according to 7:3*,Y*) Establishing a function model according to training input data and corresponding output data: y ═ f (x), given a priori to the gaussian process (x) is: (X) -GP (mu (X), k (X, X')), wherein mu is a mean function, k is a kernel function, and the kernel function adopts an RBF kernel function as follows:wherein X' is the kernel function center, sigma is the width parameter of the function, the radial action range of the function is controlled, and because the training data and the test data belong to the same distribution, the combined Gaussian distribution of the training data and the test data is obtained:according to training input and output data, the input and output relation of a Gaussian process regression prediction model can be obtained by utilizing a kernel function and combining joint Gaussian distribution, and finally, test input data X is utilized*Input model to obtain predicted valueThe hyper-parameter x based on the enhanced crow search algorithm optimized Gaussian process regression model comprises a width parameter sigma of a kernel function and a regularization parameter C, namely x is { sigma, C }.
Step four: and optimizing the hyper-parameters of the p-nitrophenol reduction catalyst prediction model based on Gaussian process regression by adopting an enhanced crow search algorithm to obtain the p-nitrophenol reduction catalyst prediction model based on the enhanced crow search algorithm optimized Gaussian process regression.
Improving the crow search algorithm to obtain an enhanced crow search algorithm, wherein the enhanced crow search algorithm comprises the following steps: tracking flight phase and disaggregation updating process of the crow;
in the crow tracking flight stage, the crow tracking flight is split into two stages according to probability, and the crow flight strategy in the first stage is as follows: x is the number ofi,iter+1=xi,iter+ri×fli,iter×(mj,iter-xi,iter) In the formula xi,iter+1Is the value of the post-location update hyper-parameter, xi,iterIs the value of the over-parameter, r, for the current iteration positioniIs a uniformly distributed random number, fl, within 0 to 1i,iterIs the flying radius of crow, mj,iterIs a memory location of crow;
the shrink wrap strategy in the second stage is formulated as follows:in the formulagbest is the best individual in the current population,a is linearly decreasing from 2 to 0 and r is [0,1 ]]Uniformly distributed random numbers within;
the solution set update process probability formula is as follows:where f (x) is a fitness function, x is a current location over-parameter value, and x' isWhen the delta f is less than or equal to 0, the updated super parameter value is accepted as the next initial super parameter value, and when the delta f is more than or equal to 0, the probability r of accepting the updated super parameter value is calculated to generate [0,1]If r is less than or equal to s, the updated super parameter value is accepted as the next initial super parameter value, otherwise, the original super parameter value is still taken as the next initial super parameter value,
the method for carrying out hyper-parameter optimization on the established Gaussian process regression prediction model by using the enhanced crow search algorithm specifically comprises the following steps:
initializing parameter setting of an enhanced crow searching algorithm, the number of crow populations and a memory position of the crow;
by a fitness functionEvaluating fitness value of each individual crow's foot, wherein Y*Representing the true output corresponding to test input data X,representing test input data X*An expected output based on a gaussian process regression model;
updating the positions of the particles by using a flight strategy formula of the crow, evaluating a fitness function value of the new position, updating the super parameter value of the current position of the crow according to the simulated annealing probability, and updating the memory position of the crow;
and repeating the previous step until the maximum iteration number is reached or the fitness value meets the requirement of error precision, and stopping the optimization process.
Step five: and (3) predicting input data of other elements in the 63 elements except 12 elements by adopting a well-established p-nitrophenol reduction catalyst prediction model based on an enhanced crow search algorithm optimized Gaussian process regression, completing the prediction of the catalytic performance of all the elements, and screening to obtain the element with the best catalytic performance.
Claims (3)
1. A4-NP reduction catalyst modeling prediction method based on Gaussian process regression of ECSA is characterized by comprising the following steps:
the method comprises the following steps: obtaining original data to form a data set of a p-nitrophenol reduction catalyst prediction model based on enhanced crow search algorithm optimized Gaussian process regression, wherein the original data comprises 16 physicochemical properties of oxidation enthalpy of 63 chemical elements, first ionization energy, second ionization energy, electric dipole polarizability, boiling point, melting point, heat of solution, heat of vaporization, heat conductivity, density, ionic radius, atomic weight, electronegativity, specific heat capacity, covalent radius and ionic valence;
step two: processing the original data: carrying out normalization processing on original data required by an experiment, then carrying out KPCA (kernel principal component analysis) dimension reduction on the normalized data, reducing 16-dimensional data to a dimension containing 80-85% of main components of the 16-dimensional data, carrying out mean shift clustering on elements by utilizing the similarity of data types, dividing 63 elements into m classes, selecting s representative elements in the m classes, thus obtaining processed input data for establishing a prediction model, and carrying out a chemical experiment by utilizing the s elements to obtain a reaction activity index of a corresponding catalyst so as to obtain required output data;
step three: establishing a p-nitrophenol reduction catalyst prediction model based on Gaussian process regression by utilizing the processed input and output data;
step four: optimizing the hyper-parameters of a p-nitrophenol reduction catalyst prediction model based on Gaussian process regression by adopting a crow search algorithm to obtain the p-nitrophenol reduction catalyst prediction model based on the crow search algorithm optimized Gaussian process regression;
step five: and (3) predicting input data of other elements in the 63 elements except the s elements by adopting a well-established p-nitrophenol reduction catalyst prediction model based on an enhanced crow search algorithm optimized Gaussian process regression, completing the prediction of the catalytic performance of all the elements, and screening to obtain the element with the best catalytic performance.
2. The method of claim 1, wherein the ECSA-based 4-NP reduction catalyst is based on gaussian process regressionThe model prediction method is characterized in that a Gaussian process regression model is established by utilizing processed input data and output data, and the specific implementation method comprises the following steps: the s input data and the output data are divided into training data (X, Y) and test data (X)*,Y*) Establishing a function model according to training input data and corresponding output data: y ═ f (x), given a priori to the gaussian process (x) is: (X) -GP (mu (X), k (X, X')), wherein mu is a mean function, k is a kernel function, and the kernel function adopts an RBF kernel function as follows:wherein X' is the kernel function center, sigma is the width parameter of the function, the radial action range of the function is controlled, and because the training data and the test data belong to the same distribution, the combined Gaussian distribution of the training data and the test data is obtained:according to training input and output data, the input and output relation of a Gaussian process regression prediction model can be obtained by utilizing a kernel function and combining joint Gaussian distribution, and finally, test input data X is utilized*Input model to obtain predicted valueThe hyper-parameter x based on the enhanced crow search algorithm optimized Gaussian process regression model comprises a width parameter sigma of a kernel function and a regularization parameter C, namely x is { sigma, C }.
3. The modeling and predicting method for the 4-NP reduction catalyst based on Gaussian process regression of ECSA as claimed in claim 2, wherein said improving the crow search algorithm to obtain the enhanced crow search algorithm comprises: tracking flight phase and disaggregation updating process of the crow;
in the crow tracking flight stage, the crow tracking flight is split into two stages according to probability, and the crow flight strategy in the first stage is as follows: x is the number ofi,iter+1=xi,iter+ri×fli,iter×(mj,iter-xi,iter) In the formula xi,iter+1Is the value of the post-location update hyper-parameter, xi,iterIs the value of the over-parameter, r, for the current iteration positioniIs a uniformly distributed random number, fl, within 0 to 1i,iterIs the flying radius of crow, mj,iterIs a memory location of crow;
the shrink wrap strategy in the second stage is formulated as follows:in the formulagbestIs the optimal individual in the current population,a is linearly decreasing from 2 to 0 and r is [0,1 ]]Uniformly distributed random numbers within;
the solution set update process probability formula is as follows:wherein f (x) is a fitness function, x is a current position super-parameter value, x 'is an updated super-parameter value, T is an initial temperature, and let Δ f ═ f (x') -f (x), when Δ f is not more than 0, the updated super-parameter value is accepted as a next initial super-parameter value, when Δ f is not more than 0, a probability r of accepting the updated super-parameter value is calculated, and [0,1 ] is generated]If r is less than or equal to s, the updated super parameter value is accepted as the next initial super parameter value, otherwise, the original super parameter value is still taken as the next initial super parameter value,
the method for carrying out hyper-parameter optimization on the established Gaussian process regression prediction model based on the enhanced crow search algorithm specifically comprises the following steps:
initializing parameter setting of an enhanced crow searching algorithm, the number of crow populations and a memory position of the crow;
by a fitness functionEvaluating fitness value of each individual crow's foot, wherein Y*Representing test input data X*The corresponding real output of the real-world output,representing test input data X*An expected output based on a gaussian process regression model;
updating the positions of the particles by using a flight strategy formula of the crow, evaluating a fitness function value of the new position, updating the super parameter value of the current position of the crow according to the simulated annealing probability, and updating the memory position of the crow;
and repeating the previous step until the maximum iteration number is reached or the fitness value meets the requirement of error precision, and stopping the optimization process.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910851343.1A CN110750858B (en) | 2019-09-10 | 2019-09-10 | 4-NP reduction catalyst modeling prediction method based on ECSA Gaussian process regression |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910851343.1A CN110750858B (en) | 2019-09-10 | 2019-09-10 | 4-NP reduction catalyst modeling prediction method based on ECSA Gaussian process regression |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110750858A true CN110750858A (en) | 2020-02-04 |
CN110750858B CN110750858B (en) | 2020-07-21 |
Family
ID=69276241
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910851343.1A Active CN110750858B (en) | 2019-09-10 | 2019-09-10 | 4-NP reduction catalyst modeling prediction method based on ECSA Gaussian process regression |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110750858B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112307559A (en) * | 2020-10-23 | 2021-02-02 | 西北工业大学 | AUV (autonomous Underwater vehicle) shape optimization method based on improved crow algorithm |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106677763A (en) * | 2016-12-29 | 2017-05-17 | 沈阳工业大学 | Oil well dynamic liquid level prediction method based on dynamic integrated modeling |
CN109190023A (en) * | 2018-08-15 | 2019-01-11 | 深圳信息职业技术学院 | The method, apparatus and terminal device of Collaborative Recommendation |
CN109325616A (en) * | 2018-09-01 | 2019-02-12 | 哈尔滨工程大学 | It is a kind of to be returned based on Gaussian process and the fine particle that combines of glowworm swarm algorithm is predicted and source tracing method |
CN109658988A (en) * | 2018-12-20 | 2019-04-19 | 中海石油炼化有限责任公司 | A kind of Hydrobon catalyst performance prediction method |
CN109800517A (en) * | 2019-01-24 | 2019-05-24 | 闽江学院 | Improved reverse modeling method for magnetorheological damper |
CN109979544A (en) * | 2019-03-28 | 2019-07-05 | 上海交通大学 | The cyclical stability prediction model and its modeling method of cathode Pt-M catalyst |
-
2019
- 2019-09-10 CN CN201910851343.1A patent/CN110750858B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106677763A (en) * | 2016-12-29 | 2017-05-17 | 沈阳工业大学 | Oil well dynamic liquid level prediction method based on dynamic integrated modeling |
CN109190023A (en) * | 2018-08-15 | 2019-01-11 | 深圳信息职业技术学院 | The method, apparatus and terminal device of Collaborative Recommendation |
CN109325616A (en) * | 2018-09-01 | 2019-02-12 | 哈尔滨工程大学 | It is a kind of to be returned based on Gaussian process and the fine particle that combines of glowworm swarm algorithm is predicted and source tracing method |
CN109658988A (en) * | 2018-12-20 | 2019-04-19 | 中海石油炼化有限责任公司 | A kind of Hydrobon catalyst performance prediction method |
CN109800517A (en) * | 2019-01-24 | 2019-05-24 | 闽江学院 | Improved reverse modeling method for magnetorheological damper |
CN109979544A (en) * | 2019-03-28 | 2019-07-05 | 上海交通大学 | The cyclical stability prediction model and its modeling method of cathode Pt-M catalyst |
Non-Patent Citations (4)
Title |
---|
ARMIN JAVIDI 等: "Enhanced crow search algorithm for optimum design of structures", 《APPLIED SOFT COMPUTING JOURNAL》 * |
S. ARUMUGAM 等: "RSM and Crow Search Algorithm-Based Optimization of Ultrasonicated Transesterification Process Parameters on Synthesis of Polyol Ester-Based Biolubricant", 《ARABIAN JOURNAL FOR SCIENCE AND ENGINEERING》 * |
赵超凡: "基于数据挖掘的甲烷化催化剂建模研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅰ辑》 * |
韩晓霞 等: "一种软计算混合策略在多相催化剂建模与预测中的应用", 《太原理工大学学报》 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112307559A (en) * | 2020-10-23 | 2021-02-02 | 西北工业大学 | AUV (autonomous Underwater vehicle) shape optimization method based on improved crow algorithm |
Also Published As
Publication number | Publication date |
---|---|
CN110750858B (en) | 2020-07-21 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Wu et al. | Ultra‐short‐term multi‐step wind power forecasting based on CNN‐LSTM | |
Wang et al. | cPSO-CNN: An efficient PSO-based algorithm for fine-tuning hyper-parameters of convolutional neural networks | |
Manzoor et al. | On entropy measures of molecular graphs using topological indices | |
Kudela et al. | Recent advances and applications of surrogate models for finite element method computations: a review | |
Bennett et al. | Autonomous chemical science and engineering enabled by self-driving laboratories | |
Li et al. | A novel double incremental learning algorithm for time series prediction | |
Pedrycz et al. | Identifying core sets of discriminatory features using particle swarm optimization | |
Kovács et al. | Machine-learning prediction of infrared spectra of interstellar polycyclic aromatic hydrocarbons | |
Kyriacou et al. | Efficient PCA-driven EAs and metamodel-assisted EAs, with applications in turbomachinery | |
CN110750858B (en) | 4-NP reduction catalyst modeling prediction method based on ECSA Gaussian process regression | |
Bi et al. | Self-adaptive teaching-learning-based optimizer with improved RBF and sparse autoencoder for high-dimensional problems | |
Chu et al. | Reliability based optimization with metaheuristic algorithms and Latin hypercube sampling based surrogate models | |
Liu et al. | FOX-NAS: Fast, on-device and explainable neural architecture search | |
Martinez-de-Pison et al. | A comparative study of six model complexity metrics to search for parsimonious models with GAparsimony R Package | |
Yi et al. | New feature analysis-based elastic net algorithm with clustering objective function | |
CN114334028A (en) | Method, system and storage medium for generating indefinite 3D structure based on graph neural network | |
Shaddeli et al. | BFRA: a new binary hyper-heuristics feature ranks algorithm for feature selection in high-dimensional classification data | |
CN113052367A (en) | Method for efficiently predicting stability of perovskite based on integrated machine learning | |
Whyte et al. | Surrogate model optimization of a ‘micro core’pwr fuel assembly arrangement using deep learning models | |
Wang et al. | A class of order-based genetic algorithm for flow shop scheduling | |
Li et al. | An adaptive surrogate-assisted particle swarm optimization for expensive problems | |
Örkçü et al. | A hybrid applied optimization algorithm for training multi-layer neural networks in data classification | |
Guo et al. | Dynamic neural network structure: A review for its theories and applications | |
Shi et al. | Multi‐objective optimization and its application in materials science | |
Elghamrawy et al. | A hybrid Genetic–Grey Wolf Optimization algorithm for optimizing Takagi–Sugeno–Kang fuzzy systems |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right | ||
TR01 | Transfer of patent right |
Effective date of registration: 20210519 Address after: 030002 No. a1116, 115 Keji street, Taiyuan Xuefu Park, Shanxi comprehensive reform demonstration zone, Taiyuan City, Shanxi Province Patentee after: SHANXI JIETAIDA COAL CHEMICAL INDUSTRY ENGINEERING Co.,Ltd. Address before: 030024 No. 79 West Main Street, Taiyuan, Shanxi, Yingze Patentee before: Taiyuan University of Technology |