Blast furnace scheme design is very important, since it directly affects the performance, cost and configuration of the blast furnace. An evaluation approach to furnace scheme design was brought forward based on artif...Blast furnace scheme design is very important, since it directly affects the performance, cost and configuration of the blast furnace. An evaluation approach to furnace scheme design was brought forward based on artificial neural network. Ten independent parameters which determined a scheme design were proposed. The improved threelayer BP network algorithm was used to build the evaluation model in which the 10 independent parameters were taken as input evaluation indexes and the degree to which the scheme design satisfies the requirements of the blast furnace as output. It was trained by the existing samples of the scheme design and the experts＇ experience, and then tested by the other samples so as to develop the evaluation model. As an example, it is found that a good scheme design of blast furnace can be chosen by using the evaluation model proposed.展开更多
RBF model,a new type of artificial neural network model was developed to design the content of carbon in low-alloy engineering steels.The errors of the ANN model are：MSE 0.052 1,MSRE 17.85%,and VOF 1.932 9.The result...RBF model,a new type of artificial neural network model was developed to design the content of carbon in low-alloy engineering steels.The errors of the ANN model are：MSE 0.052 1,MSRE 17.85%,and VOF 1.932 9.The results obtained are satisfactory.The method is a powerful aid for designing new steels.展开更多
In this paper, the method based on uniform design and neural network is proposed to model the complex system. In order to express the system characteristics all round, uniform design method is used to choose the model...In this paper, the method based on uniform design and neural network is proposed to model the complex system. In order to express the system characteristics all round, uniform design method is used to choose the modeling samples and obtain the overall information of the system;for the purpose of modeling the system or its characteristics, the artificial neural network is used to construct the model. Experiment indicates that this method can model the complex system effectively.展开更多
Artificial neural network（ANN） and full factorial design assisted atrazine（AT） multiple regression adsorption model（AT-MRAM） were developed to analyze the adsorption capability of the main components in the surf...Artificial neural network（ANN） and full factorial design assisted atrazine（AT） multiple regression adsorption model（AT-MRAM） were developed to analyze the adsorption capability of the main components in the surficial sediments（SSs）. Artificial neural network was used to build a model（the determination coefficient square r2 is 0.9977） to describe the process of atrazine adsorption onto SSs, and then to predict responses of the full factorial design. Based on the results of the full factorial design, the interactions of the main components in SSs on AT adsorption were investigated through the analysis of variance（ANOVA）, F-test and t-test. The adsorption capability of the main components in SSs for AT was calculated via a multiple regression adsorption model（MRAM）. The results show that the greatest contribution to the adsorption of AT on a molar basis was attributed to Fe/Mn（–1.993 μmol/mol）. Organic materials（OMs） and Fe oxides in SSs are the important adsorption sites for AT, and the adsorption capabilities are 1.944 and 0.418 μmol/mol, respectively. The interaction among the non-residual components（Fe, Mn oxides and OMs） in SSs interferes in the adsorption of AT that shouldn’t be neglected, revealing the significant contribution of the interaction among non-residual components to controlling the behavior of AT in aquatic environments.展开更多
An optimal design approach of high order FIR digital filter is developed based on the algorithm of neural networks with cosine basis function . The main idea is to minimize the sum of the square errors between the amp...An optimal design approach of high order FIR digital filter is developed based on the algorithm of neural networks with cosine basis function . The main idea is to minimize the sum of the square errors between the amplitude response of the desired FIR filter and that of the designed by training the weights of neural networks, then obtains the impulse response of FIR digital filter . The convergence theorem of the neural networks algorithm is presented and proved, and the optimal design method is introduced by designing four kinds of FIR digital filters , i.e., low-pass, high-pass, bandpass , and band-stop FIR digital filter. The results of the amplitude responses show that attenuation in stop-bands is more than 60 dB with no ripple and pulse existing in pass-bands, and cutoff frequency of passband and stop-band is easily controlled precisely .The presented optimal design approach of high order FIR digital filter is significantly effective.展开更多
A new product conceptual design approach is put forward based on Hopfield neural networks models. By research on the mechanisms of Hopfield neural networks, the associative simulation approaches are proposed. The appr...A new product conceptual design approach is put forward based on Hopfield neural networks models. By research on the mechanisms of Hopfield neural networks, the associative simulation approaches are proposed. The approach is given by Hebb learn- ing law, Hopfield neural networks and crossover and mutation. The calculating models and the calculating formulas for the concep- tual design are put forward. Finally, an example for the conceptual design of a solar energy lamp is given. The better results are ob- tained in the conceptual design.展开更多
In this study, an application of artificial neural network （ANN） has been presented in modeling and studying the effect of compounding variables on abrasion behavior of rubber formulations. Three case studies were c...In this study, an application of artificial neural network （ANN） has been presented in modeling and studying the effect of compounding variables on abrasion behavior of rubber formulations. Three case studies were carried out in which the experiment data were collected according to classical response surface designs. Besides developing the ANN models, we developed response surface methodology （RSM） to confirm the ANN predictions. A simple relation was employed for determination of relative importance of each variable according to ANN models. It was shown through these case studies that ANN models delivered very good data fitting and their simulating curves could help the researchers to better understand the abrasion behavior.展开更多
Neural networks are being used to construct meta-models in numerical simulation of structures.In addition to network structures and training algorithms,training samples also greatly affect the accuracy of neural netwo...Neural networks are being used to construct meta-models in numerical simulation of structures.In addition to network structures and training algorithms,training samples also greatly affect the accuracy of neural network models.In this paper,some existing main sampling techniques are evaluated,including techniques based on experimental design theory, random selection,and rotating sampling.First,advantages and disadvantages of each technique are reviewed.Then,seven techniques are used to generate samples for training radial neural networks models for two benchmarks:an antenna model and an aircraft model.Results show that the uniform design,in which the number of samples and mean square error network models are considered,is the best sampling technique for neural network based meta-model building.展开更多
Corrosion in complex coupling environments is an important issue in corrosion field, because it is difficult to take into account a large number of environment factors and their interactions. Design of Experiment(DOE)...Corrosion in complex coupling environments is an important issue in corrosion field, because it is difficult to take into account a large number of environment factors and their interactions. Design of Experiment(DOE) can present a methodology to deal with this difficulty, although DOE is not commonly spread in corrosion field. Thus, modeling corrosion of Ni-Cr-Mo-V steel in deep sea environment was performed in order to provide example demonstrating the advantage of DOE. In addition, an artificial neural network mapping using back-propagation method was developed for Ni-Cr-Mo-V steel such that the ANN model can be used to predict polarization curves under different complex sea environments without experimentation. Furthermore, roles of environment factors on corrosion of Ni-Cr-Mo-V steel in deep sea environment were discussed.展开更多
In the optimum design of the self-propelled combine gearbox, the knowledge of ANN is applied to overcome slow calculation and low design efficiency. We did the normative approach of the charts information,then solved ...In the optimum design of the self-propelled combine gearbox, the knowledge of ANN is applied to overcome slow calculation and low design efficiency. We did the normative approach of the charts information,then solved the difficult problems in the design process and get satisfactory results. We also completed three-dimensional design of the gearbox in order to verify the rationality of the design visually.展开更多
Using neural networks for supervised learning means learning a function that maps input <em>x</em> to output <em>y</em>. However, in many applications, the inverse learning is also wanted, <...Using neural networks for supervised learning means learning a function that maps input <em>x</em> to output <em>y</em>. However, in many applications, the inverse learning is also wanted, <em>i.e.</em>, inferring <em>y</em> from <em>x</em>, which requires invertibility of the learning. Since the dimension of input is usually much higher than that of the output, there is information loss in the forward learning from input to output. Thus, creating invertible neural networks is a difficult task. However, recent development of invertible learning techniques such as normalizing flows has made invertible neural networks a reality. In this work, we applied flow-based invertible neural networks as generative models to inverse molecule design. In this context, the forward learning is to predict chemical properties given a molecule, and the inverse learning is to infer the molecules given the chemical properties. Trained on 100 and 1000 molecules, respectively, from a benchmark dataset QM9, our model identified novel molecules that had chemical property values well exceeding the limits of the training molecules as well as the limits of the whole QM9 of 133,885 molecules, moreover our generative model could easily sample many molecules (<em>x</em> values) from any one chemical property value (<em>y</em> value). Compared with the previous method in the literature that could only optimize one molecule for one chemical property value at a time, our model could be trained once and then be sampled any multiple times and for any chemical property values without the need of retraining. This advantage comes from treating inverse molecule design as an inverse regression problem. In summary, our main contributions were two: 1) our model could generalize well from the training data and was very data efficient, 2) our model could learn bidirectional correspondence between molecules and their chemical properties, thereby offering the ability to sample any number of molecules from any <em>y</em> values. In conclusion, our findings revealed the efficiency and effectiveness of using invertible neural networks as generative models in inverse molecule design.展开更多
In order to improve turbine internal efficiency and lower manufacturing cost, a new highly loaded rotating blade has been developed. The 3D optimization design method based on artificial neural network and genetic alg...In order to improve turbine internal efficiency and lower manufacturing cost, a new highly loaded rotating blade has been developed. The 3D optimization design method based on artificial neural network and genetic algorithm is adopted to construct the blade shape. The blade is stacked by the center of gravity in radial direction with five sections. For each blade section, independent suction and pressure sides are constructed from the camber line using Bezier curves. Three-dimensional flow analysis is carried out to verify the performance of the new blade. It is found that the new blade has improved the blade performance by 0.5%. Consequently, it is verified that the new blade is effective to improve the turbine internal efficiency and to lower the turbine weight and manufacturing cost by reducing the blade number by about 15%.展开更多
This research work investigated comparative studies of expert system design and control of crude oil distillation column (CODC) using artificial neural networks based Monte Carlo (ANNBMC) simulation of random processe...This research work investigated comparative studies of expert system design and control of crude oil distillation column (CODC) using artificial neural networks based Monte Carlo (ANNBMC) simulation of random processes and artificial neural networks (ANN) model which were validated using experimental data obtained from functioning crude oil distillation column of Port-Harcourt Refinery, Nigeria by MATLAB computer program. Ninety percent (90%) of the experimental data sets were used for training while ten percent (10%) were used for testing the networks. The maximum relative errors between the experimental and calculated data obtained from the output variables of the neural network for CODC design were 1.98 error % and 0.57 error % when ANN only and ANNBMC were used respectively while their respective values for the maximum relative error were 0.346 error % and 0.124 error % when they were used for the controller prediction. Larger number of iteration steps of below 2500 and 5000 were required to achieve convergence of less than 10-7?for the training error using ANNBMC for both the design of the CODC and controller respectively while less than 400 and 700 iteration steps were needed to achieve convergence of 10-4?using ANN only. The linear regression analysis performed revealed the minimum and maximum prediction accuracies to be 80.65% and 98.79%;and 98.38% and 99.98% when ANN and ANNBMC were used for the CODC design respectively. Also, the minimum and maximum prediction accuracies were 92.83% and 99.34%;and 98.89% and 99.71% when ANN and ANNBMC were used for the CODC controller respectively as both methodologies have excellent predictions. Hence, artificial neural networks based Monte Carlo simulation is an effective and better tool for the design and control of crude oil distillation column.展开更多
The model describing the dependence of the mechanical properties on the chemical composition and as deformation techniques of tungsten heavy alloy is established by the method of improved the backpropagation neural ne...The model describing the dependence of the mechanical properties on the chemical composition and as deformation techniques of tungsten heavy alloy is established by the method of improved the backpropagation neural network. The mechanical properties＇ parameters of tungsten alloy and deformation techniques for tungsten alloy are used as the inputs. The chemical composition and deformation amount of tungsten alloy are used as the outputs. Then they are used for training the neural network. At the same time, the optimal number of the hidden neurons is obtained through the experiential equations, and the varied step learning method is adopted to ensure the stability of the training process. According to the requirements for mechanical properties, the chemical composition and the deformation condition for tungsten heavy alloy can be designed by this artificial neural network system.展开更多
The coating on the electrodes contains many kinds of raw materials which affect significantly on the mechanical properties of deposited metals. It is still a problem how to predict and control the mechanical propertie...The coating on the electrodes contains many kinds of raw materials which affect significantly on the mechanical properties of deposited metals. It is still a problem how to predict and control the mechanical properties of deposited metals directly according to the components of coating on the electrodes. In this paper an electrode intelligent design system is developed by means of fuzzy neural network technology and genetic algorithm,, dynamic link library, object linking and embedding and multithreading. The front-end application and customer interface of the system is realized by using visual C ＋＋ program language and taking SQL Server 2000 as background database. It realizes series functions including automatic design of electrode formula, intelligent prediction of electrode properties, inquiry of electrode information, output of process report based on normalized template and electronic storage and search of relative files.展开更多
Due to the complex chemical composition of nickel ores, the requests for the decrease of production costs, and the increase of nickel extraction in the existing depletion of high-grade sulfide ores around the world, c...Due to the complex chemical composition of nickel ores, the requests for the decrease of production costs, and the increase of nickel extraction in the existing depletion of high-grade sulfide ores around the world, computer modeling of nickel ore leaching process be- came a need and a challenge. In this paper, the design of experiments （DOE） theory was used to determine the optimal experimental design plan matrix based on the D optimality criterion. In the high-pressure sulfuric acid leaching （HPSAL） process for nickel laterite in ＂Rudjinci＂ ore in Serbia, the temperature, the sulfuric acid to ore ratio, the stirring speed, and the leaching time as the predictor variables, and the degree of nickel extraction as the response have been considered. To model the process, the multiple linear regression （MLR） and response surface method （RSM）, together with the two-level and four-factor full factorial central composite design （CCD） plan, were used. The proposed re- gression models have not been proven adequate. Therefore, the artificial neural network （ANN） approach with the same experimental plan was used in order to reduce operational costs, give a better modeling accuracy, and provide a more successful process optimization. The model is based on the multi-layer neural networks with the back-propagation （BP） learning algorithm and the bipolar sigmoid activation function.展开更多
With the increase in research on AI(Artificial Intelligence),the importance of DL(Deep Learning)in various fields,such as materials,biotechnology,genomes,and new drugs,is increasing significantly,thereby increasing th...With the increase in research on AI(Artificial Intelligence),the importance of DL(Deep Learning)in various fields,such as materials,biotechnology,genomes,and new drugs,is increasing significantly,thereby increasing the number of deep-learning framework users.However,to design a deep neural network,a considerable understanding of the framework is required.To solve this problem,a GUI(Graphical User Interface)-based DNN(Deep Neural Network)design tool is being actively researched and developed.The GUI-based DNN design tool can design DNNs quickly and easily.However,the existing GUI-based DNN design tool has certain limitations such as poor usability,framework dependency,and difficulty encountered in changing GUI components.In this study,a deep learning algorithm that solves the problem of poor usability was developed using a template to increase the accessibility for users.Moreover,the proposed tool was developed to save and share only the necessary parts for quick operation.To solve the framework dependency,we applied ONNX(Open Neural Network Exchange),which is an exchange standard for neural networks,and configured it such that DNNs designed with the existing deep-learning framework can be imported.Finally,to address the difficulty encountered in changing GUI components,we defined and developed the JSON format to quickly respond to version updates.The developed DL neural network designer was validated by running it with KISTI’s supercomputer-based AI Studio.展开更多
Based on the method of artificial neural network, a new approach has been devised to predict the mechanical property of E4303 electrode. The outlined predication model for determining the mechanical property of electr...Based on the method of artificial neural network, a new approach has been devised to predict the mechanical property of E4303 electrode. The outlined predication model for determining the mechanical property of electrode was built upon the production data. The research leverages a back propagation algorithm as the neural network’s learning rule. The result indicates that there are positive correlations between the predicted results and the practical production data. Hence, using the neural network, predication of electrode property can be realized. For the first time, this research provides a more scientific method for designing electrode.展开更多
Organic matters（OMs） and their oxidization products often influence the fate and transport of heavy metals in the subsurface aqueous systems through interaction with the mineral surfaces. This study investigates the...Organic matters（OMs） and their oxidization products often influence the fate and transport of heavy metals in the subsurface aqueous systems through interaction with the mineral surfaces. This study investigates the ethanol（EtO H）-mediated As（Ⅲ） adsorption onto Zn-loaded pinecone（PC） biochar through batch experiments conducted under Box–Behnken design. The effect of EtO H on As（Ⅲ） adsorption mechanism was quantitatively elucidated by fitting the experimental data using artificial neural network and quadratic modeling approaches. The quadratic model could describe the limiting nature of EtO H and pH on As（Ⅲ） adsorption,whereas neural network revealed the stronger influence of Et OH（64.5%） followed by pH（20.75%）and As（Ⅲ） concentration（14.75%） on the adsorption phenomena. Besides, the interaction among process variables indicated that Et OH enhances As（Ⅲ） adsorption over a pH range of2 to 7, possibly due to facilitation of ligand–metal（Zn） binding complexation mechanism.Eventually, hybrid response surface model–genetic algorithm（RSM–GA） approach predicted a better optimal solution than RSM, i.e., the adsorptive removal of As（Ⅲ）（10.47 μg/g） is facilitated at 30.22 mg C/L of Et OH with initial As（Ⅲ） concentration of 196.77 μg/L at pH 5.8. The implication of this investigation might help in understanding the application of biochar for removal of various As（Ⅲ） species in the presence of OM.展开更多
基金Provincial Natural Science Foundation of Sichuan Province of China (04JY029-003-2)
文摘Blast furnace scheme design is very important, since it directly affects the performance, cost and configuration of the blast furnace. An evaluation approach to furnace scheme design was brought forward based on artificial neural network. Ten independent parameters which determined a scheme design were proposed. The improved threelayer BP network algorithm was used to build the evaluation model in which the 10 independent parameters were taken as input evaluation indexes and the degree to which the scheme design satisfies the requirements of the blast furnace as output. It was trained by the existing samples of the scheme design and the experts＇ experience, and then tested by the other samples so as to develop the evaluation model. As an example, it is found that a good scheme design of blast furnace can be chosen by using the evaluation model proposed.
文摘RBF model,a new type of artificial neural network model was developed to design the content of carbon in low-alloy engineering steels.The errors of the ANN model are：MSE 0.052 1,MSRE 17.85%,and VOF 1.932 9.The results obtained are satisfactory.The method is a powerful aid for designing new steels.
文摘In this paper, the method based on uniform design and neural network is proposed to model the complex system. In order to express the system characteristics all round, uniform design method is used to choose the modeling samples and obtain the overall information of the system;for the purpose of modeling the system or its characteristics, the artificial neural network is used to construct the model. Experiment indicates that this method can model the complex system effectively.
基金Supported by the National Natural Science Foundation of China(No.50879025)
文摘Artificial neural network（ANN） and full factorial design assisted atrazine（AT） multiple regression adsorption model（AT-MRAM） were developed to analyze the adsorption capability of the main components in the surficial sediments（SSs）. Artificial neural network was used to build a model（the determination coefficient square r2 is 0.9977） to describe the process of atrazine adsorption onto SSs, and then to predict responses of the full factorial design. Based on the results of the full factorial design, the interactions of the main components in SSs on AT adsorption were investigated through the analysis of variance（ANOVA）, F-test and t-test. The adsorption capability of the main components in SSs for AT was calculated via a multiple regression adsorption model（MRAM）. The results show that the greatest contribution to the adsorption of AT on a molar basis was attributed to Fe/Mn（–1.993 μmol/mol）. Organic materials（OMs） and Fe oxides in SSs are the important adsorption sites for AT, and the adsorption capabilities are 1.944 and 0.418 μmol/mol, respectively. The interaction among the non-residual components（Fe, Mn oxides and OMs） in SSs interferes in the adsorption of AT that shouldn’t be neglected, revealing the significant contribution of the interaction among non-residual components to controlling the behavior of AT in aquatic environments.
基金This project was supported by the National Natural Science Foundation of China (50277010)Doctoral Special Fund of Ministry of Education (20020532016) and Fund of Outstanding Young Scientist of Hunan University.
文摘An optimal design approach of high order FIR digital filter is developed based on the algorithm of neural networks with cosine basis function . The main idea is to minimize the sum of the square errors between the amplitude response of the desired FIR filter and that of the designed by training the weights of neural networks, then obtains the impulse response of FIR digital filter . The convergence theorem of the neural networks algorithm is presented and proved, and the optimal design method is introduced by designing four kinds of FIR digital filters , i.e., low-pass, high-pass, bandpass , and band-stop FIR digital filter. The results of the amplitude responses show that attenuation in stop-bands is more than 60 dB with no ripple and pulse existing in pass-bands, and cutoff frequency of passband and stop-band is easily controlled precisely .The presented optimal design approach of high order FIR digital filter is significantly effective.
基金Partially Supported by National Natural Science Foundation of China(No.50975033,No.60875046)Education Office of Liaoning Province(No.LR2013060)Natural Science Foundation of Liaoning Province(No.2013020123)
文摘A new product conceptual design approach is put forward based on Hopfield neural networks models. By research on the mechanisms of Hopfield neural networks, the associative simulation approaches are proposed. The approach is given by Hebb learn- ing law, Hopfield neural networks and crossover and mutation. The calculating models and the calculating formulas for the concep- tual design are put forward. Finally, an example for the conceptual design of a solar energy lamp is given. The better results are ob- tained in the conceptual design.
文摘In this study, an application of artificial neural network （ANN） has been presented in modeling and studying the effect of compounding variables on abrasion behavior of rubber formulations. Three case studies were carried out in which the experiment data were collected according to classical response surface designs. Besides developing the ANN models, we developed response surface methodology （RSM） to confirm the ANN predictions. A simple relation was employed for determination of relative importance of each variable according to ANN models. It was shown through these case studies that ANN models delivered very good data fitting and their simulating curves could help the researchers to better understand the abrasion behavior.
基金Specialized Research Fund for the Doctoral Program of Higher Education,China (No.20010227012)
文摘Neural networks are being used to construct meta-models in numerical simulation of structures.In addition to network structures and training algorithms,training samples also greatly affect the accuracy of neural network models.In this paper,some existing main sampling techniques are evaluated,including techniques based on experimental design theory, random selection,and rotating sampling.First,advantages and disadvantages of each technique are reviewed.Then,seven techniques are used to generate samples for training radial neural networks models for two benchmarks:an antenna model and an aircraft model.Results show that the uniform design,in which the number of samples and mean square error network models are considered,is the best sampling technique for neural network based meta-model building.
基金the financial support of the National Natural Science Foundation of China(No.51371182)the National Program for the Young Top-notch Professionals and the Fundamental Research Funds for the Central Universities(N170205002)
文摘Corrosion in complex coupling environments is an important issue in corrosion field, because it is difficult to take into account a large number of environment factors and their interactions. Design of Experiment(DOE) can present a methodology to deal with this difficulty, although DOE is not commonly spread in corrosion field. Thus, modeling corrosion of Ni-Cr-Mo-V steel in deep sea environment was performed in order to provide example demonstrating the advantage of DOE. In addition, an artificial neural network mapping using back-propagation method was developed for Ni-Cr-Mo-V steel such that the ANN model can be used to predict polarization curves under different complex sea environments without experimentation. Furthermore, roles of environment factors on corrosion of Ni-Cr-Mo-V steel in deep sea environment were discussed.
基金Heilongjiang Province Postdoctoral Scientific Research Fund
文摘In the optimum design of the self-propelled combine gearbox, the knowledge of ANN is applied to overcome slow calculation and low design efficiency. We did the normative approach of the charts information,then solved the difficult problems in the design process and get satisfactory results. We also completed three-dimensional design of the gearbox in order to verify the rationality of the design visually.
文摘Using neural networks for supervised learning means learning a function that maps input <em>x</em> to output <em>y</em>. However, in many applications, the inverse learning is also wanted, <em>i.e.</em>, inferring <em>y</em> from <em>x</em>, which requires invertibility of the learning. Since the dimension of input is usually much higher than that of the output, there is information loss in the forward learning from input to output. Thus, creating invertible neural networks is a difficult task. However, recent development of invertible learning techniques such as normalizing flows has made invertible neural networks a reality. In this work, we applied flow-based invertible neural networks as generative models to inverse molecule design. In this context, the forward learning is to predict chemical properties given a molecule, and the inverse learning is to infer the molecules given the chemical properties. Trained on 100 and 1000 molecules, respectively, from a benchmark dataset QM9, our model identified novel molecules that had chemical property values well exceeding the limits of the training molecules as well as the limits of the whole QM9 of 133,885 molecules, moreover our generative model could easily sample many molecules (<em>x</em> values) from any one chemical property value (<em>y</em> value). Compared with the previous method in the literature that could only optimize one molecule for one chemical property value at a time, our model could be trained once and then be sampled any multiple times and for any chemical property values without the need of retraining. This advantage comes from treating inverse molecule design as an inverse regression problem. In summary, our main contributions were two: 1) our model could generalize well from the training data and was very data efficient, 2) our model could learn bidirectional correspondence between molecules and their chemical properties, thereby offering the ability to sample any number of molecules from any <em>y</em> values. In conclusion, our findings revealed the efficiency and effectiveness of using invertible neural networks as generative models in inverse molecule design.
文摘In order to improve turbine internal efficiency and lower manufacturing cost, a new highly loaded rotating blade has been developed. The 3D optimization design method based on artificial neural network and genetic algorithm is adopted to construct the blade shape. The blade is stacked by the center of gravity in radial direction with five sections. For each blade section, independent suction and pressure sides are constructed from the camber line using Bezier curves. Three-dimensional flow analysis is carried out to verify the performance of the new blade. It is found that the new blade has improved the blade performance by 0.5%. Consequently, it is verified that the new blade is effective to improve the turbine internal efficiency and to lower the turbine weight and manufacturing cost by reducing the blade number by about 15%.
文摘This research work investigated comparative studies of expert system design and control of crude oil distillation column (CODC) using artificial neural networks based Monte Carlo (ANNBMC) simulation of random processes and artificial neural networks (ANN) model which were validated using experimental data obtained from functioning crude oil distillation column of Port-Harcourt Refinery, Nigeria by MATLAB computer program. Ninety percent (90%) of the experimental data sets were used for training while ten percent (10%) were used for testing the networks. The maximum relative errors between the experimental and calculated data obtained from the output variables of the neural network for CODC design were 1.98 error % and 0.57 error % when ANN only and ANNBMC were used respectively while their respective values for the maximum relative error were 0.346 error % and 0.124 error % when they were used for the controller prediction. Larger number of iteration steps of below 2500 and 5000 were required to achieve convergence of less than 10-7?for the training error using ANNBMC for both the design of the CODC and controller respectively while less than 400 and 700 iteration steps were needed to achieve convergence of 10-4?using ANN only. The linear regression analysis performed revealed the minimum and maximum prediction accuracies to be 80.65% and 98.79%;and 98.38% and 99.98% when ANN and ANNBMC were used for the CODC design respectively. Also, the minimum and maximum prediction accuracies were 92.83% and 99.34%;and 98.89% and 99.71% when ANN and ANNBMC were used for the CODC controller respectively as both methodologies have excellent predictions. Hence, artificial neural networks based Monte Carlo simulation is an effective and better tool for the design and control of crude oil distillation column.
文摘The model describing the dependence of the mechanical properties on the chemical composition and as deformation techniques of tungsten heavy alloy is established by the method of improved the backpropagation neural network. The mechanical properties＇ parameters of tungsten alloy and deformation techniques for tungsten alloy are used as the inputs. The chemical composition and deformation amount of tungsten alloy are used as the outputs. Then they are used for training the neural network. At the same time, the optimal number of the hidden neurons is obtained through the experiential equations, and the varied step learning method is adopted to ensure the stability of the training process. According to the requirements for mechanical properties, the chemical composition and the deformation condition for tungsten heavy alloy can be designed by this artificial neural network system.
文摘The coating on the electrodes contains many kinds of raw materials which affect significantly on the mechanical properties of deposited metals. It is still a problem how to predict and control the mechanical properties of deposited metals directly according to the components of coating on the electrodes. In this paper an electrode intelligent design system is developed by means of fuzzy neural network technology and genetic algorithm,, dynamic link library, object linking and embedding and multithreading. The front-end application and customer interface of the system is realized by using visual C ＋＋ program language and taking SQL Server 2000 as background database. It realizes series functions including automatic design of electrode formula, intelligent prediction of electrode properties, inquiry of electrode information, output of process report based on normalized template and electronic storage and search of relative files.
文摘Due to the complex chemical composition of nickel ores, the requests for the decrease of production costs, and the increase of nickel extraction in the existing depletion of high-grade sulfide ores around the world, computer modeling of nickel ore leaching process be- came a need and a challenge. In this paper, the design of experiments （DOE） theory was used to determine the optimal experimental design plan matrix based on the D optimality criterion. In the high-pressure sulfuric acid leaching （HPSAL） process for nickel laterite in ＂Rudjinci＂ ore in Serbia, the temperature, the sulfuric acid to ore ratio, the stirring speed, and the leaching time as the predictor variables, and the degree of nickel extraction as the response have been considered. To model the process, the multiple linear regression （MLR） and response surface method （RSM）, together with the two-level and four-factor full factorial central composite design （CCD） plan, were used. The proposed re- gression models have not been proven adequate. Therefore, the artificial neural network （ANN） approach with the same experimental plan was used in order to reduce operational costs, give a better modeling accuracy, and provide a more successful process optimization. The model is based on the multi-layer neural networks with the back-propagation （BP） learning algorithm and the bipolar sigmoid activation function.
基金This research was supported by the KISTI Program(No.K-20-L02-C05-S01)the EDISON Program through the National Research Foundation of Korea(NRF)(No.NRF-2011-0020576).A grant was also awarded by the Ministry of Science and ICT(MSIT)under the Program for Returners for R&D.
文摘With the increase in research on AI(Artificial Intelligence),the importance of DL(Deep Learning)in various fields,such as materials,biotechnology,genomes,and new drugs,is increasing significantly,thereby increasing the number of deep-learning framework users.However,to design a deep neural network,a considerable understanding of the framework is required.To solve this problem,a GUI(Graphical User Interface)-based DNN(Deep Neural Network)design tool is being actively researched and developed.The GUI-based DNN design tool can design DNNs quickly and easily.However,the existing GUI-based DNN design tool has certain limitations such as poor usability,framework dependency,and difficulty encountered in changing GUI components.In this study,a deep learning algorithm that solves the problem of poor usability was developed using a template to increase the accessibility for users.Moreover,the proposed tool was developed to save and share only the necessary parts for quick operation.To solve the framework dependency,we applied ONNX(Open Neural Network Exchange),which is an exchange standard for neural networks,and configured it such that DNNs designed with the existing deep-learning framework can be imported.Finally,to address the difficulty encountered in changing GUI components,we defined and developed the JSON format to quickly respond to version updates.The developed DL neural network designer was validated by running it with KISTI’s supercomputer-based AI Studio.
文摘Based on the method of artificial neural network, a new approach has been devised to predict the mechanical property of E4303 electrode. The outlined predication model for determining the mechanical property of electrode was built upon the production data. The research leverages a back propagation algorithm as the neural network’s learning rule. The result indicates that there are positive correlations between the predicted results and the practical production data. Hence, using the neural network, predication of electrode property can be realized. For the first time, this research provides a more scientific method for designing electrode.
基金supported by the research funds from the University of Ulsan in South Korea during the financial year 2012–2013
文摘Organic matters（OMs） and their oxidization products often influence the fate and transport of heavy metals in the subsurface aqueous systems through interaction with the mineral surfaces. This study investigates the ethanol（EtO H）-mediated As（Ⅲ） adsorption onto Zn-loaded pinecone（PC） biochar through batch experiments conducted under Box–Behnken design. The effect of EtO H on As（Ⅲ） adsorption mechanism was quantitatively elucidated by fitting the experimental data using artificial neural network and quadratic modeling approaches. The quadratic model could describe the limiting nature of EtO H and pH on As（Ⅲ） adsorption,whereas neural network revealed the stronger influence of Et OH（64.5%） followed by pH（20.75%）and As（Ⅲ） concentration（14.75%） on the adsorption phenomena. Besides, the interaction among process variables indicated that Et OH enhances As（Ⅲ） adsorption over a pH range of2 to 7, possibly due to facilitation of ligand–metal（Zn） binding complexation mechanism.Eventually, hybrid response surface model–genetic algorithm（RSM–GA） approach predicted a better optimal solution than RSM, i.e., the adsorptive removal of As（Ⅲ）（10.47 μg/g） is facilitated at 30.22 mg C/L of Et OH with initial As（Ⅲ） concentration of 196.77 μg/L at pH 5.8. The implication of this investigation might help in understanding the application of biochar for removal of various As（Ⅲ） species in the presence of OM.
