Feedforward multi layer neural networks have very strong mapping capability that is based on the non linearity of the activation function, however, the non linearity of the activation function can cause the multiple ...Feedforward multi layer neural networks have very strong mapping capability that is based on the non linearity of the activation function, however, the non linearity of the activation function can cause the multiple local minima on the learning error surfaces, which affect the learning rate and solving optimal weights. This paper proposes a learning method linearizing non linearity of the activation function and discusses its merits and demerits theoretically.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
In this paper, we use the global search characteristics of genetic algorithms to help search the weight space of the neurons in the cascade-correlation architecture. The cascade-correlation learning architecture is a ...In this paper, we use the global search characteristics of genetic algorithms to help search the weight space of the neurons in the cascade-correlation architecture. The cascade-correlation learning architecture is a technique of training and building neural networks that starts with a simple network of neurons and adds additional neurons as they are needed to suit a particular problem. In our approach, instead ofmodifying the genetic algorithm to account for convergence problems, we search the weight-space using the genetic algorithm and then apply the gradient technique of Quickprop to optimize the weights. This hybrid algorithm which is a combination of genetic algorithms and cascade-correlation is applied to the two spirals problem. We also use our algorithm in the prediction of the cyclic oxidation resistance of Ni- and Co-base superalloys.展开更多
In terms of 34-year monthly mean temperature series in 1946-1979,the multi-level maPPing model of neural netWork BP type was applied to calculate the system's fractual dimension Do=2'8,leading tO a three-level...In terms of 34-year monthly mean temperature series in 1946-1979,the multi-level maPPing model of neural netWork BP type was applied to calculate the system's fractual dimension Do=2'8,leading tO a three-level model of this type with ixj=3x2,k=l,and the 1980 monthly mean temperture predichon on a long-t6rm basis were prepared by steadily modifying the weighting coefficient,making for the correlation coefficient of 97% with the measurements.Furthermore,the weighhng parameter was modified for each month of 1980 by means of observations,therefore constrcuhng monthly mean temperature forecasts from January to December of the year,reaching the correlation of 99.9% with the measurements.Likewise,the resulting 1981 monthly predictions on a long-range basis with 1946-1980 corresponding records yielded the correlahon of 98% and the month-tO month forecasts of 99.4%.展开更多
A new type of ANN (Artificial Neural Network) structure is introduced, and a nonlinear transformation of the original features is proposed so as to improve the learning covergence of the neural network. This kind of i...A new type of ANN (Artificial Neural Network) structure is introduced, and a nonlinear transformation of the original features is proposed so as to improve the learning covergence of the neural network. This kind of improved ANN is then used to analyse the transient stability of two real power systems. The results show that this method possesses better effectiveness and high convergence speed.展开更多
This paper analyses the error sources in neural network prediction. The relationship between prediction error and quality of training sets is revealed. The influence of quality of training sets on the performance of a...This paper analyses the error sources in neural network prediction. The relationship between prediction error and quality of training sets is revealed. The influence of quality of training sets on the performance of an artificial neural network(ANN) applied in time series prediction is discussed. A numerical criterion called degree of consistency(DCT) defined from the statistical point of view for evaluating quality of training sets is introduced. Some simulation results and corresponding suggestions are presented along with the new criterion in order to properly select the training sets for neural network training.展开更多
The results of an expert system of lanthanide intermetallic compounds using artificial neural networks and chemical bond parameter method were reported. Two pattern recognition neural models, one for prediction of the...The results of an expert system of lanthanide intermetallic compounds using artificial neural networks and chemical bond parameter method were reported. Two pattern recognition neural models, one for prediction of the occurrence of 1 : 1 lanthanide intermetallic compounds with CsClstructure and the other for prediction of congruent or incongruent melting types, were developed. Four regression neural models were also developed for prediction of melting point of these compounds. In order to get rid of overfitting, cross-vahdation method was used for the neural models. And satisfactory results were obtained in all of the neural models in this paper.展开更多
In the design of deflection coil, we have to determine those values of descriptive parameters of the deflection coil that will give us a required performance. In this paper, an artificial neural network is used in th...In the design of deflection coil, we have to determine those values of descriptive parameters of the deflection coil that will give us a required performance. In this paper, an artificial neural network is used in the design of deflection coil. It is shown that the artificial neural network is indeed possible to develop well-trained networks for designing a particular deflection coil.展开更多
There are put forward in this paper the principle of using artificial neural networks to determine the featureswhich reflect seismotectonics and seismicity of the potential focal regions and the method to divide quant...There are put forward in this paper the principle of using artificial neural networks to determine the featureswhich reflect seismotectonics and seismicity of the potential focal regions and the method to divide quantitativelythe potential focal regions. The calculation of the actual data of North China area shows that it may fully reveal the relations between the potential focal regions and their controling characteristics by the method, of which the principle is concise and convenient to apply, the calculated results reasonable, the division in a meticulous way and the achievement practicable.展开更多
Artificial neural networks(ANNs)have been investigated for application to robotic welding process.Two types of the ANN models are described.The first is a static modeling approach for the pre-setting of robotic weldin...Artificial neural networks(ANNs)have been investigated for application to robotic welding process.Two types of the ANN models are described.The first is a static modeling approach for the pre-setting of robotic welding parameters, and the other is a dynamic modelling for real time feedback control of robotic welding.These models map the relationship between the weld bead geometry and welding process parameters.Some basic concepts relating to neural networks are discussed. The performance of neural networks for modelling is discussed and evaluated by using actual robotic welding data.It is concluded that neural network is capable of modeling readily and quickly a multivariable welding process and the accuracy of neural networks modelling is comparable with the accuracy achieved by the statistical scheme. The choice between ANN and statistical models will depend on the application and control strategy used.展开更多
This paper discusses the modeling method of time series with neural network. In order to improve the adaptability of direct multi-step prediction models, this paper proposes a method of combining the temporal differen...This paper discusses the modeling method of time series with neural network. In order to improve the adaptability of direct multi-step prediction models, this paper proposes a method of combining the temporal differences methods with back-propagation algorithm for updating the parameters continuously on the basis of recent data. This method can make the neural network model fit the recent characteristic of the time series as close as possible, therefore improves the prediction accuracy. We built models and made predictions for the sunspot series. The prediction results of adaptive modeling method are better than that of non-adaptive modeling methods.展开更多
Using expert systems in intelligent CAD of electrical machines have limitations such as knowledge acquisition bottlenecks and matching conflict, combinatorial explosion, and endless recursion in the reasoning process....Using expert systems in intelligent CAD of electrical machines have limitations such as knowledge acquisition bottlenecks and matching conflict, combinatorial explosion, and endless recursion in the reasoning process. This paper discusses the principle of a hybrid system of a neural network and an expert system (HNNES), i.e., knowledge representation, reasoning mechanism, and knowledge acquisition based on neural networks. An architecture of HNNES is presented in consideration of the feature of the design of electrical machines.展开更多
文摘Feedforward multi layer neural networks have very strong mapping capability that is based on the non linearity of the activation function, however, the non linearity of the activation function can cause the multiple local minima on the learning error surfaces, which affect the learning rate and solving optimal weights. This paper proposes a learning method linearizing non linearity of the activation function and discusses its merits and demerits theoretically.
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.
文摘In this paper, we use the global search characteristics of genetic algorithms to help search the weight space of the neurons in the cascade-correlation architecture. The cascade-correlation learning architecture is a technique of training and building neural networks that starts with a simple network of neurons and adds additional neurons as they are needed to suit a particular problem. In our approach, instead ofmodifying the genetic algorithm to account for convergence problems, we search the weight-space using the genetic algorithm and then apply the gradient technique of Quickprop to optimize the weights. This hybrid algorithm which is a combination of genetic algorithms and cascade-correlation is applied to the two spirals problem. We also use our algorithm in the prediction of the cyclic oxidation resistance of Ni- and Co-base superalloys.
文摘In terms of 34-year monthly mean temperature series in 1946-1979,the multi-level maPPing model of neural netWork BP type was applied to calculate the system's fractual dimension Do=2'8,leading tO a three-level model of this type with ixj=3x2,k=l,and the 1980 monthly mean temperture predichon on a long-t6rm basis were prepared by steadily modifying the weighting coefficient,making for the correlation coefficient of 97% with the measurements.Furthermore,the weighhng parameter was modified for each month of 1980 by means of observations,therefore constrcuhng monthly mean temperature forecasts from January to December of the year,reaching the correlation of 99.9% with the measurements.Likewise,the resulting 1981 monthly predictions on a long-range basis with 1946-1980 corresponding records yielded the correlahon of 98% and the month-tO month forecasts of 99.4%.
文摘A new type of ANN (Artificial Neural Network) structure is introduced, and a nonlinear transformation of the original features is proposed so as to improve the learning covergence of the neural network. This kind of improved ANN is then used to analyse the transient stability of two real power systems. The results show that this method possesses better effectiveness and high convergence speed.
文摘This paper analyses the error sources in neural network prediction. The relationship between prediction error and quality of training sets is revealed. The influence of quality of training sets on the performance of an artificial neural network(ANN) applied in time series prediction is discussed. A numerical criterion called degree of consistency(DCT) defined from the statistical point of view for evaluating quality of training sets is introduced. Some simulation results and corresponding suggestions are presented along with the new criterion in order to properly select the training sets for neural network training.
文摘The results of an expert system of lanthanide intermetallic compounds using artificial neural networks and chemical bond parameter method were reported. Two pattern recognition neural models, one for prediction of the occurrence of 1 : 1 lanthanide intermetallic compounds with CsClstructure and the other for prediction of congruent or incongruent melting types, were developed. Four regression neural models were also developed for prediction of melting point of these compounds. In order to get rid of overfitting, cross-vahdation method was used for the neural models. And satisfactory results were obtained in all of the neural models in this paper.
文摘In the design of deflection coil, we have to determine those values of descriptive parameters of the deflection coil that will give us a required performance. In this paper, an artificial neural network is used in the design of deflection coil. It is shown that the artificial neural network is indeed possible to develop well-trained networks for designing a particular deflection coil.
文摘There are put forward in this paper the principle of using artificial neural networks to determine the featureswhich reflect seismotectonics and seismicity of the potential focal regions and the method to divide quantitativelythe potential focal regions. The calculation of the actual data of North China area shows that it may fully reveal the relations between the potential focal regions and their controling characteristics by the method, of which the principle is concise and convenient to apply, the calculated results reasonable, the division in a meticulous way and the achievement practicable.
文摘Artificial neural networks(ANNs)have been investigated for application to robotic welding process.Two types of the ANN models are described.The first is a static modeling approach for the pre-setting of robotic welding parameters, and the other is a dynamic modelling for real time feedback control of robotic welding.These models map the relationship between the weld bead geometry and welding process parameters.Some basic concepts relating to neural networks are discussed. The performance of neural networks for modelling is discussed and evaluated by using actual robotic welding data.It is concluded that neural network is capable of modeling readily and quickly a multivariable welding process and the accuracy of neural networks modelling is comparable with the accuracy achieved by the statistical scheme. The choice between ANN and statistical models will depend on the application and control strategy used.
文摘This paper discusses the modeling method of time series with neural network. In order to improve the adaptability of direct multi-step prediction models, this paper proposes a method of combining the temporal differences methods with back-propagation algorithm for updating the parameters continuously on the basis of recent data. This method can make the neural network model fit the recent characteristic of the time series as close as possible, therefore improves the prediction accuracy. We built models and made predictions for the sunspot series. The prediction results of adaptive modeling method are better than that of non-adaptive modeling methods.
文摘Using expert systems in intelligent CAD of electrical machines have limitations such as knowledge acquisition bottlenecks and matching conflict, combinatorial explosion, and endless recursion in the reasoning process. This paper discusses the principle of a hybrid system of a neural network and an expert system (HNNES), i.e., knowledge representation, reasoning mechanism, and knowledge acquisition based on neural networks. An architecture of HNNES is presented in consideration of the feature of the design of electrical machines.
