• The Transactions of the Korea Information Processing Society
• /
• v.6 no.5
• /
• pp.1393-1402
• /
• 1999

This paper proposes an efficient principal feature extraction of the image data using neural networks of a new learning algorithm. The proposed learning algorithm is a backpropagation(BP) algorithm based on the steepest descent and dynamic tunneling. The BP algorithm based on the steepest descent is applied for high-speed optimization, and the BP algorithm based on the dynamic tunneling is also applied for global optimization. Converging to the local minimum by the BP algorithm of steepest descent, the new initial weights for escaping the local minimum is estimated by the BP algorithm of dynamic tunneling. The proposed algorithm has been applied to the 3 image data of 12${\times}$12pixels and the Lenna image of 128${\times}$128 pixels respectively. The simulation results shows that the proposed algorithm has better performances of the convergence and the feature extraction, in comparison with those using the Sanger method and the Foldiak method for single-layer neural networks and the BP algorithm for multilayer neural network.

• Journal of the Korean Society of Industry Convergence
• /
• v.18 no.4
• /
• pp.259-266
• /
• 2015

This study presents new scheme for various walking pattern of biped robot under the limitted enviroments. We show that the neural network is significantly more attractive intelligent controller design than previous traditional forms of control systems. A multilayer backpropagation neural network identification is simulated to obtain a learning control solution of biped robot. Once the neural network has learned, the other neural network control is designed for various trajectory tracking control with same learning-base. The main advantage of our scheme is that we do not require any knowledge about the system dynamic and nonlinear characteristic, and can therefore treat the robot as a black box. It is also shown that the neural network is a powerful control theory for various trajectory tracking control of biped robot with same learning-vase. That is, we do net change the control parameter for various trajectory tracking control. Simulation and experimental result show that the neural network is practically feasible and realizable for iterative learning control of biped robot.

• Journal of Biomedical Engineering Research
• /
• v.17 no.3
• /
• pp.387-394
• /
• 1996

This paper presents a Neural Network Identification(NNI) method for modeling of highly complicated nonlinear and time varing human system with a pneumatically driven mock circulatory system of Left Ventricular Assist Device(LVAD). This system consists of electronic circuits and pneumatic driving circuits. The initiation of systole and the pumping duration can be determined by the computer program. The line pressure from a pressure transducer inserted in the pneumatic line was recorded System modeling is completed using the adaptively trained backpropagation learning algorithms with input variables, heart rate(HR), systole-diastole rate(SDR), which can vary state of system. Output parameters are preload, afterload which indicate the systemic dynamic characteristics. Consequently, the neural network shows good approximation of nonlinearity, and characteristics of left Ventricular Assist Device. Our results show that the neural network leads to a significant improvement in the modeling of highly nonlinear Left Ventricular Assist Device.

• Nuclear Engineering and Technology
• /
• v.53 no.10
• /
• pp.3275-3285
• /
• 2021

A Nuclear Power Plant (NPP) is a complex dynamic system-of-systems with highly nonlinear behaviors. In order to control the plant operation under both normal and abnormal conditions, the different systems in NPPs (e.g., the reactor core components, primary and secondary coolant systems) are usually monitored continuously, resulting in very large amounts of data. This situation makes it possible to integrate relevant qualitative and quantitative knowledge with artificial intelligence techniques to provide faster and more accurate behavior predictions, leading to more rapid decisions, based on actual NPP operation data. Data-driven models (DDM) rely on artificial intelligence to learn autonomously based on patterns in data, and they represent alternatives to physics-based models that typically require significant computational resources and might not fully represent the actual operation conditions of an NPP. In this study, a feed-forward backpropagation artificial neural network (ANN) model was trained to simulate the interaction between the reactor core and the primary and secondary coolant systems in a pressurized water reactor. The transients used for model training included perturbations in reactivity, steam valve coefficient, reactor core inlet temperature, and steam generator inlet temperature. Uncertainties of the plant physical parameters and operating conditions were also incorporated in these transients. Eight training functions were adopted during the training stage to develop the most efficient network. The developed ANN model predictions were subsequently tested successfully considering different new transients. Overall, through prompt prediction of NPP behavior under different transients, the study aims at demonstrating the potential of artificial intelligence to empower rapid emergency response planning and risk mitigation strategies.

• Journal of Intelligence and Information Systems
• /
• v.22 no.2
• /
• pp.127-142
• /
• 2016

Deep learning model is a kind of neural networks that allows multiple hidden layers. There are various deep learning architectures such as convolutional neural networks, deep belief networks and recurrent neural networks. Those have been applied to fields like computer vision, automatic speech recognition, natural language processing, audio recognition and bioinformatics where they have been shown to produce state-of-the-art results on various tasks. Among those architectures, convolutional neural networks and recurrent neural networks are classified as the supervised learning model. And in recent years, those supervised learning models have gained more popularity than unsupervised learning models such as deep belief networks, because supervised learning models have shown fashionable applications in such fields mentioned above. Deep learning models can be trained with backpropagation algorithm. Backpropagation is an abbreviation for "backward propagation of errors" and a common method of training artificial neural networks used in conjunction with an optimization method such as gradient descent. The method calculates the gradient of an error function with respect to all the weights in the network. The gradient is fed to the optimization method which in turn uses it to update the weights, in an attempt to minimize the error function. Convolutional neural networks use a special architecture which is particularly well-adapted to classify images. Using this architecture makes convolutional networks fast to train. This, in turn, helps us train deep, muti-layer networks, which are very good at classifying images. These days, deep convolutional networks are used in most neural networks for image recognition. Convolutional neural networks use three basic ideas: local receptive fields, shared weights, and pooling. By local receptive fields, we mean that each neuron in the first(or any) hidden layer will be connected to a small region of the input(or previous layer's) neurons. Shared weights mean that we're going to use the same weights and bias for each of the local receptive field. This means that all the neurons in the hidden layer detect exactly the same feature, just at different locations in the input image. In addition to the convolutional layers just described, convolutional neural networks also contain pooling layers. Pooling layers are usually used immediately after convolutional layers. What the pooling layers do is to simplify the information in the output from the convolutional layer. Recent convolutional network architectures have 10 to 20 hidden layers and billions of connections between units. Training deep learning networks has taken weeks several years ago, but thanks to progress in GPU and algorithm enhancement, training time has reduced to several hours. Neural networks with time-varying behavior are known as recurrent neural networks or RNNs. A recurrent neural network is a class of artificial neural network where connections between units form a directed cycle. This creates an internal state of the network which allows it to exhibit dynamic temporal behavior. Unlike feedforward neural networks, RNNs can use their internal memory to process arbitrary sequences of inputs. Early RNN models turned out to be very difficult to train, harder even than deep feedforward networks. The reason is the unstable gradient problem such as vanishing gradient and exploding gradient. The gradient can get smaller and smaller as it is propagated back through layers. This makes learning in early layers extremely slow. The problem actually gets worse in RNNs, since gradients aren't just propagated backward through layers, they're propagated backward through time. If the network runs for a long time, that can make the gradient extremely unstable and hard to learn from. It has been possible to incorporate an idea known as long short-term memory units (LSTMs) into RNNs. LSTMs make it much easier to get good results when training RNNs, and many recent papers make use of LSTMs or related ideas.

• Proceedings of the Korean Society for Agricultural Machinery Conference
• /
• 2017.04a
• /
• pp.135-135
• /
• 2017

Greenhouse have been developed to provide the plants with good environmental conditions for cultivation crop, two major factors of which are the inside air temperature and humidity. The inside temperature are influenced by the heating systems, ventilators and for systems among others, which in turn are geverned by some type of controller. Likewise, humidity environment is the result of complex mass exchanges between the inside air and the several elements of the greenhouse and the outside boundaries. Most of the existing models are based on the energy balance method and heat balance equation for modelling the heat and mass fluxes and generating dynamic elements. However, greenhouse are classified as complex system, and need to make a sophisticated modeling. Furthermore, there is a difficulty in using classical control methods for complex process system due to the process are non linear and multi-output(MIMO) systems. In order to predict the time evolution of conditions in certain greenhouse as a function, we present here to use of recurrent neural networks(RNN) which has been used to implement the direct dynamics of the inside temperature and inside humidity of greenhouse. For the training, we used algorithm of a backpropagation Through Time (BPTT). Because the environmental parameters are shared by all time steps in the network, the gradient at each output depends not only on the calculations of the current time step, but also the previous time steps. The training data was emulated to 13 input variables during March 1 to 7, and the model was tested with database file of March 8. The RMSE of results of the temperature modeling was $0.976^{\circ}C$, and the RMSE of humidity simulation was 4.11%, which will be given to prove the performance of RNN in prediction of the greenhouse environment.

• The Transactions of the Korea Information Processing Society
• /
• v.4 no.7
• /
• pp.1749-1758
• /
• 1997

This paper suggests the development of dynamic forecasting model for short-term power demand based on Radial Basis Function Network and Pal's GLVQ algorithm. Radial Basis Function methods are often compared with the backpropagation training, feed-forward network, which is the most widely used neural network paradigm. The Radial Basis Function Network is a single hidden layer feed-forward neural network. Each node of the hidden layer has a parameter vector called center. This center is determined by clustering algorithm. Theatments of classical approached to clustering methods include theories by Hartigan(K-means algorithm), Kohonen(Self Organized Feature Maps %3A SOFM and Learning Vector Quantization %3A LVQ model), Carpenter and Grossberg(ART-2 model). In this model, the first approach organizes the load pattern into two clusters by Pal's GLVQ clustering algorithm. The reason of using GLVQ algorithm in this model is that GLVQ algorithm can classify the patterns better than other algorithms. And the second approach forecasts hourly load patterns by radial basis function network which has been constructed two hidden nodes. These nodes are determined from the cluster centers of the GLVQ in first step. This model was applied to forecast the hourly loads on Mar. $4^{th},\;Jun.\;4^{th},\;Jul.\;4^{th},\;Sep.\;4^{th},\;Nov.\;4^{th},$ 1995, after having trained the data for the days from Mar. $1^{th}\;to\;3^{th},\;from\;Jun.\;1^{th}\;to\;3^{th},\;from\;Jul.\;1^{th}\;to\;3^{th},\;from\;Sep.\;1^{th}\;to\;3^{th},\;and\;from\;Nov.\;1^{th}\;to\;3^{th},$ 1995, respectively. In the experiments, the average absolute errors of one-hour ahead forecasts on utility actual data are shown to be 1.3795%.