• International Journal of Naval Architecture and Ocean Engineering
• /
• v.4 no.3
• /
• pp.199-210
• /
• 2012

The recent researches on the automatic berthing control problems have used various kinds of tools as a control method such as expert system, fuzzy logic controllers and artificial neural network (ANN). Among them, ANN has proved to be one of the most effective and attractive options. In a marine context, the berthing maneuver is a complicated procedure in which both human experience and intensive control operations are involved. Nowadays, in most cases of berthing operation, auxiliary devices are used to make the schedule safer and faster but none of above researches has taken into account. In this study, ANN is applied to design the controllers for automatic ship berthing using assistant devices such as bow thruster and tug. Using back-propagation algorithm, we trained ANN with set of teaching data to get a minimal error between output values and desired values of four control outputs including rudder, propeller revolution, bow thruster and tug. Then, computer simulations of automatic berthing were carried out to verify the effectiveness of the system. The results of the simulations showed good performance for the proposed berthing control system.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1998.10a
• /
• pp.173-179
• /
• 1998

This paper proposes a new call admission control scheme utilizing an inverse fuzzy vector quantizer(IFVQ) and neural net, which combines benefits of IFVQ and flexibilities of FCM(Fuzzy-C-Menas) arithmatics, to decide whether a requested call that is not trained in learning phase to be connected or not. The system generates the estimated traffic pattern of the cell stream of a new call, using feasible/infeasible patterns in codebook, fuzzy membership values that represent the degree to which each pattern of codebook matches input pattern, and FCM arithmatics. The input to the NN is the vector consisted of traffic parameters which is the means and variances of the number of cells arriving inthe interval. After training(using error back propagation algorithm), when the NN is used for decision making, the decision as to whether to accept or reject a new call depends on whether the output is greater or less then decision threshold(+0.5). This method is a new technique for call admi sion control using the membership values as traffic parameter which declared to CAC at the call set up stage, and is valid for a very general traffic model in which the calls of a stream can belong to an unlimited number of traffic classes. Through the simmulation. it is founded the performance of the suggested method outforms compared to the conventional NN method.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 1999.11a
• /
• pp.244-247
• /
• 1999

in this paper, we propose the recognition system of face using polynomial coefficients to recognize fact images using neural network. The system consists of following steps. First step, the sizes of fare images is reduced sizes of input images to 1/4 using wavelet transform. Second step, the polynomial coefficients is obtained from low frequency coefficient matrix after 3 level wavelet transform. Third step, polynomial coefficients is normalized. The of range of normalization is from -1 to 1. Last, Face images is trained and recognized using neural network with error back propagation algorithm.

• Proceedings of the Korean Institute of Navigation and Port Research Conference
• /
• v.1
• /
• pp.491-494
• /
• 2006

Kalman filtering (KF) is hard to be applied to the SINS (Strap-down Inertial Navigation System)/RDSS (Radio Determination Satellite Service) integrated navigation system directly because the time delay of RDSS positioning in active mode is random. BP (Back-Propagation) Neuron computing as a powerful technology of Artificial Neural Network (ANN), is appropriate to solve nonlinear problems such as the random time delay of RDSS without prior knowledge about the mathematical process involved. The new algorithm betakes a BP neural network (BPNN) and velocity feedback to aid KF in order to overcome the time delay of RDSS positioning. Once the BP neural network was trained and converged, the new approach will work well for SINS/RDSS integrated navigation. Dynamic vehicle experiments were performed to evaluate the performance of the system. The experiment results demonstrate that the horizontal positioning accuracy of the new approach is 40.62 m (1 ${\sigma}$), which is better than velocity-feedback-based KF. The experimental results also show that the horizontal positioning error of the navigation system is almost linear to the positioning interval of RDSS within 5 minutes. The approach and its anti-jamming analysis will be helpful to the applications of SINS/RDSS integrated systems.

• The Transactions of the Korean Institute of Electrical Engineers
• /
• v.40 no.2
• /
• pp.224-233
• /
• 1991

The multilayer expanson of single layer NN (Neural Network) was needed to solve the linear seperability problem as shown by the classic example using the XOR function. The EBP (Error Back Propagation ) learning rule is often used in multilayer Neural Networks, but it is not without its faults: 1)D.Rimmelhart expanded the Delta Rule but there is a problem in obtaining Ca from the linear combination of the Weight matrix N between the hidden layer and the output layer and H, wich is the result of another linear combination between the input pattern and the Weight matrix M between the input layer and the hidden layer. 2) Even if using the difference between Ca and Da to adjust the values of the Weight matrix N between the hidden layer and the output layer may be valid is correct, but using the same value to adjust the Weight matrixd M between the input layer and the hidden layer is wrong. Recognitron III was proposed to solve these faults. According to simulation results, since Recognitron III does not learn the three layer NN itself, but divides it into several single layer NNs and learns these with learning patterns, the learning time is 32.5 to 72.2 time faster than EBP NN one. The number of patterns learned in a EBP NN with n input and output cells and n+1 hidden cells are 2**n, but n in Recognitron III of the same size. [5] In the case of pattern generalization, however, EBP NN is less than Recognitron III.

• International Journal of Automotive Technology
• /
• v.6 no.2
• /
• pp.95-105
• /
• 2005

Diesel engines have low specific fuel consumption, but high particulate emissions, mainly soot. Diesel soot is suspected to have significant effects on the health of living beings and might also affect global warming. Hence stringent measures have been put in place in a number of countries and will be even stronger in the near future. Diesel engines require either advanced integrated exhaust after treatment systems or modified engine models to meet the statutory norms. Experimental analysis to study the emission characteristics is a time consuming affair. In such situations, the real picture of engine control can be obtained by the modeling of trend prediction. In this article, an effort has been made to predict emissions smoke and NO$_{x}$ using cylinder combustion derived parameters and diesel particulate filter data, with artificial neural network techniques in MATLAB environment. The model is based on three layer neural network with a back propagation learning algorithm. The training and test data of emissions were collected from experimental set up in the laboratory for different loads. The network is trained to predict the values of emission with training values. Regression analysis between test and predicted value from neural network shows least error. This approach helps in the reduction of the experimentation required to determine the smoke and NO$_{x}$ for the catalyst coated filters.

• Proceedings of the KIEE Conference
• /
• 2001.07d
• /
• pp.2286-2288
• /
• 2001

In this paper, a drive system of induction motor with high performance is realized on the viewpoint of the design and experiment, using the DSP (TMS320F240). The speed controller for induction motor drive system is designed on the basis of a neuro-fuzzy network. The neuro-fuzzy controller acts as a feed-forward controller that provides the right control input for the plant and accomplishes error back-propagation algorithm through the network. The proposed network is used to achieve the high speedy calculation of the space vector PWM (Pulse Width Modulation) and to build the neuro-fuzzy control algorithm, for the real-time control. The proposed neuro-fuzzy algorithm on the basis of DSP shows that experimental results have good performance for the precise speed control of an induction motor drive system. It is confirmed that the proposed controller could provide more improved control performance than conventional v/f vector controllers through the experiment.

• Proceedings of the KIEE Conference
• /
• 2001.07d
• /
• pp.2726-2728
• /
• 2001

This paper describes an optimal synthesis method of binary neural network(BNN) for an approximation problem of a circular region using a newly proposed learning algorithm[7] Our object is to minimize the number of connections and neurons in hidden layer by using a Newly Expanded and Truncated Learning Algorithm(NETLA) for the multilayer BNN. The synthesis method in the NETLA is based on the extension principle of Expanded and Truncated Learning(ETL) and is based on Expanded Sum of Product (ESP) as one of the boolean expression techniques. And it has an ability to optimize the given BNN in the binary space without any iterative training as the conventional Error Back Propagation(EBP) algorithm[6] If all the true and false patterns are only given, the connection weights and the threshold values can be immediately determined by an optimal synthesis method of the NETLA without any tedious learning. Futhermore, the number of the required neurons in hidden layer can be reduced and the fast learning of BNN can be realized. The superiority of this NETLA to other algorithms was proved by the approximation problem of one circular region.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.10 no.3
• /
• pp.554-559
• /
• 2009

This paper proposes a new method of weighted template matching fur machine-printed numeral recognition. The proposed weighted template matching, which emphasizes the feature of a pattern using adaptive Hamming distance on local feature areas, improves the recognition rate while template matching processes an input image as one global feature. The experiment compares confusion matrices of the template matching, error back propagation neural network classifier, and the proposed weighted template matching respectively. The result shows that the proposed method improves fairly the recognition rate of the machine-printed numerals.

• Communications of the Korean Mathematical Society
• /
• v.16 no.3
• /
• pp.445-458
• /
• 2001

The shape and the distribution of material construction of the scatterer may be obtained from its scattered fields by the iterative inversion in the spectral domain. The illposedness, the resolution, and the uniqueness of the inversion are the key problems in the inversion and inter-related. The illposedness is shown to be caused by the evanescent modes which carries and amplifies exponentially the measurement errors in the back-propagation of the measured scattered fields. By filtering out all the evanescent modes in the cost functional defined as the squared difference between the measured and the calculated spatial spectrum of the scattered fields from the iteratively chosen medium parameters of the scatterer, one may regularize the illposedness of the inversion in the expense of the resolution. There exist many local minima of the cost functional for the inversion of the large and the high-contrast scatterer and the hybrid algorithm combining the genetic algorithm and the Levenberg-Marquardt algorithm is shown to find efficiently its global minimum. The resolution of reconstruction obtained by keeping all the propating modes and filtering out the evanescent modes for the regularization becomes 0.5 wavelength. The super resolution may be obtained by keeping the evanescent modes when the measurement error and instance, respectively, are small and near.