• 유체기계공업학회:학술대회논문집
• /
• 2006.08a
• /
• pp.367-370
• /
• 2006

Performances of multiple surrogate models are evaluated in a turbomachinery blade shape optimization. The basic models, i.e., Response Surface Approximation, Kriging and Radial Basis Neural Network models as well as weighted average models are tested for shape optimization. Global data based errors for each surrogates are used to calculate the weights. These weights are multiplied with the respective surrogates to get the final weighted average models. The design points are selected using three level fractional factorial D-optimal designs. The present approach can help address the multi-objective design on a rational basis with quantifiable cost-benefit analysis.

• Journal of the Korean Society of Manufacturing Process Engineers
• /
• v.10 no.2
• /
• pp.16-25
• /
• 2011

A Turbo-impeller is widely used in industries as well as in aero engines. Its design technology has been developed since the early 20th century. However, the final configuration of the impeller depends on the designers. In this study, a whole design process was introduced and an optimization method to design an impeller was studied in order to design a better impeller without influence by designers. In particular, as the Artificial Neural Network was applied to the optimization, the computational time for the optimization was equivalent to the time consumed by the gradient method and its result was guaranteed as the optimum in the whole design domain. Using this method, any impeller can be improved by selecting design variables after measuring profiles of the impeller.

• Computers and Concrete
• /
• v.11 no.4
• /
• pp.337-350
• /
• 2013

The sensitivity of compressive strength of no-slump concrete to its ingredient materials and proportions, necessitate the use of robust models to guarantee both estimation and generalization features. It was known that the problem of compressive strength prediction owes high degree of complexity and uncertainty due to the variable nature of materials, workmanship quality, etc. Moreover, using the chemical and mineral additives, superimposes the problem's complexity. Traditionally this property of concrete is predicted by conventional linear or nonlinear regression models. In general, these models comprise lower accuracy and in most cases they fail to meet the extrapolation accuracy and generalization requirements. Recently, artificial intelligence-based robust systems have been successfully implemented in this area. In this regard, this paper aims to investigate the use of optimized support vector machine (SVM) to predict the compressive strength of no-slump concrete and compare with optimized neural network (ANN). The results showed that after optimization process, both models are applicable for prediction purposes with similar high-qualities of estimation and generalization norms; however, it was indicated that optimization and modeling with SVM is very rapid than ANN models.

• Smart Structures and Systems
• /
• v.29 no.5
• /
• pp.677-691
• /
• 2022

Water cycle algorithm (WCA) has been a very effective optimization technique for complex engineering problems. This study employs the WCA for simultaneous prediction of heating load (L_H) and cooling load (L_C) in residential buildings. This algorithm is responsible for optimally tuning a neural network (NN). Utilizing 614 records, the behavior of the L_H and L_C is explored and the captured knowledge is then used to predict for 154 unanalyzed building conditions. Since the WCA is a population-based algorithm, different numbers of the searching agents were tested to find the most optimum configuration. It was observed that the best solution is discovered by 500 agents. A comparison with five newly-developed benchmark optimizers, namely equilibrium optimizer (EO), multi-tracker optimization algorithm (MTOA), slime mould algorithm (SMA), multi-verse optimizer (MVO), and electromagnetic field optimization (EFO) revealed that the WCANN predicts the desired parameters with considerably larger accuracy. Obtained root mean square errors (1.4866, 2.1296, 2.8279, 2.5727, 2.5337, and 2.3029 for the L_H and 2.1767, 2.6459, 3.1821, 2.9732, 2.9616, and 2.6890 for the L_C) indicated that the most reliable prediction was presented by the proposed model. The EFONN, however, provided a more time-effective solution. Lastly, an explicit predictive formula was elicited from the WCANN.

• Journal of the Korean Society of Hazard Mitigation
• /
• v.10 no.6
• /
• pp.81-87
• /
• 2010

In order to calculate the accurate runoff from a basin, nonlinearity in the relationship between rainfall and runoff has to be considered. Many runoff calculation models assume the linearity in the relationship or are too complicated to be analyzed. Therefore, the storage function method has been used in the prediction of flood because of the simplicity of the model. The storage function method has five parameters with related to the basin and rainfall characteristics which can be estimated by the empirical trial and error method. To optimize these parameters, regression method or optimization techniques such as genetic algorithm have been used, however, it is not easy to optimize them because of the complexity of the method. In this study, the metamodel is proposed to estimate those model parameters. The metamodel is the combination of artificial neural network and genetic algorithm. The model is consisted of two stages. In the first stage, an artificial neural network is constructed using the given rainfall-runoff relationship. In the second stage, the parameters of the storage function method are estimated using genetic algorithm and the trained artificial neural network. The proposed metamodel is applied in the Peong Chang River basin and the results are presented.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.59 no.2
• /
• pp.436-444
• /
• 2010

In this study, we introduce and discuss a concept of a granular-oriented radial basis function neural networks (GRBF NNs). In contrast to the typical architectures encountered in radial basis function neural networks(RBF NNs), our main objective is to develop a design strategy of GRBF NNs as follows : (a) The architecture of the network is fully reflective of the structure encountered in the training data which are granulated with the aid of clustering techniques. More specifically, the output space is granulated with use of K-Means clustering while the information granules in the multidimensional input space are formed by using a so-called context-based Fuzzy C-Means which takes into account the structure being already formed in the output space, (b) The innovative development facet of the network involves a dynamic reduction of dimensionality of the input space in which the information granules are formed in the subspace of the overall input space which is formed by selecting a suitable subset of input variables so that the this subspace retains the structure of the entire space. As this search is of combinatorial character, we use the technique of genetic optimization to determine the optimal input subspaces. A series of numeric studies exploiting some nonlinear process data and a dataset coming from the machine learning repository provide a detailed insight into the nature of the algorithm and its parameters as well as offer some comparative analysis.

• KIEAE Journal
• /
• v.17 no.5
• /
• pp.69-76
• /
• 2017

Purpose: This study aimed at developing an Artificial Neural Network (ANN) model for predicting the amount of cooling energy consumption of the variable refrigerant flow (VRF) cooling system by the different set-points of the control variables, such as supply air temperature of air handling unit (AHU), condenser fluid temperature, condenser fluid pressure, and refrigerant evaporation temperature. Applying the predicted results for the different set-points, the control algorithm, which embedded the ANN model, will determine the most energy efficient control strategy. Method: The ANN model was developed and tested its prediction accuracy by using matrix laboratory (MATLAB) and its neural network toolbox. The field data sets were collected for the model training and performance evaluation. For completing the prediction model, three major steps were conducted - i) initial model development including input variable selection, ii) model optimization, and iii) performance evaluation. Result: Eight meaningful input variables were selected in the initial model development such as outdoor temperature, outdoor humidity, indoor temperature, cooling load of the previous cycle, supply air temperature of AHU, condenser fluid temperature, condenser fluid pressure, and refrigerant evaporation temperature. The initial model was optimized to have 2 hidden layers with 15 hidden neurons each, 0.3 learning rate, and 0.3 momentum. The optimized model proved its prediction accuracy with stable prediction results.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.21 no.2
• /
• pp.117-124
• /
• 2021

This study is a study to recognize hand movements using a neural network for wrist rehabilitation. The rehabilitation of the hand aims to restore the function of the injured hand to the maximum and enable daily life, occupation, and hobby. It is common for a physical therapist, an occupational therapist, and a security tool maker to form a team and approach a doctor for a hand rehabilitation. However, it is very inefficient economically and temporally to find a place for treatment. In order to solve this problem, in this study, patients directly use smart devices to perform rehabilitation treatment. Using this will be very helpful in terms of cost and time. In this study, a wrist rehabilitation dataset was created by collecting data on 4 types of rehabilitation exercises from 10 persons. Hand gesture recognition was constructed using a neural network. As a result, the accuracy of 93% was obtained, and the usefulness of this system was verified.

• Design & Manufacturing
• /
• v.17 no.4
• /
• pp.1-7
• /
• 2023

In the fabrication of curved multi-display glass for automotive use, the surface roughness of the mold is a critical quality factor. However, the difficulty in detecting micro-cutting signals in a micro-machining environment and the absence of a standardized model for predicting micro-cutting forces make it challenging to intuitively infer the correlation between cutting variables and actual surface roughness under machining conditions. Consequently, current practices heavily rely on machining condition optimization through the utilization of cutting models and experimental research for force prediction. To overcome these limitations, this study employs a surface roughness prediction formula instead of a cutting force prediction model and converts the surface roughness prediction formula into experimental data. Additionally, to account for changes in surface roughness during machining runtime, the theory of position variables has been introduced. By leveraging artificial neural network technology, the accuracy of the surface roughness prediction formula model has improved by 98%. Through the application of artificial neural network technology, the surface roughness prediction formula model, with enhanced accuracy, is anticipated to reliably perform the derivation of optimal machining conditions and the prediction of surface roughness in various machining environments at the analytical stage.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.7 no.2
• /
• pp.132-137
• /
• 2007

Experimental software data capturing the essence of software projects (expressed e.g., in terms of their complexity and development time) have been a subject of intensive modeling. In this study, we introduce a new category of Hybrid Fuzzy Neural Networks (gHFNN) and discuss their comprehensive design methodology. The gHFNN architecture results from highly synergistic linkages between Fuzzy Neural Networks (FNN) and Polynomial Neural Networks (PNN). We develop a rule-based model consisting of a number of "if-then" statements whose antecedents are formed in the input space and linked with the consequents (conclusion pats) formed in the output space. In this framework, FNNs contribute to the formation of the premise part of the overall network structure of the gHFNN. The consequences of the rules are designed with the aid of genetically endowed PNNs. The experiments reported in this study deal with well-known software data such as the NASA dataset. In comparison with the previously discussed approaches, the proposed self-organizing networks are more accurate and yield significant generalization abilities.