• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.25 no.2
• /
• pp.173-181
• /
• 2001

This paper presents an artificial life algorithm which is remarkable in the area of engineering for functions optimization. As artificial life organisms have a sensing system, they can find the resource which they want to find and metabolize. And the characteristics of artificial life are emergence and dynamic interaction with environment. In other words, the micro-interaction with each other in the artificial lifes group results in emergent colonization in the whole system. In this paper, therefore, artificial life algorithm by using above characteristics is employed into functions optimization. The optimizing ability and convergent characteristics of this proposed algorithm is verified by using three test functions. The numerical results also show that the proposed algorithm is superior to genetic algorithms and immune algorithms for the multimodal functions.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.33 no.8
• /
• pp.753-759
• /
• 2009

Feed-forward neural networks have been widely used as function approximation tools in the context of global approximate optimization. In the present study, a wavelet neural network (WNN) which is based on wavelet transform theory is suggested as an alternative to a traditional back-propagation neural network (BPN). The basic theory of wavelet neural network is briefly described, and approximation performance is tested using a nonlinear multimodal function and a composite rotor blade analysis problem. Laplacian of Gaussian function, Mexican function, and Morlet function are considered during the construction of WNN architectures. In addition, approximation results from WNN are compared with those from BPN.

• 제어로봇시스템학회:학술대회논문집
• /
• 1995.10a
• /
• pp.18-21
• /
• 1995

In this paper we describe a new method for multimodal function optimization using genetic algorithms(GAs). We propose adaptation rules for GA parameters such as population size, crossover probability and mutation probability. In the self organizing genetic algorithm(SOGA), SOGA parameters change according to the adaptation rules. Thus, we do not have to set the parameters manually. We discuss about SOGA and those of other approaches for adapting operator probabilities in GAs. The validity of the proposed algorithm will be verified in a simulation example of system identification.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.26 no.6
• /
• pp.1008-1015
• /
• 2002

This paper proposes a combinatorial method to compute the global and local solutions of optimization problem. The present hybrid algorithm is the synthesis of a genetic algorithm and a local concentrate search algorithm (simplex method). The hybrid algorithm is not only faster than the standard genetic algorithm, but also gives a more accurate solution. In addition, this algorithm can find both the global and local optimum solutions. An optimization result is presented to demonstrate that the proposed approach successfully focuses on the advantages of global and local searches. Three numerical examples are also presented in this paper to compare with conventional methods.

• Proceedings of the KIEE Conference
• /
• 2011.07a
• /
• pp.826-827
• /
• 2011

The selection of optimum parameters in electromagnetic design usually requires optimization of multimodal, non linear functions. This leads to extensive calculations which pose a huge inconvenience in the design process. This paper proposes a novel algorithm for dealing efficiently with this issue. Through the use of contour line concept coupled with Kriging, the algorithm finds out all the peaks in the problem domain with as few function calls as possible. The proposed algorithm is applied to the magnet shape optimization of an axial flux permanent magnet synchronous machine and the cogging torque was reduced to 79.8% of the initial one.

• Journal of Institute of Control, Robotics and Systems
• /
• v.9 no.2
• /
• pp.168-175
• /
• 2003

A new algorithm based on Genetic Algorithms is proposed f3r solving process optimization problems formulated in MINLP, GDP and hybrid MINLP/GDP. This work is focused especially on the design of the Genetic Algorithm suitable to handle disjunctive programming with the same level of MINLP handling capability. Hybridization with the Simulated Annealing is experimented and many heuristics are adopted. Real and binary coded Genetic Algorithm initiates the global search in the entire search space and at every stage Simulated Annealing makes the candidates to climb up the local hills. Multi-Niche Crowding method is adopted as the multimodal function optimization technique. and the adaptation of probabilistic parameters and dynamic penalty systems are also implemented. New strategies to take the logical variables and constraints into consideration are proposed, as well. Various test problems selected from many fields of process systems engineering are tried and satisfactory results are obtained.

• Journal of Computing Science and Engineering
• /
• v.9 no.2
• /
• pp.39-50
• /
• 2015

Genetic algorithm (GA) and particle swarm optimization (PSO) are two excellent approaches to multimodal optimization problems. However, slow convergence or premature convergence readily occurs because of inappropriate and inflexible evolution. In this paper, a novel optimization algorithm with a flexible grid optimization (FGO) is suggested to provide adaptive trade-off between exploration and exploitation according to the specific objective function. Meanwhile, a uniform agents array with adaptive scale is distributed on the gird to speed up the calculation. In addition, a dominance centroid and a fitness center are proposed to efficiently determine the potential guides when the population size varies dynamically. Two types of subregion division strategies are designed to enhance evolutionary diversity and convergence, respectively. By examining the performance on four benchmark functions, FGO is found to be competitive with or even superior to several other popular algorithms in terms of both effectiveness and efficiency, tending to reach the global optimum earlier. Moreover, FGO is evaluated by applying it to a parameter decision in a least squares support vector machine (LS-SVM) to verify its practical competence.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.26 no.9
• /
• pp.1806-1813
• /
• 2002

It is well known that a crack has an important effect on the dynamic behavior of a structure. This effect depends mainly on the location and depth of the crack. To identify the location and depth of a crack in a structure, a classical optimization technique was adopted by previous researchers. That technique overcame the difficulty of finding the intersection point of the superposed contours that correspond to the eigenfrequency caused by the crack presence. However, it is hard to select a trial solution initially for optimization because the defined objective function is heavily multimodal. A method is presented in this paper, which uses continuous evolutionary algorithms(CEAs). CEAs are effective for solving inverse problems and implemented on PC clusters to shorten calculation time. With finite element model of the structure to calculate eigenfrequencies, it is possible to formulate the inverse problem in optimization format. CEAs are used to identify the crack location and depth minimizing the difference from the measured frequencies. We have tried this new idea on a simple beam structure and the results are promising with high parallel efficiency over about 94%.

• Proceedings of the KSME Conference
• /
• 2001.11a
• /
• pp.533-538
• /
• 2001

This paper describes a combinational method to compute the global and local solutions of optimization problems. The present hybrid algorithm uses both a genetic algorithm and a local concentrate search algorithm (e. g simplex method). The hybrid algorithm is not only faster than the standard genetic algorithm but also supplies a more accurate solution. In addition, this algorithm can find the global and local optimum solutions. The present algorithm can be supplied to minimize the resonance response (Q factor) and to yield the critical speeds as far from the operating speed as possible. These factors play very important roles in designing a rotor-bearing system under the dynamic behavior constraint. In the present work, the shaft diameter, the bearing length, and clearance are used as the design variables.

• Journal of Intelligence and Information Systems
• /
• v.25 no.1
• /
• pp.163-177
• /
• 2019

As smartphones are getting widely used, human activity recognition (HAR) tasks for recognizing personal activities of smartphone users with multimodal data have been actively studied recently. The research area is expanding from the recognition of the simple body movement of an individual user to the recognition of low-level behavior and high-level behavior. However, HAR tasks for recognizing interaction behavior with other people, such as whether the user is accompanying or communicating with someone else, have gotten less attention so far. And previous research for recognizing interaction behavior has usually depended on audio, Bluetooth, and Wi-Fi sensors, which are vulnerable to privacy issues and require much time to collect enough data. Whereas physical sensors including accelerometer, magnetic field and gyroscope sensors are less vulnerable to privacy issues and can collect a large amount of data within a short time. In this paper, a method for detecting accompanying status based on deep learning model by only using multimodal physical sensor data, such as an accelerometer, magnetic field and gyroscope, was proposed. The accompanying status was defined as a redefinition of a part of the user interaction behavior, including whether the user is accompanying with an acquaintance at a close distance and the user is actively communicating with the acquaintance. A framework based on convolutional neural networks (CNN) and long short-term memory (LSTM) recurrent networks for classifying accompanying and conversation was proposed. First, a data preprocessing method which consists of time synchronization of multimodal data from different physical sensors, data normalization and sequence data generation was introduced. We applied the nearest interpolation to synchronize the time of collected data from different sensors. Normalization was performed for each x, y, z axis value of the sensor data, and the sequence data was generated according to the sliding window method. Then, the sequence data became the input for CNN, where feature maps representing local dependencies of the original sequence are extracted. The CNN consisted of 3 convolutional layers and did not have a pooling layer to maintain the temporal information of the sequence data. Next, LSTM recurrent networks received the feature maps, learned long-term dependencies from them and extracted features. The LSTM recurrent networks consisted of two layers, each with 128 cells. Finally, the extracted features were used for classification by softmax classifier. The loss function of the model was cross entropy function and the weights of the model were randomly initialized on a normal distribution with an average of 0 and a standard deviation of 0.1. The model was trained using adaptive moment estimation (ADAM) optimization algorithm and the mini batch size was set to 128. We applied dropout to input values of the LSTM recurrent networks to prevent overfitting. The initial learning rate was set to 0.001, and it decreased exponentially by 0.99 at the end of each epoch training. An Android smartphone application was developed and released to collect data. We collected smartphone data for a total of 18 subjects. Using the data, the model classified accompanying and conversation by 98.74% and 98.83% accuracy each. Both the F1 score and accuracy of the model were higher than the F1 score and accuracy of the majority vote classifier, support vector machine, and deep recurrent neural network. In the future research, we will focus on more rigorous multimodal sensor data synchronization methods that minimize the time stamp differences. In addition, we will further study transfer learning method that enables transfer of trained models tailored to the training data to the evaluation data that follows a different distribution. It is expected that a model capable of exhibiting robust recognition performance against changes in data that is not considered in the model learning stage will be obtained.