The Experts below are selected from a list of 23490 Experts worldwide ranked by ideXlab platform
Witold Pedrycz - One of the best experts on this subject based on the ideXlab platform.
-
optimized face recognition algorithm using radial basis function neural networks and its practical applications
Neural Networks, 2015Co-Authors: Sunghoon Yoo, Witold PedryczAbstract:In this study, we propose a hybrid method of face recognition by using face region information extracted from the detected face region. In the preprocessing part, we develop a hybrid approach based on the Active Shape Model (ASM) and the Principal Component Analysis (PCA) algorithm. At this step, we use a CCD (Charge Coupled Device) camera to acquire a facial image by using AdaBoost and then Histogram Equalization (HE) is employed to improve the quality of the image. ASM extracts the face contour and image shape to produce a personal profile. Then we use a PCA method to reduce dimensionality of face images. In the recognition part, we consider the improved Radial Basis Function Neural Networks (RBF NNs) to identify a unique pattern associated with each person. The proposed RBF NN architecture consists of three functional modules realizing the condition phase, the conclusion phase, and the inference phase completed with the help of fuzzy rules coming in the standard 'if-then' format. In the formation of the condition part of the fuzzy rules, the input space is partitioned with the use of Fuzzy C-Means (FCM) clustering. In the conclusion part of the fuzzy rules, the connections (weights) of the RBF NNs are represented by four kinds of polynomials such as constant, linear, quadratic, and reduced quadratic. The values of the Coefficients are determined by running a gradient descent method. The output of the RBF NNs model is obtained by running a fuzzy inference method. The essential design parameters of the network (including learning rate, Momentum Coefficient and fuzzification Coefficient used by the FCM) are optimized by means of Differential Evolution (DE). The proposed P-RBF NNs (Polynomial based RBF NNs) are applied to facial recognition and its performance is quantified from the viewpoint of the output performance and recognition rate.
-
optimized face recognition algorithm using radial basis function neural networks and its practical applications
Neural Networks, 2015Co-Authors: Sungkwun Oh, Witold PedryczAbstract:In this study, we propose a hybrid method of face recognition by using face region information extracted from the detected face region. In the preprocessing part, we develop a hybrid approach based on the Active Shape Model (ASM) and the Principal Component Analysis (PCA) algorithm. At this step, we use a CCD (Charge Coupled Device) camera to acquire a facial image by using AdaBoost and then Histogram Equalization (HE) is employed to improve the quality of the image. ASM extracts the face contour and image shape to produce a personal profile. Then we use a PCA method to reduce dimensionality of face images. In the recognition part, we consider the improved Radial Basis Function Neural Networks (RBF NNs) to identify a unique pattern associated with each person. The proposed RBF NN architecture consists of three functional modules realizing the condition phase, the conclusion phase, and the inference phase completed with the help of fuzzy rules coming in the standard 'if-then' format. In the formation of the condition part of the fuzzy rules, the input space is partitioned with the use of Fuzzy C-Means (FCM) clustering. In the conclusion part of the fuzzy rules, the connections (weights) of the RBF NNs are represented by four kinds of polynomials such as constant, linear, quadratic, and reduced quadratic. The values of the Coefficients are determined by running a gradient descent method. The output of the RBF NNs model is obtained by running a fuzzy inference method. The essential design parameters of the network (including learning rate, Momentum Coefficient and fuzzification Coefficient used by the FCM) are optimized by means of Differential Evolution (DE). The proposed P-RBF NNs (Polynomial based RBF NNs) are applied to facial recognition and its performance is quantified from the viewpoint of the output performance and recognition rate.
-
polynomial based radial basis function neural networks p rbf nns realized with the aid of particle swarm optimization
Fuzzy Sets and Systems, 2011Co-Authors: Sungkwun Oh, Witold Pedrycz, Byoungjun ParkAbstract:In this study, we design polynomial-based radial basis function neural networks (P-RBF NNs) based on a fuzzy inference mechanism. The essential design parameters (including learning rate, Momentum Coefficient and fuzzification Coefficient of the underlying clustering method) are optimized by means of the particle swarm optimization. The proposed P-RBF NNs dwell upon structural findings about training data that are expressed in terms of a partition matrix resulting from fuzzy clustering in this case being the fuzzy C-means (FCM). The network is of functional nature as the weights between the hidden layer and the output are some polynomials. The use of the polynomial weights becomes essential in capturing the nonlinear nature of data encountered in regression or classification problems. From the perspective of linguistic interpretation, the proposed network can be expressed as a collection of ''if-then'' fuzzy rules. The architecture of the networks discussed here embraces three functional modules reflecting the three phases of input-output mapping realized in rule-based architectures, namely condition formation, conclusion creation, and aggregation. The proposed classifier is applied to some synthetic and machine learning datasets, and its results are compared with those reported in the previous studies.
Sungkwun Oh - One of the best experts on this subject based on the ideXlab platform.
-
optimized face recognition algorithm using radial basis function neural networks and its practical applications
Neural Networks, 2015Co-Authors: Sungkwun Oh, Witold PedryczAbstract:In this study, we propose a hybrid method of face recognition by using face region information extracted from the detected face region. In the preprocessing part, we develop a hybrid approach based on the Active Shape Model (ASM) and the Principal Component Analysis (PCA) algorithm. At this step, we use a CCD (Charge Coupled Device) camera to acquire a facial image by using AdaBoost and then Histogram Equalization (HE) is employed to improve the quality of the image. ASM extracts the face contour and image shape to produce a personal profile. Then we use a PCA method to reduce dimensionality of face images. In the recognition part, we consider the improved Radial Basis Function Neural Networks (RBF NNs) to identify a unique pattern associated with each person. The proposed RBF NN architecture consists of three functional modules realizing the condition phase, the conclusion phase, and the inference phase completed with the help of fuzzy rules coming in the standard 'if-then' format. In the formation of the condition part of the fuzzy rules, the input space is partitioned with the use of Fuzzy C-Means (FCM) clustering. In the conclusion part of the fuzzy rules, the connections (weights) of the RBF NNs are represented by four kinds of polynomials such as constant, linear, quadratic, and reduced quadratic. The values of the Coefficients are determined by running a gradient descent method. The output of the RBF NNs model is obtained by running a fuzzy inference method. The essential design parameters of the network (including learning rate, Momentum Coefficient and fuzzification Coefficient used by the FCM) are optimized by means of Differential Evolution (DE). The proposed P-RBF NNs (Polynomial based RBF NNs) are applied to facial recognition and its performance is quantified from the viewpoint of the output performance and recognition rate.
-
a study on three phase partial discharge pattern classification with the aid of optimized polynomial radial basis function neural networks
The Transactions of the Korean Institute of Electrical Engineers, 2013Co-Authors: Sungkwun OhAbstract:In this paper, we propose the pattern classifier of Radial Basis Function Neural Networks(RBFNNs) for diagnosis of 3-phase partial discharge. Conventional methods map the partial discharge/noise data on 3-PARD map, and decide whether the partial discharge occurs or not from 3-phase or neutral point. However, it is decided based on his own subjective knowledge of skilled experter. In order to solve these problems, the mapping of data as well as the classification of phases are considered by using the general 3-PARD map and PA method, and the identification of phases occurring partial discharge/noise discharge is done. In the sequel, the type of partial discharge occurring on arbitrary random phase is classified and identified by fuzzy clustering-based polynomial Radial Basis Function Neural Networks(RBFNN) classifier. And by identifying the learning rate, Momentum Coefficient, and fuzzification Coefficient of FCM fuzzy clustering with the aid of PSO algorithm, the RBFNN classifier is optimized. The virtual simulated data and the experimental data acquired from practical field are used for performance estimation of 3-phase partial discharge pattern classifier.
-
polynomial based radial basis function neural networks p rbf nns realized with the aid of particle swarm optimization
Fuzzy Sets and Systems, 2011Co-Authors: Sungkwun Oh, Witold Pedrycz, Byoungjun ParkAbstract:In this study, we design polynomial-based radial basis function neural networks (P-RBF NNs) based on a fuzzy inference mechanism. The essential design parameters (including learning rate, Momentum Coefficient and fuzzification Coefficient of the underlying clustering method) are optimized by means of the particle swarm optimization. The proposed P-RBF NNs dwell upon structural findings about training data that are expressed in terms of a partition matrix resulting from fuzzy clustering in this case being the fuzzy C-means (FCM). The network is of functional nature as the weights between the hidden layer and the output are some polynomials. The use of the polynomial weights becomes essential in capturing the nonlinear nature of data encountered in regression or classification problems. From the perspective of linguistic interpretation, the proposed network can be expressed as a collection of ''if-then'' fuzzy rules. The architecture of the networks discussed here embraces three functional modules reflecting the three phases of input-output mapping realized in rule-based architectures, namely condition formation, conclusion creation, and aggregation. The proposed classifier is applied to some synthetic and machine learning datasets, and its results are compared with those reported in the previous studies.
Li Hao Feng - One of the best experts on this subject based on the ideXlab platform.
-
suppression of lift fluctuations on a circular cylinder by inducing the symmetric vortex shedding mode
Journal of Fluids and Structures, 2015Co-Authors: Yaguang Liu, Li Hao FengAbstract:Abstract The flow around a stationary circular cylinder modified by two synthetic jets positioned at the mean separation points is numerically studied. The Reynolds number based on the free-stream velocity and the circular cylinder diameter is Re=500. The focus is to present a novel way to suppress the lift fluctuations by changing the vortex shedding mode, and thus particular attention is paid to the interactions between the synthetic jets and wake shear layers and the resulting vortex dynamics. The overall influences of both Momentum Coefficient and excitation frequency are discussed. In some simulated cases, the vortex lock-on phenomenon is discovered, which causes the typical Karman type vortex shedding to be converted into the symmetric shedding modes, leading to the complete suppression of lift fluctuations. In other cases, the asymmetric shedding mode still dominates the wake evolution. Detailed vortical evolution for each typical wake pattern is analyzed to reveal the control mechanism. Additionally, the control effectiveness is evaluated, indicating that the present control strategy contributes an effective way to suppress the lift fluctuations and reduce the mean drag.
-
modification of a circular cylinder wake with synthetic jet vortex shedding modes and mechanism
European Journal of Mechanics B-fluids, 2014Co-Authors: Li Hao Feng, Jinjun WangAbstract:Abstract The wake behind a circular cylinder is modified by a synthetic jet positioned at the front stagnation point. The flow field is measured with a time-resolved particle image velocimetry (PIV) system, and the proper orthogonal decomposition (POD) and λ c i methods are used to analyze the vortex dynamics. The synthetic jet vortex pair is induced near the exit orifice periodically and then moves upstream. The interaction between the synthetic jet and the oncoming flow gives rise to an envelope formed upstream of the circular cylinder, which acts as a virtual aerodynamic shape. It is found that the envelope can be categorized into the periodic closed envelope and the quasi-steady open envelope, leading to different shedding modes for the wake around the circular cylinder. In the present investigation, six kinds of vortex shedding modes under synthetic jet control have been classified as natural Karman vortex mode, bistable state mode I, symmetric mode, bistable state mode II, antisymmetric mode with shortened vortex formation length, vortex generation close to the rear stagnation point. The vortex dynamics analysis indicates that the wake vortex trajectory, vortex circulation, and convection velocity at the vortex core all exhibit regular variations for these typical shedding modes. The formation mechanisms for these shedding modes have been further revealed, which present some novel formation processes in comparison with the natural Karman vortex street. Moreover, the effects of the synthetic jet Momentum Coefficient and excitation frequency on the control are also compared, which suggests that the type of the front envelope is most important for the vortex shedding modes.
-
proper orthogonal decomposition analysis of vortex dynamics of a circular cylinder under synthetic jet control
Physics of Fluids, 2011Co-Authors: Li Hao Feng, Jinjun Wang, Chong PanAbstract:Vortex dynamics of a circular cylinder controlled by a synthetic jet positioned at the back stagnation point is experimentally investigated using particle image velocimetry (PIV) technique. The proper orthogonal decomposition (POD) method is adopted to present the variations of the POD energy, mode, Coefficient, corresponding dominant frequency, and the reconstructed spanwise vorticity. It is found that the dominant dimensionless control parameters should be the synthetic jet stroke length L0/D, where D is the diameter of the experimental circular cylinder, and the equivalent Momentum Coefficient Cμ. For the same stroke length L0/D=3.3, the states of the wake vortex shedding are determined by the Momentum Coefficient. They can be categorized into three groups summarizing all the parameters tested: antisymmetric Karman vortex shedding mode (Cμ≤0.027), vortex synchronization with shedding modes varying between the symmetric and antisymmetric ones (0.061≤Cμ≤0.109), and vortex synchronization with symmetric s...
Michael Amitay - One of the best experts on this subject based on the ideXlab platform.
-
The Flow Field of a Dynamically Pitching Finite Span S809 Blade and its Control
32nd ASME Wind Energy Symposium, 2014Co-Authors: Keith Taylor, Chia M. Leong, Michael AmitayAbstract:The ultimate purpose of this study is to improve the lifetime of wind turbine blades and to reduce costs associated with replacing damaged blades, such that wind energy can compete more economically. The specific goal of this paper is to understand the flow field around a dynamically pitching finite span wind turbine blade and subsequently reduce any unsteady loading by controlling the flow field around the blade using synthetic jet actuators. Experiments were conducted at a Reynolds number of 220,000, where the aerodynamic forces and moments were measured using a six-component load cell and the flow field was measured using Stereoscopic Particle Image Velocimetry (SPIV). Force measurements showed that unsteady loading was present during a dynamic pitch and this unsteady loading can be either enhanced (at low Momentum Coefficients) or reduced (at high Momentum Coefficients) through active flow control. SPIV measurements showed that there is a sudden change of the separation point from the trailing edge to the leading edge within a small range of angles of attack when dynamic stall happens. This phenomenon can be either delayed to a higher angle of attack or mitigated depending on the Momentum Coefficient. The baseline flow field around a 2-D and a 3-D model is also compared, and the flow at the mid-span though can be considered nominally 2-D for both models; however, there are major differences in the flow fields for both cases which can be attributed to tip vibration of the 3-D model.
-
active control of flow separation and structural vibrations of wind turbine blades
Wind Energy, 2010Co-Authors: Victor Maldonado, John Farnsworth, William Gressick, Michael AmitayAbstract:The feasibility of using synthetic jet actuators to enhance the performance of wind turbine blades was explored in wind tunnel experiments on a small scale model blade. Using this technique, the global flow field over the blade was altered such that flow separation was mitigated. Consequently, this resulted in a significant decrease in the vibration of the blade. Global flow measurements were conducted, where the moments and forces on the blade were measured using a six component wall-mounted load cell. The effect of the actuation was also examined on the surface static pressure at two spanwise locations; near the blade's root and near its tip. In addition, Particle Image Velocimetry (PIV) technique was used to quantify the flow field over the blade. Using synthetic jets, the flow over the blade was either fully or partially reattached, depending on the angle of attack, and the Reynolds number. Furthermore, the changes induced on the moments and forces, as well as on the blade vibrations were found to be proportionally controllable by either changing the Momentum Coefficient, the number of synthetic jets used, or by the driving waveform. Finally, a proof-of-concept closed-loop control system was developed to test the ability of using synthetic jet actuators to restore and maintain flow attachment and reduce the vibrations in the blade during dynamic pitching maneuvers. The control system demonstrated the ability of synthetic jet actuators to reduce blade vibrations during dynamic motion depending upon their control concept, which might be either pitch or active stall control. Copyright © 2009 John Wiley & Sons, Ltd.
-
active control of a free jet using a synthetic jet
International Journal of Heat and Fluid Flow, 2008Co-Authors: David A Tamburello, Michael AmitayAbstract:Abstract The control of an axisymmetric free jet ( Re U e = 6600 ) using a single synthetic jet was investigated experimentally. The interaction was examined for a range of Momentum Coefficients, Strouhal numbers, and synthetic jet orientations (with respect to the main jet). To better explore the complex flow field resulting from the interaction, a rendering technique was used where three-dimensional flow fields were calculated from multiple two-dimensional measurement planes. The synthetic jet deflects the majority of the main jet flow away from it, while drawing some of the flow back toward it. Also, the synthetic jet is shown to appreciably raise the main jet’s turbulent quantities, suggesting that mixing has been enhanced. Using triple decomposition, it was shown that the random and coherent motions have similar contributions to the turbulent stresses near the interaction region; whereas the coherent motions prevail farther downstream (and along the shear layers). Measurements of the streamwise vorticity showed that the interaction results in the formation of counter-rotating streamwise vortices, similar to the effect of passive tabs. The size and strength of these structures can be controlled by changing the synthetic jet’s Momentum Coefficient, actuation frequency, or orientation. At low Momentum Coefficients, the largest effect is obtained for a Strouhal number of 0.32; while at higher Momentum Coefficients saturation is obtained due to the high excitation level. A steady control jet, which only utilizes the direct impact mechanism, results in vectoring and a deep penetration into the main jet. However, it yields decreased spreading compared to a synthetic jet with the same Momentum Coefficient.
-
aerodynamic performance modification of the stingray uav at low angles of attack
25th AIAA Applied Aerodynamics Conference, 2007Co-Authors: John Farnsworth, John Vaccaro, Michael AmitayAbstract:Active flow control using fluidic actuators, via synthetic jets and steady blowing jets, was used to provide control power for trimming the Stingray UAV in the longitudinal (pitch) and lateral (roll) directions at low angles of attack. Using this technique, the pitching and roll moments were altered such that the effect is similar to that of a deflection of conventional control effectors in trim. The control effectiveness of the flow control on the aerodynamic performance of the Stingray UAV was investigated experimentally in a wind tunnel. Global flow measurements were conducted, where the moments and forces on the vehicle were measured using a six component sting balance. The effect of the actuation was also examined on the surface static pressure at two spanwise locations. In addition, Particle Image Velocimetry (PIV) technique was used to quantify the velocity vector field over the model, both the global flow field as well as the localized interaction domain near the synthetic jet orifice. The synthetic jets were able to alter the local streamlines through the formation of a quasi-steady interaction region on the suction surface of the Stingray UAV’s wing. Phase locked PIV data was acquired to provide insight into the growth, propagation, and decay of the synthetic jet impulse and its interaction with the cross-flow. The changes induced on the moments and forces can be proportionally controlled by either changing the Momentum Coefficient or by driving the synthetic jets with a pulse modulation waveform. This can lead the way for future development of closed-loop control models.
-
aerodynamic flow control over an unconventional airfoil using synthetic jet actuators
AIAA Journal, 2001Co-Authors: Michael Amitay, Douglas R Smith, Valdis Kibens, David Parekh, Ari GlezerAbstract:Control of flow separation on an unconventional symmetric airfoil using synthetic (zero net mass flux) jet actuators is investigated in a series of wind tunnel tests. The symmetric airfoil comprises the aft portion of a NACA four-digit series airfoil and a leading edge section that is one-half of a round cylinder. The experiments are conducted over a range of Reynolds numbers between 3.1 × 10 5 and 7.25 × 10 5 . In this range, the flow separates near the leading edge at angles of attack exceeding 5 deg. When synthetic jet control is applied near the leading edge, upstream of the separation point, the separated flow reattaches completely for angles of attack up to 17.5 deg and partially for higher angles of attack. The effect of the actuation frequency, actuator location, and Momentum Coefficient is investigated for different angles of attack. The Momentum Coefficient required to reattach the separated flow decreases as the actuators are placed closer to the separation point. In some cases, reattachment is also achieved when the actuators are placed downstream of the stagnation point on the pressure side of the airfoil
Yewei Huang - One of the best experts on this subject based on the ideXlab platform.
-
Effects of steady wake-jets on subcritical cylinder flow
Experimental Thermal and Fluid Science, 2019Co-Authors: Donglai Gao, Guanbin Chen, Wenli Chen, Yewei HuangAbstract:We investigate the effects of active wake-jets (characterized by a dimensionless jet Momentum Coefficient C μ) on the suppression of aerodynamic forces and the manipulation of wake flow topology behind a cylindrical model through wind tunnel tests. The active jets are positioned at the rear stagnation points of the cylindrical test model. The experimental campaign is conducted at a subcritical Reynolds number of = × Re 3.33 10 4. The surface pressure distributions around the bare and controlled cylinders are obtained by using a pressure measurement system. Apart from pressure measurements, we also obtain the streamwise and spanwise flow structures around the circular cylinder with different C μ (including = C 0 μ) by employing the particle image velo-cimetry (PIV) technique. Pressure measurement results demonstrate that the lift force acting on the cylindrical test model is greatly reduced and drag decreased with the implementation of active wake-jets. Besides, it is found that a higher C μ contributes to a better control effectiveness in unsteady lift forces but not necessarily a better drag reduction. PIV measurement results indicate that the mechanism of the active jet control scheme is to impose steady and symmetric perturbations into the unsteady and asymmetric flows in the cylinder wake. Owing to the dynamic competition of the wake-jet flow and shear layer flows, the vortex shedding pattern behind the controlled cylinder is greatly modified, vortex formation length significantly elongated and the fluctuations of aerodynamic forces conceivably suppressed.