Browsing by Author "Lu, YC"

Now showing 1 - 12 of 12

A neural fuzzy system with fuzzy supervised learning
(1996-10-01)
Lin, CT
;
Lu, YC
;
交大名義發表
;
電控工程研究所
;
National Chiao Tung University
A neural fuzzy system learning with fuzzy training data (fuzzy if-then rules) is proposed in this paper, This system is able to process and learn numerical information as well as linguistic information, At first, we propose a five-layered neural network for the connectionist realization of a fuzzy inference system, The connectionist structure can house fuzzy logic rules and membership functions for fuzzy inference, We use a-level sets of fuzzy numbers to represent linguistic information. The inputs, outputs, and weights of the proposed network can be fuzzy numbers of any shape, Furthermore, they can be hybrid of fuzzy numbers and numerical numbers through the use of fuzzy singletons, Based on interval arithmetics, a fuzzy supervised learning algorithm; is developed for the proposed system, It extends the normal supervised learning techniques to the learning problems where only linguistic teaching signals are available, The fuzzy supervised learning scheme can train the proposed system with desired fuzzy input-output pairs which are fuzzy numbers instead of the normal numerical values, With fuzzy supervised learning, the proposed system can be used for rule base concentration to reduce the number of rules in a fuzzy rule base, Simulation results are presented to illustrate the performance and applicability of the proposed system.
A neural fuzzy system with fuzzy supervised learning
(1996-10-01)
Lin, CT
;
Lu, YC
;
電控工程研究所
;
Institute of Electrical and Control Engineering
A neural fuzzy system learning with fuzzy training data (fuzzy if-then rules) is proposed in this paper, This system is able to process and learn numerical information as well as linguistic information, At first, we propose a five-layered neural network for the connectionist realization of a fuzzy inference system, The connectionist structure can house fuzzy logic rules and membership functions for fuzzy inference, We use a-level sets of fuzzy numbers to represent linguistic information. The inputs, outputs, and weights of the proposed network can be fuzzy numbers of any shape, Furthermore, they can be hybrid of fuzzy numbers and numerical numbers through the use of fuzzy singletons, Based on interval arithmetics, a fuzzy supervised learning algorithm; is developed for the proposed system, It extends the normal supervised learning techniques to the learning problems where only linguistic teaching signals are available, The fuzzy supervised learning scheme can train the proposed system with desired fuzzy input-output pairs which are fuzzy numbers instead of the normal numerical values, With fuzzy supervised learning, the proposed system can be used for rule base concentration to reduce the number of rules in a fuzzy rule base, Simulation results are presented to illustrate the performance and applicability of the proposed system.
Encoding ASK labeled CSRZ-DPSK payload by using only one dual-drive Mach-Zehnder modulator with enhanced label performance
(2005-10-01)
Peng, WR
;
Lu, YC
;
Chen, J
;
Chi, S
;
光電工程學系
This theoretical study presents a novel modulation scheme which simultaneously encodes the amplitude shift keying label on the carrier-suppressed return-to-zero differential phase shift keying payload using only one dual-drive Mach-Zehnder modulator. Compared with the previous scheme, in which three optical modulators are required, we offer a simple, compact, and cost-effective method. We theoretically evaluate and simulate the performance differences between the two schemes. It is found that the sensitivity of the label is 2 dB better than the previous scheme and the payload, with optimized extinction ratio in the electroabsorption eraser, has almost the same performance after 120-km standard single-mode fiber transmission.
Femtosecond fluorescence dynamics of rotation-restricted azobenzenophanes: new evidence on the mechanism of trans -> cis photoisomerization of azobenzene
(2005-03-17)
Lu, YC
;
Diau, EWG
;
Rau, H
;
應用化學系分子科學碩博班
;
Institute of Molecular science
The ultrafast relaxation dynamics of two rotation-restricted (azobenzeno-2S-phane and azobenzeno-4S-phane) and one rotation-free (4,4'-dimethylazobenzene) azobenzene derivatives were investigated using femtosecond fluorescence up-conversion on both S-1(n,pi*) and S-2(pi,pi*) excitations. On S-2 excitation, pulse-limited kinetics with a decay coefficient of similar to 100 fs corresponding to ultrafast S-2-> S-1 relaxation is found to be common for all molecules under investigation regardless of the molecular structure. This indicates that a direct rotational. relaxation on the S-2 surface is unfavorable. On S-1 excitation, we observed biphasic fluorescence decay with a femtosecond component attributed to the decay of the Franck-Condon state prepared by excitation and a picosecond component attributed to the deactivation of the relaxed molecule on the S, surface. This picosecond component is slowed by at least a factor of 2 for the rotation-restricted 2S-bridged molecule compared to that of the rotation-free molecule; for the even stronger rotation-restricted azobenzeno-4S-phane, the decrease is by a factor of 10. These differences in deactivation suggest that the relaxed states and probably the trajectories for rotation-free and rotation-restricted molecules are different on the S, surface, which should be important for the quantum yield of photoisomerization.
Femtosecond fluorescence dynamics of trans-azobenzene in hexane on excitation to the S-1(n,pi*) state
(2002-10-01)
Lu, YC
;
Chang, CW
;
Diau, EWG
;
應用化學系
;
Department of Applied Chemistry
Femtosecond fluorescence dynamics of trans-azobenzene in hexane have been investigated on excitation to the S-1(n,pi*) state at 432 nm using the up-conversion technique. Two transient components were observed to represent the fast and the slow S-1 fluorescence dynamics in the wavelength range of 550-732 nm. Based on the results obtained from recent ab initio calculations (Ishikawa, T.; Noro, T.; Shoda, T., J. Chem. Phys. 2001, 115, 7503), a dynamical picture is given in the following. Upon initial excitation to the S, state, the excited molecule is moving away from the first detection window within the observed 200-300 fs. The structural relaxation from the Franck-Condon region may be responsible for the observed fast S-1 dynamics with the driving force being the CNNC twisting motion along the rotational coordinate. For the rest of the motion on the S-1 global potential surface, the excited molecule may search for the S-0/S-1 conical intersection for an efficient internal conversion to the ground state. The nuclear motions for the observed slow S-1 dynamics not only involve the CNNC torsional coordinate but also the other degrees of freedom such as the CNN bending coordinate on the multidimensional S-1 potential energy surface. The whole electronic relaxation process occurs within the observed 1-2 ps. The slow S-1 dynamics were found to vary with the fluorescence wavelengths due to the influence of the solvent-induced vibrational relaxation in the S-1 state; the vibrational relaxation should occur on a time scale comparable to the time scale of the electronic relaxation (S-1-->S-0 internal conversion).
Improved SPM tolerance and cost-effective phase-modulation duobinary transmission over 230-km standard single-mode fiber using a single Mach-Zehnder modulator
(2005-12-01)
Lu, YC
;
Chen, J
;
Feng, KM
;
Yeh, PC
;
Huang, TY
This letter proposes and experimentally demonstrates a cost-effective phase-modulation duobinary modulation scheme for improving self-phase modulation tolerance using a single modulator. Successful transmission over 230 km of standard single-mode fiber was achieved.
Improvement in leakage current and breakdown field of Cu-comb capacitor using a silicon oxycarbide dielectric barrier
(2004)
Chiang, CC
;
Ko, IH
;
Chen, MC
;
Wu, ZC
;
Lu, YC
This work investigates in the first place, the improvement in leakage current and breakdown field of the copper metal-insulator-semiconductor (Cu-MIS) capacitor with a plasma-enhanced chemical vapor deposited (PECVD) amorphous silicon oxycarbide (alpha-SiCO, k = 3.7) dielectric barrier. This is followed by investigating the improvement in leakage current and breakdown field of the Cu-comb capacitor with a carbon-doped low-k PECVD organosilicate glass (k = 3) as the intermetal dielectric and an alpha-SiCO dielectric film as the Cu cap barrier. The leakage current and breakdown field of Cu-MIS and Cu-comb capacitors are dependent on the species of the dielectric barrier. The Cu-MIS and Cu-comb capacitors with an alpha-SiCO dielectric barrier exhibit a leakage current at least three orders of magnitude smaller than those with an amorphous silicon carbide (alpha-SiC, k = 4.4) dielectric barrier at an applied electric field of 1.6 MV/cm between 25 and 250 degreesC. Moreover, the breakdown field of the Cu-MIS and Cu-comb capacitors with an alpha-SiCO dielectric barrier, measured at 200 degreesC, are 60 and 25%, respectively, higher than that of the capacitors with an alpha-SiC barrier. The decreased leakage current and increased breakdown field of the Cu-MIS and Cu-comb capacitors with an alpha-SiCO dielectric barrier are attributed to the higher density, oxygen-improved film property, nonsemiconductor behavior, and lower fringe- or surface-electric field of the alpha-SiCO dielectric film. (C) 2004 The Electrochemical Society.
Leakage and breakdown mechanisms in cu damascene with a bilayer-structured a-SiCN/a-SiC dielectric barrier
(2003)
Chiang, CC
;
Ko, IH
;
Chen, MC
;
Wu, ZC
;
Lu, YC
This work investigates the leakage and breakdown mechanisms in the Cu damascene structure with a carbon-doped low-k PECVD organosilicate glass (OSG, k=3) as the intermetal dielectric (IMD) and an alpha-SiCN(k=5)/alpha-SiC(k=4) bilayer-structured dielectric film as the Cu-cap barrier. It is found that the leakage mechanism between Cu lines is dependent on the thickness ratio of the alpha-SiCN/alpha-SiC bilayer barrier in the Cu damascene structure. In the Cu damascene using an alpha-SiCN/alpha-SiC bilayer barrier of 40nm/10nm or 30nm/20nm bilayer thickness, the large leakage current (Frenkel-Poole emission) between Cu lines is attributed to the plenty of interfacial defects, such as cracks, voids, traps or dangling bonds at the alpha-SiC/OSG interface, which are generated by the larger tensile force of the thicker alpha-SiC film. On the other hand, the breakdown field and TDDB (time-dependent-dielectric-breakdown) lifetime of the Cu damascene reveal little dependence on the thickness ratio of the alpha-SiCN/alpha-SiC bilayer barrier, and the observed breakdown of the Cu damascene structure is presumably due to dielectric breakdown of the bulk OSG layer.
Leakage and breakdown mechanisms of Cu comb capacitors with bilayer-structured alpha-SiCN/alpha-SiC Cu-cap barriers
(2004-02-01)
Chiang, CC
;
Ko, IH
;
Chen, MC
;
Wu, ZC
;
Lu, YC
This work investigates the leakage and breakdown mechanisms in copper (Cu) comb capacitors with carbon-doped low-k plasma-enhanced chemical vapor deposited organosilicate glass (OSG; k = 3) as the intermetal dielectric and an alpha-SiCN (k = 5)/ alpha-SiC (k = 4) bilayer-structured dielectric film as the Cu-cap barrier. The leakage mechanism between Cu lines is dependent on the thickness ratio of the alpha-SiCN/alpha-SiC bilayer barrier. Using an alpha-SiCN/alpha-SiC bilayer barrier of 40 nm/10 nm or 30 nm/20 nm bilayer thickness, the increased leakage current (Frenkel-Poole emission) between Cu lines is attributed to the large number of interfacial defects, such as cracks, voids, traps or dangling bonds at the alpha-SiC/OSG interface, which are generated by the larger tensile force of the thicker alpha-SiC film. The Cu comb capacitor with an alpha-SiCN (50 nm)/alpha-SiC (2 nm) bilayer barrier exhibits a much smaller leakage current. The breakdown field and time-dependent dielectric breakdown lifetime of the Cu comb capacitor reveal little dependence on the thickness ratio of the alpha-SiCN/alpha-SiC bilayer barrier, and the observed breakdown of the Cu comb capacitor is presumably due to dielectric breakdown of the bulk OSG layer. (C) 2004 The Electrochemical Society.
Optimal implementation of miniature piezoelectric panel speakers using the Taguchi method and genetic algorithm
(2004-07-01)
Bai, MR
;
Lu, YC
;
機械工程學系
;
Department of Mechanical Engineering
A thin and miniature piezoelectric panel speaker has been developed in this paper. The system is modeled using the energy method in conjunction with the finite element method. The electrical system, mechanical system and acoustic loading are considered in combined fashion. The genetic algorithm (GA) and the Taguchi method are employed to achieve optimal designs with low fundamental frequency and high acoustic output. The designs resulting from the optimization are then implemented and evaluated by experiments. These experiments were carried out on the basis of quantitative and subjective performance measures. The experimental results indicate that the piezoelectric panel speakers are capable of producing comparable acoustic output with significantly less electrical power than the conventional voice-coil panel speakers.
Photoisomerization dynamics of azobenzene in solution with S-1 excitation: A femtosecond fluorescence anisotropy study
(2004-08-18)
Chang, CW
;
Lu, YC
;
Wang, TT
;
Diau, EWG
;
應用化學系
Measurements of anisotropy of femtosecond fluorescence after direct excitation of the S-1(npi(*)) state of azobenzene in hexane and ethylene glycol solutions have been carried out to address the controversy about inversion and rotation in the mechanism of photoisomerization. The observed anisotropies in hexane decay to a nonzero asymptotic level with a relaxation period the same as that for slow decay of the corresponding biexponential transient; this effect is attributed to involvement of the out-of-plane CNNC-torsional motion on approach to a twisted conical intersection along the "rotation channel" that depolarizes the original in-plane transition moment. In contrast, when the rotational channel becomes substantially hindered in ethylene glycol, the anisotropies show no discernible decay feature, but the corresponding transients show prominent decays attributed to involvement of in-plane symmetric motions; the latter approach a planar-sloped conical intersection along a "concerted inversion channel" for efficient internal conversion through vibronic coupling. The proposed mechanism is consistent with theoretical calculations and rationalizes both results on quantum yields and ultrafast observations.
Physical and barrier properties of PECVD amorphous silicon-oxycarbide from trimethylsilane and CO2
(2004)
Chiang, CC
;
Ko, IH
;
Chen, MC
;
Wu, ZC
;
Lu, YC
This work investigates the thermal stability and physical and barrier properties of amorphous silicon-carbide (alpha-SiC) and amorphous silicon-oxycarbide (alpha-SiCO) dielectric barriers deposited by plasma-enhanced chemical vapor deposition (PECVD) using trimethylsilane (3MS) precursor and He carrier gas. Films were deposited without and with various CO2 flow rates. The dielectric constant of the alpha-SiCO films decreased with increasing CO2 flow rate. Increasing CO2 flow rate also promotes better thermal stability, higher breakdown field, lower leakage current, and superior resistance to Cu diffusion through the films. The improved barrier property is attributed to the denser and less porous structure of the alpha-SiCO dielectric barrier upon CO2 addition. The alpha-SiCO barrier films deposited with the large (1200 Seem) CO2 flow rate exhibit the low k value of 3.7, thermal stability up to 550degreesC, room-temperature breakdown field of 8 MV/cm and leakage current densities of 10(-7) to 10(-6) A/cm(2) at 3 MV/cm, and a superb Cu barrier property. (C) 2004 The Electrochemical Society.