Guangzhou, China

Guangdong University of Technology is a key provincial university in Guangzhou, Guangdong, China. The university offers a wide range of courses in engineering, science, technology, management, liberal arts and law, with major emphasis on the study of engineering. Wikipedia.


Time filter

Source Type

Wang L.J.,Guangdong University of Technology
International Journal of Machine Learning and Cybernetics | Year: 2011

Multiple nearest neighbor classifier system (MNNCS) is a popular method to relax the curse of dimensionality. In previous work, most of the MNNCSs are designed by random methods. Random methods may generate unstable component classifiers. In order to relax the randomness, large amount of component classifiers are needed. This paper first extends nearest neighbor classifier into fuzzy nearest neighbor classifier, and proposes a new multiple fuzzy nearest neighbor classifier system based on mutual information and fuzzy integral, called MIFI-MFNNCS. MIFI-MFNNCS adopts target perturbation. Target perturbation decomposes the original classification problem into several sub-problems, where one sub-problem represents one class data. Each sub-problem is described by the relevant data and features. Then it is classified by one component classifier. Therefore, the number of component classifiers can be fixed and reduced. For one component classifier, data may be selected according to its class. And feature is needed to be selected by mutual information. Mutual information can reduce the uncertainty of each component classifier. Feature selection by mutual information in MIFI-MFNNCS may be less affected by the interaction among different classes. The diversity decisions from sub-problem classifiers are combined by fuzzy integral to get the final decision. Here we propose a new method to compute density value according to mutual information, which is a simple method. To demonstrate the performance of the proposed MIFI-MFNNCS, we perform experimental comparisons using five UCI datasets. The results of component classifiers in MIFI-MFNNCS for Ionosphere are shown and analyzed. MIFI-MFNNCS is compared with (1) NNC (2) NNC after feature selection by mutual information (MI-FS-NNC). In multiple fuzzy nearest neighbor classifier system (MFNNCS), mutual information is compared with attribute bagging. And three combination methods are compared, including fuzzy integral, majority voting rule and average. The experimental results show that the accuracy of MIFI-MFNNCS is better than other methods. And mutual information is superior to attribute bagging. Fuzzy integral shows a better performance than majority voting rule and average. © 2010 Springer-Verlag.


Wang Q.,Guangdong University of Technology
Optics Communications | Year: 2012

To completely eliminate the silhouette problem that inherently exists in the earlier interference-based encryption scheme with two phase-only masks (POMs), we propose a simple new encryption method based on optical interference of one random POM and two analytically obtained POMs. Different from the previous methods which require time-consuming iterative computation or postprocessing of the POMs for silhouette removal, our method can resolve the problem during the production of the POMs based on interference principle. Information associated with the original image is smoothed away by modulation of the random POM. Illegal deciphers cannot retrieve the primitive image using only one or two of the POMs. Incorporated with the linear phase blend operation, our method can provide higher robustness against brute force attacks. Simulation results are presented to support the validity and feasibility of our method. © 2012 Elsevier B.V. All rights reserved.


Wang B.,Guangdong University of Technology
Journal of Physics B: Atomic, Molecular and Optical Physics | Year: 2011

We describe high-efficiency two-port beam splitters of total internal reflection (TIR) gratings, where incident waves are reflected in the 0th and -1st orders with the same energy, respectively. To achieve high efficiency and uniformity between two orders, the grating duty cycle and depth are optimized using rigorous coupled wave analysis for a laser wavelength of 800 nm. With the optimized grating profiles, the TIR grating can work as a 50/50 beam splitter for not only TE or TM polarization but also both TE and TM polarization. Efficiencies can reach nearly 50% × 2 = 100% in the 0th and -1st reflected orders. It indicates that high efficiency can be obtained based on a TIR grating instead of complicated structures. For high efficiency of a TIR grating and a high damage threshold of fused silica, the presented two-port beam splitter gratings should be useful elements in numerous practical applications, especially in high laser power systems. © 2011 IOP Publishing Ltd.


Peng S.,Guangdong University of Technology
Nonlinear Analysis: Real World Applications | Year: 2010

Bidirectional associative memory (BAM) model is considered with the introduction of continuously distributed delays in the leakage (or forgetting) terms. By using continuation theorem in coincidence degree theory and the Lyapunov functional, some very verifiable and practical algebraic mean delay dependent criteria on the existence and global attractive periodic solutions are derived. © 2009 Elsevier Ltd. All rights reserved.


Patent
CAS South China Sea Institute of Oceanology and Guangdong University of Technology | Date: 2015-02-05

An in-situ and on-line acoustic measuring system for natural gas flux at the hydrocarbon seeps at a seafloor includes a seepage tent and a flow measuring channel. The flow measuring channel includes a lower bubble breaking channel, an ultrasonic transducer measuring channel and an upper bubble breaking channel. The lower bubble breaking channel communicates with the seepage tent provided with bubble breaking grids. Lower and upper bubble breaking devices, arranged in a bubble rising direction, are respectively mounted in the lower and upper bubble breaking channels. One side of the ultrasonic transducer measuring channel is fixedly connected with an acoustic wave demultiplexer, and the other side is fixedly connected with flat receiving transducers receiving transmitting acoustic waves generated by an acoustic wave branching unit. Acoustic wave probes, also used for receiving the transmitting acoustic waves generated by the acoustic wave demultiplexer, are arranged in the ultrasonic transducer measuring channel.


The present disclosure relates to a macro-micro composite grating ruler measuring system based on conversion and amplification in vertical and horizontal directions. The macro-micro composite grating ruler includes a grating ruler, a macro-micro reading system moving with respect to the grating ruler, and a counting and image processing module. The macro-micro reading system faces grating strip datum and is parallel to the grating ruler. The system further includes a measuring reference line. The measuring reference line obtained by the image sensor together with grating strips forms an image overlap in the counting and image processing module. The measuring reference line and the grating strip jointly include an angle . With the foregoing configuration, the present invention is compatible with the existing incremental grating rulers and absolute grating rulers, so is highly applicable.


Patent
Dongguan South China Design And Innovation Institute and Guangdong University of Technology | Date: 2015-08-10

The invention proposed a novel hot pressing flowing sintering method to fabricate textured ceramics. The perfectly 2-dimensional textured Si3N4 ceramics (Lotgering orientation factor fL 0.9975) were fabricated by this method. During the initial sintering stage, the specimen flowed along the plane which is perpendicular to the hot pressing direction under pressure, through the controlling of the graphite die movement. The rod-like -Si3N4 nuclei was easily to texture during the flowing process, due to the small size of the -Si3N4 nuclei and the high porosity of the flowing specimen. After aligned, the -Si3N4 grains grew along the materials flowing direction with little constraint. textured Si3N4 ceramics fabricated by this invention also showed high aspect ratio. Compared to the conventional hot-forging technique which contained the sintering and forging processes, hot pressing flowing sintering proposed is simpler and lower cost to fabricate textured Si3N4.


Patent
Guangdong University of Technology | Date: 2014-09-24

The present disclosure involves occasions where precise two-dimensional motion takes place, and is applicable to XY motion stages for precise displacement compensation. The present disclosure particularly involves a stiffness-frequency adjustable XY micromotion stage based on stress stiffening, which includes X-direction and Y-direction motion sub-stages and corresponding drivers and a micromotion working table. The micromotion stage uses membrane sets that have tension levels thereof adjusted by bolts as a flexible hinge, so as to achieve independent adjustment of the vibration frequency of the XY micromotion stage. The present disclosure implements the foregoing configuration based on prestressed membrane, so the frequency is adjustable. The inherent frequency of the micromotion stage can be adjusted before or during operation according to various working conditions and driving frequency. The two feed motion direction are perpendicular so as to prevent the micromotion working table from coupling during two-dimensional motion.


Various embodiments relate to a method of planning asymmetric variable acceleration based on non-linear finite element dynamic response simulation. The planning method involves obtaining solution of a non-linear finite element model positioning process that has kinematic freedom and adopts a parameterized motion function as its boundary condition; determining whether post-driving amplitude of an execution end satisfies positioning precision, and if it does not, continuing getting solution, and if it is, adjusting an energy decay time; determining whether a target response time is minimum, and if it is, verifying the set motion parameter as optimal, and if it is not, calculating a gradient and a step size of the motion parameter, and resetting the motion parameter for solution. The present disclosure utilizes this method to plan high-speed high-acceleration motion for mechanisms that are affected by non-linear factors such as large flexible deformation and require precise positioning.


Patent
Guangdong University of Technology | Date: 2015-01-28

The present invention is a method of assisted mounting and error compensation for absolute grating ruler. It comprises the following: (1) when mounting a CMOS sensor and a grating ruler body, the CMOS sensor reads an upper and a lower sample windows, and due to an angle existing between the grating ruler body and the CMOS sensor, a difference exists between the numbers of the upper and lower sample windows, and by continually adjusting the grating ruler body or the CMOS sensor, the code reading difference is made minimal so that the angle existing between the grating ruler body and the CMOS sensor is zeroed; (2) when mounting the grating ruler body and a mechanic housing, it is moved by a fixed displacement in a motion direction, and a grating encoding reading is recorded and an error compensating amount is obtained which serves as error compensation value in an actual motion to correct a cumulative error introduced by the angle between the grating ruler body and the motion direction. The invention may automatically calculate the deviation angles between the CMOS sensor and the grating ruler body and correct parallelism errors. Meanwhile, it also provides an error compensation in the case that the parallelism errors between the grating ruler body and the reference direction cannot be corrected.

Loading Guangdong University of Technology collaborators
Loading Guangdong University of Technology collaborators