Pattern recognition methods have become a powerful tool for segmentation in the sense that they are capable of automatically building a segmentation model from training images. However, they present several difficulti...
详细信息
ISBN:
(纸本)9781424479948
Pattern recognition methods have become a powerful tool for segmentation in the sense that they are capable of automatically building a segmentation model from training images. However, they present several difficulties, such as requirement of a large set of training data, robustness to imaging conditions not present in the training set, and complexity of the search process. In this paper we tackle the second problem by using a deep belief network learning architecture, and the third problem by resorting to efficient searching algorithms. As an example, we illustrate the performance of the algorithm in lip segmentation and tracking in video sequences. Quantitative comparison using different strategies for the search process are presented. We also compare our approach to a state-of-the-art segmentation and tracking algorithm. The comparison show that our algorithm produces competitive segmentation results and that efficient search strategies reduce ten times the run-complexity.
An advanced aerodynamic design optimization system for steam turbine stages considering rotor blade tip leakage and blade end-wall non-axis-symmetric contouring has been developed. Using this system, fluid dynamic opt...
详细信息
ISBN:
(纸本)9780791844021
An advanced aerodynamic design optimization system for steam turbine stages considering rotor blade tip leakage and blade end-wall non-axis-symmetric contouring has been developed. Using this system, fluid dynamic optimizations were carried out for a steam turbine stage with stator and rotor blades. The system includes parametric modeling of blade and end-wall contouring, evaluation system with in-house or package CFD software and optimization strategy module. The designs of rotor blade and hub end-wall surface in a typical large-scale high-pressure steam turbine stage were optimized in order to know this design optimization impact on enhancing the stage efficiency. Results show that: from the current well designed high pressure steam turbine stage, the demonstrated efficiency enhancement with the present optimum design is around 0.2% under consideration of rotor tip leakage. Design cycle could be greatly shortened by parallel optimization algorithm and cluster PC, and especially four days could be sufficient for an optimization with one thousand iterations on 20 CPUs of 2.0G cluster PC.
We present a new updating parameters (UPs) selection method to tackle the bottleneck created by having too many UPs and limited measured data in model updating processing. While the model updating is performed by para...
详细信息
ISBN:
(纸本)9780791848982
We present a new updating parameters (UPs) selection method to tackle the bottleneck created by having too many UPs and limited measured data in model updating processing. While the model updating is performed by parameter optimization, an ill-conditioned numerical problem may be encountered or the reliability of the result may be unacceptable if too many parameters are used. The selection of UPs thus becomes a key issue, especially for long-span bridges with finite element models that should be divided into at least hundreds of element numbers. A new method is introduced to reduce the number of UPs and retain their physical significance. In this method, original UPs are described by a few macro-parameters based on shape functions. The model subsequently is updated by a normal optimization algorithm, such as the first-order optimization method. Based on a bridge with a three-span continuous beam and a long-span tie-arch, the optimal effects are investigated, with or without a shape function and using different types of shape functions. The results indicate that the effect of the modal updating based on a shape function is more robust than without shape function and the effect of a linear shape function is better than that of a constant value shape function.
In this contribution, we propose a Distributed Concatenated Irregular Convolutional Coded Unity Rate Coded and Space Time Trellis Coded (DC-IrCC-URC-STTC) scheme for cooperative communications employing multiple singl...
详细信息
ISBN:
(纸本)9781424435746
In this contribution, we propose a Distributed Concatenated Irregular Convolutional Coded Unity Rate Coded and Space Time Trellis Coded (DC-IrCC-URC-STTC) scheme for cooperative communications employing multiple single-antenna relays. Each coding arrangement is designed for efficient decoding convergence by employing non-binary Extrinsic Information Transfer (EXIT) charts. The source node transmits two-stage IrCC-URC encoded symbols to two relays during the first transmission period. Then each relay performs two-stage URC-IrCC decoding and re-encodes the information bits using the three-stage concatenated coding scheme IrCC-URC-STTC. However, only the signals of one predefined STTC output are transmitted from each relay to the destination node during the second transmission period. At the destination, the received signals are treated as if they arrived from one transmitter employing STTC using two co-located transmit antennas. The simulation results show that the Monte-Carlo based performance of the DC-IrCC-URC-STTC scheme exactly matches the EXIT-chart analysis. It also performs closely to its idealised counterpart that assumes perfect decoding at the relays when an EXIT-chart based optimal algorithm is applied for selecting relays.
We study a certain route configuration problem via optimization theory. We consider the optimal bit error rate (BER) and transmission rate allocations on each hop, subject to overall BER and delay constraints for a de...
详细信息
We study a certain route configuration problem via optimization theory. We consider the optimal bit error rate (BER) and transmission rate allocations on each hop, subject to overall BER and delay constraints for a designated route. The pivot of the problem lies in the delay constraint, which divides the problem into two cases-the loose and the tight delay case. In the former, analytical solutions are obtained by applying the Karush-Kuhn-Tucker (KKT) theorem. Specifically, we discover in this case that for a given target BER, the optimum solutions are only related to the hop lengths in the route. When the delay constraint is tight, a mapping can be used to reduce the dimension of the problem by a factor of two;a numerical optimization algorithm has to be used to find the optimum. Simulation results show that by optimally configuring a chosen route, substantial energy savings could be obtained, especially under tight delay constraints. Simulation also reveals that a performance limit is reached as the number of hops increases. A parameter determining this limit is defined, and physical explanations are given accordingly.
A procedure based on the principles of inverse analysis has been employed to identify the elastic material parameters and the strength characteristics of epoxy carbon fiber reinforced composite laminates. By adopting ...
详细信息
ISBN:
(纸本)9780791849156
A procedure based on the principles of inverse analysis has been employed to identify the elastic material parameters and the strength characteristics of epoxy carbon fiber reinforced composite laminates. By adopting a classical optimization algorithm, elasticity parameters are chosen by minimizing the mean squared error between the values predicted by the classical theory of elastic laminates and the experimentally observed ones. The average modulus of the laminate has been calculated from elastic parameters of laminas by implementing the classical equations of anisotropic elasticity in a subroutine, in order to speed up the optimization process. In a similar fashion, even though under strongly simplifying hypothesis, the strength characteristic parameters of laminas have been identified by inverse analysis, with the aim of assessing the strength of the composite laminates in the case of complex stress state.
In this work, a novel synthesis technique for the design of linear uniformly fed sparse arrays has been presented. Such method, based on a fully analytical procedure, is able to mimic a given array factor mask without...
详细信息
ISBN:
(纸本)9781424449682
In this work, a novel synthesis technique for the design of linear uniformly fed sparse arrays has been presented. Such method, based on a fully analytical procedure, is able to mimic a given array factor mask without using computationally demanding optimization algorithms. As an example, an array of 51 antennas having aperture L*max = 24.5 λo has been designed in order to synthesize a very complex radiation mask useful in a large variety of applications.
We investigate feature selection methods, which have been applied to automatic kinds of compressed audio classification systems. It is based on attribute dependency for feature optimization and modified SVM (Support V...
详细信息
ISBN:
(纸本)9781424458653
We investigate feature selection methods, which have been applied to automatic kinds of compressed audio classification systems. It is based on attribute dependency for feature optimization and modified SVM (Support Vector Machine) for classifier. In this paper, we present a new method for feature selection based on priori knowledge by removing both irrelevant and redundant features, and it still retains sufficient information for classification purpose. Experiments on compressed audio category classification indicated that when using this proposed method to select the optimal feature subset and combing with the modified SVM classifier, we could get better efficiency up to 90%, even 10% higher to the total feature sets.
Alarmingly low pulp prices in early 2009 left pulp and paper mills across North America desperate for any way to improve thin profit margins. One solution that continues to gain popularity among the industry is improv...
详细信息
ISBN:
(纸本)9781424474264
Alarmingly low pulp prices in early 2009 left pulp and paper mills across North America desperate for any way to improve thin profit margins. One solution that continues to gain popularity among the industry is improved energy management systems for cogeneration systems, which use steam for two purposes - to provide heat for the pulping process, and to generate electricity for sale to regional providers. This paper presents an energy optimization algorithm for use in a pulp and paper mill cogeneration system. The algorithm is applicable to a number of popular mill configurations, power sale contracts, and fuel purchasing scenarios. Results are presented in the form of two case studies.
A new design of a machine is developed which performs contactless actuation in four axis - three translational and one rotational - with high accuracy. Remarkable on the concept is the moving part of the machine which...
详细信息
ISBN:
(纸本)9781424470594
A new design of a machine is developed which performs contactless actuation in four axis - three translational and one rotational - with high accuracy. Remarkable on the concept is the moving part of the machine which is free of contacts, enabling free z-rotation. The components of the device are optimized for different requirements. The optimization algorithm is a combination of an evolution strategy (ES) and an approximation method based on the Response Surface Methodology (RSM). The design and optimization is supported by numerical simulations using the volume integral equation method (VIEM) combined with the fast multipole method (FMM).
暂无评论