Search

The semiconductor production data is typically complex, nonlinear and high-dimension, and the traditional post-sampling quality inspection often have large errors and will bring the economic losses caused by defective products. Based on machine learning technologies, this research is aimed to establish the proper model to predict semiconductor quality in advance. Based on the requirements of actual production, the BP neural network and AdaBoost algorithm are combined, and a new BP-AdqBoost model is proposed after optimizing the AdaBoost algorithm. The data of LCD Monitor production was analyzed. Then the improved BP-AdqBoost model, BP neural network, and the unmodified BP-AdaBoost prediction results are compared by prediction accuracy and reliability. The comparison shows that the improved BP-AdqBoost model can not only improve the prediction accuracy, but also strengthen the prediction reliability, which would be useful to the practical semiconductor production.

Nowadays off-line path planning technology is widely used to keep autonomous unmanned aeriel vehi-cle(U AV) safety and reach the given target efficiently at the same time. But in unknown, dynamic or changeable environments, a real-time reactive obstacle avoidance method is necessary. This paper presents an obstacle avoidance method using fuzzy control for unmanned helicopter(UMH), a kind of UAV with unique characters. The flight process is made up of several mutually exclusive fight modes. The environment is detected in flight, and the desicion is made simultinously considering the current flight mode and environment detection. Finally the fuzzy controller is applied to an precisely identified UMH model to test the full closed-loop controller. Additonal strategies are used to avoid local loitering and make the fuzzy contrller keep on working.

High-fidelity reconstruction of complete objects is known as a particularly challenging problem. In this paper, we provide a system for object modeling with Kinect and rotation platform. Our modeling solution is divided into three steps: (1) system calibration, (2) Point cloud real time registration,(3) Mesh surface reconstruction and automatic texture mapping. We propose a easy-to-use calibration method for the first step, and based on the calibration result, a automatic texture-independent 3D point clouds registration method is presented. Finally, triangulate meshing and automatic texture mapping technique are used to recreate complete mesh model from the integrated point clouds. Our system is validated by the modeling of several real-world objects, and all the results show its effectiveness.

Path planning is one of the most studied problems in the field of robotics, unmanned aerial vehicles (uavs), vehicle navigation and fields like these. The majority algorithms of path planning produce possible paths of grid graph, and then apply in problems such as classical graph route searching. Astar algorithm, Hierarchical Path-Finding A-star (HPA*) and Jump Point Search (JPS) algorithms are studied in this paper to compare the maze searching capacity and different search maps' efficiency. We also propose an improved JPS algorithm in symmetric grid graph. In this paper, we compare their search time and efficiency. Conducted experiment shows that by adopting the same benchmarks our algorithm not only tackles the congestion problem, but also solves the problem more efficiently. Experiments validated the improved JPS algorithm and results show it outperforms the other algorithms.

The 4H-SiC Schottky diode with 2-step mesa junction termination extension (JTE) structure has been investigated and optimized using SILVACO device simulator. Comparisons between different JTE structures of breakdown voltage and electric field crowding for Schottky diodes have been made. Simulation results show that the Space Modulated two-zone JTE has the highest breakdown voltage which is about 97% of the ideal 1D parallel plane conditions. With the novel 2-step MESA Single Implant JTE, the breakdown voltage could achieve the same as the parallel plane breakdown votlage. The influences of the surface charge (Qs) and the oxide passivation on the breakdown characteristics of 4H-SiC Schottky diode with JTE are also investigated. A reduced sensitivity of breakdown voltage with respect to P-implant doping concentration is obtained for a novel 2-step mesa JTE with an additional P-type guard ring.

Research in mechanical characteristics and material parameters of MEMS thin-films is significant to the MEMS device. In-situ measuring these parameters such as Young's modulus and stress gradient is essential to promote the performance of MEMS device. A center-anchored circular plate adopted as the test structure which is used to in-situ measure the material parameters of MEMS thin-films by a non-contact method has been present here. The advantages have been arranged. The influence to the resonance frequency with different dimensions of the anchor also has been studied.

In this article, a Multi-Regression Model is established to analyze the interrelations among those influencing factors related to the capital adequacy ratio (CAR) of banks based on some cross section data of 17 commercial banks in China from 2000–2006. It is found that the CAR is in positive correlation with bank capital and profit, assets income ability, historical influence and securities market but in negative correlation with the loan concentration ratio of banks.

In order to enhance the complexity and pseudo-randomness of encryption scheme, we presented a novel parameter variable chaotic system in this paper. Experimental results illustrated that our algorithm outperforms several classical chaotic systems. Moreover, we also proposed a class of dual scrambling image encryption algorithm employing the variable parameter chaos system to scramble pixel positions as well as values. Simulation and analysis demonstrated that our algorithm is more resistant against brutefroce attack with remarkable performance, which stands for high security and efficiency.

The existing chaos optimization algorithms were almost based on Logistic map. However, the probability density function of chaotic sequences for Logistic map is a Chebyshev type function, which may affect the global searching capacity and computational efficiency of chaos optimization algorithm. In this paper, firstly, a new chaotic sequences with Skew Tent map (STM) is established, and is improved by its iterative optimization property. Then, the Skew Tent map (STM) is introduced to perform the chaotic search. An adaptive chaos embedded particle swarm optimization algorithm combined with STM (STMACPSO) is proposed subsequently. The convergence speed and global optimal value of the presented algorithm are thus improved. Finally, The experiments with complex and Multi-dimensional functions demonstrate that STMACPSO outperforms the original CPSO in the global searching ability and convergence rate.

Chaos particle swarm optimization (CPSO) can not guarantee the population multiplicity and the optimized ergodicity, because its algorithm parameters are still random numbers in form. This paper proposes a new adaptive chaos embedded particle swarm optimization (ACEPSO) algorithm that uses chaotic maps to substitute random numbers of the classical PSO algorithm so as to make use of the properties of stochastic and ergodicity in chaotic search and introduces an adaptive inertia weight factor for each particle to adjust its inertia weight factor adaptively in response to its fitness, which can overcome the drawbacks of CPSO algorithm that is easily trapped in local optima. The experiments with complex and Multi-dimensional functions demonstrate that ACEPSO outperforms the original CPSO in the global searching ability and convergence rate.

Rayleigh criterion, based on wave optics, is imposed as a limit on the resolution of optical devices, but resolution beyond Rayleigh criterion has been reported. The current optical analysis and optical design is based on geometrical optics or wave optics. Hence we think that the resolution beyond Rayleigh criterion is owing to the limitations of the theoretical foundations. Also given to the need of analysis of crosstalk for some kinds of stereoscopic display, we applied path integral, instead of geometrical optics or physical optics, to the analysis of crosstalk. Crosstalk can be discussed in the form of probability amplitude and a constraint of time, not intensity of light. Meanwhile, various kinds of aberrations, being discussed as discrete entities now, may be integrated to be discussed under the frame of path integral. Finally, the distribution of crosstalk of this kind of stereoscopic display changes over time. Due to wave particle duality, we suggest that in this form, experiments should be designed to get a revised criterion, better than Rayleigh criterion, as a new limit on the resolution of optical devices.

An improved level set framework for fast segmentation based on single parameter is presented. The traditional level set methods for image segmentation need inevitably too many parameters adjustment and they have usually lower computationally implementation. To solve this problem, the proposed method improves the C-V PDE model by adding a penalized energy term and replacing the dirac function with the norm of level set function gradient. Besides, only the parameter of the length term is reserved in the model and an evolution criterion is introduced for the value rules of this single parameter. The experimental results of synthesized and biomedical images show that the new method is faster and more robust. Moreover, the new method has more extensive adaptability on account of the zero level set function being set anyplace freely and the single parameter adjustment convenience.

In order to obtain the optimum proportion and preparation conditions of fly ash (FA)-based aggregate for purifying phosphorus, orthogonal experiment was made on the mixture tests about the main raw materials—FA, the pore-forming material—dry sewage sludge (DSS) from the municipal sewage plant, the expanding agent—FGD gypsum from the thermal power plants and the binding agent—clay. And the impacts of main factors (drying time, presintering temperature, presintering time, sintering temperature, sintering time)on characteristics of aggregate (water absorption, apparent porosity, bulk density, weight loss in hydrochloric acid ) and the purification effects on low concentration phosphorus were studied. The results showed that the optimum proportion of fly ash to additives was that fly ash accounted for about 71%, dry sewage sludge about 25%, FGD gypsum about 2%, clay about 2%; The optimum preparation conditions were as follows: drying time 2h, without presintering, sintering temperature 850 ~ 900 ? and sintering time 24min. Under these conditions, the characteristics of aggregate were obtained as follows: water absorption 26.97 %, apparent porosity 48.35%, bulk density 1.79g/m3, weight loss in hydrochloric acid 4.46%, and the compressive strength absolutely meet the requirements of transport and use. And after 24h, the removal rate for 5mg / L phosphorus solution is 85.50%.

Reduced set method is an important approach to speed up support vector machine (SVM) training on large data sets. Existing works mainly focused on selecting patterns near the decision boundary for SVM training by applying clustering, nearest neighbor algorithm and so on. However, on very large data sets, these algorithms require huge computational overhead, and thus the total running time is still enormous. In this paper, an intuitive geometric method is developed to select convex hull samples in the feature space for SVM training, which has a time complexity that is linear with training set size n. Experiments on real data sets show that the proposed method not only preserves the generalization performance of the result SVM classifiers, but outperforms existing scale-up methods in terms of training time and number of support vectors.

One of the key factors that limit support vector machines (SVMs) application in large sample problems is that the large-scale quadratic programming (QP) that arises from SVMs training cannot be easily solved via standard QP technique. The sequential minimal optimization (SMO) is current one of the major methods for solving SVMs. This method, to a certain extent, can decrease the degree of difficulty of a QP problem through decomposition strategies, however, the high training price for saving memory space must be endured. In this paper, an algorithm in the light of the idea of parallel computing based on Symmetric multiprocessor (SMP) machine is improved. The new technique has great advantage in terms of speediness when applied to problems with large training sets and high dimensional spaces without reducing generalization performance of SVMs. .

A distinctive topology configuration can be obtained when the design domain has a similar size in all directions for a plane or three-dimensional problem. However, it is difficult to solve the topology optimization problem for a long-and-thin structure with variable cross-section using conventional algorithm. In order to solve this problem, a new method for topology optimization of similar periodic structures is proposed, based on the ideas of the variable displacement constraint limits. To satisfy the similar periodic constraint, the design domain is divided into a certain number of similar sub-regions which have the same number of elements. Relations of the relevant elements of different sub-regions are established. A mathematical programming formula and its procedure are proposed, being based on the idea of the independent, continuous and mapping method. And a set of displacement approximate formulation is derived and a new topology optimization method is developed and implemented. Several simulation examples show that the proposed method is of validity and effectiveness.

Sizing and shape structural optimization problems are normally stated in terms of a minimum weight approach with constraints that limit the maximum allowable stresses and displacements. However, topology structural optimization problems have been usually stated in terms of a maximum stiffness (minimum compliance) approach. In this kind of formulations, while no stress or displacement constraints are taken into account. Moreover in some FEM minimum weight topology optimization method with stress constraints formulation, transferred stress constraint functions cannot completely embody stress constraint requirements. In this paper, we build an equivalent optimization model for the topological optimization problem with the objective function being the structural weight and only stress constraints. In this model all element stress constraints of the structure being optimized under a load case are replaced by its most potential active stress constraint and average stress constraint. In order to make the stress constraint approximations hold true during an optimization process, we propose a solving strategy of varying stress limits. And a set of stress sensitivity formulation is derived and a new topology optimization method is developed and implemented. Several simulation examples show that stress sensitivity computation cost may be greatly reduced and there is not any objective oscillation phenomenon, and verify that the proposed method is of validity and effectiveness.

A novel optical fiber pressure sensor based on Fabry-Perot (FP) interferometer and phase demodulation method is described. MEMS techniques and the common communicational components are used to fabricate the sensor. The principles of pressure measurement and sensor design have been introduced. Phase demodulation method based on Fourier transformation is explored, which can reduce errors resulting from intensity variation of light source. Experimental results demonstrate that the sensor has reasonable linearity, sensitivity and a wide pressure measurement range from 0.1MPa to 3MPa.

Ion noise in output signal in klystron was obtained by particle in cell method. The phase fluctuation in output signal is caused by the change of the beam current DC velocity which is affected by the change of the ion amount. The phase of output signal changed with the change of beam electron amount.

This article modified the Tunneladder circuit expecting to widen the bandwidth and hope to carry on its merits. In this modified Tunneladder circuit, 'forward-wave' was assumed. Cross-section resembling the idealized unmodified circuit of Tunneladder is analyzed. Through making the longitudinal electronic fields satisfying the boundary condition at one point along the boundary, the dispersing equation and interaction impedance expression of Tunnelplane was derived. Good accordance between theory and HFSS of dispersion characteristics of Tunnelplane is acquired.

The paper presents an approach to characterize the material properties of multi-layered films using the electrostatic pull-in test structure. The structure is composed of a multi-layered fixed-fixed beam with different widths. An analytical model for the pull-in voltage of the multi-layered fixed-fixed beam with different widths is first presented. The model is then fitted using numerical analysis to improve its accuracy further. Based on this model, a novel in-situ method for simultaneously extracting Young's modulus and the residual stress of each layer for the multi-layered films is reported. Validation and accuracy of the extracting method have been confirmed by FEM simulation and experiments.

With the rapid development of ITS (intelligent transportation systems) technology and its application on urban expressway, the concept of highway level of service has been changing. The traffic condition with high speed and large volume is considered as good service quality. Based on the re-understanding of the concept of level of service, this paper selects non-dimensional traffic pressure and non-dimensional density as performance measures which embody the new concept well. After separating level of service of expressway into four classes, establishing each grey class whitening function and determining the corresponding threshold value, the service levels of expressway segment then have been determined. The validation of field data collected on segments of the 2/sup nd/, 3/sup rd/, 4/sup th/ and 5/sup th/ Urban Ring Road of Beijing indicated that the methodology put forward in this paper was correct and feasible.

It is a challenging task to provide Quality of Service (QoS) control for a shared high-speed downlink packet access (HSDPA) wireless channel. In this paper, we first propose a new dynamic resource control framework integrated with adaptive modulation and coding (AMC) and hybrid automatic repeat request (H-ARQ) to support class-based multimedia applications over HSDPA wireless channels. Then we present a new scheduling algorithm, Delay-sensitive Dynamic Fair Queueing (DSDFQ), to meet delay requirements of multimedia applications as well as maintain high network efficiency. The proposed approach can easily adapt to load fluctuations of different traffic classes and varying wireless channel conditions caused by user mobility, fading and shadowing. Performance evaluation shows the advantage of our proposed approach.