Search

Course scheduling is closely related to the allocation of teaching resources in colleges and universities, which is an important problem in teaching tasks. It belongs to the NP-complete problem of typical combinatorial optimization, which needs to consider many constraints. Based on the learning of various algorithms and usage feedback of existing scheduling systems, this paper adopts a genetic algorithm which can flexibly consider constraints and has a huge optimization space to construct a course scheduling system. In the system, we design a special fitness function and use the value of this function to measure the degree that the scheduling result of a certain course satisfies the constraint. At the same time, the design of fitness function facilitates the addition of more constraints. In addition, we add a roulette algorithm with skip and filter functions to make the matching of Course-Teacher-Class and locations more reasonable. In conclusion, through the improved genetic algorithm, the system has accomplished many works, such as the convenient addition of constraints, the reduction of data space complexity and the optimization of chromosome genetic strategy. As a result, we realize a widely used course scheduling system, which is conducive to the efficient development of teaching work.

Intelligent warehouse management system (IWMS) attracts lots of interest by increasing the productivity and efficiency, minimizing the number of human workers and decreasing errors. As an important part of IWMS, warehouse portal in-out registration still face with enormous challenge. Usually, multiple UHF RFID reader antennas are installed on the portal, in-out registration can be performed by estimating the locations of RFID tags which are attached on the goods. Even though the precise localization accuracy is not required, it's still hard to distinguish the import and export directions. In this paper, we analyzed the fluctuation of received signal strength indicator (RSSI) backscattered from the RFID tags and proposed a movement direction judging method based on double-tag with an exact spacing distance. Experimental results have shown that the double-tag method can distinguish the moving direction exactly in portal environments and realize efficient in-out registration.

RF signals are vulnerable to multipath, noise and interference. Most localization algorithms are trying to create a similarity based relationship between RF signal measurements and propagation distance or tag location. So it's a huge challenge to realize precise localization for real applications. In this paper, we propose a novel passive UHF RFID tag localization algorithm based on RF signal fluctuation, FBL algorithm. Not considering similarity, we use signal fluctuation as the neighboring criteria. According to our experimental results, with a metal shelf covered with multiple fixed tags, FBL algorithm can estimate 15 grid-level locations with an average error of about 32 cm. As the grid spacing distance was about 35 cm and the accuracy of traditional k nearest neighbor (KNN) algorithm was about 72 cm, the results showed that FBL algorithm significantly reduced the location estimation error.

Target size is an important indicator for reconstructing the contour of the radar target. Early accurate prediction of target size is beneficial for subsequent target evaluation. In our paper, we propose a method to predict the target size for radar target using machine learning based on its monostatic Radar Cross Section (RCS). Our prediction scheme comprises a deep neural network (DNN) model with a multi-hidden layer. The experimental results show that the proposed approach achieves 100% prediction accuracies for the height and radius values of the cylinder target.

The sample size directly affects the accuracy of the item parameter estimation of the online test. We simulate several tests comprised of 5 items, 9 items, 21 items and 41 items with different difficulties. Then, the Monte Carlo method is used to simulate candidates’ answers to test items by Rasch response model. From this, we calculate the error metric of the test item distribution under different sample sizes. By comparing the real data with simulation data, we find that the fitting effect is good.

This paper proposes a new Distributed Bayesian Algorithm (DBA) for data fault detection in order to address problems of data fault in wireless sensor networks. Although many methods for data fault detection have been proposed, the diagnosis accuracy is still very low especially in the situation of the densely deployed WSN with a large number of faulty nodes. In DBA, the Bayesian network is introduced to calculate the fault probability of sensor nodes, and the fault probability will be adjusted by exploiting the border nodes to improve the accuracy of fault probability and avoid the negative impact of the large number of faulty nodes. A comprehensive simulation has been made to compare the performance of DBA and Distributed Fault Detection (DFD). The simulation results show DBA can improve the fault detection accuracy significantly even in the situation with a large number of faulty nodes.

Light trapping induced by plasmonic metal nanoparticles is of extensive interest for photovoltaics. In this paper, plasmonic effect of silver nanoparticles embedded in dielectric layers hybrid structures on c-Si photoactive absorbing material is presented. Dielectric SiN x and SiO 2 layers were considered and deposited by plasma enhanced chemical vapor deposition (PECVD) and electron beam evaporation (EBE) respectively. Silver nanoparticles were prepared by annealing of thermally evaporated silver thin films ranging from 5 to 25 nm in N 2 gas atmosphere at 260 °C for an hour. For comparative study, thicknesses of each layer including silver thin flim, SiN x spacer layer and c-Si wafer were varied to attempt preliminary judgement of the appropriate structure. It is shown that reflection near bandgap of Si can be reduced while silver nanoparticles are fabricated on SiN x layer and covered by SiO 2 layer. By comparing the morphologies of nanoparticle arrays resulted from different thicknesses of silver thin film, it is indicated that nanoparticle arrays with large average sizes and high surface coverage on the front surface can greatly increase undesirable reflection in visible wavelength region in spite of their advantage in reflection reducing at longer wavelength. In addition, light trapping of the hybrid structures for varying thicknesses of SiN x spacer layer and c-Si wafer were also investigated respectively. Results suggest that hybrid plasmonic structures (HPS) can play an even greater part as c-Si wafers become thinner.

This paper firstly sums up the main security threats to data confidentiality in cloud computing environment, then classifies and introduces the existing access control models according to the core strategy they have adopted. Lastly this paper points out the main challenges facing access control in cloud computing environment and the subsequent research directions.

It is summarized the research hotspots of flue gas desulfurization (FGD) and analyzed the advantages and disadvantages of Limestone-Gypsum wet flue gas desulfurization (WFGD) and microbic desulfurization methods. Put forward a new idea of combining the Limestone-Gypsum WFGD, microbial method with catalytic oxidation of transition metal (Fe3+/Fe2+ galvanic couple) reasonably. Then established the model of the new type method and analyzed the mechanism and feasibility deeply, promoted the development of FGD technic.

Fast and accurate estimation of background model in video sequences is a basic task in many computer vision and video analysis applications. Underwater vision is a new area and the background of underwater has special quality such as unstable light spot, water ripple. To this end, this paper proposes an algorithm based on Kalman Filter, which is applied to the estimation of dynamic underwater background with a static monitoring camera of swimming pool's bottom. Experimental on several underwater video sequences performing the model can efficiently adapt to the environmental of underwater.

The effect of the tilted substrate on the lithography patterns in maskless Laser Interference Lithography (MLIL) was numerically and experimentally studied. The experiments showed that the period of interference patterns becomes larger with increasing tilt angles of the substrate, as is in good agreement with theoretical analysis. It was also shown that there is no evident change when the rotational angle was less than 3 degree in our experiment. The significance of our analytical results for MLIL and the fabrication of integrated circuit was discussed.

2600 years ago, YinYang was born in the ancient China with the theme "Taiji (one matter) produces YinYang; YinYang produces four forms; the four forms lead to eight diagrams; and the eight diagrams can have 64 direct combinations." This theme can be called YinYang 1-2-4-8-64. An equilibrium-based scientific decoding of YinYangl-2-4-8-64 is presented which states "Equilibrium (the one matter) has two poles (YinYang); the two poles can be mapped to a bipolar lattice with four bipolar truth values (four forms); the four values lead to eight bipolar dynamic operators (eight diagrams); and the eight operators can have 64=82 direct biological or neurobiological or any dynamic combinations. " The scientific decoding results in a bipolar dynamic logic (BDL) that satisfies four pairs of dual dynamic DeMorgan's laws and a YinYang bipolar universal modus ponens. Bipolar granular neurobiological computing with BDL is introduced.

The notion of a rough set was originally proposed by Pawlak (1982). Later on, there is a fast growing interest in this theory. In this paper we present a more general approach to the generalization of rough sets. Specifically, generalized formulation has been studied by using a binary relation on two universes without any restriction on the cardinality. The algebraic properties of generalized rough sets are given, and the extension principle for crisp sets is explained as the upper approximations of rough sets. The relationship between the extension principle for fuzzy sets and the upper approximations of fuzzy rough sets is investigated.

Combining rough set theory and self-organizing feature map (SOFM) neural network, a method was presented for the car's plate character recognition. The features of the training samples were extracted to build up the decision table; the discretization algorithm of decision attributes was proposed based on the clustering ability of SOFM network; the rough set theory was applied to reduce the decision table. Finally, the reduced decision attributes were used to construct neural network recognizing machine. The method can reduce the numbers of attributes in the decision table, simplify the structure of neural network and improve the ability of generality. The experiment results of the car's plate character recognition show that the algorithms are practical and effective.