Concerning the problem that the local feature and rotation invariant feature cannot be jointly paid attention to in traditional attention-based neural networks, a multi-branch neural network model based weakly supervised fine-grained image classification method was proposed. Firstly, the lightweight Class Activation Map (CAM) network was utilized to localize the local region with potential semantic information, and the residual network ResNet-50 with deformable convolution and Oriented Response Network (ORN) with rotation invariant coding were designed. Secondly, the pre-trained model was employed to initialize the feature networks respectively, and the original image and the above regions were input to fine-tune the model. Finally, the three intra-branch losses and between-branch losses were combined to optimize the entire network, and the classification and prediction were performed on the test set. The proposed method achieves the classification accuracies of 87.7% and 90.8% on CUB-200-2011 dataset and FGVC_Aircraft dataset respectively, which are increased by 1.2 percentage points, and 0.9 percentage points respectively compared with those of the Multi-Attention Convolutional Neural Network (MA-CNN) method. On Aircraft_2 dataset, the proposed method reaches 91.8% classification accuracy, which is 4.1 percentage points higher than that of ResNet-50. The experimental results show that the proposed method improves the accuracy of weakly supervised fine-grained image classification effectively.

In view of the shortcomings of the current path planning strategies for multiple robots, such as high path coupling, long total path, long waiting time for collision avoidance, and the resulting problems of low system robustness and low robot utilization, a multi-robot path planning algorithm based on 3D spatiotemporal maps and motion decomposition was proposed. Firstly, the dynamic temporary obstacles in time dimension were generated according to the existing path set and the current robots' positions, and were expanded into 3D search space together with the static obstacles. Secondly, in the 3D search space, the total time of path motion was divided into three parameters:motion time, turning time, and in-situ dwell time, and the conditional depth first search strategy was used to calculate the set of all paths from the starting node to the target node that met the parameter requirements. Finally, all paths in the path set were traversed. For each path, the actual total time consumption was calculated. If the difference between the actual total time consumption and the theoretical total time consumption of a path was less than the specified maximum error, the path was considered as the shortest path. Otherwise, the remaining paths were continued to traverse. And if the differences between the actual total time and the theoretical total time of all paths in the set were greater than the maximum error, the parameters needed to be adjusted dynamically, and then the initial steps of algorithm were continued to execute. Experimental results show that, the path planned by the proposed algorithm has the advantages of short total length, less running time, no collision and high robustness, and the proposed algorithm can solve the problem of completing continuous random tasks by multi-robot system.

When the smart grid phasor measurement equipment competes for limited network communication resources, the data packets will be delayed or lost due to uneven resource allocation, which will affect the accuracy of power system state estimation. To solve this problem, a Sampling Awareness Weighted Round Robin (SAWRR) scheduling algorithm was proposed. Firstly, according to the characteristics of Phasor Measurement Unit (PMU) sampling frequency and packet size, a weight definition method based on mean square deviation of PMU traffic flow was proposed. Secondly, the corresponding iterative loop scheduling algorithm was designed for PMU sampling awareness. Finally, the algorithm was applied to the PMU sampling transmission model. The proposed algorithm was able to adaptively sense the sampling changes of PMU and adjust the transmission of data packets in time. The simulation results show that compared with original weighted round robin scheduling algorithm, SAWRR algorithm reduces the scheduling delay of PMU sampling data packet by 95%, halves the packet loss rate and increases the throughput by two times. Applying SAWRR algorithm to PMU data transmission is beneficial to ensure the stability of smart grid.

Aiming at the problems of limited monitoring area and difficult target locating in video security surveillance system, a radar-guided video linkage monitoring model was established with the characteristics of wide radar monitoring range and freedom from optical conditions. On this basis, a target location algorithm and a multi-target selection algorithmm were proposed. Firstly, according to the target information detected by radar, the corresponding camera azimuth and pitch angle of a moving target in the system linkage model were automatically calculated so that the target could be accurately locked, monitored and tracked by camera in real-time. Then, with multiple targets appearing in the surveillance scene, the multi-target selecting algorithm was used for data weighted fusion of discrete degree of target, radial velocity of target and the distance between target and camera to select the target with the highest priority for intensive monitoring. The experimental results show that, the locating accuracy of the proposed target location algorithm for pedestrians and vehicles can reach 0.94 and 0.84 respectively, which can achieve accurate target location. Moreover, the proposed multi-target selection algorithm can effectively select the best monitoring target in complex environment, and has good robustness and real-time performance.

The problem of Total Variation (TV)-based image denoising was considered. Since the traditional serial computation speed based on Central Processing Unit (CPU) was low, a parallel computation based on Graphics Processing Unit (GPU) was proposed. The dual model of the total variation-based image denoising was derived and the relationship between the primal variable and the dual variable was considered. The projected gradient method was applied to solve the dual model. Numerical results obtained by CPU and GPU show that the algorithm implemented by GPU is more efficient than that by CPU, and with the increasing of image size, the advantage of GPU parallel computing is more outstanding.

To improve the reliability of the Wireless Sensor and Actuator Network (WSAN), an optimal design method based on Genetic Algorithm (GA) for WSAN fault detection filter was proposed. In system modeling, the influence of the wireless network transmission delay on network control system was modeled as an external noise, the composite optimization index which is composed of sensitivity and robustness was made as the design goal of fault detection filter, and the optimization objective was made as the core of GA—the fitness function. At the same time, according to the numerical characteristics of optimization objective in WSAN, the corresponding real coding, uniform mutation, arithmetic crossover and other processing methods were selected to speed up the convergence rate, meanwhile taking the accuracy of the calculation results into account. The optimized filter design mentioned herein, not only restrains the noise signal, but also amplifies the fault signal. Finally, the effectiveness of the proposed design is demonstrated by the results of Matlab/OMNET++ hybrid simulations.

To increase the diversity among the selected members to enhance the performance of the ensemble system, an ensemble Extreme Learning Machine (ELM) based on the selection of members similarity named EELMBSMS was proposed. Firstly, some candidate ELMs with high classification ability were selected. Then, Particle Swarm Optimization (PSO) algorithm was used to select the optimal subset of the ensemble members according to the similarity among the members. The diversity of the selected members was improved by selecting those ELMs with low similarity, which improved the classification performance of the ensemble system effectively. The selected ELMs obtained better performance with different integration rules. The experimental results on four UCI datasets verify that EELMBSMS has better stability and better generalization than some classical ensemble extreme learning machines.

To solve the routing confusion of the conventional hierarchical layout algorithm in the large-scale program, based on the Sugiyama hierarchical layout algorithm, this paper proposed an optimized algorithm for channel routing, which dynamically adjusted the number of channel according to the program size. In order to solve the low efficiency and lines overlap, the algorithm built functional relationships between channel number and program size. And by using the generalized tensor balance algorithm to reduce the crossings and realize the artistic layout. The algorithm also gave the corresponding line distribution and application strategy in accordance with the relative positional relationship between the calling nodes to achieve the ordered routing. The experimental results show that the algorithm has greater layout efficiency. It can reduce the crossings effectively, realize clear layout and is easy to implement.

We presented a novel approach for empty space skipping for object-order volume rendering. A two-staged space skipping was introduced: the first stage applied bricking on a regular grid, and the second stage used octree to reach a finer granularity. The approach further took into account that rendered volume may exceed the available texture memory and allow fast runtime changes of transfer function. For the bottleneck that heavy workload was assigned to CPU in our approach, a novel algorithm was proposed. The algorithm efficiently computed slicing for texture based volume render in Graphic Processing Unit (GPU); balanced the workload between CPU and GPU. Combining the two approaches above, we can render large volume data efficiently without compromising the image quality.

A kind of method of web data mining on association rule,authoritative Web pages and Weblog items with quantum Grover search algorithm and probability were given in this paper,and then the algorithm to be optimal in mining a large volumes of unsorted and unstructured Web data was dicussed.

A development framework for human-oriented workflow management system was designed and implemented. A case of specific scenario using this developement framework was given. The framework was divided into three layers, business object layer, runtime layer, and business entity layer. Business object layer provided efficient and concise API; Runtime layer set up a manageable runtime environment to make application run reliably; Business entity layer stored and maintained business object. Developing application with the framework will decrease the development complexity and make system maintain easier.

A novel algorithm for mining user navigation pattern with incremental clustering was presented. Firstly, a new method for expressing user interest was introduced to construct user profile object. Based on the basic concept of ant colony clustering, artificial ants were used to pick up or drop down object to implement clustering by analyzing the similarity with other local regional objects and. Then a mechanism of decomposing clusters was used to form new clusters when users'interests changed. Experimental results show that the method can adaptively and efficiently achieve incremental clustering.

This paper illustrated a new approach and environment to describe and simulate the embedded system. Firstly, auther constructed the CDM models to describe the design requirements of the specified system. Secondly, in order to get the aim of verification of a specified system, according to the converting rules, the approach automatic converted the requirements documents to the CDM models and SystemC codes to simulate a specified system. Finally, an application example using the above approach was illustrated.