Transmission lines are an important part of the power system, and the normal operation of the transmission line is a key step to ensure the stable operation of the power system. The failure of transmission line can cause interruption of power supply and cause serious economic losses. This work uses Convolutional Neural Networks (CNN) to detect and locate the transmission line faults. In the model, a time window is used to perform sliding sampling on the fault data to generate the data set, which not only speeds up the training process, but also retains the temporal correlation of the data. Two CNNs are used to achieve fault detection and location, respectively. In addition, Gaussian white noise is added to the test set to evaluate the anti-interference ability of the model. The results show that the model has good performance and high stability for transmission line fault detection and location.

Deep learning methods have exhibited remarkable perforamance for fault diagnosis in power grid transmission lines. However, specific power systems often demand dedicated classifiers. Alterations to the structure of grid can negatively impact the accuracy of the previously trained models. Retraining a new model will cost a lot of time and sacrifice more economic resources. In such cases, transfer learning is able to transfer the knowledge acquired from a previous task to a new target task. Consequently, this notably decreases the training costs. In this study, a convolutional neural network (CNN) with transfer learning is implemented for fault diagnosis in AC/DC hybrid power systems with various structures. Through a series of simulations, feasibility and efficiency of this method for fault diagnosis are proved. This method also shows strong robustness in fault detection with varying fault inception angle, fault resistance, fault location, and system frequency shuffling. Moreover, the improved CNN model in this paper exhibits powerful generalization ability on test data. The aforementioned advantages illustrate the potentiality of this method for fault diagnosis in real power grid.

In the modern society, physical infrastructure and information technology are inseparable. The concept of cyber-physical systems (CPSs) is then proposed and has been widely concerned by researchers in recent years. Various models have been introduced to meet the actual needs. In one type of those models, the physical part and the cyber part are coupled and influence each other. The cyber part monitors and controls the physical part, but at the same time it may also bring certain harm; the fault in one part may also be transmitted to the other and affect the performance of the counterpart. Here, this paper introduces an asymmetric interdependent model and studies how the coupling pattern of CPSs affects the system robustness. In addition, the coupling pattern of CPSs is optimized by using the simulated annealing (SA) algorithm to reduce the performance loss. The results of this paper may find applications in the future CPS planning.

Under the variety of information and communication technologies, the control center can securely and reliably operate power grid during the system disturbances and natural events. Specifically, when the initial failures are detected by cyber monitoring, the system will timely take emergency protecting operations to restrain the spreading of failures. Although cyber network makes the system more robust, they also increase the risk from the cyber network. In this paper, we firstly consider a control strategy into the cascading failures model. Then, we study how the cyber attack - False negatives attack confuses the control center, and makes the control center can not take emergency operation timely, causing the failure continue spreading. We propose an optimization problem to model the above assumptions and, by solving the optimization problem, study the impact of different attack scenarios on power system.

Graph embedding has been widely used to process various downstream tasks on large-scale graphs, i.e. node classification, community detection and link prediction. Among various embedding methods, Deep Graph Infomax (DGI) is a newly proposed method which achieves excellent performance in node classification. However, such outstanding achievement may cause the over-mining issue of user privacy and the robustness of this embedding method is still unexplored. In this paper, we investigate how to disturb the node classification accuracy of DGI from an attacker's perspective. We propose a novel attack method called Cross Entropy Attack (CEA), which aims to make target nodes be misclassified by DGI model with only limited edges being modified. By slightly changing the topological structure of a graph, CEA can successfully interfere with the accuracy of node classification in an unsupervised manner. Experiment results show that the proposed CEA obviously outperforms two baseline methods in terms of both Misclassified Rate (MR) and Average Modified of Edge (AME).

The theory of complex networks has been used in recent studies on robustness of power grids. In most of previous studies, however, the outage condition of nodes or links was not practical enough. Precisely, the outage of a node or link only depends on whether it is overloaded. While in real networks, a variety of factors such as aging of equipments or bad weather will also cause elements to fail. In this paper, we consider a probability failure model, which takes those factors into consideration. Then, we apply the model to study the robustness of power grids, and compare the results with those under the previous ideal model.

Network theory has been demonstrated as an effective approach for better understanding and analysis of software systems from a systematic perspective. In this paper, we develop a directed and weighted software dependency network model to analyse software systems from a complex network perspective. In particular, to measure the "High Cohesion and Low Coupling" nature of object-oriented software systems, we propose to use a directed and weighted modularity index, which can better reflect the cohesion of software systems. Experiments are conducted on a series of open source object-oriented software systems with various scales, and the results show that the directed and weighted modularity can better characterize software systems with different cohesion.

Real-world networks feature weights of interactions, where link weights often represent some physical attributes. In many situations, to recover the missing data or predict the network evolution, we need to predict link weights in a network. In this paper, we first proposed a series of new centrality indices for links in line graph. Then, utilizing these line graph indices, as well as a number of original graph indices, we designed three supervised learning methods to realize link weight prediction both in the networks of single layer and multiple layers, which perform much better than several recently proposed baseline methods. We found that the resource allocation index (RA) plays a more important role in the weight prediction than other topological properties, and the line graph indices are at least as important as the original graph indices in link weight prediction. In particular, the success application of our methods on Yelp layered network suggests that we can indeed predict the offline co-foraging behaviors of users just based on their online social interactions, which may open a new direction for link weight prediction algorithms, and meanwhile provide insights to design better restaurant recommendation systems.

This paper studies the mathematical modelling of Internet congestion control. Differently to previous models, which consider either the link capacity or the node processing capability as the constraint, here we take both of them into account, i.e., the aggregate flow rate on a link cannot exceed the link capacity and the aggregate flow rate at a node is limited by the node processing capability. A decentralized primal-dual algorithm, assuming the asynchronous case, is proposed to solve the congestion control problem and its convergence is proven. Using the proposed algorithms we show the bottleneck of the network performance when these two kinds of constraints are unbalanced. Further, we explore the effect of relating the capacities to network structural properties.

With the tremendous increase of the Internet traffic, achieving the best performance with limited resources is becoming an extremely urgent problem. In order to address this concern, in this paper, we build an optimization problem which aims to maximize the total utility of traffic flows with the capacity constraint of nodes and links in the network. Based on Duality Theory, we propose an iterative algorithm which adjusts the rates of traffic flows and capacity of nodes and links simultaneously to maximize the total utility. Simulation results show that our algorithm performs better than the NUP algorithm on BA and ER network models, which has shown to get the best performance so far. Since our research combines the topology information with capacity constraint, it may give some insights for resource allocation in real communication networks.

The continuous growing number of applications in current communication networks highlights the necessity for efficient resource allocation strategies. Associated with the current widely-used TCP Reno and Vegas protocols, we build an optimization problem to find the optimal resource allocation strategy. The optimization problem is solved by using the Lagrangian dual method and the exact solution is derived. This optimal strategy can guarantee the globally optimal traffic performance compared with other resource allocation strategies.

Assortativity is an important structural characteristic of complex networks. In this paper, we propose a simple method to generate a network with the desired assortative coefficient but still keep the degree distribution unchanged. We simulate cases with different assortative coefficients and calculate the packet drop probability when the buffer size is limited. The simulation results indicate that assortativity has a significant effect on the traffic performance. Our study helps to understand how the network structure influences the traffic dynamics in complex networks.

We study the tolerance of complex networks to attacks. Due to the large network scale, it is almost impossible for an attacker to have the complete topology information about the whole network. Thus, an efficient attack strategy based on partial degree information is proposed. Using the generating function method we give the exact solution for the attack strategy. A theoretical scale-free random network and the real Internet data are considered as examples, and the results clearly show the performance degradation due to the lack of topology information. By comparing to previous strategies, the efficiency of our strategy is demonstrated.

The problem of traffic congestion in complex networks is studied. Two kinds of complex network structures, namely random graphs and scale-free networks, are considered. In terms of the structure of connection, random graphs are homogeneous networks whereas the scale-free networks are heterogeneous networks. For both types of networks, we introduce an additional scale-free feature in the load generation process such that a small number of nodes are more heavily loaded than others. A traffic model similar to the routing algorithm in computer networks is used in our simulation study. We show how the network structures and parameters influence the traffic congestion status.