Search

Community detection is crucial for analyzing the structure of social networks and extracting hidden information from them. The goal is to find groups of nodes (communities) with high intra-group and low inter-group communications. This problem is NP-hard, and most existing algorithms are global with high computational complexity, especially for large networks. Recently, local methods with acceptable computational complexity have been developed, but many have low accuracy and are non-deterministic. This paper introduces a new local algorithm, LCD-SN, which identifies communities based on first- and second-degree neighbor nodes. Unlike other local algorithms, LCD-SN is highly accurate, definitive, and not dependent on initial seed nodes. Additionally, a new index is proposed to determine the importance of network nodes using their local characteristics (first- and second-degree neighbors). Using this index, LCD-SN first identifies important nodes, forms initial communities with these nodes and their first-degree neighbors, and then obtains final communities through post-processing. Experiments show that LCD-SN is effective in identifying communities in social networks.

Partitioning water distribution networks (WDNs) into district metered areas offers benefits including reduced nonrevenue water and simplified pressure management. However, current research in this field tends to narrowly focus on the topological relationships among nodes, often overlooking the influence of pressure reducing valves (PRVs) and pump stations on clustering results. To address this limitation, this study introduces a topology-distance-based clustering (TDBC) method that enhances the accuracy of partitioning by explicitly considering the impact of PRVs and pump stations. In the TDBC method, the Floyd algorithm is initially employed to construct a topological distance matrix that quantifies the degree of node connectivity. By amplifying topological distances for links including PRVs and pump stations, their effect on clustering results is incorporated. Subsequently, nodes are clustered using the K-means algorithm based on the resulting topology-distance matrix. The proposed TDBC approach is applied to four network cases, and its outcomes are compared with those of two traditional methods. The comparative analysis indicates that the TDBC algorithm achieves precise partitioning results for networks incorporating PRVs or pump stations, while ensuring a harmonious balance between modularity and the uniformity of the partitioning results, even in networks with complex structures and highly interconnected loops. HIGHLIGHTS Incorporation of PRVs and pump stations: The TDBC method introduces a novel approach by explicitly incorporating the influence of pressure reducing valves (PRVs) and pump stations into the clustering process.; Topological distance-based clustering: The TDBC method shifts from traditional geographical proximity-based methods to a topological distance-based approach.; Improved clustering uniformity.;

Distributed architectures wherein multiple decision-making units are employed to coordinate their decision-making/actions based on real-time communication have become increasingly important for monitoring processes that have large scales and complex structures. Typically, the development of a distributed monitoring scheme involves two key steps, that is, the decomposition of the process into subsystems, and the design of local monitors based on the configured subsystem models. In this article, we propose a distributed process monitoring approach that tackles both steps for large-scale processes. A data-driven process decomposition approach is proposed by leveraging community structure detection to divide variables into subsystems optimally via finding a maximal value of the metric of modularity. A two-layer distributed monitoring scheme is developed where local monitors are designed based on the configured subsystems of variables using canonical correlation analysis. Inner-subsystem interactions and inter-subsystem interactions are tackled by the two layers separately, such that the sensitivity of this monitoring scheme to certain types of faults is improved. We utilize a numerical example to illustrate the effectiveness and superiority of the proposed method. It is then applied to a simulated wastewater treatment process. [ABSTRACT FROM AUTHOR]

• A distributed framework for simultaneous state and parameter estimation. • Subsystem decomposition considering parameters using community detection. • A distributed MHE scheme for maximum information extraction. In this paper, we consider distributed simultaneous state and parameter estimation for a class of nonlinear systems, for which the augmented model comprising both the states and the parameters is only partially observable. Specifically, we first illustrate how the sensitivity analysis (SA) can select variables for simultaneous state and parameter estimation. Then, a community structure detection (CSD) based process decomposition method is proposed for dividing the entire system into interconnected subsystems as the basis of distributed estimation. Next, we develop local moving horizon estimators based on the configured subsystem models, and the local estimators communicate with each other to exchange their estimates. Finally, an SA and CSD based distributed moving horizon estimation (DMHE) mechanism is proposed. The effectiveness of the proposed approach is illustrated using a chemical process consisting of four connected reactors. [ABSTRACT FROM AUTHOR]

In this paper, a new fine-grained evolutionary model, called CCLA-EM, is proposed for solving the optimization problems, which greatly overcomes the premature convergence problem of the existing evolutionary algorithms. In the proposed model, a combination of an evolutionary algorithm with a cellular learning automaton is used. The population individuals are distributed on the cells of a cellular learning automaton. Each individual interacts and cooperates with the individuals of neighboring cells to reach the global optimum. Distributing the population individuals on the cells of a cellular learning automaton allows the parallel implementation of the proposed model. Also, in different stages of the proposed model, numbers generated by a chaotic process are used instead of random ones. The use of numbers generated by a chaotic process leads to a complete search of the search space and hence avoids being trapped in local optima. Experiments on various benchmarks of the community structure detection problem indicate the superiority of the proposed model to the well-known algorithms GA-net and ICLA-net. [ABSTRACT FROM AUTHOR]