An Advanced Graph Embedding Framework with Node Embedding to Design Water Pipeline Networks
Alvida Mustika Rukmi, Ary Mazharuddin Shiddiqi, Ahmad Saikhu, Muhammad Jefri Fransiska, Leo Mazeau · 2023
This paper introduces a novel approach to developing pipeline networks using graph embedding techniques. The proposed framework involves creating a graph representation of the pipeline network and embedding it into a lower-dimensional space. By node embedding, as a part of graph embedding, our approach applies the relation and closeness between nodes to obtain network optimization by minimum costs while satisfying the demand points represented by nodes on the graph. While node embedding generation, the approach uses a random walk process to generate sequences of nodes and a skipgram process to train neighbors of a node. Spectral and Kruskal’s algorithm supports the optimization of network layout. This contribution adds to the existing body of research on graph embedding and its application in network optimization, offering a new perspective on the development of pipeline networks. Our results highlight the potential of graph embedding as an effective tool for engineers and planners in designing pipeline networks that meet community needs, minimize environmental impact, and maximize sustainability.