Predicting the Evolution Process of Infrastructure Networks with an NSIPA Link Prediction Method

Infrastructure networks, such as power grids, have been developing rapidly. There is a growing demand of predicting the future evolution process of power grids in the planning of infrastructure systems. As the structure of power grids evolves to be more complex, the grids can be studied under complex network theory by extracting structural properties. Accordingly, the prediction of evolution process of power grids can be regarded as a link prediction problem. To correspond with the actual self-organization regulations of power grids, a novel link prediction model based on preferential attachment is proposed in this paper. After experimenting on several real world power grids, results show that our model outperforms other six popular models in various kinds of realistic networks. It is demonstrated as a possible methodology of solving the prediction problem of power grid evolution.