Optimal planning of power distribution networks with fault-tolerant configuration - Dataset

The dataset contains five test networks. For each test network, there is a certain number of nodes located in the city center, with a higher spatial density, and there are other nodes located in the suburbs, with a lower spatial density. The load demand of each SS is calculated using two different Gaussian probability distributions, one associated with SSs located in the city center and the other with SSs located in the suburbs. The edges connecting each pair of nodes have a cost proportional to the Euclidean distance between the nodes, to which a random component uniformly distributed in a range of values proportional to the Euclidean distance is added. In each network, a natural obstacle that may be present in a real case, such as a river, is simulated using a piecewise linear function. Whenever an edge crosses the obstacle, an additional cost is added to the connection, much greater than that associated with the Euclidean distance. A weighted graph is thus obtained (the weights are the costs of edges) and the total cost of an arc connecting two nodes s and t is defined as the cost of the shortest path starting from source node s and ending at target node t. Lastly, for a random number of node triplets in the network, the triangle inequality is intentionally violated, so that the sum of the costs of two connections is less than the cost of the third connection.