Abstract: A system and a computer-implemented method for training a learning algorithm for siting a new energy generation facility comprises executing simultaneous simulations over multiple electrical power generation capacities by accessing land parcel characteristics, including electricity transmission and market demand data. A clustering algorithm produces a first training set with expected electricity output for subsets of land parcels. Combined scores are calculated based on cumulative land size, transmission length, and projected market revenue, ranked by highest score, and stored by lowest construction cost. The method identifies changes in land parcel characteristics, determines if the changes exceed a threshold, and, if so, re-trains the clustering algorithm by re-executing the steps using the updated characteristics as a second training set.

Abstract: Disclosed is a system and method for identifying various combinations of parcels of land with sufficient transmission, resources, market demand, and available land to build a wind farm or solar farm. Different criteria associated with each parcel of land include land characteristics including size, ownership, tree coverage, elevations, terrain, buildable land, transmission characteristics including substation hardware costs, network upgrades, general tie-in costs, market characteristics including historical locational marginal pricing (LMPs), and resource characteristics including net capacity factor (NCF) of wind or solar. In one example, the present invention identifies clusters of land parcels that minimize the number of land owners, maximize buildable land, minimize transmission cost, maximize NCF, and maximize nodal LMPs.