| CPC G06F 16/9024 (2019.01) [G06Q 10/06316 (2013.01); G06Q 50/06 (2013.01); H04W 24/02 (2013.01)] | 7 Claims |
|
1. A non-transitory computer-readable medium comprising:
storing software comprising instructions executable by one or more computers which, upon such execution, cause the one or more computers to:
create a graph representing an electric vehicle (EV) charging network, the graph comprising a plurality of nodes interconnected by a plurality of edges, each edge of the plurality of edges corresponding to a relationship between two nodes;
categorize at least a first edge and a second edge of the plurality of edges as high impact edges;
perform a centrality assessment of the graph, wherein categorizing the first edge and the second edge is additionally based on an output of the centrality assessment, and
wherein performing the centrality assessment includes determining a number of edges that are outgoing from a node that corresponds to an asset and determining a centrality of the asset based on the number of edges that are outgoing from the node that corresponds to the asset, wherein the asset includes content in an actuatable pop-up window, and wherein the performing the centrality assessment includes at least one of:
determining the centrality of the asset includes determining a number of unique edges that are outgoing from the node that corresponds to the asset and determining the centrality of the asset based on the number of unique edges that are outgoing from the node that corresponds to the asset;
determining the centrality of the asset includes providing a prompt for a user to indicate a type of asset and determining the centrality of the asset based on the type of asset indicated by the user through the prompt;
perform a link utility index assessment of the graph, wherein categorizing the first edge and the second edge is additionally based on an output of the link utility index assessment, wherein the performing the link utility index assessment comprising:
identifying a number of new EV locations to be considered for a region which is under construction,
calculating a shortest path between each node of the plurality of nodes and store the shortest path as output in an edge list, and
calculate a link occurrence for each new EV location from the shortest path stored in the edge list;
calculate a base graph entropy for the graph;
remove the first edge from the graph to yield a first modified graph;
calculate a first entropy reduction factor based on the first modified graph;
remove the second edge from the graph to yield a second modified graph;
calculate a second entropy reduction factor based on the second modified graph;
determine an impact of each of the first edge and the second edge of the plurality of edges on the electric vehicle (EV) charging network by virtually removing one edge at a time based on the first entropy reduction factor and the second entropy reduction factor;
train a machine learning model using differential network entropy calculated using the first entropy reduction factor and the second entropy reduction factor, wherein training the machine learning model comprises:
reduce data to a minimal feature set by performing dimensionality reduction using an unsupervised training procedure,
apply a classification technique to the minimum feature set, wherein the classification technique includes a support vector machine (SVM) classifier, and
train, using a supervised training procedure, the machine learning model relative to the unsupervised training procedure by receiving input to reduce an amount of time and an amount of processing resources;
apply the machine learning model based on differential network entropy calculated using the first entropy reduction factor and the second entropy reduction factor, to select the first edge as one edge of the plurality of edges in the graph associated with the highest impact includes ranking each edge of the plurality of edges in order of impact on the electric vehicle (EV) charging network based on a relationships between each of the edges and a role of the edge in the electric vehicle (EV) charging network, wherein applying the machine learning model comprises instructions to:
receive a user request for identifying a new location for an electric vehicle (EV) charging station in a target region;
extract, using an existing EV location extractor, existing EV charging station locations in the target region based on selected map area by the user and filtering data based on the selected map area;
generate a map of the target region from the selected map area in which the existing electric vehicle (EV) charging station locations are indicated, wherein the map generation evaluates the target region based on a network map analyzer;
convert the map to the graph created to represent the electric vehicle (EV) charging network; and
display, via a user interface with the one or more computers, a recommendation for the optimal location of the-a new electric vehicle (EV) charging station, in which a first relationship in the electric vehicle (EV) charging network that corresponds to the first edge is identified as having the highest impact on the electric vehicle (EV) charging network.
|