Distribution grid topology identification encoding known toplogial information
A computer-implemented method for identifying a topology of a power distribution grid having a number of transformers includes acquiring measurement signals of one or more electrical quantities pertaining to nodes of the power distribution grid. A graph representation is generated using the measurement signals and grid topological information, wherein the measurement signals pertaining to respective nodes are used to derive node features and the grid topological information is used to encode edges representing certain and uncertain connections between the nodes. The graph representation is processed using a graph neural network to classify the nodes and output a mapping of each of the nodes to one of the transformers, whereby a status of the uncertain connections is determined.
1 . A computer-implemented method for identifying a topology of a power distribution grid comprising a plurality of transformers, the method comprising:
acquiring measurement signals of one or more electrical quantities pertaining to a plurality of nodes of the power distribution grid,
generating a graph representation using the measurement signals and grid topological information, wherein the measurement signals pertaining to respective nodes are used to derive node features and wherein the grid topological information is used to encode edges representing connections between the nodes, the encoding configured to distinguish certain connections from uncertain connections, and
processing the graph representation using a graph neural network to classify the nodes and output a mapping of each of the nodes to one of the transformers, whereby a status of the uncertain connections is determined,
wherein the measurement signals are continuously sampled over a moving time window to detect power distribution grid topology changes in real time.
2 . The method according to claim 1 , wherein the edges of the graph representations are encoded by assigning edge weights to respective edges, the edge weights defined by a first weight assigned to edges representing certain connections between adjacent a second weight assigned to edges representing uncertain connections, the second weight being less than the first weight.
3 . The method according to claim 1 , wherein the power distribution grid comprises a number of switching devices, wherein uncertain connections are represented by edges that are realized via a switching device, and wherein the determined status of the uncertain connections is indicative of a status of the switching devices.
4 . The method according to claim 1 , wherein the graph neural network is trained on a dataset comprising a number of graph representations generated using measurement signals collected over different time windows and the grid topological information.
5 . The method according to claim 4 , wherein the graph neural network is trained in a supervised or semi-supervised learning process wherein at least some of the nodes in the dataset are assigned ground truth labels indicative of a transformer mapping.
6 . The method according to claim 5 , wherein the nodes assigned ground truth labels during the training of the graph neural network include “transformer nodes” that are directly connected to a primary or secondary side of the transformers.
7 . The method according to claim 5 , wherein the nodes assigned ground truth labels during the training of the graph neural network includes nodes having known associations with transformers that are not impacted by uncertain connections.
8 . The method according to claim 1 ,
wherein the measurement signals pertaining to respective nodes comprise time series data, and
wherein the node features are determined by processing the time series data pertaining to the respective nodes to determine spectral embeddings that define respective feature vectors for each node, wherein the spectral embeddings are determined based on computing a similarity measure between the time series data pertaining to each pair of nodes of the plurality of nodes.
9 . The method according to claim 1 ,
wherein the measurement signals pertaining to respective nodes comprise time series data,
wherein the node features are defined by data samples of the time series data pertaining to the respective nodes, and
wherein the graph neural network includes a recurrent neural network-based graph neural network.
10 . The method according to claim 1 , wherein the measurement signals are associated with multiple electrical quantities defining multiple channels of time series data, wherein the node features are defined by concatenating feature vectors derived from the individual channels.
11 . The method according to claim 1 , wherein the one or more electrical quantities being measured include active power and voltage at the nodes.
12 . The method according to claim 1 , wherein the one or more electrical quantities being measured include only active power at the nodes.
13 . The method according to claim 1 , wherein the power distribution grid comprises measured and unmeasured nodes, wherein the graph representations include only measured nodes.
14 . A non-transitory computer-readable storage medium including instructions that, when processed by a computing system, configure the computing system to perform the method according to claim 1 .
15 . A system for identifying a topology of a power distribution grid comprising a plurality of transformers, the system comprising:
measurement devices for communicating measurement signals of one or more electrical quantities pertaining to a plurality of nodes of the power distribution grid, and
a computing system, comprising:
one or more processors, and
a memory storing algorithmic modules executable by the one or more processors, the algorithmic modules comprising:
a graph generator engine configured to generate a graph representation using the measurement signals and grid topological information, wherein the measurement signals pertaining to respective nodes are used to derive node features and wherein the grid topological information is used to encode edges representing connections between the nodes, the encoding configured to distinguish certain connections from uncertain connections, and
a node-transformer mapping engine configured to utilize a graph neural network for processing the graph representation to classify the nodes and output a mapping of each of the nodes to one of the transformers, whereby a status of the uncertain connections is determined,
wherein the measurement signals are continuously sampled over a moving time window to detect power distribution grid topology changes in real time.