Scalably generating distribution grid topology

1 . A method for generating a composite representation of an electric power grid, the method comprising:

obtaining two or more datasets comprising data indicating a presence of or operation of assets of the electric power grid in a geographical region that has the electric power grid, each dataset including data of a particular kind, wherein the assets comprise utility poles and connecting lines between the utility poles;

identifying one or more respective predictive models for each dataset, each predictive model being configured to process the respective kind of data of the respective dataset to generate a corresponding output that specifies a respective partial representation of the electric power grid;

generating, using each of the predictive models that have been identified for the two or more datasets, respective partial representations of the electric power grid that each characterize a different electrical aspect of the electric power grid in the geographical region, wherein the generating comprises:

generating, using a first predictive model that is configured as a convolutional neural network, predicted locations of the utility poles included in the assets based on processing a first dataset that comprises imagery data of the geographical region, and

assigning the predicted locations of the utility poles as locations of anchor points for a composite representation of the electric power grid; and

aggregating, in accordance with a set of aggregation rules, the respective partial representations to generate the composite representation of the electric power grid, wherein the composite representation of the electric power grid comprises information about the locations of the anchor points and information about connections between the anchor points, wherein aggregating comprises removing a particular output of one predictive model based on output generated by at least one other predictive model such that the particular output is excluded from the composite representation of the electric power grid.

2 . The method of claim 1 , wherein obtaining two or more datasets comprises obtaining two or more of the first dataset, a second dataset, a third dataset, or a fourth dataset, wherein:

the second dataset comprises sensor data taken in the geographical region, the second dataset comprising one or more categories of sensor data, the sensor data in each category including sensor measurements of assets of the electric power grid in the geographical region;

the third dataset comprises LIDAR data taken in the geographical region, the third dataset comprising one or more categories of LIDAR data; and

the fourth dataset comprises utility data taken in the geographical region, the fourth dataset comprising one or more categories of utility data.

3 . The method of claim 2 , wherein obtaining two or more datasets comprises obtaining the first dataset, the second dataset, the third dataset, and the fourth dataset.

4 . The method of claim 2 , wherein the one or more categories of sensor data comprise distribution line sensor data, smart meter readings, equipment submeter readings, consumer device readings, including readings from sensors for home accessories and mobile devices, standalone sensor data, including readings from sensors in wall plugs or breakers, or ground-based field sensor data.

5 . The method of claim 2 , wherein the one or more categories of LIDAR data comprise data obtained by respective LIDAR sensors on-board one or more moving ground or aerial vehicles.

6 . The method of claim 2 , wherein the one or more categories of utility data comprise data specifying utility-recorded asset locations, data collected through supervisory control and data acquisition (SCADA) system, or data specifying input from line crews.

7 . The method of claim 2 , wherein the one or more categories of sensor data or utility data comprise voltage, current, or electromagnetic field intensity measurements.

8 . The method of claim 1 , wherein:

the electric power grid comprises one or more electric power transmission networks and one or more electric power distribution networks; and

generating the representation of the electric power grid comprises generating a representation of feeders within the electric power distribution networks, and generating a representation of high voltage power lines that connect one or more power generators to one or more substations within the electric power transmission networks.

9 . The method of claim 1 , further comprising:

providing the generated representation of the electric power grid for display on a user device.

10 . The method of claim 1 , wherein the at least one respective partial representation further comprises a representation of categories of the assets.

11 . The method of claim 1 , wherein generating the corresponding output that specifies the partial representation of the electric power grid comprises generating data that identifies respective physical characteristics of the assets of the electric power grid in the geographical region.

12 . The method of claim 1 , wherein generating the corresponding output that specifies the partial representation of the electric power grid comprises generating data that identifies respective connections between respective assets of the electric power grid.

13 . The method of claim 1 , wherein the imagery data comprise one or more of: satellite imagery, aerial imagery, drone imagery, hyperspectral imagery, infrared imagery, or depth map.

14 . The method of claim 13 , wherein the composite representation comprises a topological representation of the electric power grid.

15 . The method of claim 14 , further comprising using the topological representation to direct electricity to a first area instead of a second area.

16 . The method of claim 1 , wherein the set of aggregation rules comprises one or more disambiguation rules, and wherein aggregating the respective partial representations comprises disambiguating a corresponding output of each predictive model in accordance with the one or more disambiguation rules, and wherein the one or more disambiguation rules comprise rejecting outputs of the predictive models that violate one or more assets placement rules.

17 . The method of claim 1 ,

removing the particular output of the one predictive model based on outputs generated by the at least one other predictive model comprises removing the particular output when outputs from at least two other predictive models that agree with one another and do not agree with the particular output.

18 . The method of claim 1 , further comprising:

receiving a user input specifying one or more modifications to the composite representation; and

regenerating the composite representation by incorporating the modifications.

19 . A system comprising one or more computers and one or more storage devices storing instructions that when executed by the one or more computers cause the one or more computers to perform operations for generating a composite representation of an electric power grid, the operations comprising:

obtaining two or more datasets comprising data indicating a presence of or operation of assets of the electric power grid in a geographical region that has the electric power grid, each dataset including data of a particular kind, wherein the assets comprise utility poles and connecting lines between the utility poles;

identifying one or more respective predictive models for each dataset, each predictive model being configured to process the respective kind of data of the respective dataset to generate a corresponding output that specifies a respective partial representation of the electric power grid;

generating, using each of the predictive models that have been identified for the two or more datasets, respective partial representations of the electric power grid that each characterize a different electrical aspect of the electric power grid in the geographical region, wherein the generating comprises:

generating, using a first predictive model that is configured as a convolutional neural network, predicted locations of the utility poles included in the assets based on processing a first dataset that comprises imagery data of the geographical region, and

assigning the predicted locations of the utility poles as locations of anchor points for a composite representation of the electric power grid; and

aggregating, in accordance with a set of aggregation rules, the respective partial representations to generate the composite representation of the electric power grid, wherein the composite representation of the electric power grid comprises information about the locations of the anchor points and information about connections between the anchor points, wherein aggregating comprises removing a particular output of one predictive model based on output generated by at least one other predictive model such that the particular output is excluded from the composite representation of the electric power grid.

20 . The system of claim 19 , wherein obtaining two or more datasets comprises obtaining two or more of the first dataset, a second dataset, a third dataset, or a fourth dataset, wherein:

the second dataset comprises sensor data taken in the geographical region, the second dataset comprising one or more categories of sensor data, the sensor data in each category including sensor measurements of assets of the electric power grid in the geographical region;

the third dataset comprises LIDAR data taken in the geographical region, the third dataset comprising one or more categories of LIDAR data; and

the fourth dataset comprises utility data taken in the geographical region, the fourth dataset comprising one or more categories of utility data.

21 . The system of claim 19 , wherein the composite representation comprises a topological representation of the electric power grid.

22 . One or more non-transitory computer storage media storing instructions that when executed by one or more computers cause the one or more computers to perform operations for generating a composite representation of an electric power grid, the operations comprising:

obtaining two or more datasets comprising data indicating a presence of or operation of assets of the electric power grid in a geographical region that has the electric power grid, each dataset including data of a particular kind, wherein the assets comprise utility poles and connecting lines between the utility poles;

identifying one or more respective predictive models for each dataset, each predictive model being configured to process the respective kind of data of the respective dataset to generate a corresponding output that specifies a respective partial representation of the electric power grid;

generating, using each of the predictive models that have been identified for the two or more datasets, respective partial representations of the electric power grid that each characterize a different electrical aspect of the electric power grid in the geographical region, wherein the generating comprises:

generating, using a first predictive model that is configured as a convolutional neural network, predicted locations of the utility poles included in the assets based on processing a first dataset that comprises imagery data of the geographical region, and

assigning the predicted locations of the utility poles as locations of anchor points for a composite representation of the electric power grid; and

aggregating, in accordance with a set of aggregation rules, the respective partial representations to generate the composite representation of the electric power grid, wherein the composite representation of the electric power grid comprises information about the locations of the anchor points and information about connections between the anchor points, wherein aggregating comprises removing a particular output of one predictive model based on output generated by at least one other predictive model such that the particular output is excluded from the composite representation of the electric power grid.