Machine learning for optimization of power distribution to electric vehicle charging ports

1. A method for dynamically controlling distribution of power amongst a plurality of charging station ports that includes a particular port, comprising:

obtaining input data that includes a sequence of time intervals and a charging ramp-up time of a vehicle that is connected to the particular port in a particular time interval of the sequence of time intervals;

training, by reinforcement learning that is based on the input data, one or more machine learning engines to determine a particular power distribution, among the plurality of charging station ports, wherein the reinforcement learning comprises determining a scalar number for each objective in a plurality of objectives that include:

maximizing a number of vehicles serviced in the particular time interval,

maximizing a daily power output, and

minimizing a use of peak-time power; and

configuring the plurality of charging station ports according to the particular power distribution to:

decrease the power allocated to the particular port during a peak time and

increase the power to the particular port during a non-peak time;

wherein the method is performed by one or more computers.

2. The method of claim 1 wherein the plurality of charging station ports includes

at least one charging station port on a single-port charging station.

3. The method of claim 1 wherein the plurality of charging station ports includes

at least multiple charging station ports located on a single multi-port charging station.

4. The method of claim 1 , wherein the input data includes non-user data that comprises at least one of:

data concerning an owner or an associated business of the one or more charging stations; or

data concerning a third party.

5. The method of claim 1 , wherein at least some of the input data corresponds to one or more features, and wherein the one or more machine learning engines are trained based on data corresponding to the one or more features, wherein the one or more features include at least one of:

a current state of a vehicle battery to be charged;

information regarding a user's current trip;

a current time of day;

a current cost of energy;

an energy source for charging; or

individual charger capacity for each of one or more charging stations associated with the plurality of charging station ports.

6. The method of claim 1 , wherein the input data comprises at least one of:

real-time data from one or more input devices located at one or more charging stations associated with the plurality of charging station ports;

user identifying information from a card reader, a mobile device display, or a wireless signal;

current trip information received from a mobile device using an application program interface (API);

user or vehicle identifying information from image recognition using one or more sensors at the one or more charging stations, or by receiving wireless radio signals transmitted by a vehicle; or

outputs from other machine learning engines.

7. The method of claim 1 , wherein the particular power distribution specifies a maximum charging rate for each of the plurality of charging station ports, or a percentage of a power budget for each of the plurality of charging station ports.

8. The method of claim 1 , wherein the plurality of objectives comprise at least one of:

maximizing a revenue of one or more charging stations associated with the plurality of charging station ports; or

ensuring vehicles are charged sufficiently to reach a next charging station in a current route.

9. The method of claim 1 , wherein the one or more machine learning engines are trained based on data corresponding to one or more features, and wherein the one or more features include at least one of:

locations of a plurality of charging stations, associated with the plurality of charging station ports, where charging power is at least partially from solar panels;

locations of a plurality of charging stations, associated with the plurality of charging station ports, where charging power is at least partially from local battery storage;

types of home charging mechanisms associated with one or more current charging station users, including charging speed and battery storage size; or

cost of electricity at each time interval, at each of a plurality of charging stations associated with the plurality of charging station ports.

10. The method of claim 1 , wherein the reinforcement learning (RL) comprises obtaining observation inputs at recurring time intervals, wherein the observation inputs include

control values including a percentage of power directed to each of the plurality of charging station ports.

11. The method of claim 1 , further comprising processing, from the input data, at least one of user data or non-user data through the one or more machine learning engines to further generate and configure at least one of:

parameters of one or more charging stations associated with the plurality of charging station ports, including pricing, for influencing driver behavior; or

an energy source selection for the one or more charging stations.

12. One or more computing devices configured to dynamically control distribution of power amongst a plurality of charging station ports that includes a particular port, by:

obtaining input data that includes a sequence of time intervals and a charging ramp-up time of a vehicle that is connected to the particular port in a particular time interval of the sequence of time intervals;

training, by reinforcement learning that is based on the input data, one or more machine learning engines to determine a particular power distribution, among the plurality of charging station ports, wherein the reinforcement learning comprises determining a scalar number for each objective in a plurality of objectives that include:

maximizing a number of vehicles serviced in the particular time interval,

maximizing a daily power output, and

minimizing a use of peak-time power; and

configuring the plurality of charging station ports according to the particular power distribution to:

decrease the power allocated to the particular port during a peak time and

increase the power to the particular port during a non-peak time.

13. A non-transitory computer readable medium comprising instructions executable by a processor to dynamically control distribution of power amongst a plurality of charging station ports that includes a particular port, the instructions comprising instructions for:

obtaining input data that includes a sequence of time intervals and a charging ramp-up time of a vehicle that is connected to the particular port in a particular time interval of the sequence of time intervals;

training, by reinforcement learning that is based on the input data, one or more machine learning engines to determine a particular power distribution, among the plurality of charging station ports, wherein the reinforcement learning comprises determining a scalar number for each objective in a plurality of objectives that include:

maximizing a number of vehicles serviced in the particular time interval,

maximizing a daily power output, and

minimizing a use of peak-time power; and

configuring the plurality of charging station ports according to the particular power distribution to:

decrease the power allocated to the particular port during a peak time and

increase the power to the particular port during a non-peak time.

14. The non-transitory computer-readable media of claim 13 wherein the plurality of charging station ports includes at least one charging station port on a single-port charging station.

15. The non-transitory computer-readable media of claim 13 wherein the plurality of charging station ports includes at least multiple charging station ports located on a single multi-port charging station.

16. The non-transitory computer-readable media of claim 13 , wherein the input data includes non-user data, and the non-user data that comprises at least one of:

data concerning an owner or an associated business of the one or more charging stations; or

data concerning a third party.

17. The non-transitory computer-readable media of claim 13 , wherein

at least some of the input data corresponds to one or more features, and wherein the one or more machine learning engines are trained based on data corresponding to the one or more features, wherein the one or more features include at least one of:

a current state of a vehicle battery to be charged;

information regarding a user's current trip;

a current time of day;

a current cost of energy;

an energy source for charging; or

individual charger capacity for each of one or more charging stations associated with the plurality of charging station ports.

18. The non-transitory computer-readable media of claim 13 , wherein the input data comprises at least one of:

real-time data from one or more input devices located at one or more charging stations associated with the plurality of charging station ports;

user identifying information from a card reader, a mobile device display, or a wireless signal;

current trip information received from a mobile device using an application program interface (API);

user or vehicle identifying information from image recognition using one or more sensors at the one or more charging stations, or by receiving wireless radio signals transmitted by a vehicle; or

outputs from other machine learning engines.

19. The non-transitory computer-readable media of claim 13 , wherein the particular power distribution specifies a maximum charging rate for each of the plurality of charging station ports, or a percentage of a power budget for each of the plurality of charging station ports.

20. The non-transitory computer-readable media of claim 13 , wherein the plurality of objectives comprise at least one of:

maximizing a revenue of one or more charging stations associated with the plurality of charging station ports; or

ensuring vehicles are charged sufficiently to reach a next charging station in a current route.

21. The non-transitory computer-readable media of claim 13 , wherein the one or more machine learning engines are trained based on data corresponding to one or more features, and wherein the one or more features include at least one of:

locations of a plurality of charging stations, associated with the plurality of charging station ports, where charging power is at least partially from solar panels;

locations of a plurality of charging stations, associated with the plurality of charging station ports, where charging power is at least partially from local battery storage;

types of home charging mechanisms associated with one or more current charging station users, including charging speed and battery storage size; or

cost of electricity at each time interval, at each of a plurality of charging stations associated with the plurality of charging station ports.

22. The non-transitory computer-readable media of claim 13 , wherein the reinforcement learning (RL) comprises obtaining observation inputs at recurring time intervals, wherein the observation inputs include:

control values including a percentage of power directed to each of the plurality of charging station ports.

23. The non-transitory computer-readable media of claim 13 , wherein the instructions are configured to process, from the input data, at least one of user data or non-user data through the one or more machine learning engines to further generate and configure at least one of:

parameters of one or more charging stations associated with the plurality of charging station ports, including pricing, for influencing driver behavior; or

an energy source selection for the one or more charging stations.