Energy flow prediction for electric systems including photovoltaic solar systems
Methods, systems, and computer storage media are disclosed for determining electric energy flow predictions for electric systems including photovoltaic solar systems. In some examples, a method is performed by a computer system and includes supplying a consumption time series and a predicted production time series for an electric system to a machine-learning predictor trained during a prior training phase using electric energy consumption training data and photovoltaic production training data. The consumption time series has a first data resolution, and the electric energy consumption training data and the photovoltaic production training data have a second data resolution greater than the first data resolution. The method includes determining, using an output of the machine-learning predictor, a predicted import time series of electric import values each specifying an amount of electric energy predicted to be imported by the electric system with a prospective photovoltaic solar system installed.
1. A method comprising:
receiving electric energy flow data for a photovoltaic (PV) electric system at a first data resolution from one or more sources;
supplying the electric energy flow data to a machine-learning predictor trained during a prior training phase using energy flow training data comprises at a second data resolution greater than the first data resolution, the electric energy flow data comprises energy consumption time series and predicted production time series, wherein the energy flow training data and the predicted production time series data include energy values spaced apart by a plurality of time intervals at the second data resolution such that a first plurality of time intervals are spaced apart by a first time difference, a second plurality of time intervals are spaced apart by a second time difference, and wherein the first time difference is greater than the second time difference;
receiving an output responsive to the supplied electric flow energy data from the machine-learning predictor;
determining, using the output of the machine-learning predictor, an energy flow prediction at the first data resolution;
exporting the energy flow prediction for display on a user device.
2. The method of claim 1 , wherein receiving electric energy flow data from one or more sources includes receiving electric energy flow data from a user device, monitoring system associated with the PV electric system, an electric meter coupled to the PV electric system, a utility computer system, or a combination thereof.
3. The method of claim 1 , wherein the energy consumption time series has a lower resolution than a metering resolution used by an electric utility at a site associated with the PV electric system.
4. The method of claim 1 , wherein the determined energy flow prediction at the first data resolution is a predicted import time series of electric import values each specifying an amount of electric energy predicted to be imported by the PV electric system.
5. The method of claim 1 , wherein the machine-learning predictor is trained during the prior training phase with electric energy flow training data downsampled to the first resolution, and wherein the machine-learning predictor is configured to build a model to minimize self-consumption error between electric energy flow training data and downsampled electric energy flow training data.
6. The method of claim 5 , wherein the electric energy flow training data is downsampled to a first resolution used by a utility metering system.
7. The method of claim 1 , wherein exporting the energy flow prediction for display on a user device comprises presenting to a user, in a graphical user interface, a target physical configuration of the PV electric system.
8. The method of claim 1 , wherein exporting the energy flow prediction for display on a user device comprises presenting a predicted energy cost reduction by the PV electric system.
9. The method of claim 8 , wherein the predicted energy cost reduction is predicted based on the energy flow prediction for the PV electrical system and a utility rate structure.
10. An electronic device, comprising processing circuitry configured to:
transmit electric energy flow data for a photovoltaic (PV) electric system at a first data resolution to a server;
the server being configured to:
supply the electric energy flow data to a machine-learning predictor trained during a prior training phase using energy flow training data at a second data resolution greater than the first data resolution, the electric energy flow data comprises energy consumption time series and predicted energy production time series, wherein the energy flow training data and the predicted production time series data include energy values spaced apart by a plurality of time intervals at the second data resolution such that a first plurality of time intervals are spaced apart by a first time difference, a second plurality of time intervals are spaced apart by a second time difference, and wherein the first time difference is greater than the second time difference; and,
generate an energy flow prediction at the first data resolution, and
display the energy flow prediction generated by the server.
11. The method of claim 10 , wherein the energy consumption time series has a lower resolution than a metering resolution used by an electric utility at a site associated with the PV electric system.
12. The method of claim 10 , wherein the energy flow prediction generated by the server is a predicted import time series of electric import values each specifying an amount of electric energy predicted to be imported by the PV electric system.
13. The method of claim 10 , wherein the machine-learning predictor is trained during the prior training phase with electric energy flow training data downsampled to the first resolution, and wherein the machine-learning predictor is configured to build a model to minimize self-consumption error between electric energy flow training data and downsampled electric energy flow training data.
14. The method of claim 13 , wherein the electric energy flow training data is downsampled to a first resolution used by a utility metering system.
15. The method of claim 10 , wherein the processing circuitry is further configured to display a target physical configuration of the PV electric system.
16. The method of claim 10 , wherein the processing circuitry is further configured to display a predicted energy cost reduction by the PV electric system.
17. The method of claim 16 , wherein the predicted energy cost reduction is predicted based on the energy flow prediction for the PV electrical system and a utility rate structure.