Prognostics and health management of photovoltaic systems

Patent metadata

Field	Value
Publication number	US-9939485-B1
Application number	US-201314023296-A
Country	US
Kind code	B1
Filing date	Sep 10, 2013
Priority date	Nov 14, 2012
Publication date	Apr 10, 2018
Grant date	Apr 10, 2018

Abstract

The various technologies presented herein relate to providing prognosis and health management (PHM) of a photovoltaic (PV) system. A PV PHM system can eliminate long-standing issues associated with detecting performance reduction in PV systems. The PV PHM system can utilize an ANN model with meteorological and power input data to facilitate alert generation in the event of a performance reduction without the need for information about the PV PHM system components and design. Comparisons between system data and the PHM model can provide scheduling of maintenance on an as-needed basis. The PHM can also provide an approach for monitoring system/component degradation over the lifetime of the PV system.

First claim

What is claimed is:

1. A system that facilitates analyzing an output of a photovoltaic (PV) system, comprising:
a modeling component configured to receive data associated with at least one parameter relating to current operation of the PV system, and further configured to output an anticipated operating metric relating to the at least one parameter;
an operating component configured to receive data relating to an output of the PV system during the current operation of the PV system, and further configured to output an actual operating metric relating to the output of the PV system;
a comparison component configured to determine whether a difference exists between the value of the anticipated operating metric and the actual operating metric; and
an evaluation component configured to determine, in response to the comparison component determining that a difference exists between the value of the anticipated operating metric and the actual operating metric, whether the difference exceeds a threshold, the threshold being a defined value, percentage, or mathematical formula;
wherein:
the modeling component comprises an artificial neural network (ANN) and the comparison component is configured to compute at least some of the differences between anticipated and actual operating metrics by determining model residuals of the ANN;
the artificial neural network has been configured based upon training data obtained prior to the current operation of the PV system;
for determining whether the operating metric difference exceeds a threshold, the evaluation component considers metrics that relate to a time window of a first defined duration; and
the evaluation component is further configured to determine for a time window of a second defined duration shorter than the first defined duration whether the ANN model residuals indicate a fault condition.

2. The system of claim 1 , further comprising an alarm component configured to activate an alarm in response to the evaluation component determining that the difference exceeds the threshold.

3. The system of claim 1 , wherein the evaluation component is further configured to compare training data with the data relating to an output of the PV system during the current operation of the PV system to identify a cause for the difference between the value of the anticipated operating metric and the actual operating metric.

4. The system of claim 1 , wherein the at least one parameter is at least one of a plane of array (POA) irradiance, a wind speed, an ambient air temperature, a direct current (DC) power, a DC current, a DC voltage, an alternating current (AC) power, an AC current, or an AC voltage.

5. The system of claim 1 , wherein the actual operating metric relating to the output of the PV system relates to at least one of a plane of array (POA) irradiance, a wind speed, an ambient air temperature, a direct current (DC) power, a DC current, a DC voltage, an alternating current (AC) power, an AC current, or an AC voltage.

6. A method, comprising:
receiving data associated with at least one parameter relating to a current operating condition of an energy system;
generating, from the data, an anticipated operating metric relating to the at least one parameter;
receiving data relating to an output of the energy system during the current operating condition of the energy system;
generating an actual operating metric relating to the output of the energy system;
determining whether a difference exists between the anticipated operating metric and the actual operating metric; and
when a difference is determined to exist between the anticipated operating metric and the actual operating metric, determining whether the difference exceeds a threshold, the threshold being a defined value, percentage, or mathematical formula;
wherein:
the determining whether a difference exists and the determining whether the difference exceeds a threshold are based at least in part on model residuals of an artificial neural network (ANN) that has been configured based upon training data obtained prior to the current operation of the energy system;
the steps of determining whether a difference exists and determining whether the difference exceeds a threshold are performed using anticipated operating metrics and actual operating metrics that relate to a time window of a first defined duration; and
the method further comprises determining, for a time window of a second defined duration shorter than the first defined duration, whether the ANN model residuals indicate a fault condition.

7. The method of claim 6 , wherein the energy system is at least one of a photovoltaic (PV)-based energy system, a wind-based energy system, a hydropower-based energy system, a tidal-based energy based system, a wave-based energy based system, a geothermal-based energy based system, a biomass-based energy based system, a renewable energy-based energy based system, a nuclear energy based system, or a coal-based energy system.

8. The method of claim 6 , further comprising, in response to determining the difference exceeds a threshold, activating an alarm.

9. The method of claim 6 , wherein the at least one parameter is at least one of a plane of array (POA) irradiance, a wind speed, an ambient air temperature, a direct current (DC) power, a DC current, a DC voltage, an alternating current (AC) power, an AC current, or an AC voltage.

10. The method of claim 6 , wherein the actual operating metric relating to the output of the PV system relates to at least one of a plane of array (POA) irradiance, a wind speed, an ambient air temperature, a direct current (DC) power, a DC current, a DC voltage, an alternating current (AC) power, an AC current, or an AC voltage.

11. A computer-readable storage medium comprising instructions that, when executed by a processor, cause the processor to perform acts comprising:
receiving data associated with at least one parameter relating to a current operating condition of a photovoltaic (PV) system;
generating, from the data, an anticipated operating metric relating to the at least one parameter;
receiving data relating to an output of the PV system during the current operating condition of the PV system; and
generating an actual operating metric relating to the output of the PV system;
determining whether a difference exists between the anticipated operating metric and the actual operating metric; and
when a difference is determined to exist between the anticipated operating metric and the actual operating metric, determining whether the difference exceeds a threshold, the threshold being a defined value, percentage, or mathematical formula;
wherein:
the determining whether a difference exists and the determining whether the difference exceeds a threshold are based at least in part on model residuals of an artificial neural network (ANN) that has been configured based upon training data obtained prior to the current operation of the energy system;
the steps of determining whether a difference exists and determining whether the difference exceeds a threshold are performed using anticipated operating metrics and actual operating metrics that relate to a time window of a first defined duration; and
the acts further comprise determining, for a time window of a second defined duration shorter than the first defined duration, whether the ANN model residuals indicate a fault condition.

12. The system of claim 1 , wherein the evaluation component is configured to base determinations of fault conditions on probability distributions of the ANN model residuals, and wherein said probability distributions are based on the training data.

13.

Assignees

• Sandia Corp
• Nat Tech & Eng Solutions Sandia Llc

Inventors

• Johnson Jay
• Riley Daniel

Classifications

• H02J2101/24

  Photovoltaics · CPC title

• G01R31/2605 Primary

  Physics · mapped topic

• G08B21/182 Primary

  Level alarms, e.g. alarms responsive to variables exceeding a threshold · CPC title

• H02S50/10

  Testing of PV devices, e.g. of PV modules or single PV cells (testing of semiconductor devices during manufacturing {H10P74/00}) · CPC title

• H02S50/00 Primary

  Monitoring or testing of PV systems, e.g. load balancing or fault identification · CPC title