Systems and methods for photovoltaic fault detection using a feedback-enhanced positive unlabeled learning

Patent metadata

Field	Value
Publication number	US-2022138631-A1
Application number	US-202117517447-A
Country	US
Kind code	A1
Filing date	Nov 2, 2021
Priority date	Nov 3, 2020
Publication date	May 5, 2022
Grant date	—

Abstract

Various embodiments of a system and associated method for identifying and classifying faults in a photovoltaic array using relatively little labeled data are described herein. In particular, the system builds on existing PU classification techniques by addition of a feedback loop that enables classification of limited operational data of a photovoltaic array by expanding a plurality of features within the operational data based on a learned importance of each feature.

First claim

What is claimed is:

1 . A system, comprising:
a processor in communication with a memory, the memory including instructions, which, when executed, cause the processor to:
provide a classification model to classify a node of a plurality of nodes, wherein each node is associated with a plurality of features and wherein the classification model defines a plurality of weights;
select a plurality of most influential features of the plurality of features based on a magnitude of each weight of the plurality of weights of the classification model;
increase a dimensionality of each feature of the selected plurality of most influential features to obtain an expanded feature space;
minimize one or more features of the expanded feature space that do not substantially contribute to classification of the node of the plurality of nodes; and
apply the classification model to the expanded feature space.

2 . The system of claim 1 , further comprising:
a photovoltaic array including a plurality of photovoltaic panels, wherein each photovoltaic panel or group of photovoltaic panels is associated with a respective node of the plurality of nodes at a point in time; and one or more sensors in operative communication with the photovoltaic system and the processor, wherein the one or more sensors is operable to capture operational data for the photovoltaic system and communicate the operational data to the processor.

3 . The system of claim 2 , wherein the memory further includes instructions which, when executed, cause the processor to:
receive, by the processor, operational data from the one or more sensors, wherein the operational data includes photovoltaic data including the plurality of features for each node of the plurality of nodes of the photovoltaic array.

4 . The system of claim 1 , wherein the memory further includes instructions which, when executed, cause the processor to:
classify the node of the plurality of nodes into a first class or a second class based on operational data associated with the node at a point in time, wherein the first class is indicative of a fault and wherein the second class is indicative of no fault.

5 . The system of claim 1 , wherein the classification model is a Modified Logistic Regression classifier.

6 . The system of claim 5 , wherein the Modified Logistic Regression classifier defines a weighted combination of the plurality of features, wherein each feature of the plurality of features is associated with a respective weight of the plurality of weights.

7 . The system of claim 1 , wherein the memory further includes instructions which, when executed, cause the processor to:
learn the classification model including the plurality of weights based on the operational data.

8 . The system of claim 1 , wherein the dimensionality of each feature is increased by adding p-level polynomial combinations of the selected plurality of most influential features to the dataset.

9 . The system of claim 1 , wherein the step of increasing the dimensionality of each feature of the selected plurality of most influential features introduces a non-linear decision boundary into a feature space descriptive of the plurality of features of the dataset.

10 . The system of claim 1 , wherein the memory further includes instructions which, when executed, cause the processor to:
receive an amount k of most influential features to be selected.

11 . The system of claim 1 , wherein the step of minimizing one or more features of the expanded feature space that do not substantially contribute to classification is achieved using a dimensionality reduction algorithm.

12 . The system of claim 11 , wherein the dimensionality reduction algorithm is a Principal Component Analysis algorithm.

13 . The system of claim 1 , wherein the step of minimizing one or more features of the expanded feature space that do not substantially contribute to classification is achieved using a regularization algorithm.

14 . The system of claim 1 , further comprising:
receiving a value indicative of a level of polynomial enhancements (p).

15 . The system of claim 1 , wherein the dataset includes operational data for a photovoltaic array, wherein the photovoltaic array includes the plurality of nodes and wherein each node of the plurality of nodes of the photovoltaic array is associated with the plurality of features at a point in time.

16 . A system, comprising:
a photovoltaic array including a plurality of photovoltaic panels, wherein each photovoltaic panel or group of photovoltaic panels is associated with a respective node of the plurality of nodes, wherein each node of the plurality of nodes is associated with operational data at a point in time; a processor in communication with a memory and the photovoltaic array, the memory including instructions, which, when executed, cause the processor to:
provide a classification model to classify a node of the plurality of nodes based on the operational data,
wherein each node of the plurality of nodes is associated with a plurality of features and wherein the classification model defines a plurality of weights;
select a plurality of most influential features of the plurality of features based on a magnitude of each weight of the plurality of weights of the initial classification model;
increase a dimensionality of each feature of the selected plurality of most influential features to obtain an expanded feature space;
minimize one or more features of the expanded feature space that do not substantially contribute to classification of the node of the plurality of nodes; and
apply the classification model to the expanded feature space.

17 . The system of claim 16 , wherein the classification model is a Modified Logistic Regression classifier.

18 . The system of claim 17 , wherein the Modified Logistic Regression classifier defines a weighted combination of the plurality of features, wherein each feature is associated with a respective weight of the plurality of weights.

19 . The system of claim 1 , wherein the dimensionality of each feature is increased by adding p-level polynomial combinations of the selected plurality of most influential features to the dataset.

20 . The system of claim 1 , wherein the step of minimizing one or more features of the expanded feature space that do not substantially contribute to classification is achieved using a dimensionality reduction algorithm.

21 . A method, comprising:
providing, by a processor, a classification model to classify a node of a plurality of nodes based on operational data, wherein each node is associated with a plurality of features and wherein the classification model defines a plurality of weights; selecting a plurality of most influential features of the plurality of features based on a magnitude of each weight of the plurality of weights of the initial classification model; increasing a dimensionality of each feature of the selected plurality of most influential features to obtain an expanded feature space; minimizing one or more features of the expanded feature space that do not substantially contribute to classification of the node of the plurality of nodes; and applying the classification model to the expanded feature space.

22 . The method of claim 21 , further comprising:
receiving, by the processor, operational data from the one or more sensors, wherein the operational data includes photovoltaic data including the plurality of features

Assignees

• Jaskie Kristen
• Martin Joshua
• Spanias Andreas
• Univ Arizona State

Inventors

• Jaskie Kristen
• Martin Joshua
• Spanias Andreas

Classifications

• 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

• G06N20/10 Primary

  using kernel methods, e.g. support vector machines [SVM] · CPC title

• G06N20/00 Primary

  Machine learning · CPC title