Predicting near-future photovoltaic generation

Patent metadata

Field	Value
Publication number	US-9053439-B2
Application number	US-201213631480-A
Country	US
Kind code	B2
Filing date	Sep 28, 2012
Priority date	Sep 28, 2012
Publication date	Jun 9, 2015
Grant date	Jun 9, 2015

Abstract

Systems and automated methods for predicting photovoltaic (PV) generation are provided. Weather forecast data and present-day and historical PV generation data are provided to respective predictors. The predictors derive weighted predictions that are used to calculate a Bayesian model average. Near-future generation by the PV system is predicted using the Bayesian model average. Production rates, worker scheduling, hours of operation and other planning decisions can be made in accordance with the predicted near-future generation.

First claim

What is claimed is:

1. A method performed using a computer, comprising:
deriving, by a processor of the computer, a first weighted prediction using a trained naïve Bayesian predictor and using weather forecast data for a present day, wherein the first weighted prediction predicts future electrical generation by a photovoltaic system;
deriving, by the processor of the computer, a second weighted prediction using a k-nearest neighbor predictor and using present day power generatio…

Assignees

• Hewlett Packard Development Co

Inventors

• Marwah Manish
• Arlitt Martin
• Chakraborty Prithwish
• Ramakrishnan Naren

Classifications

• G06Q10/04 Primary

  Physics · mapped topic

• F24J2200/04

  Mechanical Engineering · mapped topic

• H02S50/00

  Electricity · mapped topic

• F24S2201/00

  Mechanical Engineering · mapped topic

• Y02E10/50

  Cross-Sectional Technologies · mapped topic