Showing papers by "Shivkumar Kalyanaraman published in 2019"

A Deep Learning Approach to Solar-Irradiance Forecasting in Sky-Videos

Abstract: Ahead-of-time forecasting of incident solar-irradiance on a panel is indicative of expected energy yield and is essential for efficient grid distribution and planning. Traditionally, these forecasts are based on meteorological physics models whose parameters are tuned by coarse-grained radiometric tiles sensed from geo-satellites. This research presents a novel application of deep neural network approach to observe and estimate short-term weather effects from videos. Specifically, we use time-lapsed videos (sky-videos) obtained from upward facing wide-lensed cameras (sky-cameras) to directly estimate and forecast solar irradiance. We introduce and present results on two large publicly available datasets obtained from weather stations in two regions of North America using relatively inexpensive optical hardware. These datasets contain over a million images that span for 1 and 12 years respectively, the largest such collection to our knowledge. Compared to satellite based approaches, the proposed deep learning approach significantly reduces the normalized mean-absolute-percentage error for both nowcasting, i.e. prediction of the solar irradiance at the instance the frame is captured, as well as forecasting, ahead-of-time irradiance prediction for a duration for upto 4 hours.

A deep learning approach to solar-irradiance forecasting in sky-videos

Abstract: Ahead-of-time forecasting of incident solar-irradiance on a panel is indicative of expected energy yield and is essential for efficient grid distribution and planning. Traditionally, these forecasts are based on meteorological physics models whose parameters are tuned by coarse-grained radiometric tiles sensed from geo-satellites. This research presents a novel application of deep neural network approach to observe and estimate short-term weather effects from videos. Specifically, we use time-lapsed videos (sky-videos) obtained from upward facing wide-lensed cameras (sky-cameras) to directly estimate and forecast solar irradiance. We introduce and present results on two large publicly available datasets obtained from weather stations in two regions of North America using relatively inexpensive optical hardware. These datasets contain over a million images that span for 1 and 12 years respectively, the largest such collection to our knowledge. Compared to satellite based approaches, the proposed deep learning approach significantly reduces the normalized mean-absolute-percentage error for both nowcasting, i.e. prediction of the solar irradiance at the instance the frame is captured, as well as forecasting, ahead-of-time irradiance prediction for a duration for upto 4 hours.

Internet of things-enabled solar photovoltaic health monitoring and advising related thereto

Abstract: Methods, systems, and computer program products are provided herein in connection with IoT-enabled solar PV health monitoring and advising related thereto. A computer-implemented method includes obtaining current-voltage samples corresponding to solar photovoltaic modules by triggering switch circuitry between (i) an inverter attributed to the solar photovoltaic modules and (ii) a current-voltage tracer; detecting one or more anomalies in the obtained current-voltage samples by applying machine learning techniques to the obtained current-voltage samples; automatically performing a root cause analysis on the detected anomalies by (i) converting the obtained current-voltage samples to sequential data, (ii) applying a sequence classifier to the sequential data, and (iii) identifying a pre-determined anomaly class comparable to the sequential data based on the application of the sequence classifier; and automatically generating and outputting a suggestion for remedial action based on the identified pre-determined anomaly class.

Photonic harvesting string configuration

Abstract: One embodiment provides a method, including: receiving photovoltaic output from a solar module, the solar module comprising a plurality of solar panels electronically connected together by a plurality of strings, wherein the photovoltaic output corresponding to any given one of the strings is limited by the lowest performing solar panel in said given one of the strings; determining a reflection profile for the solar module, wherein the reflection profile describes an illumination pattern of light incident onto the solar module; and increasing, based upon the determined reflection profile, the photovoltaic output of the solar module to match a requested photovoltaic output, wherein the increasing comprises electrically reassigning one or more solar panels within the solar module from one string to a different string, such that each string electrically ties together solar panels that produce a photovoltaic output within a predetermined range of each other.

Dynamically controlled curved solar reflectors for flexible photovoltaic generation

Abstract: One embodiment provides a method, including: determining a light profile of light falling onto a solar module, wherein the light profile identifies (i) a position of the light with respect to the solar module and (ii) the intensity of the light with respect to solar panels within the solar module; identifying at least one solar panel within the solar module having partial light coverage; and changing the light profile by shaping the reflection of the light onto the solar module created by a flexible curved reflector in proximity to the solar module, thereby increasing the amount of light falling onto said at least one solar panel within the solar module; the changing the light profile comprising modifying the geometry of the flexible curved reflector by activating at least one actuator to move at least a portion of the flexible curved reflector.