SlideShare a Scribd company logo
1 of 55
AUTOMATION OF DEMAND
RESPONSE MANAGEMENT
M.V.N.VINAY BALAKRISHNA
AM.EN.U4ME12137
PROJECT GUIDE:
K.SIVASELVI
BUSINESS CONTEXT
 In US, the consumption of electricity is highest in the
summer due to the increased use of electrical appliances
 Regional utilities implement DR programs and shed / clip
electricity by offering financial incentives to end
customers as per standard rules and regulations
 Events are called when the forecasted electrical load
exceeds 96% of summer peak load
 The aim is to build an automated model that will
accurately predict occurrences of events with least
manual intervention
BUSINESS CONTEXT – SGQ
SGQ (State-Gap-Question) representing the current state, the gap
preventing us from reaching the final state, the key question that must
be answered to reach the final state and the final state
DEMAND RESPONSE (DR)
 Reduction in usage of electricity by end customers, in
response to the events called by regional utility by
curtailing peak electricity load
 Involves generation, transmission and distribution of
electric load through proper planning & scheduling
 They’re generally implemented in the summer as usage
of electrical appliances is at its peak
 The incentives offered are proportional to the amount of
electricity that was clipped
DEMAND RESPONSE (DR)
 Maximum of 6 events can be called for the summer that
is for 4 months
 Customers are informed of events in advance via
Curtailment Service Providers (CSPs) who are third party
vendors
 Hence, customers efficiently manage electrical usage
and subsequently, their costs
OBJECTIVES
 To predict events based on the past 10 years historical
data for a client in Philadelphia considering all factors
such as weather, humidity, precipitation etc.
 Develop a reliable framework by process flow
automation through real time collection, creating a
repository for storage and creating a visualization for
the client
FORECASTING ACCURACY
 The forecasted error is the difference between actual and the
forecasted values for a particular period of time
 The error is measured via MAPE (Mean Absolute Percentage Error)
 MAPE usually measures accuracy as a percentage:
N
M = 1/N * ∑ |(At-Ft)/At| * 100
t = 1
where At – Actual value at t,
Ft – Forecasted value at t and
N – No. of data points
FACTORS INFLUENCING POWER CONSUMPTION
 Atmospheric pressure
 Temperature
 Wind speed
 Precipitation
 Humidity
 Number of residents
 Consumption pattern of electrical appliances
FORECASTING TECHNIQUES
 It’s the process of making reasonable prediction by taking
the historical and present data into account
 There is a fair amount of risk and uncertainty that is
associated with forecasting and prediction
 For getting the most accurate forecasts, the data must
be updated
FORECASTING MODELS CONSIDERED
 There are 3 models - 3 hour, 12 hour and 24 hour look-
ahead model
 The 3 hour look ahead model depicts the forecasted
values for the next 3 hours, and similarly for the other 2
models
 The 3 hour look-ahead model is the best model as its
MAPE is 1.30% compared to 2.05% and 2.40% of the 12
and 24 hour models respectively
GENERALIZED ADDITIVE MODEL (GAM)
 GAM is used to develop the forecasting equations
 A simple GAM relationship can be written as:
g(E(Y)) = b0 + f1(X1) + f2(X2) + ….. +
fm(Xm)
GENERALIZED ADDITIVE MODEL (GAM)
 Here,
 X1, X2, …, Xm - m predictor variables
 Y - dependent variable
 E(Y) - mean of dependent variable
 b0 - regression coefficient
 f1, f2, …, fm - functions of the m predictors
 g - link function forming a relationship between the
predictor variables and the dependent variable
MOVING AVERAGES
 It creates a set of averages of various subsets of the
entire data set, to create few derived variables
 The process is explained below:
 The first element of the moving average series is
calculated by taking the average of the initial fixed
subset of the number series
 The subset is ‘shifted forward’ by excluding the first
number in the series and repeating the same process
again, over the entire data series
MOVING AVERAGES
 So, this creates a new subset of numbers
 A line connecting all the averages is called the moving
average
 In simpler terms, a moving average is a set of points,
each of which, is the average of a larger data set
R SHINY
 R Shiny is an R package that is used to build interactive
web applications in R
 Shiny web-applications are programmed using R without
the need to use HTML, CSS, or JavaScript for
development, although Shiny is flexible to incorporate
the same if the user desires (for developing customized
visualizations)
 It’s the visualization tool that was used to create the
dashboard
R SHINY - FEATURES
 Open source
 Generate interactive visualizations
 Easy to use
 Designed to build applications to analyze and/or visualize
data
PROCESS FLOW
Data
Gathering
Data
Manipulation
Visualization
Forecasting
models
FLOW DIAGRAM
PROCESS FLOW
Activity diagram depicting the general process flow
USE CASE DIAGRAM
SEQUENCE DIAGRAM
WEB SCRAPING – LOAD DATA
 The instantaneous live-feed load is scraped from the PJM
website every 5 minutes
 The HTML table which contains the load data is
converted into a data frame in R
 From the list of 18 utilities (or zones) present in PJM’s
website, the data for the required zone is extracted
 Appropriate quality checks are implemented sequentially
when scraping real-time load data
WEB SCRAPING – LOAD DATA
 If the server crashes, NULL / missing values are replaced
by the average of the load values of the past week for
that corresponding hour, depending on how close that
value will be to the forecasted values
 Outlier treatment is done to check if the scraped values
lie within a particular range. If not, it’s replaced by
either bound depending on which is close to this value
 Only the maximum load value of an hour is stored in a
CSV file. Hence, there’ll be 24 pts in a particular day
WEB SCRAPING – WEATHER DATA
 The weather data is scraped from 2 different sources:
 PULSE API for scraping instantaneous weather data
every 5 minutes
 ENCast API for scraping forecasted weather data once
in an hour
 As the 2 sets of weather data are present in JSON
format, they’re converted to the CSV format
WEB SCRAPING – WEATHER DATA
 The same set of quality checks that were performed on
the load data are also done on the weather data so that
there aren’t any discrepancies in the data
 The average of the temperature and relative humidity
and maximum wind speed and temperature on an hourly
basis are extracted
 The ENCast API gives the forecasted weather data for the
next 6 days
SCRAPPING REAL TIME DATA
RAW LOAD DATA
RAW WEATHER DATA
MERGING OF DATASETS
 The 3 datasets, i.e., the instantaneous load data, the
instantaneous weather data and the forecasted weather
data are merged into a single CSV file based on the
primary key (date and time)
CREATION OF DERIVED VARIABLES
 THI (Temperature-Humidity Index) is calculated based on
temperature and humidity
 Other derived variables include:
 THI Square
 THI moving averages for 4 and 6 hours
 12 hour prior load
 Moving averages for 12 and 6 hour
 Maximum temperature
 Maximum squared temperature
FORECASTING MODEL
 A 3 hour, 12 hour and 24 hour look-ahead model are each
executed each hour
 The graph contains 24 data points that correspond to the
forecasted load in a day
 The first 3 data points in the graph indicate the best
forecast as the MAPE is considerably lesser than the
other two models
 The next 9 data points are taken from the 12 hour look-
ahead model
FORECASTING MODEL
 The last 12 data points are taken from the 24 hour look-
ahead model
 This continues until all the 24 data points are taken from
the 3 hour model as it’s most accurate
 The last 3 rows from the 3 hour model, last 9 from 12
hour model and last 12 from the 24 hour model are used
to form the respective sets of equations for the different
models
 The expected MAPE (Mean Absolute Percentage Error) of
the forecasting model is less than 2%
FORECASTING MODEL
 For example, if one wants to predict the next day’s load
at 8.00 AM, he / she must have the
 3 hour prior load and weather data to be fed into the
3 hour model
 12 hour prior load and weather data to be fed into the
12 hour model and
 24 hour prior load and weather data to be fed into the
24 hour model
FORECASTED DATA STORAGE
 Three separate CSV files are created, one for each model
containing columns like date, time, load recorded, wind
speed, relative humidity, etc
 These files are then merged into a single CSV file based
on the primary key – date and time
VISUALIZATION – LOAD FORECAST TAB
 There are three lines in the graph that indicate the
actual load line, forecasted load line and threshold Line
 Text inputs for entering the summer peak load and the
threshold % are provided because these 2 parameters
change each year
 A reactive threshold line is present on the graph that
corresponds to 96% of the summer peak load
VISUALIZATION – LOAD FORECAST TAB
 A graph showing the actual and the forecasted load
values with the x-axis being the 24 hours of a day and the
y-axis being the load values
 A data table which has columns – Actual load, Actual load
as % of summer peak load, Forecasted load and
Forecasted load as % of summer peak load
 An option for the client to enter the date, duration of
event and thereby, add an event to a log
 An option for the client to download the load summary
data table is also available
VISUALIZATION – LOAD FORECAST TAB
Default view of the load forecast tab
VISUALIZATION – LOAD FORECAST TAB
Entering the date of occurrence of event
VISUALIZATION – LOAD FORECAST TAB
Entering the duration of the event
VISUALIZATION – LOAD FORECAST TAB
Adding an event to the event log
VISUALIZATION – LOAD FORECAST TAB
Event log details
VISUALIZATION – LOAD FORECAST TAB
Downloading the load summary
VISUALIZATION – LOAD FORECAST TAB
Load summary downloaded as a CSV file
VISUALIZATION – HISTORICAL LOAD TAB
 An option for the client to select any date and see the corresponding plot
and data table is available
 Both the data table and graph contain 24 points each
 An option for the client to download the historical load data table is also
available
VISUALIZATION – HISTORICAL LOAD TAB
Default view of the historical load tab
VISUALIZATION – HISTORICAL LOAD TAB
Selecting a date for viewing its corresponding plot and summary
VISUALIZATION – HISTORICAL LOAD TAB
The corresponding plot and summary of the selected day are displayed
VISUALIZATION – HISTORICAL LOAD TAB
Downloading the historical load for that day
VISUALIZATION – HISTORICAL LOAD TAB
Historical Load downloaded as a CSV file
VISUALIZATION – EVENT LOG TAB
 An event log data table that displays the dates and the duration of the
events that have occurred
 The event log is sorted so that the most recent events appear at the top of
the table
VISUALIZATION – EVENT LOG TAB
Default view of the event log tab
CONCLUSION
 OUTCOME
The client achieves faster and efficient event
prediction through this automated model
 BEHAVIOUR
The client predicts the occurrence of events with
greater reliability and less manual intervention
 INSIGHTS
A reliable model for event prediction is developed and
automated
THANK YOU!

More Related Content

What's hot

A review
A reviewA review
A reviewedzam
 
Universal Engineering Model for Cooling Towers
Universal Engineering Model for Cooling TowersUniversal Engineering Model for Cooling Towers
Universal Engineering Model for Cooling TowersIJERA Editor
 
Brussels airport forecast
Brussels airport  forecast Brussels airport  forecast
Brussels airport forecast Mohammed Awad
 
Class 10 mathematical modeling of continuous stirred tank reactor systems (...
Class 10   mathematical modeling of continuous stirred tank reactor systems (...Class 10   mathematical modeling of continuous stirred tank reactor systems (...
Class 10 mathematical modeling of continuous stirred tank reactor systems (...Manipal Institute of Technology
 
Ian_McLeod_Masters_Portfolio_Rough
Ian_McLeod_Masters_Portfolio_RoughIan_McLeod_Masters_Portfolio_Rough
Ian_McLeod_Masters_Portfolio_RoughIan McLeod
 
Machine Learning - Time Series Part 2
Machine Learning - Time Series Part 2Machine Learning - Time Series Part 2
Machine Learning - Time Series Part 2Rupak Roy
 
Machine Learning - Time Series
Machine Learning - Time Series Machine Learning - Time Series
Machine Learning - Time Series Rupak Roy
 
Optimal combinaison of CFD modeling and statistical learning for short-term w...
Optimal combinaison of CFD modeling and statistical learning for short-term w...Optimal combinaison of CFD modeling and statistical learning for short-term w...
Optimal combinaison of CFD modeling and statistical learning for short-term w...Jean-Claude Meteodyn
 
Assessment of a diagnostic procedure for the monitoring and control of indust...
Assessment of a diagnostic procedure for the monitoring and control of indust...Assessment of a diagnostic procedure for the monitoring and control of indust...
Assessment of a diagnostic procedure for the monitoring and control of indust...Manuel Stefanato
 
Work Order Summary From Mp2 Database2003
Work Order Summary From Mp2 Database2003Work Order Summary From Mp2 Database2003
Work Order Summary From Mp2 Database2003Mbisset
 
CAD -- CUSUM Anomaly Detection
CAD -- CUSUM Anomaly DetectionCAD -- CUSUM Anomaly Detection
CAD -- CUSUM Anomaly DetectionFarkas Kinga
 
Energy-Efficient Reduce-and-Rank Using Input-Adaptive Approximations
Energy-Efficient Reduce-and-Rank Using Input-Adaptive ApproximationsEnergy-Efficient Reduce-and-Rank Using Input-Adaptive Approximations
Energy-Efficient Reduce-and-Rank Using Input-Adaptive ApproximationsJAYAPRAKASH JPINFOTECH
 
Showing metric value in BSM Service Health KPI's and SHA
Showing metric value in BSM Service Health KPI's and SHAShowing metric value in BSM Service Health KPI's and SHA
Showing metric value in BSM Service Health KPI's and SHAkobyah
 
Load foecasting and power procurement planning in power sector
Load foecasting and power procurement planning in power sectorLoad foecasting and power procurement planning in power sector
Load foecasting and power procurement planning in power sectorManish Kumar
 
Benefit Assessment Plan: Sub-Metering Buildings
Benefit Assessment Plan: Sub-Metering BuildingsBenefit Assessment Plan: Sub-Metering Buildings
Benefit Assessment Plan: Sub-Metering BuildingsChaitali Bose Roy
 
Real-time simulation of the iterative calculation of the satellite based surf...
Real-time simulation of the iterative calculation of the satellite based surf...Real-time simulation of the iterative calculation of the satellite based surf...
Real-time simulation of the iterative calculation of the satellite based surf...Ramesh Dhungel
 

What's hot (20)

Time_Series_Assignment
Time_Series_AssignmentTime_Series_Assignment
Time_Series_Assignment
 
2015 NWA poster
2015 NWA poster2015 NWA poster
2015 NWA poster
 
A review
A reviewA review
A review
 
Class 13 p & i diagram
Class 13   p & i diagramClass 13   p & i diagram
Class 13 p & i diagram
 
Universal Engineering Model for Cooling Towers
Universal Engineering Model for Cooling TowersUniversal Engineering Model for Cooling Towers
Universal Engineering Model for Cooling Towers
 
Brussels airport forecast
Brussels airport  forecast Brussels airport  forecast
Brussels airport forecast
 
Class 10 mathematical modeling of continuous stirred tank reactor systems (...
Class 10   mathematical modeling of continuous stirred tank reactor systems (...Class 10   mathematical modeling of continuous stirred tank reactor systems (...
Class 10 mathematical modeling of continuous stirred tank reactor systems (...
 
Ian_McLeod_Masters_Portfolio_Rough
Ian_McLeod_Masters_Portfolio_RoughIan_McLeod_Masters_Portfolio_Rough
Ian_McLeod_Masters_Portfolio_Rough
 
Machine Learning - Time Series Part 2
Machine Learning - Time Series Part 2Machine Learning - Time Series Part 2
Machine Learning - Time Series Part 2
 
Machine Learning - Time Series
Machine Learning - Time Series Machine Learning - Time Series
Machine Learning - Time Series
 
Optimal combinaison of CFD modeling and statistical learning for short-term w...
Optimal combinaison of CFD modeling and statistical learning for short-term w...Optimal combinaison of CFD modeling and statistical learning for short-term w...
Optimal combinaison of CFD modeling and statistical learning for short-term w...
 
GENETIK
GENETIKGENETIK
GENETIK
 
Assessment of a diagnostic procedure for the monitoring and control of indust...
Assessment of a diagnostic procedure for the monitoring and control of indust...Assessment of a diagnostic procedure for the monitoring and control of indust...
Assessment of a diagnostic procedure for the monitoring and control of indust...
 
Work Order Summary From Mp2 Database2003
Work Order Summary From Mp2 Database2003Work Order Summary From Mp2 Database2003
Work Order Summary From Mp2 Database2003
 
CAD -- CUSUM Anomaly Detection
CAD -- CUSUM Anomaly DetectionCAD -- CUSUM Anomaly Detection
CAD -- CUSUM Anomaly Detection
 
Energy-Efficient Reduce-and-Rank Using Input-Adaptive Approximations
Energy-Efficient Reduce-and-Rank Using Input-Adaptive ApproximationsEnergy-Efficient Reduce-and-Rank Using Input-Adaptive Approximations
Energy-Efficient Reduce-and-Rank Using Input-Adaptive Approximations
 
Showing metric value in BSM Service Health KPI's and SHA
Showing metric value in BSM Service Health KPI's and SHAShowing metric value in BSM Service Health KPI's and SHA
Showing metric value in BSM Service Health KPI's and SHA
 
Load foecasting and power procurement planning in power sector
Load foecasting and power procurement planning in power sectorLoad foecasting and power procurement planning in power sector
Load foecasting and power procurement planning in power sector
 
Benefit Assessment Plan: Sub-Metering Buildings
Benefit Assessment Plan: Sub-Metering BuildingsBenefit Assessment Plan: Sub-Metering Buildings
Benefit Assessment Plan: Sub-Metering Buildings
 
Real-time simulation of the iterative calculation of the satellite based surf...
Real-time simulation of the iterative calculation of the satellite based surf...Real-time simulation of the iterative calculation of the satellite based surf...
Real-time simulation of the iterative calculation of the satellite based surf...
 

Similar to Automated DR Event Prediction

Statistical analysis software article - jan 2016
Statistical analysis software article - jan 2016Statistical analysis software article - jan 2016
Statistical analysis software article - jan 2016Kyma AS
 
Statistical analysis software article - jan 2016
Statistical analysis software article  - jan 2016Statistical analysis software article  - jan 2016
Statistical analysis software article - jan 2016Peter Elvingdal
 
Electric Load Forecasting
Electric Load ForecastingElectric Load Forecasting
Electric Load Forecastinginventy
 
Objectives ApproachExpected Results.docx
Objectives ApproachExpected Results.docxObjectives ApproachExpected Results.docx
Objectives ApproachExpected Results.docxamit657720
 
Objectives ApproachExpected Results.docx
Objectives ApproachExpected Results.docxObjectives ApproachExpected Results.docx
Objectives ApproachExpected Results.docxvannagoforth
 
Short-term load forecasting with using multiple linear regression
Short-term load forecasting with using multiple  linear regression Short-term load forecasting with using multiple  linear regression
Short-term load forecasting with using multiple linear regression IJECEIAES
 
Short term power forecasting Awea 2014
Short term power forecasting Awea 2014Short term power forecasting Awea 2014
Short term power forecasting Awea 2014Jean-Claude Meteodyn
 
meteodynWT meso coupling downscaling regional planing
meteodynWT meso coupling downscaling regional planingmeteodynWT meso coupling downscaling regional planing
meteodynWT meso coupling downscaling regional planingJean-Claude Meteodyn
 
Estimation of precipitation during the period of south west monsoon
Estimation of precipitation during the period of south west monsoonEstimation of precipitation during the period of south west monsoon
Estimation of precipitation during the period of south west monsoonIAEME Publication
 
A Novel Technique in Software Engineering for Building Scalable Large Paralle...
A Novel Technique in Software Engineering for Building Scalable Large Paralle...A Novel Technique in Software Engineering for Building Scalable Large Paralle...
A Novel Technique in Software Engineering for Building Scalable Large Paralle...Eswar Publications
 
POWER SYSTEM OPERATION AND CONTROL. load forecasting - introduction, methodo...
POWER SYSTEM OPERATION AND CONTROL. load forecasting -  introduction, methodo...POWER SYSTEM OPERATION AND CONTROL. load forecasting -  introduction, methodo...
POWER SYSTEM OPERATION AND CONTROL. load forecasting - introduction, methodo...Jobin Abraham
 
An Enhanced Support Vector Regression Model for Weather Forecasting
An Enhanced Support Vector Regression Model for Weather ForecastingAn Enhanced Support Vector Regression Model for Weather Forecasting
An Enhanced Support Vector Regression Model for Weather ForecastingIOSR Journals
 
IRJET - Intelligent Weather Forecasting using Machine Learning Techniques
IRJET -  	  Intelligent Weather Forecasting using Machine Learning TechniquesIRJET -  	  Intelligent Weather Forecasting using Machine Learning Techniques
IRJET - Intelligent Weather Forecasting using Machine Learning TechniquesIRJET Journal
 
Feinberg.ppt
Feinberg.pptFeinberg.ppt
Feinberg.pptBakriBuga
 
Ecm time series forecast
Ecm time series forecastEcm time series forecast
Ecm time series forecastAyapparaj SKS
 
A WIND POWER PREDICTION METHOD BASED ON BAYESIAN FUSION
A WIND POWER PREDICTION METHOD BASED ON BAYESIAN FUSIONA WIND POWER PREDICTION METHOD BASED ON BAYESIAN FUSION
A WIND POWER PREDICTION METHOD BASED ON BAYESIAN FUSIONcsandit
 
Load types, estimation, grwoth, forecasting and duration curves
Load types, estimation, grwoth, forecasting and duration curvesLoad types, estimation, grwoth, forecasting and duration curves
Load types, estimation, grwoth, forecasting and duration curvesAzfar Rasool
 
An overview of electricity demand forecasting techniques
An overview of electricity demand forecasting techniquesAn overview of electricity demand forecasting techniques
An overview of electricity demand forecasting techniquesAlexander Decker
 

Similar to Automated DR Event Prediction (20)

Statistical analysis software article - jan 2016
Statistical analysis software article - jan 2016Statistical analysis software article - jan 2016
Statistical analysis software article - jan 2016
 
Statistical analysis software article - jan 2016
Statistical analysis software article  - jan 2016Statistical analysis software article  - jan 2016
Statistical analysis software article - jan 2016
 
Electric Load Forecasting
Electric Load ForecastingElectric Load Forecasting
Electric Load Forecasting
 
Objectives ApproachExpected Results.docx
Objectives ApproachExpected Results.docxObjectives ApproachExpected Results.docx
Objectives ApproachExpected Results.docx
 
Objectives ApproachExpected Results.docx
Objectives ApproachExpected Results.docxObjectives ApproachExpected Results.docx
Objectives ApproachExpected Results.docx
 
Short-term load forecasting with using multiple linear regression
Short-term load forecasting with using multiple  linear regression Short-term load forecasting with using multiple  linear regression
Short-term load forecasting with using multiple linear regression
 
Marvuglia
MarvugliaMarvuglia
Marvuglia
 
Short term power forecasting Awea 2014
Short term power forecasting Awea 2014Short term power forecasting Awea 2014
Short term power forecasting Awea 2014
 
meteodynWT meso coupling downscaling regional planing
meteodynWT meso coupling downscaling regional planingmeteodynWT meso coupling downscaling regional planing
meteodynWT meso coupling downscaling regional planing
 
I02095257
I02095257I02095257
I02095257
 
Estimation of precipitation during the period of south west monsoon
Estimation of precipitation during the period of south west monsoonEstimation of precipitation during the period of south west monsoon
Estimation of precipitation during the period of south west monsoon
 
A Novel Technique in Software Engineering for Building Scalable Large Paralle...
A Novel Technique in Software Engineering for Building Scalable Large Paralle...A Novel Technique in Software Engineering for Building Scalable Large Paralle...
A Novel Technique in Software Engineering for Building Scalable Large Paralle...
 
POWER SYSTEM OPERATION AND CONTROL. load forecasting - introduction, methodo...
POWER SYSTEM OPERATION AND CONTROL. load forecasting -  introduction, methodo...POWER SYSTEM OPERATION AND CONTROL. load forecasting -  introduction, methodo...
POWER SYSTEM OPERATION AND CONTROL. load forecasting - introduction, methodo...
 
An Enhanced Support Vector Regression Model for Weather Forecasting
An Enhanced Support Vector Regression Model for Weather ForecastingAn Enhanced Support Vector Regression Model for Weather Forecasting
An Enhanced Support Vector Regression Model for Weather Forecasting
 
IRJET - Intelligent Weather Forecasting using Machine Learning Techniques
IRJET -  	  Intelligent Weather Forecasting using Machine Learning TechniquesIRJET -  	  Intelligent Weather Forecasting using Machine Learning Techniques
IRJET - Intelligent Weather Forecasting using Machine Learning Techniques
 
Feinberg.ppt
Feinberg.pptFeinberg.ppt
Feinberg.ppt
 
Ecm time series forecast
Ecm time series forecastEcm time series forecast
Ecm time series forecast
 
A WIND POWER PREDICTION METHOD BASED ON BAYESIAN FUSION
A WIND POWER PREDICTION METHOD BASED ON BAYESIAN FUSIONA WIND POWER PREDICTION METHOD BASED ON BAYESIAN FUSION
A WIND POWER PREDICTION METHOD BASED ON BAYESIAN FUSION
 
Load types, estimation, grwoth, forecasting and duration curves
Load types, estimation, grwoth, forecasting and duration curvesLoad types, estimation, grwoth, forecasting and duration curves
Load types, estimation, grwoth, forecasting and duration curves
 
An overview of electricity demand forecasting techniques
An overview of electricity demand forecasting techniquesAn overview of electricity demand forecasting techniques
An overview of electricity demand forecasting techniques
 

Automated DR Event Prediction

  • 1. AUTOMATION OF DEMAND RESPONSE MANAGEMENT M.V.N.VINAY BALAKRISHNA AM.EN.U4ME12137 PROJECT GUIDE: K.SIVASELVI
  • 2. BUSINESS CONTEXT  In US, the consumption of electricity is highest in the summer due to the increased use of electrical appliances  Regional utilities implement DR programs and shed / clip electricity by offering financial incentives to end customers as per standard rules and regulations  Events are called when the forecasted electrical load exceeds 96% of summer peak load  The aim is to build an automated model that will accurately predict occurrences of events with least manual intervention
  • 3. BUSINESS CONTEXT – SGQ SGQ (State-Gap-Question) representing the current state, the gap preventing us from reaching the final state, the key question that must be answered to reach the final state and the final state
  • 4. DEMAND RESPONSE (DR)  Reduction in usage of electricity by end customers, in response to the events called by regional utility by curtailing peak electricity load  Involves generation, transmission and distribution of electric load through proper planning & scheduling  They’re generally implemented in the summer as usage of electrical appliances is at its peak  The incentives offered are proportional to the amount of electricity that was clipped
  • 5. DEMAND RESPONSE (DR)  Maximum of 6 events can be called for the summer that is for 4 months  Customers are informed of events in advance via Curtailment Service Providers (CSPs) who are third party vendors  Hence, customers efficiently manage electrical usage and subsequently, their costs
  • 6. OBJECTIVES  To predict events based on the past 10 years historical data for a client in Philadelphia considering all factors such as weather, humidity, precipitation etc.  Develop a reliable framework by process flow automation through real time collection, creating a repository for storage and creating a visualization for the client
  • 7.
  • 8. FORECASTING ACCURACY  The forecasted error is the difference between actual and the forecasted values for a particular period of time  The error is measured via MAPE (Mean Absolute Percentage Error)  MAPE usually measures accuracy as a percentage: N M = 1/N * ∑ |(At-Ft)/At| * 100 t = 1 where At – Actual value at t, Ft – Forecasted value at t and N – No. of data points
  • 9. FACTORS INFLUENCING POWER CONSUMPTION  Atmospheric pressure  Temperature  Wind speed  Precipitation  Humidity  Number of residents  Consumption pattern of electrical appliances
  • 10. FORECASTING TECHNIQUES  It’s the process of making reasonable prediction by taking the historical and present data into account  There is a fair amount of risk and uncertainty that is associated with forecasting and prediction  For getting the most accurate forecasts, the data must be updated
  • 11. FORECASTING MODELS CONSIDERED  There are 3 models - 3 hour, 12 hour and 24 hour look- ahead model  The 3 hour look ahead model depicts the forecasted values for the next 3 hours, and similarly for the other 2 models  The 3 hour look-ahead model is the best model as its MAPE is 1.30% compared to 2.05% and 2.40% of the 12 and 24 hour models respectively
  • 12. GENERALIZED ADDITIVE MODEL (GAM)  GAM is used to develop the forecasting equations  A simple GAM relationship can be written as: g(E(Y)) = b0 + f1(X1) + f2(X2) + ….. + fm(Xm)
  • 13. GENERALIZED ADDITIVE MODEL (GAM)  Here,  X1, X2, …, Xm - m predictor variables  Y - dependent variable  E(Y) - mean of dependent variable  b0 - regression coefficient  f1, f2, …, fm - functions of the m predictors  g - link function forming a relationship between the predictor variables and the dependent variable
  • 14. MOVING AVERAGES  It creates a set of averages of various subsets of the entire data set, to create few derived variables  The process is explained below:  The first element of the moving average series is calculated by taking the average of the initial fixed subset of the number series  The subset is ‘shifted forward’ by excluding the first number in the series and repeating the same process again, over the entire data series
  • 15. MOVING AVERAGES  So, this creates a new subset of numbers  A line connecting all the averages is called the moving average  In simpler terms, a moving average is a set of points, each of which, is the average of a larger data set
  • 16. R SHINY  R Shiny is an R package that is used to build interactive web applications in R  Shiny web-applications are programmed using R without the need to use HTML, CSS, or JavaScript for development, although Shiny is flexible to incorporate the same if the user desires (for developing customized visualizations)  It’s the visualization tool that was used to create the dashboard
  • 17. R SHINY - FEATURES  Open source  Generate interactive visualizations  Easy to use  Designed to build applications to analyze and/or visualize data
  • 20. PROCESS FLOW Activity diagram depicting the general process flow
  • 23. WEB SCRAPING – LOAD DATA  The instantaneous live-feed load is scraped from the PJM website every 5 minutes  The HTML table which contains the load data is converted into a data frame in R  From the list of 18 utilities (or zones) present in PJM’s website, the data for the required zone is extracted  Appropriate quality checks are implemented sequentially when scraping real-time load data
  • 24. WEB SCRAPING – LOAD DATA  If the server crashes, NULL / missing values are replaced by the average of the load values of the past week for that corresponding hour, depending on how close that value will be to the forecasted values  Outlier treatment is done to check if the scraped values lie within a particular range. If not, it’s replaced by either bound depending on which is close to this value  Only the maximum load value of an hour is stored in a CSV file. Hence, there’ll be 24 pts in a particular day
  • 25. WEB SCRAPING – WEATHER DATA  The weather data is scraped from 2 different sources:  PULSE API for scraping instantaneous weather data every 5 minutes  ENCast API for scraping forecasted weather data once in an hour  As the 2 sets of weather data are present in JSON format, they’re converted to the CSV format
  • 26. WEB SCRAPING – WEATHER DATA  The same set of quality checks that were performed on the load data are also done on the weather data so that there aren’t any discrepancies in the data  The average of the temperature and relative humidity and maximum wind speed and temperature on an hourly basis are extracted  The ENCast API gives the forecasted weather data for the next 6 days
  • 30. MERGING OF DATASETS  The 3 datasets, i.e., the instantaneous load data, the instantaneous weather data and the forecasted weather data are merged into a single CSV file based on the primary key (date and time)
  • 31. CREATION OF DERIVED VARIABLES  THI (Temperature-Humidity Index) is calculated based on temperature and humidity  Other derived variables include:  THI Square  THI moving averages for 4 and 6 hours  12 hour prior load  Moving averages for 12 and 6 hour  Maximum temperature  Maximum squared temperature
  • 32. FORECASTING MODEL  A 3 hour, 12 hour and 24 hour look-ahead model are each executed each hour  The graph contains 24 data points that correspond to the forecasted load in a day  The first 3 data points in the graph indicate the best forecast as the MAPE is considerably lesser than the other two models  The next 9 data points are taken from the 12 hour look- ahead model
  • 33. FORECASTING MODEL  The last 12 data points are taken from the 24 hour look- ahead model  This continues until all the 24 data points are taken from the 3 hour model as it’s most accurate  The last 3 rows from the 3 hour model, last 9 from 12 hour model and last 12 from the 24 hour model are used to form the respective sets of equations for the different models  The expected MAPE (Mean Absolute Percentage Error) of the forecasting model is less than 2%
  • 34. FORECASTING MODEL  For example, if one wants to predict the next day’s load at 8.00 AM, he / she must have the  3 hour prior load and weather data to be fed into the 3 hour model  12 hour prior load and weather data to be fed into the 12 hour model and  24 hour prior load and weather data to be fed into the 24 hour model
  • 35. FORECASTED DATA STORAGE  Three separate CSV files are created, one for each model containing columns like date, time, load recorded, wind speed, relative humidity, etc  These files are then merged into a single CSV file based on the primary key – date and time
  • 36.
  • 37. VISUALIZATION – LOAD FORECAST TAB  There are three lines in the graph that indicate the actual load line, forecasted load line and threshold Line  Text inputs for entering the summer peak load and the threshold % are provided because these 2 parameters change each year  A reactive threshold line is present on the graph that corresponds to 96% of the summer peak load
  • 38. VISUALIZATION – LOAD FORECAST TAB  A graph showing the actual and the forecasted load values with the x-axis being the 24 hours of a day and the y-axis being the load values  A data table which has columns – Actual load, Actual load as % of summer peak load, Forecasted load and Forecasted load as % of summer peak load  An option for the client to enter the date, duration of event and thereby, add an event to a log  An option for the client to download the load summary data table is also available
  • 39. VISUALIZATION – LOAD FORECAST TAB Default view of the load forecast tab
  • 40. VISUALIZATION – LOAD FORECAST TAB Entering the date of occurrence of event
  • 41. VISUALIZATION – LOAD FORECAST TAB Entering the duration of the event
  • 42. VISUALIZATION – LOAD FORECAST TAB Adding an event to the event log
  • 43. VISUALIZATION – LOAD FORECAST TAB Event log details
  • 44. VISUALIZATION – LOAD FORECAST TAB Downloading the load summary
  • 45. VISUALIZATION – LOAD FORECAST TAB Load summary downloaded as a CSV file
  • 46. VISUALIZATION – HISTORICAL LOAD TAB  An option for the client to select any date and see the corresponding plot and data table is available  Both the data table and graph contain 24 points each  An option for the client to download the historical load data table is also available
  • 47. VISUALIZATION – HISTORICAL LOAD TAB Default view of the historical load tab
  • 48. VISUALIZATION – HISTORICAL LOAD TAB Selecting a date for viewing its corresponding plot and summary
  • 49. VISUALIZATION – HISTORICAL LOAD TAB The corresponding plot and summary of the selected day are displayed
  • 50. VISUALIZATION – HISTORICAL LOAD TAB Downloading the historical load for that day
  • 51. VISUALIZATION – HISTORICAL LOAD TAB Historical Load downloaded as a CSV file
  • 52. VISUALIZATION – EVENT LOG TAB  An event log data table that displays the dates and the duration of the events that have occurred  The event log is sorted so that the most recent events appear at the top of the table
  • 53. VISUALIZATION – EVENT LOG TAB Default view of the event log tab
  • 54. CONCLUSION  OUTCOME The client achieves faster and efficient event prediction through this automated model  BEHAVIOUR The client predicts the occurrence of events with greater reliability and less manual intervention  INSIGHTS A reliable model for event prediction is developed and automated