demand prediction for rural area

voltaireblingData Management

Nov 20, 2013 (3 years and 10 months ago)

108 views

Data Mining based electricity
demand prediction for rural area
Dr. Sonali Agarwal

Assistant Professor

IIIT
-

Allahabad, India

Outline


Energy overview and Electricity


Status of Electricity in Rural Area


Electricity Demand Prediction through Data Mining


Proposed model using Smart metering


Conclusion


Introduction


Energy

plays

a

significant

role

in

the

economic

and

technological

advancement

of

modern

society

and

plays

crucial

role

in

human

life

standard
.


Approximately

65
%

of

India

lives

in

rural

areas

and

faces

tremendous

hardship
.



Rural energy is not getting the importance it deserves.


There

is

a

need

to

increase

the

access

of

the

rural

poor

to

affordable,

reliable,

safe

and

high

quality

energy

to

strengthen

their

self
-
reliance

and

empowerment
.



It

must

improve

the

quality

of

their

environment

(starting

with

their

immediate

environment

in

their

households)
.




3

Power Generated


How does it
flow? Where does it go?

Transmission System



4% losses

Unavoidable Distribution

Losses
-

15%

Theft
-
35% losses

29 units

Billed to consumer



53 units

Energy usefully consumed



42 units (max.)

100 units

96 units


82 units

Generating

Station

20% loss

Rural Electrification


The

objective

of

Rural

Electrification

is

to

provide

100
%

electrification

of

all

villages

and

habitations

in

the

country,

electricity

access

to

all

households,

free

of

cost

electricity

connection

to

BPL
.

The

various

schemes

are
:


Kutir

Jyoti

Program (
KJP
) ,


Pradhan

Mantri

Gramodaya

Yojna

(
PMGY
),


Rajiv Gandhi
Grameen

Vidyutikaran

Yojna
,


Jawaharlal Nehru National Solar Mission scheme.

Need of Load Management


These

schemes

are

used

for

rural

electrification

but

not

sufficient

to

provide

electricity

for

24

hours
.


This

paper

presents

how

to

manage

the

load

and

forecast

the

electricity

of

rural

areas

according

to

rainy,

winter

and

summer

seasons
.



The

main

issue

is

predicting

the

load

for

upcoming

month

or

months

in

the

rural

area

because

there

is

some

particular

time

period

in

which

the

need

of

electricity

is

very

high

for

agriculture

purpose
.



Load Forecasting


Load

forecasting

is

very

important

for

electric

utilities

in

a

competitive

environment

created

by

the

electric

industry

:


Energy Demand and Resources


The electrical loads of the area are classified as
domestic, agricultural, commercial and street light.


Domestic Loads:
TV, fan, Tube light, fridge, water heater
and motor for Pumping.


Agricultural Loads:
Motors for water pumping.


Commercial Loads:
Schools, Shops, Flour mill, Small
scale industries, and village panchayat office buildings.

Load Survey of Rural Area


The load survey of the rural areas is achieved by
taking the interview of
sarpanch
, school teachers,
members of rural area etc.


Number of villages of rural area


Number of houses of rural area


Population of rural area


Demand of domestic load


Demand of street lighting load


Demand of commercial load


Demand of agriculture load


Average energy consumptions


Others demand.

Data Mining


Data

Mining

is

a

new

methodology

that

came

into

existing

in

the

middle

of

1990
’s
.



Data

mining

is

a

powerful

tool

that

provides

support

for

fetching

previously

unknown

pattern

and

useful

information

from

huge

dataset
.



Data

mining

is

applied

on

the

historical

data

which

is

used

for

making

prediction
.

Data Mining


There

are

many

techniques

of

Data

mining
:


Clustering,



Classification



Association

Rule

Mining
.


In

this

particular

domain,



Support

Vector

Machine,



Artificial

Neural

Network



Genetic

Algorithm



can

be

used

for

predicting

the

future

requirement

of

energy
.



Data Mining


This

needs

to

develop

a

Central

Data

Warehouse

of

electricity

data
.



Data

mining

technology

gives

the

electricity

pattern

used

by

particular

area

or

customer
.



These

patterns

which

are

derived

from

the

data

warehouse,

may

further

used

for

predicting

the

future

trend

of

the

energy

requirement

of

any

particular

region

or

customer
.

Data Mining for Electricity Demand
Prediction





contd…

Data Mining for Electricity Demand
Prediction in Agriculture



Rice

is

one

of

the

important

grains

of

India

which

is

mostly

produced

in

the

month

of

august

and

required

more

water

as

compare

to

other

grains
.



So,

to

fulfill

the

demand

of

water

there

is

a

need

of

electricity
.



The

demand

for

electricity

is

high

in

this

month

because

every

farmer

needs

watering

through

the

tube

well

to

his

field

for

rice
.



So

sufficient

amount

of

electricity

is

essential

for

this

particular

time
.


Example of Electric load
Classification



This

proposed

model

helps

for

any

country

to

make

decision

about

the

rate

of

consumption

and

rate

of

production

for

next

year
.



The

patterns

will

be

different

for

every

region

and

also

depends

upon

the

location

of

region,

time,

type

of

work

and

agriculture

type
.



Benefits of Electricity Demand
Prediction





Smart Metering concept


Electricity

meters

measures

the

consumption

and

need

of

electricity

of

any

type

of

customer
.



These

two

types

of

the

meters

are


i
)

Electromechanical

Meters


ii)

Electronic

Meters

Electromechanical Meters


Electromechanical

Meters
:

This

type

of

meter

uses

an

aluminum

disc
.

The

revolution

of

this

disc

at

particular

speed

is

counted

which

are

propositional

to

electricity

consumption

in

that

particular

house

or

company
.



Electronic Meters


Electronic

Meters
:

This

is

an

advance

type

of

electricity

measurement

way
.

This

meter

measures

the

load

and

supply

parameter

like

maximum

need,

power

factors

etc
.

it

shows

the

reading

on

LED

/LCD
.

This

meter

is

useful

for

us

because

it

can

transmit

the

reading

data

to

any

other

remote

place
.


Smart meter


Smart

meter

is

a

modified

version

of

previous

electronics

meter
.



This

meter

can

be

programmed

which

can

store

the

consumption

data

in

it
.


The

Data

could

be

sent

to

the

Central

Data

Warehouse

using

(wired/wireless)

networking

infrastructure

.



This

provides

two

way

communications

between

the

Central

Data

Warehouse

and

meter

machine
.



The

data,

which

is

stored,

can

be

achieved

from

Central

Data

Warehouse

or

from

meter

also
.



Smart meter


The

functional

aspects

of

the

meter

are

shown

in

the

figure
:




Smart meter


A

smart

meter

has

three

elementary

functions


Measure

the

consumed

electricity

by

the

customer

(or

generated

by

the

power

station)


Remotely

accessibility

of

customer

for

switch

off
.


Control

on

electricity

consumption

through

the

network

.


This

meter

is

attached

with

network

through

the

modem

for

the

communication
.

It

can

be

use

wireless/wired
.



All

appliances’

consumption

can

be

store

individually

and

can

control

directly
.



Smart meter


It

provides

the

up
-
to

date

information

of

electricity

consumption

and

can

show

the

bill

of

the

any

particular

day

or

till

date
.



The

detailed

data

is

available

on

the

data

center

by

which

a

prediction

can

be

done

for

future

trend
.


Proposed model


The

proposed

model

gives

the

pattern

of

load

in

any

area

for

upcoming

year

or

particular

month
.



For

this

we

need

some

historical

data

of

the

usage

of

electricity

for

particular

region

or

for

particular

customer
.



By

this

way

we

can

predict

the

consumption

pattern

for

that

specified

customer

or

area

for

upcoming

year

or

month
.



Data

mining

can

provide

an

efficient

way

for

the

load

shedding

so

that

the

need

of

electricity

for

any

upcoming

month

or

year

can

be

scheduled

using

backup

or

some

other

way
.

Central smart
meter data
warehouse

Home 1

Home 2

Home 3

Internet

Customer

Data analyst

Energy
management
authority

Proposed model


The

rules

which

are

generated

from

data

mining

algorithm

are

the

base

for

next

prediction
.

So

using

the

patterns

we

can

guess

upcoming

requirements

of

energy

with

scientific

approach
.

The

steps

associated

with

data

mining

are

given

below
:


a)

Selecting

the

central

data

base



b)

Prepare

a

data

warehouse

from

central

database
.


c)

Apply

pre
-
processing

on

data



d)

Apply

data

mining

approach

on

this

data
.


e)

Decision

making

on

the

electricity

need

/balancing
.


The

data

of

any

particular

area

or

customer

can

be

fetched

from

the

system
.

Data communication in smart
metering scenario

1,Consumers
:
access to the
metered data

2,Smart


metering

devices
:

sense the

consumed

energy

3, Grid operator/supplier
: load balancing

4,Communication
network:
be
secured

5,Electricity


producer:

sells the

electricity

to customers

6 ,Aggregator:
producing the relevant
and needed figures

Smart Meter Analytics addresses the needs and challenges

Billing
Management

Customer
Management

Outage
Management

Meter
Configuration
Data

Network

CSR

IVR

Web

Grid Operations

Managed Meter
Solution

Customer Bill

Smart Meter

(metering and device management)

Managed MDM Solution

Customer Service Channels

Meter Data
Management

Meter
Data

Analytics

Outage
Data

Outage
Restoration
Status

Work
Management

Demand
Management

(DSM)

Distribution
Management

(DMS)

Dynamic
Pricing data



Load
Management
Data

Supply
Management

Online Bill
Presentment

Asset
Management

Service Provision

Meter
Management

Managed CIS
Solution

New AMI functions

Legacy functions

OMS/DSM

Management

Managed
OMS/DMS

Solution

1

2

3

6

5

4

7

8

9

10

Smart Meter Data has a
pervasive impact across all
utility grid operation and
back
-
office systems




This represents an
opportunity to create a
broad set of IBM service
solutions


and leverage the
Smarter Analytics portfolio

Residential and C and I

Data Acquisition

Analytics solution development requires several interacting
design steps

Streaming data

Text data

Multi
-
dimensional

Time series

Geo spatial

Relational

Data mining
and
statistics

Optimization
and
simulation

Fuzzy

matching

Network

algorithms

Composition and

Packaging

Core Analytics

Filtering and

Extraction Validation

Social network

Video

and
image

Semantic
analysis

Business Rules Engine

Data Evaluation and Fusion

Algorithm Composition and Invention

Testing and Execution Optimization



Deployment

New
algorithms

34

Load pattern outcomes for
spring


Load pattern outcomes for
summer


Load pattern outcomes for
winter


Outcomes of the proposed
model


Above

Data

mining

steps

provide

the

pattern

of

consumption

of

energy

of

the

particular

area
.



These

patterns

are

very

useful

for

predicting

the

demand

of

energy

of

the

upcoming

the

time
.



These

patterns

will

play

important

role

for

managing

the

energy

requirement

of

the

rural

area

with

the

effective

management

of

the

energy
.


Outcomes of the proposed
model


balancing

between

available

energy

and

needed

energy
:

The

technique

provides

a

better

way

to

meet

the

requirement

of

energy

in

the

upcoming

year
.



This

step

gives

the

patterns

of

consumption

of

the

energy

in

that

area
.

These

patterns

focus

on

the

location,

time

and

type

of

need

of

the

rural

people
.



In

this

way,

the

rural

people

can

get

the

sufficient

energy

for

their

tasks

like

agriculture

and

power

cut

can

be

schedule

for

other

less

important

period

of

time
.



Conclusion


In

India,

Energy

demand

is

not

matched

with

the

availability

so

there

is

a

need

of

effective

energy

management
.


Energy

Management

through

Data

mining

especially

for

electricity

load

prediction

is

an

efficient

and

reliable

approach
.



This

approach

gives

better

utilization

of

energy

and

provides

a

good

level

of

satisfaction

to

the

people
.



Using

Data

Mining

techniques

it

could

be

easily

predicted

that

at

what

time

the

demand

for

electricity

would

be

high

and

using

this

information

electricity

provider

can

manage

the

supply

of

electricity

in

rural

areas
.


Conclusion


The

prediction

is

based

on

the

location

as

well

as

on

the

type

of

user

and

provides

a

better

way

to

handle

the

situation

of

energy

management
.



Data

mining

results

are

useful

for

reducing

the

electricity

problem

in

the

country
.



It

is

an

efficient

way

to

handle

the

electricity

requirement

of

the

any

area
.

Reference


Stephen
Makonin

, Lyn Bartram ,Fred
Popowich

, “A Smarter Smart
Home: Case Studies of Ambient Intelligence”, January
-
March 2013 (Vol.
12, No. 1) pp. 58
-
66


Michael
Friedewald
, Olivier
Da

Costa, Yves
Punie,Petteri

Alahuhta
,
Sirkka

Heinonen
, “Perspectives of ambient intelligence in the home
environment”,
Telematics

and Informatics 22 (2005) 221

238


Abhaya

Chandra K. and Andreas
König
, “Smart Homes, Intelligent
Kitchens, Sensate Floors
-

Ambient Intelligence”, Institute of Integrated
Sensor Systems, Kaiserslautern University and Technology, Germany


Giuseppe
Loseto
,
Floriano

Scioscia
, Michele
Ruta
, Eugenio Di
Sciascio
,
“Semantic
-
based Smart Homes: a Multi
-
Agent Approach”, OWL Web
Ontology Language.


Kevin Harris, “Smart Homes”, Spring 2005 Computational Intelligence
Seminar Series


Robert F.
Heile
, “802 Smart Grid Tutorial November 16, 2009,
ZigBee

Smart Energy”


Alexandra
-
Gwyn
Paetz
, “Living in a Smart Home ”, Dublin, 27
th

March
2012