Environment and Pollution Monitoring in Rourkela Using Machine Learning

ABSTRACT RELATED WORK MACHINE LEARNING IDL POLICE MONITORING STATION

ALGORITHM 12
11
In this presentation we have monitored
 Many previous works have been 10 10
the environment and pollution data(like 10
proposed to apply machine learning  Many previous works have been 9
SO2,NO2,RSPM/PM10,PM2.5,O3,lead,CO)
algorithms to air quality predictions. proposed to apply machine learning
in Rourkela using machine learning. 8

Machine learning is one of the most algorithms to air quality predictions.

 Kurt and Oktay [33] modeled 6
6 6
popular technique to train a model on big
geographic connections into a neural  Kurt and Oktay [33] modeled
5 5 5

data using optimizing algorithm. As there 4 4

network model and predicted daily geographic connections into a neural 4
are some algorithms which are restricted
concentration levels of SO2, CO, and network model and predicted daily
to collect the data and produce the 2
PM10 3 days in advance. concentration levels of SO2, CO, and 2
proper outputs. So here we generate 0.7 0.7

some good results about the producing PM10 3 days in advance. 0.3 0.5

 Some researchers also worked on 0

pollutants in Rourkela 1/1/2017 5/1/2017 8/1/2017 12/1/2017

predicting the concentration of air  Some researchers also worked on Date

pollutants on training neural network predicting the concentration of air SO2 NO2 CO Lead

KEYWORDS modelling to predict the hourly pollutants on training neural network REGIONAL OFFICE ORPB MONITORING STATION
producing O3 and PM10 concentration modelling to predict the hourly
Air pollutant prediction, machine learning, on the basis of previous day datasets. 16
producing O3 and PM10 concentration
data analysis
on the basis of previous day datasets. 14

12
INTRODUCTION DATA COLLECTION & GENERAL FORMULATION

POLLUTANT mg/m^3
PREPROCESSING Our goal is to predict the concentration of
10

Adverse health impacts from exposure to air pollutants of the next 3/4 day on the 8

outdoor air pollutants are complicated basis of the air pollutant data collected 6

functions of pollutant compositions and from 3 weather monitoring station of

We collected air pollutant data from Rourkela. 4
concentrations. Major outdoor air
data.gov.in and collected the data from 3
pollutants in Rourkela mostly producing 2
weather monitoring stations from 2017 Let (xi and yi) are the ith training data.
from steel plants and the chimney include
datasets preprocessing: 0
ozone (O3), particle matter (PM), sulfur 1/1/2017 5/1/2017 8/1/2017 12/1/2017
DATE
24 ∗1
dioxide (SO2), carbon monoxide (CO),
We paired the collected meteorological 𝑦𝑖 ∈ ℝ SO2 NO2 CO LEAD
nitrogen di-oxide (NO2),lead(Pb).An
data and air pollutant data on the basis of CONCLUSION REFERENCES
important reason for controlling air
date to obtain the required data for The general formulation can be expressed
pollutants specially SO2(Sulphur dioxide),
applying the machine learning methods. as We have developed efficient machine [1] Dixian Zhu 1,*, Changjie Cai 2, Tianbao Yang 1 and
which effects to premature death as well Xun Zhou 3 “A Machine Learning Approach for Air
To preprocess the data we calculated date learning methods for air pollutant
as increases of chronic heart and lung 1 𝑛 Quality Prediction: Model Regularization and
wise pollutants produces in Rourkela. prediction. We have focused on alleviating
diseases. According to WHO(World Health 𝑚𝑖𝑛𝑤 Σ𝑖=1 | 𝑓 𝑊, 𝑥𝑖 − 𝑦𝑖 |22 + 𝜙 𝑊 Optimization” Big data & cognitive computing,MDPI, pp.
Organization) Ozone can cause shortness 𝑛 model complexity by reducing the 1798-1805,February 2018.
of breath and causes coughing and number of model parameters and on
improving the performance by using a [2] Hinds, W.C. “Aerosol Technology: Properties, Behavior,
damage the lungs. Here we have put and Measurement of Airborne Particles”. John Wiley &
machine algorithms after analyzing and structured regulariser. We have also
Sons: Hoboken, NJ, USA, 2012.
reducing and data visualization. visualized the pollutants produced from
the steel plant in Rourkela by date.

ABIR SEN 518CS1007 DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING