A big step in the direction of Industrial safety: Preventing gas leakage using machine learning.
Summary
TLDRThis video features a discussion with Vinit and D, winners of the Smart India Hackathon 2023. They developed a machine learning-based solution to prevent industrial disasters caused by gas leaks. Their system detects gas leakage in real-time and predicts affected areas, helping prevent accidents like the Bhopal gas tragedy. The solution uses existing sensors and digitizes them for real-time monitoring. The team aims to expand the project by incorporating expert knowledge to enhance their dataset and improve accuracy, ultimately safeguarding lives and infrastructure in industrial settings.
Takeaways
- 🏭 The industrial sector is crucial for national progress but faces challenges, especially with safety risks such as gas leaks.
- 🎉 Vinit and D, winners of the Smart India Hackathon 2023, developed an innovative solution to address gas leaks in industrial settings.
- ⚠️ Gas leaks in factories pose serious dangers, often going unnoticed until it's too late, leading to fatalities like the Bhopal gas tragedy.
- 🚨 Their solution provides real-time notifications and alerts to factory managers and workers, preventing potential disasters.
- 📱 The app they developed digitizes existing industrial sensors, sending critical data about gas leaks directly to users.
- 🌬️ The solution includes a machine learning model that predicts gas leakage spread based on environmental factors like temperature, wind, and terrain.
- 🔧 Industry personnel and emergency services can access the app to monitor sensor data, check gas pressure, and control ventilation in real-time.
- 🌍 The app will be available on Play Store and iOS, allowing widespread access to workers and managers for preventive safety.
- 💡 Their project is currently in the testing phase, with a focus on building datasets to refine the machine learning model for accurate predictions.
- 🔄 They aim to prevent future industrial disasters by providing rapid alerts and information to save lives and safeguard infrastructure.
Q & A
What is the main issue that the winning team at Smart India Hackathon 2023 addressed?
-The team addressed the critical issue of gas leakage in industrial settings, which can cause significant harm and even fatalities if not detected and managed in time.
How did the team plan to tackle the problem of gas leakage in industries?
-They developed a software solution to digitize existing sensors in factories and provide real-time data to industry owners. Additionally, they created a machine learning model to predict the direction and extent of gas leakage.
What was the team's main motto when developing their solution?
-The team's main motto was to save lives by preventing gas leakages before they occur, rather than just fighting the consequences of such incidents.
How does the machine learning model predict the danger zone of a gas leak?
-The model takes into account various factors such as temperature, humidity, and landscape to determine how far a gas leak might spread and in which direction, thus identifying the danger zone.
What happens when a gas leakage is detected by the system?
-Upon detecting a gas leak, the system triggers an API to send data to a backend server, which then uses the machine learning model to predict the danger zone and notifies relevant stakeholders through the app and local network providers.
Is there a need for subscription to access the team's gas leakage detection application?
-No, the application is open source and will be available on the Play Store for anyone to download. It also collaborates with local network providers to notify people in the affected area without the need for an app.
How does the application help in managing different levels of gas potency?
-The application is currently being updated to manage different levels of gas potency by improving the machine learning model to consider the weight, density, and other characteristics of individual gases.
What additional features does the application offer to industrial managers?
-The application allows managers to check real-time sensor status, control sensors, and contact emergency services directly from the app. It also sends notifications to healthcare services and authorities in case of hazardous situations.
Are the team's sensors proprietary, or do they work with existing industrial sensors?
-The team's solution works with existing sensors in industrial setups, digitizing their data and making it accessible through their application for real-time monitoring and control.
What is the current status of the application, and when can it be expected to be implemented?
-The application is mostly ready, but the team is still working on improving the machine learning model with a more accurate dataset to enhance the Hazard Zone prediction. The timeline for implementation is not specified but is dependent on the completion of this dataset.
What impact does the team expect their application to have in real-world scenarios?
-The team expects their application to significantly help in preventing gas leakage incidents, thereby saving lives and protecting industrial infrastructure, similar to the Bhopal gas tragedy, from occurring.
Outlines
🏭 Introduction to Industrial Safety and Innovation
The introduction highlights the critical role of the industrial sector in national progress, emphasizing the responsibility of safeguarding infrastructure. The video introduces the guests, Vinit and D, who are part of the winning team from the Smart India Hackathon 2023. Vinit, a computer science student, introduces himself as a founder of a startup, while D shares that he is also a part-time developer for a US-based organization. Their team’s solution addresses industrial safety, particularly in the context of gas leakage prevention.
🔥 Addressing the Problem of Industrial Gas Leakage
This section dives into the core problem the team tackled: gas leakage in industrial environments, which can be deadly if undetected. The team explains that past tragedies, such as in Bihar and elsewhere, could have been prevented with timely information. Their solution aims to alert factory owners and workers before such accidents occur, leveraging software to detect and prevent issues. They also developed a machine learning model to predict the spread and danger zone of leaked gas based on factors like temperature, humidity, and the surrounding landscape.
📊 How the Solution Works: Detection and Notifications
The technical functionality of the solution is detailed here. Sensors installed in gas cylinders constantly monitor pressure and gas type. When a leak is detected, the system sends data to a machine learning server that predicts the danger zone and alerts users. Notifications are sent to industry workers, managers, and local network providers, ensuring that even the general public can be warned. The solution is open-source and available to everyone through mobile apps, without a subscription, making it accessible to a wide audience.
📱 App Features and Community Involvement
The discussion moves on to how users interact with the system. Individuals can download the app, sign up, and receive alerts based on their area. The app has features like displaying affected zones and offering real-time updates on the situation. Users can also help during emergencies by contributing skills, and the app connects directly with emergency services like hospitals and fire stations. The focus remains on ensuring that information is shared rapidly, helping prevent or mitigate industrial disasters.
🔧 Enhancing the Solution and Future Plans
This section discusses the team's efforts to further improve the application. They are working with chemists to better understand gases and their behaviors, which will refine their machine learning model. The team is manually testing different gases and environmental conditions to create a robust dataset. Once they gain more knowledge about gases, they plan to enhance the application’s accuracy and bring it into real-world industrial scenarios.
📈 Real-World Implementation and Impact
Here, the team talks about the final stages of the application’s development, focusing on real-world implementation. The app enables factory managers to monitor sensors, control equipment, and contact emergency services remotely. Existing sensors in factories are digitized through the app, allowing managers to act quickly in case of gas leaks or other hazards. The team is optimistic about the impact their solution will have, believing it can prevent tragedies like the Bhopal gas disaster by providing early warnings and helping save lives.
🎯 The Final Vision: Preventing Industrial Tragedies
The conclusion ties the project’s significance to industrial safety and its potential to prevent disasters. The team reiterates their commitment to delivering a solution that can help avoid incidents like the Bhopal gas tragedy, ensuring that people are informed and can take action in time. They express their hope that the application will soon be fully implemented and recognized for its contributions to industrial safety and public health.
Mindmap
Keywords
💡Industrial Sector
💡Smart India Hackathon 2023
💡Gas Leakage
💡Machine Learning Model
💡Danger Zone
💡Real-Time Data
💡Sensors
💡Open Source Application
💡Emergency Contact List
💡Bhopal Gas Tragedy
Highlights
The industrial sector is crucial for the nation's progress, and safeguarding its infrastructure is vital.
Vinit and D's team won the Smart India Hackathon 2023 with a project focused on enhancing industrial safety.
Their solution addresses the significant problem of gas leaks in industrial settings, which can lead to deadly consequences.
Previous industrial gas leak disasters, such as those in Bihar and Bhopal, inspired the team's project.
The team developed a system to digitize existing sensors in industries to detect gas leaks in real time.
Their machine learning model predicts the spread of leaked gas by considering factors like temperature, humidity, and the landscape.
The model maps out danger zones to warn people and factory owners of potential hazards before they occur.
The application will notify users via a mobile app and through SMS in collaboration with local network providers.
The system is open-source, accessible through both the Play Store and the iOS App Store, with no subscription required.
Users can enter their area and industry ID for personalized notifications in case of gas leaks.
The system will contact local emergency services, including hospitals, fire stations, and police, as soon as a hazard is detected.
Factory managers can control sensors remotely and check real-time data on gas pressures and sensor statuses through the app.
The team is working on improving the accuracy of the machine learning model by collaborating with chemists to enhance their gas leak predictions.
Their solution is designed to prevent incidents like the Bhopal gas tragedy from happening again by providing timely alerts and notifications.
The project demonstrates how technology and machine learning can significantly improve safety in industrial environments.
Transcripts
[Music]
[Music]
the industrial sector is the backbone of
our nation's progress and with the
progress comes the responsibility to
safeguard the infrastructure today we
are going to meet a winning team whose
Triumph in the smart India haathon 2023
is not just a victory for themselves but
a significant milestone in advancing
safety measures within our industrial
Landscapes let's meet vinit and th
welcome to the show vinit and th please
introduce yourself to us first thank you
sir thank you
sir hello sir I am Vin Jan currently
pursuing my bachelor in computer science
engineering at chandigar University I
the found and a programmer science
startup profit
and now D will talk about himself yeah
myself is D and I'm also puring my best
degree from CH University currently I'm
start I'm in like second year and I'm
also working as a part-time developer
for an US based organization
T great so you two participated along
with your team in the smart India
haathon 2023 and came up as winners so
wonderful congratulations for that now
when you came up with that winning
formula or the winning uh solution uh
let's start talking about that so before
we go on to the solution let's talk
about the problem so what is the problem
area that you you know dealt with so the
currently the problem in the industry is
that uh you know in industrial settings
there's a deadly threat that is gas
leakage and you know Industries uses
hazardous substances uh that chemicals
that are when released in air can cause
significant harm to people there were
many cases in past like Bihar and
uh in other countries as well where the
gas leaked overnight and people couldn't
even know and by the morning that caused
a lot of deaths so the industry owners
Factory people the managers they even
didn't didn't know about the gas leak
that happened and people and media got
to know way after it was too late for
the you know ambulances and
Healthcare uh facilities to be informed
so we have created a solution
that targets on fixing this and
spreading the information very fast so
that before something happens like gas
leakage or fires we can notify people
and the factory owners that has happened
indeed that is certainly a very very
critical area and who would have
forgotten Bal gas strategy back then
that tragedy was like you know that
shook the nation and I wish we had
something like the solution that you
have come up with and uh it certainly is
going to be a pathbreaking solution I'm
sure so uh please tell us about you know
how did you start working on this
particular solution and what impact do
you think is going to make we had the PS
155 which targets or mean specifies on
aial Hazard we were thinking on how we
can provide the solution which can save
life our main motto was to save life you
know it's better to uh prevent fire than
fighting fire so our main motto was to
prevent the gas leakage from happening
most of the factories already have
sensors and all installed but they are
not digitized so we have provided a
software solution to digitize those
application and provide the data to the
industry owners real time so they can
know even a slightly change in data like
gas leakage they will be notified in the
app and they can instantly go and fix it
and stop it so that is the main part
that we have we are preventing before it
is occuring but we have also created a
machine learning model which predicts
the gas leakage you know when a gas
leaks in the air it spreads in the
entire area we need a proper machine
learning model to predict for in which
direction and up to how far that gas
leak will occur so we have made a list
of factors that determines how much the
gas will leak like temperature humidity
the landscape of the area it is a fact
you know open land or it is in between
multiple buildings so we have counted
all those factors and created a machine
learning model that predicts the Zone
Danger Zone in which the gas will be
leak so people can get notified of that
area and stay away from that area okay
great that sounds really wonderful and
uh in the era of you know Bal gas
tragedy this would have certainly made a
lot of difference so now tell me how
does it really work so when a gas is
leaked the sensor triggers a uh you know
the sensor in in each suppose a gas
cylinder is there it has sensors
installed that continuously checks the
pressure in the cylinder and also have
the data which kind of gas is this and
its density and all so whenever a gas is
leaked it triggers a API to the out to
our back end and sends the data about
how much gas is leaked at what rate the
gas is leaking and the all information
about the gas and in the back end we
have a machine learning server which
python server which predicts from the
data that what gas is leaked and
according to that gas leak we predict
the uh danger zone which is around that
area Okay once that leakage is detected
and your it comes to your system then
what what happens next like do you you
know put this up this information
somewhere or a person needs to access
your platform to get this information
how does it work after that basically
once we get the notification that the
gas is leaked in our database or in our
system then first step we will do we
will predict the area and we will map
that area to the user application which
is to all the like industry workers
industry employeer manager and all also
apart from this whenever it gets sleak
in certain amount of area so we will be
also collaborating with the local
network provider like ATL PS and all so
what we will do we will notify them by
means of like SMS or some kind of
messaging app so that the common people
also can get to know that this happens
here okay so does one need to subscribe
to this kind of a thing or how is it
available to Common masses or the
industrial no there is no kind of
subscription and all and it is open
source our application will be available
on Play Store everyone can like download
that and apart from the application also
if you don't have if you do not have
access to our application you can also
get benefit of this because we will be
collaborating as I already mentioned we
will be collaborating with the local
area provider local network provider so
if some kind of trasy some kind of has
that happens to your area location so
you will be notified so here is the
application that is available to
everyone from Play store or iOS store
they can download it so once they
download it they have a page for sign up
so they can enter their details and
according to their area they will get a
industry ID so if they are living in
bopal in certain area the each and every
Industries will have their own IDs so
when they sign up they will enter the
industry ID here
so once the sign up is complete we also
take the skills of that people so in
case of suppose a fire has occurred and
healthcare service cannot reach on time
so we can contact these people and based
on their skills we can get help from
local people as
well so here is the application so it is
currently showing that a gas is leaked
in this area and this is the Red Zone
which is the danger zone and this orange
zone is the means it is kind of Hazard
Zone but outside that that is the safe
Zone basically our Landing model gives
us two part of areas one which is like
intensely affected like 100% affected
and second area is like it might be
affected or not might be affected it's
like
5050 I guess every gas leakage would
have a different level of potency so is
the application also able to manage that
a certain gas will be you know less
deadly than the other one so your hazard
areas takes that into
account
yeah that's what we are like working on
right now because we know is gas is like
it's on it's own weight it's on like
density and all so for that kind of
reason we are working on that we are
preparing our machine learning model to
work on like each and every individual
guas part so that uh in future we will
be implementing that with the
application you are trying to update a
certain feature in there so what else
are you like you know working on right
now to make this even
better yeah so right now we are like
Gathering a team for like some chemist
peoples so those peoples will help us to
understand understand more about the
guesses the factors which we are not
aware about the gases they will help us
to like know about the GES and they will
help us to prepare our data set so after
preparing the data set we will be
implementing that data set in our
machine learning model and improving it
basically right now we are like
constructing our own data set or why we
are testing different different gases we
are manually spreading them we are
manually putting the like temperature
manually putting the wind speed and all
and like testing different kind of
scenarios and preparing the data s so
once that part of knowledge comes to you
once you know you have a better
understanding of gases and how these
operate then you'll make this
application even more robust and at that
point probably you would want to you
know Implement that in the actual field
conditions am I right yes sir there are
several other features as well in the
app like if you are the manager of the
factory and you are not in the factory
so you can open the app and check your
emergency contact list and contact the
Emergency Services fast from the app on
directly also our app directly contacts
to the hospitals fire office and police
station so whenever Hazard situation
occurred app already sends the
notification to each and every you know
healthcare services and all also in the
app there the managers can see the
realtime sensor status like all the
sensors are working correctly or not and
the pressure of each gesses and
everything these data is updated
like they can control the sensors as
well from the app like if they want the
doors closed or ventilation open they
can directly do that from the app
so these sensors that we are talking
about they are the ones which are
already existing in the any industrial
setup right or are you putting planning
to put your own sensors there which are
compatible with your
application very common in the uh you
know in industry but they are all
working in offline mode they are not
digitized so we are just digitizing
those sensor and putting the data on the
application and controlling the sensors
from the
application so basically you'll be using
the existing sensors which are there in
the industrial setup already right yes
sir okay so if I may ask how much more
time do you need to you know bring this
application or bring this you know
solution to the real world scenario I
think uh hard to like give you like like
appro time Al because we are like quite
slow in the process because you need to
understand that dealing with the guesses
like it's little bit like more complex
task the most important part of the
application is the machine learning
model and that model needs a data set
and so creating that data set is the
time-taking part overall the application
is ready so to make the this Hazard Zone
more accurate we need the machine
learning model to be better okay so in
the end I would like to ask you what
impact do you expect your application to
make in the real world scenario when you
actually implement it on ground so when
the application is uh you know the live
in live it will help more mostly in
preventing any situation from occur our
main motto was that so that this never
happens such like bopal gas strategy it
should have never happened if the
application was existing people could
have got notified about the hazard that
has occurred and they could have run
away from the area so that is the main
part we will be notifying and spreading
the information very fast you know
through social media and uh
notifications and alerts yeah of course
uh rightly said I mean if you're able to
avoid those kind of tragedies there
can't be anything better which can be
given to the as a gift to the industrial
world so great guys it was really nice
talking to you and understanding the
solution that you provided for
industrialized setup of the country and
I'm sure it's going to immensely help
the infrastructure that we have built
over a period of time and which is
helping the economy of the country and
you're trying to protect that very
infrastructure wonderful I wish you all
the best and I wish that this particular
application this particular solution
sees light of the day very very soon if
not much later thank you so much for
joining us today thank and we wish you
all the best in an era where
technological advancements are shaping
Industries this team has demonstrated a
profound understanding of the critical
importance of fire safety in industrial
settings their winning project is a
testament to their Ingenuity and
commitment to addressing the unique
challenges associated with industrial
safety we will meet another ingenious
team in the coming Edition watch out for
this
space
Ver Más Videos Relacionados
Weidmüller Industrial AutoML - Profitieren Sie von Machine Learning ohne Data Science Kenntnisse
Drinking-water distribution systems | Veolia
Finapp - Life from the cosmos, Corporate Presentation
Machine Learning Predicts Floods and Landslides [2024] | AI Project
Part 8/8: ML Based Web App Firewall : Testing the IPS in Real Time
How to Use the Si2 Series Acoustic Imagers for Pressurized Leak and Mechanical Fault Detection
5.0 / 5 (0 votes)