WARNING: Submitted to HooHacks by mistake. Please don't disqualify this submission from Rowdyhacks.
What's your idea? This will be a short tagline for the project
Tankmensional is a web-based tool for college students that predicts and plots the consumed power for liquid agitation using fluid mechanics models.
We wanted to build something closely related to our studies. The four of us are studying different specialties of Engineering, so we decided to work on a Mechanical Fluids project. We are highly interested in Machine Learning for data analysis and we decided to challenge traditional models with the use of AI-based techniques. We also wanted to build a beautiful solution that would make data collection, data modelling and data representation easier and faster for students.
What it does
Tankmensional predicts the consumed powers for several kinds of mixers using linear and dimensionless models. TankMensional also provides a clean environment in which the user is able to upload experiment data to compare its results with our data modelling studies.
How we built it We developed our project using various tools. For the database we used the** MongoDB Atlas** service, being able to access it both via MongoDB Compass app and python.
The majority of this project has been developed in Python. We used the following libraries: PyMongo: For access to database ** SKlearn*: We used this library to model our data **Flask, **Bootstrap 4* and Jinja: these libraries were used to develop the app service Chart js: finally, we used Chart js to display the data and the model in the web browser
Challenges we ran into
At first we started by working in a Machine Learning model from the linear regression learner of Matlab. However, we encountered difficulties when trying to call our Matlab function from Python so we pushed forward with Python libraries.Another challenge we ran into was transferring our model from GoogleColab to python in our CPU because we got errors installing Tensorflow libraries.
As is common in hackathons time was scarce and we have been challenged at meeting the deadline.
Accomplishments that we're proud of
We have done something we find difficult in very little time.We had no previous knowledge about flask and MongoDB, noone on the team had worked on this previously.And also is the first time we have done a direct user application.
We are especially proud of having been able to develop a full-functioning web app.
What we learned
We have learned that Matlab is not the best language when it comes to connectivity with other platforms. Also that libraries of TensorFlow are not very python3.8-friendly. We have learned to program with flask libraries,dealt with MongoDB data structures .
What's next for TankMensional
We would like to** further develop** our web app to implement services as user creation and account management. Also we would like to invest some time in the web design. The following development steps would be to implement the ability to registered users which have been previously certificated to** add more data to our database*. Another development step we would like to explore is providing the ability to the user to choose between* several AI-based model fitting techniques**.