AI / ML with Agri-food Industries : Starting from Sweetest Place

AI / ML with Agri-food Industries : Starting from Sweetest Place

 

Kovaion’s New Exploration of Artificial Intelligence & Machine Learning (AI&ML) with Agri-food Industries: Starting from The Sweetest Place Indeed!

 

  • A new initiative by Kovaion with Ponni Sugars to bring out exciting value-added opportunities

 

In the quest for identification of process digitalisation opportunities in Agri-Food Industry, the Data Analytics and AI/ML (Artificial Intelligence & Machine Learning) team from Kovaion Consulting made a factory visit to Ponni Sugars. The team was taken through an end-to-end process study starting from sugar cane procurement till sugar crystal production stage.

 

In the process we made a detailed note of,

 

  • Procurement of sugar canes from specific centres with traceability information
  • Sugar Juice extraction with huge rollers,
  • Sending the by-product bagasse for paper and energy production
  • Clarification and concentration of juice for further crystallization process and finally,
  • Conversion to white sugar crystals and packaging the same.
  • After the completion of the production unit visit, we were also able to witness how the process control is done through data handling and automation.

 

With the support of the management and other senior members, we were able to identify opportunities for automation, as stated below

 

  • To make the data visualization more meaningful and relevant for the management using which they can take decisions quickly.
  • To digitalize the manual task in identifying the required slurry level during crystallization process.
  • After identifying the above use cases, our team grouped together in the task of creating sample prototypes.

 

Data Visualization:

We noticed that the reports sent to the management were:

  • More transactional and textual in nature
  • Not easily accessible from all devices/locations
  • Not real time
  • We came up with a quick resolution of transforming and modelling the sample data and produced visually appealing interactive dashboards.

 

Tools Used:

For illustration of the production and the procurement data, we utilized Microsoft Power BI to create dynamic dashboards.

 

Business Benefits:

We presented the dashboard to the management, and they were able to appreciate and accept that the newly created reports were:

  • Graphically appealing in nature
  • Easy to understand and interactive
  • Effortless to access from any location and any device
  • Simple to gather up-to-the-minute and accurate data and relay it to users as it happens
  • Helpful for making quicker decisions

 

Dashboards:

ai-ml-agri-ponni-sugars-2

Figure: Procurement Dashboard

 

ai-ml-agri-ponni-sugars-3

Figure: Plantation Dashboard

 

ai-ml-agri-ponni-sugars-4

Figure: Production Dashboard

 

ai-ml-agri-ponni-sugars-5

Figure: Transformation of Textual Data Format to an interactive dashboard format

 

Automation with Artificial intelligence and Machine Learning:

In the crystallization process, finding out the exact timing and the amount of seed/slurry (to initiate the crystal formation) to be added are very important factors.

In most of the industries these are generally determined manually by checking the texture and colour with human inspection, which requires extensive training and experience. We were also let known by the management that this process is prone to variability in manual assessment.

To make the process robust, we took a sample of the past data which had Purity and viscosity as independent variables and slurry category as the dependent variable and trained it using Logistic Regression to create a classification model. The accuracy on the test data came up close to 94%.

 

Tool Used:

The Scikit-Learn library of Python was used for building a statistical machine learning model to provide suggestions on addition of slurry amount.

 

Business Benefits:

  • Automation of the process with help of AI/ML driven techniques, to reduce manual assessment variability,
  • Determination of the optimal amount of the slurry addition to make the process, time and cost effective,
  • Reduction of cost required for experienced workers.

 

Dashboards:

ai-ml-agri-ponni-sugars-6

Figure: Slurry Level Prediction

 

Conclusion:

To make India, a global leader and self-sufficient in food & Agri sector, Indian companies need to modernize and adopt latest technologies using Data Analytics, AI&ML. The process study was an eye opener for us and has turned our focus towards this sector.

 

Authors:
Sudipta Kumar Hazra
Rituparna Dutta

AI & ML Solution

Kovaion will assist your business to implement AI & ML Services effectively.

Read More