Kovaion’s New Exploration of Artificial Intelligence & Machine Learning (AI&ML) with Agri-food Industries: Starting from The Sweetest Place Indeed!
Blog written by Sudipta Kumar Hazra & reviewed by Rituparna Dutta of Kovaion Consulting, Coimbatore
Sugar sweetens everything it touches, be it our daily food and beverages, or gulping bitter medicines easily. In the same vein, our visit to Ponni Sugars to understand the complex production process flow was effectively sugar-coated by the cordiality and hospitality of the management in allowing us and taking us through the whole process seamlessly and patiently.
With the aim to aid Agri-Food Industry with our Data Analytics & AI/ML expertise, our team embarked on a journey for the plant visit. Our team drove from Kovai to reach Ponni Sugars on the outskirts of Erode, early in the morning on 28th July 2022. On the way we had taken the winding roads through villages filled with lush greenery and the scenic landscape which kept us enthralled and engaged.
We reached the gate on time and in few minutes, we were able to reach the main conference meeting room without any hassles. We were surprised to see a long queue of trucks loaded with sugarcane waiting for their turn to offload.
To give a short introduction about Ponni Sugars, it is the brainchild of Late Shri S. Viswanathan with the goal of making the sugar industry circular and sustainable, both economically and environmentally. To dive a little deeper, along with sugar crystal production, Ponni Sugars is also utilizing the waste product (bagasse) for paper and energy production, and molasses for other by-product industries, to reduce waste disposition as much as possible.
Our visit started with a brief interaction with Mr. K. Yokanathan, President of Ponni Sugars. We were introduced to the Management team of Mr. B. Chandrasekar, Senior Vice President (Operations), Mr. R. Chandrasekaran, Deputy General Manager (Engineering) and Mr. A. Ravikumar General Manager (Engineering). We presented them with the objectives and goals of this process just to set the right expectations. Soon after, they handed us over to one of the Operations in-charge, he took us inside the processing units and provided us with a detailed insight about the process steps and also about the recently deployed automation systems which help in monitoring and guiding the various stages of white sugar crystals production from sugar cane. As we carefully walked through the plant, seeing in awe the huge crushing mills cleaning, washing & squeezing out sugar canes, we came to understand the complexity of the process on a high level.
Then, the lead Instrumentation Engineer took us through the Quality Control Rooms where they were keeping a close eye on all the process parameters through different process control softwares. We next went to the area, where concentrated sugar juice is converted to sugar crystals, which further go to the Packaging unit to get packed and go on a ride through a long journey to come to your plate directly or indirectly. We could understand the importance of configuring & maintaining various parameters like purity & viscosity and noted the manual intervention to determine the slurry levels, which could be a potential use case for Machine learning provided the past data is made available.
We also had a short visit to a different section of the plant to understand how electricity was generated using the sugarcane waste and steam, which was then used not only for the plant operations but also was supplied to the local grid.
Finally for the day, we went to the IT department, where the IT head Mr. Muthu, provided us the information on tools and technologies they use to maintain their applications and database. He was generous enough to show us the management reports format and we could sense that the format could be presented much more graphically showing the trends and insights in form of dynamic dashboards which would be easier for the senior management to understand and take decisions quickly. On our request, Mr. Muthu also provided us with a sample dataset of the procurement data and production data.
After updating the President, we wound up for the day updating him about the two use cases for which we would present him with solutions the next day afternoon. He was very helpful in providing us with a Guest House accommodation which was a short walk away at one end of the plant close to the bank of the Cauvery River.
After finishing the day in the production line, we started working on the data at the bank of the river during sunset. Our data analytics team in Chennai came to our rescue and was quick to understand the data and came up with relevant dashboards, one for procurement and another for production.
In parallel, we had to come up with a sample machine learning model to predict the slurry level based on features like Brix & Purity. We had to take some publicly available data from internet for this purpose. It was a great experience to work on a Machine learning prediction model in the midst of strolling Peacocks on one side and the sun setting slowly on the other side.
A special mention has to be said about the quality of the food offered at the guest house. It was definitely a luxury considering the remoteness of the location without which we would have struggled.
It was a long night for us as we needed to design, coordinate, validate & finetune the models & reports for both of the use cases.
The next day, we devoted the morning session showcasing our solution to Mr. Muthu, the head of IT. He guided us on the changes needed keeping in mind the interests of the management.
For illustrating the procurement & production data, we used Microsoft Power BI to come up with interactive dashboards, which could be easier for the management to monitor from any place and through any device. For providing prediction about Slurry levels, we used logistic regression Machine Learning model using Python.
Finally, we presented our Analysis and solution to the management which was very well appreciated. The management was very forthcoming with their recommendations and thoughts about probable use cases in Inventory and Stores and promised to get back to us soon after internal discussions.
To conclude, our first Food Industry visit came to an end with a satisfaction that we were able to understand the opportunities and come up with proof of concepts & prototypes very quickly which was accepted by the management. This has given us a huge confidence and a greater focus towards the automation needs in other agri-food related industries. We eagerly look forward to exploring more such Agri-Food entrepreneurs and Industries to aid them with our expertise.
