Bert Labs CEO Unveils Reinforcement Learning Role in Industrial Energy Efficiency


Discover how Rohit Kochar, CEO of Bert Labs, sheds light on the transformative potential of Bert Nova, the flagship product within the Bert Platform Solution. In an insightful interview with TimesTech, Mr. Rohit details the utilization of Reinforcement Learning (RL) to optimize energy consumption and production metrics across industries. Learn about Bert Nova’s collaboration between components, real-world scalability, stringent data security, and its outstanding impact on energy savings and production enhancements. Mr. Rohit also outlines the challenges overcome during development and paints a vivid picture of the future, emphasizing data-driven innovation and the convergence of technology for smarter, more efficient industries.

Read the full interview here:

TimesTech: How do the components of Bert Platform Solution collaborate to enhance energy efficiency and production processes for clients?

Mr. Rohit: Bert Platform Solution is Reinforcement Learning (RL) powered optimization platform. It optimizes energy (power & fuel) consumption and production related metrices (yield improvement, raw material and consumables reduction, plant maintenance cost reduction). 

In line with Reinforcement Learning Neural Network framework which defines state vectors, action vectors and reward functions, parameters/ variables are sensed/ captured by Bert Maximus IoT devices, transmitted real time by Bert Qrious wireless sensor network and data stored, processed, analysed by Bert Nova Big Datalake software platform.

As part of Bert Nova multi-tenant distributed RL/ Deep Neural Network (DNN) computing platform, Operational Digital Twins are created through integration of DNN (mathematical) models and First Principles (Physics, Chemical, Biology) Models for determining current state and predicted state (at accuracy of 99.99%). 

On the above predicted Operational Digital Twin RL, Neural Network gets trained iteratively until there is 100% convergence towards the set points in a manufacturing plant, simultaneously leading to 30% improvement in power consumption, 20% improvement in fuel consumption, 25% improvement in plant maintenance cost, 3% improvement in yield, 0.5% improvement in raw material and consumables cost.

TimesTech: Could you share real-world examples showcasing Bert Nova’s scalability and its impact on data processing?

Mr. Rohit: Bert Nova multi-tenant distributed Computing Platform Microservices are hosted on any public and / or private cloud and / or centralised on premise server of Bert Labs’ clients. This receives data from the first factory as part of POC and multiple factories globally as part of the scale-up, where the data of all factories is stored, processed, analysed and RL/ DNN computed. 

Reinforcement Learning
Reinforcement Learning

TimesTech: What measures ensure data security in Bert Nova’s architecture, especially with its multi-cloud integration?

Mr. Rohit: As we integrate and connect data from a number of devices, connect and communicate, be it in smart building, manufacturing setting, pharma clean room, automotive paint shop, cement kiln or chemical production to protect the system from outside threat and to secure it is the topmost priority.  

Security engineers who are managing Bert Nova are creative, adaptable, skilled, and have executed the following: 

  • Input Data Validation prior to accepting the data e.g., Input format checks, Checksums and Input Boundaries Checks etc.
  • Output Data Validation e.g., Out-of-Boundary Checks and Balance & Reconciliation etc. 
  • Strategic points throughout an information application system e.g., Balancing Controls and Batch or Session Controls etc.
  • Mechanism for responding to validation errors are established 
  • Solution uses server-side checks and shall not rely on client-side checks for functions that can be manipulated to steal information or compromise the app
  • Procedures are established to obtain authorization each time data is copied from production system.
  • Production data used for application system testing shall be completely erased as soon as the testing is completed.

TimesTech: Can you highlight a standout project where Bert Nova drove exceptional energy savings and production improvements?

Mr. Rohit: Bert Nova drives remarkable energy savings and production enhancements wherever it is deployed. 

E.g.: At a Caustic Chlorine manufacturing plant, at a Pharmaceutical OSD Manufacturer and at an Ammonia recovery section

By seamlessly integrating with IoT devices on the office floor, factory floor, Bert Nova collected real-time data on HVAC performance, energy consumption, and comfort level.

Through sophisticated data computations, Bert Nova identified inefficiencies and patterns that were previously overlooked. It pinpointed specific AHUs which are consuming excessive energy during non-working hours and optimized it for efficiency. Additionally, predictive maintenance models enabled proactive repairs, reducing downtime.

Please check the case studies below

Reinforcement Learning
Reinforcement Learning
Reinforcement Learning

TimesTech: With your extensive experience, what notable challenges did you overcome while developing the Bert Platform Solution?

Mr. Rohit: Creating the Bert Platform Solution posed significant challenges that we strategically addressed methodically in the last 6 years and as a team we are improvising it. Scaling and performance optimization required a dynamic architecture. Integrating diverse data sources demanded adaptable connectors. Advanced analytics integration involved complex data processing. User experience design focused on accessibility. Interconnectivity required standardized protocols. Edge computing called for low-latency optimization. We are still working on two of the major challenge, ‘change management’ and ‘market education’ to facilitate adoption. Our continuous innovation, even during the COVID induced slow period, maintained our market leadership. As we tackle these challenges head-on, we have created a transformative platform empowering industries with efficient data management and utilization.

TimesTech: What future developments do you envision for Bert Nova and the broader Bert Platform Solution?

Mr. Rohit: At Bert Labs, as part of Bert Platform Solution, we are committed to pushing the boundaries of microservice computing and data management through our flagship product, Bert Nova. Innovation as part of Bert Platform Solution – Bert Nova, is a journey and not a destination, a journey which would be lived forever. We have some exciting research, innovation and development going on in this area. 

First and foremost; we are dedicated to enhancing the scalability of this platform. The team is working on maximising multi-tenancy in context of number of different client organisations, in different industry verticals, in several manufacturing plants and commercial buildings in different geographical locations, for variety of industrial use cases.

Second; the team is relentlessly working to optimize our microservice architecture, to seamlessly handle larger and more complex datasets across industry verticals, domains, and use cases. We envision Bert Nova becoming the go-to solution for enterprises seeking to process and store massive amounts of data with unparalleled efficiency.

Third; in terms of computation, we are investing heavily at the integration of physics modelling, chemical modelling, biological modelling, and complex mathematics (in form of advanced AI., Machine Learning, Deep Neural Network, and Reinforcement Learning) capabilities. Bert Nova engine can predict trends, identify anomalies at an accuracy of 99.99%, and provide real-time, on-the-go, actionable recommendations, enabling our clients to make informed decisions in real-time and fully automated execution of these commands, leading to efficiency on energy and carbon, production, quality and safety. 

Interconnectedness between these microservices, is another crucial aspect of our vision. Bert Nova is actively working on integrations and partnerships to enable seamless data flow between different microservices within the Bert Platform Solution. Our aim is to create an ecosystem where data can be effortlessly shared and utilized across various applications, boosting efficiency and innovation across industries.

Security and privacy remain paramount. As we evolve, we are committed to staying at the forefront of data protection technologies. We are investing in advanced encryption methods, decentralized identity management, and secure multi-party computation to ensure that Bert Nova remains a trusted guardian of sensitive information.

Another exciting development is our edge computation. Bert Nova extends its capabilities to the edge, allowing for real-time data processing, AI computation and decision-making directly at the data source. This is a game-changer for industries where our clients belong, requiring instant insights and rapid actions such as, automobile paint shop oven and paint booth, chemical soda ash carbonation and ammonia recovery, chlor-alkali electrolysis process, chemical recovery sections, cement pyro process, toothpaste/ soap/ detergent manufacturing process, water/ effluent treatment plants in beverage manufacturing, highly regulated pharmaceutical clean room process, utility/ boiler/ HVAC systems. 

The synergy between Bert Nova and the broader Bert Platform Solution will usher in an era of data-driven innovation, where businesses can achieve unprecedented levels of efficiency & effectiveness, creativity – constantly setting new gold standards, impact, growth, and minimizing human interventions.

Another area where Bert Labs innovation in Bert Nova is on business continuity and disaster recovery capabilities, where data storage, data processing, data analytics & computation, will seamlessly happen four pronged; from a public cloud, from a centralised on-prem server, from edge computing devices, and from controllers/ PLCs. 

I am incredibly excited about the future of Bert Nova and the entire Bert Platform Solution. Our journey is about pioneering new frontiers in microservice computing, and we are dedicated to leading the change toward a smarter, more connected, and data-driven world.