AI-Based Digital Twins: The Modern Factory’s Newest Ally

In a multi-million dollar manufacturing plant, process flow is critical for optimum results and maximum productivity. A single bottleneck can disrupt workflow, interfere with the functioning of multiple processes and result in losses for a manufacturer. Aberdeen Research states that 82% of companies have experienced unplanned downtime, caused mostly by industrial bottlenecks. Unplanned downtime can cost a company nearly a quarter of a million dollars an hour!

Bottlenecks are caused by a number of reasons – namely, congestions during product flow, an overworked resource, a manufacturing process that lacks throughput, temporary blockcades and operations impeding production.  These examples are commonly witnessed in manufacturing plants, and the ongoing challenge for plant operators is to ensure assets are utilized to their maximum capacity without giving way to any hindrance in workflow. The Theory of Constraints validates the presence of bottlenecks fairly well – it hypothesizes that every complex system, including manufacturing processes, consists of multiple linked activities, one of which acts as a constraint upon the entire system – the constraint activity is the “weakest link in the chain”

Bangalore-based IoT startup SwitchOn has decoded how to identify this weakest link in the supply chain and consequently help companies overcome bottlenecks – by using Artificial Intelligence to predict multiple case scenarios and events, enabling real-time manufacturing intelligence from the shop-floor to the top-floor.

Founded by Aniruddha and Avra Banerjee, SwitchOn uses efficient edge computing systems to continuously analyze high-frequency data streams of vibration, temperature and energy consumption in manufacturing plants. This is done by providing “Digital Twins” of heavy assets, whose insights industries can use to monitor plant activity, predict availability and performance bottlenecks in advance.

Digital Twins and Other Technologies That Diagnose Your Plant

SwitchOn can start from legacy equipment or smart equipment to completely digitize them and expose high-frequency data patterns. SwitchOn builds Edge-Compute systems that process data on the shop-floor to reduce the operating costs and increase the vigour of products without having to work with multiple system integrators. The hardware is installed atop the equipment to retrieve data from the equipment in realtime. The company has built proprietary AI-based digital twins of multiple assets operated in the manufacturing industry to understand how the assets operate in the real-world environment. It can automatically recognize operating modes of the assets to identify and predict failures in assets as well as performance bottlenecks.

In addition, there is a scalable workflow created by SwitchOn that allows customers to collaborate on challenges that can be automatically recognised, making it simpler for customers to collaborate with stakeholders ranging from the shop-floor operator to the upper management.

 SwitchOn Raises $1 Million in Seed Funding

Earlier this week, SwitchOn made headlines for raising $1mn in seed funding from pi Ventures, India’s first Applied Artificial Intelligence, IoT and Blockchain focused early stage venture fund. Reports claim that Axilor and investors from The Chennai Angels also participated in this round. The company is expected to utilize these funds for strengthening edge-compute architecture and focus on market adoption in overseas markets in Asia Pacific and EU.

The Advent of Scalable, Market-Friendly Solutions in IoT

The fundraise for SwitchOn signals a significant turn for the automation industry in India. According to pi Ventures’ founding partner Manish Singhal, one of the key attractions of SwitchOn was their clear understanding of technology and consumer needs, as well as a minimally intrusive solution – the need of the hour in industrial IoT.

Moreover, SwitchOn’s unique algorithms tap into a combination of vibration data, electrical data and other signals to produce a “Digital Twin” of an asset – allowing for a variety of data for the benefit of the operator such as predictive maintenance, operational transparency, and machine availability.

The additional USP would also be the non-invasive nature of the product, which appears to be in vogue in the industry. According to McKinsey, advances in computer power, software development and networking technologies have made assembly, installation and maintenance of sensors and related products easier.

Plug-and-play technologies with the ability to compute numbers and generate data is a massive value add in industrial setups.

SwitchOn – A Reliable Ally in Factories

The product has already been tested in various industrial setups:

  1. Stamping Press – The product can automatically recognize the part getting stamped and the line stop reasons to provide analytics and identify defects in quality.
  2. Packing Machines – SwitchOn can recognize failures in sub-components of packing machines, and digitizes them automatically to identify failures
  3. Welding Payloads – Granular productivity analytics by recognizing the part being produced and detect high-frequency power consumption to detect quality defects.
  4. Robotic Arms – Also automatically identifies part being manufactured and predicts failures in robotic arms to digitize non-intrusively.
  5. Motors – Specific electro-mechanical failure patterns in HV-LV motors and high RPM gearboxes to predict failure patterns

 SwitchOn is incubated out of the NASSCOM Center of Excellence IoT Bangalore.

Share This Post

Leave a Reply