In a guest article for EVreporter, Vinay Gunasekaran – a tech Leader working at the intersection of batteries, motors and power electronics shares his thoughts on how Electric Vehicle ecosystem can be enhanced for all stakeholders through IoT and Artificial Intelligence (AI) solutions.
The mobility industry in India is on the cusp of an electric revolution. As customers step out of the COVID-19 pandemic, it is widely anticipated that personal mobility through electric two-wheelers would see greater demand. And with rising fuel prices, electric might also become the obvious choice for an operationally intensive three-wheeler logistics industry, especially in the first and last mile spaces.
But all that said, electric vehicles (EV) in India is still a very nascent industry. The number of electric two-wheelers sold in India in FY 19-20 is approximately 1.5 lakh units. Compare that to about 2 crore units sold for traditional IC engine based two-wheelers in India in the same period.
The level of stability we see in the IC engine based vehicles and their ecosystem today is after decades of R&D being spent in improving the performance and robustness of these vehicles, and building supporting infrastructure.
However, most of the OEMs and suppliers in the EV industry are new and in all likelihood haven’t seen a complete product life cycle yet. With rising demand and push from competitors, they would have to be really fast in evaluating components, integrating parts from suppliers and bringing the product to market. It might be imperative to launch and pilot multiple products with different features to understand customer preferences and evaluate various business models in this new industry.
Areas to benefit most from AI and IoT based models
Most EV players in India are moving towards adopting superior Li-ion battery packs for their vehicles. While Li-ion batteries have a longer life and a higher energy density giving you a farther range, they do come with challenges – the charging and discharging of these packs have to be very tightly controlled and the temperature needs to stay within limits. Li-ion packs are expensive, often costing up to 50% of the vehicle itself. Faster wear of these batteries would prove very costly. A battery management system (BMS) is usually present on board to address these challenges but those systems need to be properly evaluated and monitored.
Fortunately, a lot of these challenges could be managed by gathering onboard sensor data through IoT and running them through physics and AI-based models to evaluate the performance and robustness of the system.
I worked at MathWorks in the USA for 5 years where we helped a major automotive OEM in the Detroit area in building models to predict the health of their Li-ion battery packs.
IoT, or Internet-of-Things is a very upcoming technology landscape enabling machines to send important sensor data to a server and helping gain insights on its operation, performance and wear. IoT has been made possible through innovations that have brought down the costs of mobile computing and Internet access to near-ubiquitous levels.
You could quickly perform accelerated experiments on a few individual Li-ion cells to understand their full charge/discharge patterns, partial charge/discharge patterns, stress patterns (including thermal stress) and collect data at each step. With this data, and using physics behavioral equations, you could characterize models, integrate AI algorithms and deploy them on a server. Now, as this server monitors vehicles using IoT, these models would be able to present you insights on the exact state of charge, state of health and remaining capacity for the battery packs of the EVs.
Predictive Maintenance and Warranty models
After my stint in the US, I came back to India to head R&D at Ecozen as their VP Technology where we designed, developed and deployed 20,000+ power electronics + motors + IoT systems with remote performance monitoring and predictive diagnostics.
As a part of tender requirements for Solar Pumping systems, we were mandated to provide 5 years warranty and an aggressive 72 hour turnaround time for issues. From the design stage itself, we started using IoT data and AI-based models to evaluate system performance and extrapolate future trends to predict how these systems would start wearing under different conditions. Even post production, our models and algorithms were constantly monitoring all remote system data and immediately flagged any deviations, which could be fixed in the next production lot. Using AI and IoT, we always ensured that our warranty costs were under limits.
Even with carefully designed systems, there would still be conditions where EV components could fail. In such cases, AI algorithms, processing remote IoT data would actually be able to predict an upcoming failure. EV users can be alerted well ahead of time to get things fixed before a total breakdown, providing an enhanced customer experience.
If I look at the EV industry, I see many parallels. Being a new territory, electric vehicles require a redesign of warranty models. The technology today, including Li-ion for Indian temperature and moisture conditions is very new. Remote performance monitoring of these systems could very quickly and efficiently help in fixing issues that come up. In fact, potential customers wary of adopting new technology can be offered a sense of security and comfort through transparent monitoring of vehicle health and predictive diagnostics.
The past has been littered with cases of pilots and financial models failing, generating a poor return on investment and creating a weak public impression because of immature technology deployments even when the fundamentals on paper made sense. In the EV industry itself, the initial deployment of three-wheelers with lead-acid batteries is a great example.
As the EV industry is embarking on this journey of mass customer adoption, it is important and arguably crucial to balance the pace of deployments with great customer experience. IoT and AI solutions would provide a great way to achieve that balance – quick vehicle development and deployments without compromising on robustness and customer experience, which would help in growing the industry successfully.
About the Author
Vinay Gunasekaran has a Masters from Carnegie Mellon University and was until recently, VP Technology at Ecozen, where he led design, development and deployment of about 25,000 intelligent and IoT enabled solar cooling and pumping solutions. Prior to that, he was in the US at MathWorks helping EV and automotive companies in analyzing sensor data, controls/software development, optimizing systems and predictive diagnostics.