Learn how Bidgely’s EV disaggregation technology is unlocking your EV program potential in the latest white paper.  Explore the fundamentals of Bidgely’s EV load disaggregation and application to the demand side management of EV chargers, challenges of EV detection and estimation, user charging insights pulled from AMI data, and time of use demand response opportunities.

AI-Powered EV Detection and Estimation Helps Utilities Realize the Promise of the EV Revolution

With an all-time high global adoption rate, the proliferation of electric vehicles is set to impact commercial and residential energy demand like never before. According to the Global EV Outlook 2020, the sales of electric cars reached a record-setting 2.1 million worldwide in 2019, bringing total EV ownership to 7.2 million EVs.

Fig 1: Industry trends on the Electric Vehicle market. Image source: https://www.marketsandmarkets.com/Market-Reports/

As auto-makers increasingly turn away from internal combustion engines and focus their R&D on electric vehicles, responsibility for powering cars is shifting from the oil and refining industry to utilities and the electric grid. The challenges and the rewards to utilities are both high.

In times like these when innovation and the ability to pivot are crucial, AI-driven solutions help utilities remain future-ready. 

A More Sophisticated Understanding EV Usage and its Impacts

Increasing numbers of electric vehicles and the associated public and private charging infrastructure are making dynamic load management and demand forecasting increasingly complex. 

For example, when large numbers of EV owners in one geographic location start charging their vehicles at the same time of day, the grid experiences unanticipated demand, and in extreme cases, outages. 

With data-driven insights, however, utilities are able to anticipate how EV charging will grow over time, identify feeder lines and individual transformers that could come under strain due to EV proliferation, and pinpoint where the addition of public docking stations would yield the most benefit. 

This intelligence then quantitatively and accurately guides utility investments and timelines in connection with staged grid upgrades, charging infrastructure deployments, and targeted roll outs of EV rates or controlled chargers to encourage load shifting as EV adoption expands. Using AI, utilities can effectively balance demand vs supply and avoid major setbacks. 

Overcoming Roadblocks to EV Detection and Estimation

Because EV penetration is skewed to select geographies and public research and datasets on EV penetration is limited, developing robust EV solutions that algorithmically detect EV ownership and estimate EV usage is incredibly difficult. Because the market is transforming rapidly, it is also not easy to forecast demand, manage power load on the grid, and determine optimal, surge-preventing logistics for charging docks. And, because the fossil fuel-to-EV transformation is unprecedented for grid companies, there is no legacy playbook on which to build. 

These barriers are significant. In fact, Bidgely is one of the only companies in the world to have successfully developed patented EV detection and estimation algorithms. UtilityAI accurately identifies:

  • The number and location of EV chargers
  • Charger size for all chargers in use and the associated regional charger-size mix
  • The time and duration of EV charging activity

Learn more in this infographic.

Importantly, Bidgely achieves these insights without any ancillary hardware monitors or sensors — which have historically required time and money on the part of the utility and effort on the part of the customer in order to ensure accurate measurement. In fact, an analysis by a large utility in the northeastern US demonstrated that Bidgely’s easily scalable software solutions achieve 90% accuracy and detection in the estimation of EV load charge.

Bidgely’s EV detection and estimation algorithm requires only weather data and smart energy meter data with different sampling rates (such as 15 mins, 30 mins, and 1 hour) combined with behavioral energy consumption studies that isolate EV owner energy patterns. Behavioral patterns inform the supervised tree-based machine learning mode for EV detection, including seasonal dependency, time of usage habits, duration of charging and more. The model has been trained on users from varied geographies including North America, Europe and Australia.

Bidgely also employs state-of-the-art computer vision and deep learning techniques such as box filtering, box refinements and object detection to extract EV charging instances from user energy consumption data. The patented approach also borrows learnings from image processing methods for multiple post-processing steps. Owing to the quality of the algorithms and comprehensiveness of the training data used, the trained model is extremely robust and is able to provide an accurate sampling-rate level estimation.

Bidgely’s UtilityAI Science in Practice

Bidgely’s proprietary algorithm suite detects and distinguishes EV chargers with a high degree of accuracy. 

The following examples illustrate the capabilities of the machine learning model to identify charger type and other important insights that can be derived from EV owners. Heat maps depict the amplitude of power consumed. Color is most intense (red) for high energy values and least intense (blue) for low energy values. The heatmap displays the power consumption of 365 days, where each day is represented by one row. 

Beyond Grid Management – A Utility-Wide Multiplier Effect

Strategic infrastructure management is only one benefit of an AI-informed approach. Bidgely’s EV intelligence can be put to work to improve planning, investments and outcomes across utility operational areas, including:  

  1. Customer Satisfaction: Our EV disaggregation module detects and estimates EV power consumption through an AI-enabled solution applied to consumers’ power meter readings. This serves as a credible behavioral analysis source to EV owners and nudges them to be more mindful of their usage. 
  2. Grid Planning and Management: Bidgely’s load management and demand forecasting solutions equip utilities to solve demand-supply issues. 
  3. Energy efficiency: Using UtilityAI’s hyper-personalized recommendation engine, consumers are propelled to take actions like upgrading to more efficient EV chargers or more desirable rate-plans for EV charging, to consume less energy. 
  4. Targeting: The UtilityAI platform makes targeting EV owners with personalized insights and advice efficient and effective. 
  5. Demand Side Management: UtilityAI also easily fulfills ad-hoc requests from grid companies to facilitate detection and estimation at the most granular level.

Looking Forward

Never before have we been able to understand so much about EV impacts without the use of hardware sensors or time intensive and expensive survey efforts. Using only AMI data, machine learning and artificial intelligence, Bidgely is able to help utilities quickly, cost-effectively and accurately understand the multi-faceted impacts and opportunities associated with the electric transportation revolution. 

As the world continues to transition to electrified vehicles, this approach will serve as the foundation for successfully mitigating hurdles, resolving inefficiencies and conducting advanced analysis, planning and engagement. Electric vehicles are the future, and UtilityAI keeps our partners future-ready.

Learn how Bidgely’s EV disaggregation technology is unlocking your EV program potential in the latest white paper.  Explore the fundamentals of Bidgely’s EV load disaggregation and application to the demand side management of EV chargers, challenges of EV detection and estimation, user charging insights pulled from AMI data, and time of use demand response opportunities. 

Categories: Bidgely