Statistical Analysis of 12V Power Load Uncertainty Using Conditional Probability for SDV Energy Management

Gyeongmin Han, Sungmin Lee, JoonHee Lee, Kichang Lee, Jeonggil Ko

Keywords: SDV, Battery Management System, Load Prediction, Conditional Probability, Proactive Control

TL;DR: Leveraging past power usage history reduces 12V load prediction uncertainty by 41.7%, enabling proactive energy management for Software Defined Vehicles.

Abstract: As Software Defined Vehicle (SDV) technology advances, the surge in in-vehicle electrical loads has revealed the limitations of existing reactive battery management systems. This study statistically analyzes the utility of past power usage history to address the uncertainty of driver patterns in power load prediction for 12V battery optimization. Analyzing heater and air conditioner loads using the public VED dataset, we confirmed that using the past 60 seconds of history as an input for conditional probability distribution reduces prediction uncertainty (variance) by up to 41.7% in high-load regions. Validation using a LightGBM model demonstrated that the proposed method exhibits superior tracking performance compared to existing methods during rapid load fluctuations.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Submission Number: 6