Data-Driven Battery-Lifetime-Aware Scheduling for Electric Bus Fleets

on YouTube ↗06m

" summary

A data-driven scheduler that uses bus GPS and transaction data, K-means to extract stations, and a Prophet time-series model to predict passenger arrivals feeds a Lyapunov optimization scheduler that minimizes time-average battery degradation while controlling passenger queues. Evaluated on a 12-month Shenzhen dataset, the system yields about 14.3% to 21.7% reduction in battery degradation under comparable passenger loads, maintains comparable passenger wait times, tolerates parameter estimation error, and runs in polynomial time with deterministic performance bounds.

AI-written summary · about

" chapters (6)

1. 0:00Start
2. 0:19Introduction
3. 1:10Motivation
4. 1:51System Overview
5. 4:35Data-Driven Evaluation
6. 5:29Conclusions

" related talks

• 43mHacking your local bus system for the greater goodFlorian Harr · Mobile Era 2018
• 16mTidy Transit: Real Life Data Modeling for Public TransportationHunter Owens · rstudio::conf 2019
• 57mSmarter Tools for (Citi)Bike-SharingDavid Shmoys · Real-Time Decision Making
• 06mUnderstanding User Behavior in Car Sharing Services Through The Lens of Mobility: Mixing ...Shuai Wang, Desheng Zhang, Guang Wang, Harsh Rajkumar Vaish, Huijun Sun, Jianjun Wu · UbiComp 2021
• 21m"Capacity Prediction" instead of "Capacity Planning":Rick Boone · SREcon18 Americas
• 16mTidy Transit: Real Life Data Modeling for Public TransportationHunter Owens · rstudio::conf(2022)
• 06mD2Park: Diversified Demand-aware On-street Parking GuidanceZijian Cao, Huadong Ma, Desheng Zhang, Dong Zhao, Chen Ju · UbiComp 2021
• 18mBatchy: Batch scheduling Data Flow Graphs with Service level ObjectivesGabor Retvari, Tamás Lévai, Felicián Németh, Barath Raghavan · NSDI 2020