Skip to content

FelixJMartin/Considition-Hackathlon-2025

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

32 Commits
 
 
 
 

Repository files navigation

Considition 2025 — EV Charging Optimization

Team: Alex & Felix | Duration: One week (Nov 3–13, 2025)
Organizers: Consid AB × Vattenfall AB | Participants: 230+ teams


The Challenge

Develop an AI strategy to optimize electric vehicle charging across urban networks.

Objectives:

  • Maximize grid efficiency and renewable energy usage
  • Maintain customer satisfaction (minimize wait times, range anxiety)
  • Minimize costs (energy pricing, congestion)
  • Prioritize green charging (solar/wind availability)

Constraints: Dynamic demand, limited capacity, real-time weather, time-varying pricing


Our Solution

Three-Layer Strategy

1. State-of-Charge (SoC) Management

  • Emergency: <35% → immediate charging
  • Preventive: <65% → opportunistic charging
  • Persona-based targets: 87–97% (varies by customer type & charger speed)

2. Smart Station Selection

green_factor = base + solar_bonus + wind_boost - cloud_penalty
if congestion > 90% and soc > emergency:
    skip_station()  # Avoid queues

3. Customer Persona Modeling

  • Eco-conscious: +3% target for green stations
  • Cost-sensitive: -2% target to save money
  • Stress-averse: -2% to reduce range anxiety

Key Features

Dynamic Green Energy Scoring

Weather + time-of-day → green factor (0.0–1.0)

  • Solar peak (10am–3pm): +0.10
  • Wind strength: +0.25×wind
  • Cloud cover: -0.20×clouds

Adaptive Charging Targets

Scenario Fast (≥150kW) Slow (<150kW)
First charge 93% 90%
Regular 90% 87%
Emergency 96% 93%

Memory & Learning

  • Track charging history per customer
  • Estimate consumption rate (exponential moving average)
  • Cooldown period to prevent over-charging

📈 Example Results

Final State Distribution:
  Home: 487 | Charging: 156 | Moving: 98 | Waiting: 23

Average final SoC: 84.7%
Customers charged: 612/650 (94.2%)

Tech Stack

Python | RESTful API | Event-driven tick simulation

Core logic:

  1. Parse game state → extract customer positions/SoC
  2. Score stations → green factor × availability × speed
  3. Generate recommendations → customer→station→target SoC
  4. Submit → local validation → cloud competition

Real-World Impact

This addresses:

  • Smart city EV infrastructure planning
  • Grid stability with renewable energy
  • Fleet logistics for electric vehicles

Competition

Considition 2025: Annual AI hackathon by Consid + Vattenfall
Focus: Sustainable energy & smart cities
Format: One-week practice + final night live leaderboard


TL;DR: Built a reactive AI system that routes EVs to optimal charging stations based on battery level, customer persona, green energy availability, and congestion. Achieved 94% customer coverage with balanced cost/sustainability trade-offs.

About

Participated in considition global hackathlon nov 2025 with the aim of using "AI and machine learning, develop an algorithm to optimise EV charging in urban environments."

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages