The required input features are as follows.
Station
- Station id : dims = (1,)
- number of chargers(slow/fast) : dims = (2,)
- supply capacity : dims = (1,)
- location of station(lat/lon) : dims = (2,)
- average of taxi trip of road link within 500m radius : dims = (1,)
- proportion of road type within 500m radius(4 road type) : dims = (4,)
- proportion of district type within 500m radius(5 district type) : dims = (5,)
Sequence
- Realtime Sequence : dims = (12, 1)
- Historical Sequence : dims = (4, 1)
Time
- (Time Index, day of week, weekday) : dims = (3,)
Outputs are as follows.
- Occupancy_20, 40, 60, 120 / continuous : dims = (4, 1)
Original Repo: https://github.com/easttuna/ev-charger-occupancy-prediction
Clone git repo on EC2:
sudo yum update -y
sudo yum install git -y
git clone https://github.com/hwangpeng-sam/model-serving.git
Run shell script files to build docker image:
cd model-serving
sh setting4build.sh
sh image_build.sh
After setting aws configure, run the shell script file to push image:
(Modify region, Elastic Container Registry!)
sh image_push.sh
After create aws lambda function, connect your DB (by creating your private.db_info and inserting values into table)
Lastly, create test event as belows.
( lambda function configuration: Memory(2048 MB), Runtime(1m) )
event = {}