ADS for Wind farms is the ADS core product declined for a wind farm operation use case. You can find a tour of its functionalities in the product overview screencast.

First set the docker compose setup. This will spin up a complete environment with all the moving pieces you'll need. Namely:

• A postgres database
• A rabbitmq queue
• An elastisearch cluster
• A grafana instance
• A worker instance
• A web frontend instance

docker-compose up -d

Once everything has run correctly, run the database migration with

alembic upgrade head

Then follow the instructions on the PROVISIONING section.

Once you have your dockerised environment up and running, you can connect to localhost:5000. Log in by using the default credentials detailed in build.env.

This is valid only on a fresh installation of the platform with all migration executed

• Activate the virtual environment and export your app config

$ export APP_CONFIG=environments/local.env

From the root of the project run

$ bash provisioning/ads/provision.sh

This will populate the local database with some (fake) data

connect to the instance of postgres

$ psql -h localhost -U postgres

connect to the appropriate database

\c anomaly

From the root of the project run

$ bash provisioning/grafana/provision.sh

ADS is roughly divided in two parts: the web server and the worker. The division is reflected in the two different processes that are brought up by docker-compose.

• The web server talks to the database and the data lake, also with the queue in order to schedule a prediction. Data shown is fetched from elasticsearch and postgres, and shown on the web interface accordingly.

• The latest WF model is under the models module. Specifically, it's the models.chunkycnn.ChunkyCNN class.