How to create a new sample app#

This guide provides step-by-step instructions on how you can create a new sample app by taking Wind Turbine Anomaly Detection OR Weld Anomaly Detection sample app as a reference by configuring the generic Time Series AI Stack. Create a copy of the edge-ai-suites/manufacturing-ai-suite/industrial-edge-insights-time-series/apps/wind-turbine-anomaly-detection folder and replace the folder name with the sample app name to be created.

Steps to follow#

Configuration#

1. Data Simulators/Destinations:#

Options available:

  1. Update the OPC-UA/MQTT simulator containers (edge-ai-suites/manufacturing-ai-suite/industrial-edge-insights-time-series/simulator) as needed to ingest the needed dataset to analyze via a CSV file

  2. If you have existing OPC-UA/MQTT simulators, you can use that too. Just adjust the compose and helm template files similar to the above existing OPC-UA/MQTT simulators

  3. Directly configure Telegraf to connect to the device acting as OPC-UA server OR MQTT publisher/broker. Refer to the Telegraf plugin documentation for more details

2. Generic Time Series AI stack#

  • Data Ingestion - Telegraf

    Update the [[inputs.mqtt_consumer]] and [[inputs.opcua]] sections in edge-ai-suites/manufacturing-ai-suite/industrial-edge-insights-time-series/apps/<new-app-name>/telegraf-config/Telegraf.conf, based on the MQTT/OPC-UA configuration, the input would be read and the topics in MQTT input plugin and name_override in OPC-UA input plugin is used for writing this as measurement in InfluxDB.

    In the above same conf file, the two [[outputs.influxdb]] plugins publish the data to InfluxDB and Time Series Analytics microservice

  • Data Processing - Time Series Analytics Microservice

    The tick_scripts, models and udfs folders at edge-ai-suites/manufacturing-ai-suite/industrial-edge-insights-time-series/apps/<new-app-name>/time-series-analytics-config constitutes the User Defined Function (UDF) deployment package. The config.json at the same location is the Time Series microservice configuration.

    More details at how-to-configure-custom-udf

  • Data Visualization - Grafana

    Prepare the custom grafana dashboard (edge-ai-suites/manufacturing-ai-suite/industrial-edge-insights-time-series/apps/<new-app-name>/grafana-dashboard.yml) which gets volume mounted in the grafana microservice at edge-ai-suites/manufacturing-ai-suite/industrial-edge-insights-time-series/docker-compose.yml.

Deployment#

Note: Adjust the edge-ai-suites/manufacturing-ai-suite/industrial-edge-insights-time-series/Makefile to add the support for the new sample app name for both docker compose and helm deployments

  1. For docker compose deployment, follow the Get Started guide/

  2. For helm deployment, follow the How to deploy with helm