| ... | ... | @@ -17,42 +17,48 @@ You can check the datasheet of the used Sensor PZEM-016. |
|
|
|
* **Power factor**
|
|
|
|
* **Alarm level**
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
!\[11\](\<img src="../images/Systemarchitekture.png" alt="Systemarchitektur" width="600"/\>)
|
|
|
|
|
|
|
|
The IoT device sends this information cyclically to a topic on the MQTT broker. To understand the system architecture, please refer to the schematic diagram below.
|
|
|
|
|
|
|
|
TODO hinzufügen Systemarchitekture -\> Geht nicht mehr wiesooo?
|
|
|
|
|
|
|
|
//Bild Curves
|
|
|
|
|
|
|
|
As you can see, we will operate the coffee machine through the power box. The power box automatically connects to our WiFi router. We have created an MQTT broker on a server laptop. The data is published on a topic there. If you have no experience with MQTT communication, please read the following article: https://www.hivemq.com/blog/how-to-get-started-with-mqtt/
|
|
|
|
|
|
|
|
We receive characteristic energy profiles for the different product possibilities of the Senseo. What we aim to achieve at the end of the course is the determination of the product based on this data and a trained AI model.
|
|
|
|
|
|
|
|
//To Do 4 Kurven aus Media hinzufügen
|
|
|
|
|
|
|
|
One possibility would be to manually label these energy data with the different product categories and use this information to train a model. As you can imagine, this is very labor-intensive and error-prone. Therefore, we want to "hack" our coffee machine and use low-cost sensors to automatically obtain additional information from the machine. This will mainly be your task in the hands-on course. To achieve our goals within a week, you should come up with a strategy in advance on how the system architecture could look.
|
|
|
|
|
|
|
|
## Product Variants We Can Produce with the Machine:
|
|
|
|
|
|
|
|
```
|
|
|
|
Product A: Small coffee + low intensity
|
|
|
|
Product B: Small coffee + high intensity
|
|
|
|
Product A: Single coffee + low intensity
|
|
|
|
Product B: Single coffee + high intensity
|
|
|
|
Product C: Double coffee + low intensity
|
|
|
|
Product D: Double coffee + high intensity
|
|
|
|
```
|
|
|
|
|
|
|
|
Additional Sensors:
|
|
|
|
Synchronisation:
|
|
|
|
|
|
|
|
As you can see, we have another data source in addition to the power box. Now the synchronisation of this is crucial. We have set up a so-called NTP server service on our laptop server. The power box sends a request every 60 seconds and obtains a new time stamp. This should also be done by the coffee machine. This means that synchronisation in this project is implemented via a central server service.
|
|
|
|
|
|
|
|
Wie sie sehen haben wir neben der Strom-Box eine weitere Datenquelle. Nun ist die Synchronisation dieser entscheidend. Auf unserem Laptop-Server haben wir einen sogenannten NTP-Server-Service eingerichtet. Die Strombox sendet alle 60s eine Anfrage und zieht sich einen neune Zeitstempel. Dies sollte von der Kaffeemaschine ebenfalls erfolgen. Somit wird die Synchronisation bei diesem Projekt über einen zentralen Server-Service realisiert.
|
|
|
|
After you have hacked the coffee machine, you should receive two data sets. One is the energy history and the other is the production information.
|
|
|
|
|
|
|
|
Nachdem Sie die Kaffeemaschine gehackt haben sollten Sie zwei Datensätze erhalten. Einmal die den Energyverlauf und einmal die Produktionsinformationen.
|
|
|
|
* How can we now use this data to train an AI model?
|
|
|
|
* Which labels are necessary?
|
|
|
|
* How do we use the model?
|
|
|
|
* Remember the Kaggle course?
|
|
|
|
|
|
|
|
* Wie können wir nun diese Daten nutzen, um ein KI-Modell zu trainieren?
|
|
|
|
* Welche Labels sind notwendig?
|
|
|
|
* Wie nutzen wir das Modell?
|
|
|
|
* Erinnern Sie sich an den Kaggle Kurs?
|
|
|
|
Work out a concept of how you can realise the system. Create a schematic representation in which your concept becomes apparent.
|
|
|
|
|
|
|
|
Erarbeiten Sie ein Konzept, wie Sie das System realisieren können.
|
|
|
|
Consider in detail which steps might be necessary?
|
|
|
|
|
|
|
|
Erstellen Sie dazu eine schematische Darstellung in der ihre Konzept ersichtlich wird.
|
|
|
|
- How do I obtain two data sets from two data sources and how can I process them further to train my model?
|
|
|
|
- How do I prepare the data? Cleaning, sorting?
|
|
|
|
- Which features could be interesting for the training?
|
|
|
|
- How could I build a user interface in Node-Red to display interesting information?
|
|
|
|
- Which models would be useful for this use case?
|
|
|
|
|
|
|
|
Können Sie sich Zustände vorstellen, welche wir ebenfalls erfassen könnten? Hint: Anomalie? |
|
|
\ No newline at end of file |
|
|
|
For the practical part, we will provide a few python scripts so that even inexperienced students can successfully complete the seminar. Nevertheless, everyone should create a procedure with the individual steps of what a possible solution could look like. It's more about understanding which steps I need to take to develop a cool product and less about which function I need to call from which library. |
|
|
\ No newline at end of file |
