MCU IIoT with InductiveAutomation Ignition & AI
E Building a gingerbread house over the winter holiday is a tradition because we also build memories. Last year we put in tiny strings of Christmas lights and lamps next to the dollhouse furniture from the year before.
When the whole thing collapses, we all shout "Opa!" and laugh.
This year, join our upgrade to electronics -- open the door with an RFID card, turn the fans on automatically based on temperature, and other experiments our wives won't let us do to our real house?
Why don't we do this on our real house?
Learning this would enable us to get jobs at real industrial plants, troubleshooting ICS (Industrial Control Systems) in OT (Operational Technologies).
But for those who can't even get a paper cut in our job, it's fun to explore hands-on.
Real industrial plants (rightfully) segment what each person can access and do. The effort here gives us unfettered access to software and hardware so we can test their security and our enhancements.
With this project, we work as friends rather than adversaries who undermine each other in the workplace. For example, when asked a simple question, co-workers may say in front of others "if you don't know that, you shouldn't be here".
Videos
We will be recording a video of us building this demo system. But we'll use AI-generated faces and voices to avoid being undermined.
There are too many videos of people bragging but not many step-by-step of how they got there. Mistakes and struggles are fun.
Most importantly, information about how to integrate systems are rare because individual vendors typically focus on their own product.
OEE (Overall Equipment Effectiveness)
Our effort makes use of "lean" manufacturing and "5s" strategies along with OEE (Overall Equipment Effectiveness) techniques to track and assess the availability, productivity, and quality of our building efforts. That's because we want to use this opportunity to experience approaches while we can still freely laugh about them and see the good and bad when we use them. For example, does it really make sense for us to sort the many tiny components in labeled trays (like we do with eating utensils) before starting the build?
(1) Demo Home Electronics:
To reduce frustration of wasting time shopping for parts that don't work together, we're assembling a kit based on instructions at https://github.com/keyestudio/KS5009-Keyestudio-Smart-Home-Kit-for-ESP32
Video Description: https://www.youtube.com/watch?v=LBHK77J43Ak
Glowing user review of the house: https://www.youtube.com/watch?v=ueVqSmoEvmw
(2) Visualization:
The kits use an ESP32 micro-controller which has built-in networking, so it can send sensor telemetry to a central server for analysis and visualization. We're installing a local stand-alone InductiveAutomation Ignition server that the "Big Boys" use for SCADA control of industrial PLCs.
This step ensures that we have the skill to customize a dashboard that make trends visible in the metrics we will collect.
(3) Hardening:
We look at this project as an opportunity to analyze the security posture of electronics interacting with control systems. We aim to provide a fun way to demonstrate the control system which we have local unfettered access to explore implementing mechanisms to "harden" security
There is great concern about insecure IIoT devices, particularly by critical infrastructure utilities such as water treatment, hospitals, oil & gas, etc. Electronics manufactured in China (such as DJI drone, shipyard cranes, picture frames, etc.) are known to "phone home" data to China. ESP32 Espressif us based in China.
NOTE: Security actions are recommended by https://www.cisa.gov/sites/default/files/2024-05/defending-ot-operations-against-ongoing-pro-russia-hacktivist-activity-508c.pdf, NIST-CSF, NIST-SP800-82, ISA99 (IEC-62443), NERC CIP, API RP 1167 Alarm Management.
(4) Demo Farm Electronics:
The $52 farm kit provides 48 parts that include temperature and humidity sensors as well as detectors for light, water level, steam, soil humidity.
In a classroom, I would use the opportunity for students to find a way to build collaborative yet quickly.
A 126 page tutorial pdf is provided at a Dropbox folder. A web wiki page is also available. But one reviewer mentioned a lack of documentation in the Amazon reviews.
Glowing user video review of the kit: https://www.youtube.com/watch?v=K5SLkFsjFpE
领英推荐
See a build video at: https://www.youtube.com/watch?v=n1E-fgYWm40
(5) Closed-loop Process Controls:
Our other objective is to use sensors and actuators to automatically regulate temperature, water, and other levels. Perhaps open the window when the outside temp is more desired than indoor temps, but only if humidity is also higher in the winter and lower in the summer. Instead of videos, we want a real-world experience of "auto tune" feedback process controls. If we have time, we would conduct experiments to compare "PID" versus other process control algorithms such as MPT (Model Predictive
Being able to smoothly maintain processes would enable us to be more creative at controlling industrial equipment that make our lives comfortable today.
(6) Weather Data Variation Analysis:
Most weather maps report just the numbers. Using an industrial-strength visualization product would enable us to take wiser actions, such as closing doors based on declining atmosphere pressure that presage a storm at our location, before it actually rains.
Do you know whether your specific location is in a micro-climate that's typically few degrees hotter or colder than what's reported? We want to identify the pattern of deviation between our local station and what's reported by several weather stations, such as
"Today's temperature is forecast to be MUCH WARMER than yesterday."
We hope to use metrics on a comparative dashboard such as at: https://openweathermap.org/accuracy-and-quality which shows this:
Being able to do such analysis would enable us to better measure deviations in other sets of metrics. https://inductiveautomation.com/exchange/2674/overview
(7) AI Analysis:
We will explore AI and Machine Learning services to create models that more intelligently control our little house and farm.
(8) AI Integration:
If the system can integrate to the calendar of people in the home, the system can pre-warm or pre-cool the house when people are scheduled to be actually in the house (according to their calendars) rather than a general pattern like Nest does.
Being able to take more user-centric metrics into consideration before taking action enables us to take wiser action on any other process we want to control.
(9) AI-created Robot programs:
AI analytical can, based on data, create and download a program to control a roving robot to, as much as our current technology allow, autonomously navigate to a location to upload video and other telemetry.
Being able to control our simple cheap robot would prepare us to more efficiently work on "sophisticated" robots.
Join us!
We'll be doing all the above while playing Terminator and Christmas movies ("Home Alone") in the background.
Our little way of saving the planet, one fake gingerbread house at a time.
Maybe next year we'll add voice commands and facial recognition of Santa.
We'll be taking notes about our learnings applicable to the real world, such as the math and programming.
Our notes on PID (Proportion Integral Derivative): https://bomonike.github.io/pid
Our notes on InductiveAutomation: https://bomonike.github.io/plc
And all that, with security against unwanted intrusions :)
Working with small business to understand how their website works and works with their business
2 个月Great project! It's a fun way to illustrate that not all API calls are to a database.