Saving the Earth with #AI #IoT & #Sustainability
With buildings responsible for 27% of operational carbon emissions due to their energy usage, connecting office building occupancy to heating, ventilation, and air conditioning systems can cut that usage by as much as 30%. Zone-based climate control and motion detectors can reduce wasteful energy usage by heating or cooling buildings, floors, and spaces only when people are detected. Predicting when and where people will be within buildings with AI is even better.
IoT + AI Blueprint:
Assemble the collection of components displayed in the tables below to address the use case of reducing energy usage in buildings with smart HVAC systems. This assembly begins by selecting the appropriate device based on capabilities and the presence of constant, reliable electricity. Next, the sensor(s) needed to meet the data requirements for the use case are connected to the chosen device. Then the device’s particular power supply is dictated by the availability of electricity.
| Sensor(s) |
|---|
| Motion: Determine Occupancy by Detecting the Presence of People |
| Temperature: Detect Indoor and Outdoor Temperatures |
| Weather Forecast APIs |
| Device(s) |
|---|
| Microcontroller (MCU): Arduino, STM32, ESP32, RP2040, and Others for Low-Power Scenarios |
| Single Board Computer (SBC) with Microprocessor: Raspberry Pi, NVIDIA Jetson, Orange Pi (If Constant, Reliable Power is Available) |
| Power Options |
|---|
| Lithium-Ion Polymer (LiPo) Batteries |
| Power Over Ethernet (PoE) |
| AC/DC Power (Utility | Mains | Wall Outlet) |
Using a device SDK, on-device software is written that takes periodic readings of attached sensors and saves + sends data over appropriate networks. Depending on location, power, and bandwidth requirements, a device will have networking capabilities to send sensor data. In some cases, sensor data may be sent to a nearby Edge computing location for initial processing and filtering and then on to a Cloud computing platform. In other cases, sensor data is sent directly to an IoT platform in the Cloud.
| Embedded Software |
|---|
| Programming Languages, Frameworks, and Libraries Appropriate for the Device’s Processor, Storage, and Memory are Used to Bring the Device to Life. |
| The Chosen Programming Language Interacts with the Device SDK to Take a Sensor Reading and Capture Data from the Attached Sensor(s) and then Serialize it in JSON Format. |
| Using the Device SDK, the JSON Data is Either Batched-Up and Saved to the Local Filesystem and/or Securely Sent to the IoT Platform’s Edge or Cloud Location Along with a Unique Identifier, Access Token, Timestamp, and Optional Latitude and Longitude Values. |
| Network(s) |
|---|
| Ethernet: If Present Within a Given Space for High Bandwidth Scenarios |
| Wi-Fi: If Available for Indoor, High Bandwidth Scenarios |
| Bluetooth: For Short Range, Low Bandwidth Scenarios |
| LoRa: Create Low Bandwidth Coverage Needed to Reach Internet Access |
| Cellular: If Coverage Available and Cost-Effective for High Bandwidth Scenarios |
| Data Processing + Storage Location(s) |
|---|
| Edge: Resides Inside the Building to Capture Data and Execute Symbolic AI |
| Cloud: Captures Filtered Data Relayed from Multiple Edge Locations to Monitor More Than One Building or Property and Execute Predictive AI |
The messages containing sensor data sent from devices that flow through networks and are processed at Edge and Cloud locations must be secured throughout every step of their journey. Digital Twins that model the physical objects and processes for a particular use case are created to digitally represent the physical twins. People are critical to creation, deployment, and ongoing operations of an IoT systems and the outcomes it derives.
| Security |
|---|
| Uniquely Identify Each Device |
| TLS Encryption for Data in Transit |
| Encrypt Data at Rest |
| Validate Device Messages to Ensure they use Expected Data Format |
| Rotate Access Tokens |
| Limit IP Address Ranges |
| Detect Message Anomalies |
| Zero Trust: Reauthenticate Device Messages Through Every Step of the System |
| Digital Twin Modeling |
|---|
| Buildings |
| Floors within Buildings |
| Spaces withing Floors (Offices, Conference Rooms, Common Areas, Hallways) |
| HVAC System |
| People |
|---|
| Deploy and Maintain Solution |
| Subject Matter Experts (SMEs) Define Symbolic AI (Setpoints, KPI Ranges, Thresholds) and Automations to Perform |
| Facilities Management Personnel and HVAC Technicians Provide Systems Expertise |
| AI Engineers Create, Train, and Fine-Tune Predictive AI Models |
The creation of Symbolic AI that processes streaming data (without learning) is important for the real time operations of the system including data filtering, alarms, and automation. Predictive AI models based on the physical twins being monitored predicts future event occurrences enabling the system to perform proactive automations. The IoT Platform is an advanced type of middleware that connects devices, collects device data, stores the data, analyzes data with various types of AI, and takes actions in the form of automation with integrated systems.
| Symbolic AI |
|---|
| Pattern Match Incoming Sensor Values and Compare to Defined Setpoints, KPI Value Ranges and Thresholds to Classify Data and Execute Automation |
| Pattern Match Incoming Sensor Values and Compare to Previous Values to Filter out Duplicate Sensor Readings |
| Predictive AI |
|---|
| AI Time Series Forecasting Model Predicts When People are Most Likely to be in Buildings, Floors, or Specific Spaces |
| IoT Platform |
|---|
| IoT Platform Captures, Authenticates, and Saves Incoming Device + Sensor Data to a Message Queue |
| Background Process Takes Queued Data and Hydrates Digital Twin by Saving it to a Database Table that Mimics the Structure of the Physical Twin |
| AI Model is Trained, Fine-Tuned, and Retrained with Digital Twin Dataset of Current and Historical Data Where Properties of Twin are Mapped to AI Features |
| Hot Path Data is Sent to Symbolic AI to Facilitate Real Time Data Classification and Automation |
| Hot Path Data is Also Sent to Predictive AI Model for Inferencing to Predict Future Occurrences |
The whole point of using IoT and AI for a particular use case is to facilitate automation that drives tangible value. Creating integrations between the IoT platform and various external systems makes the automation possible. The outcome(s) reflect the value driven by putting this system in place.
| Automation |
|---|
| Green KPI: People are Fully Present Throughout the Building so Turn On HVAC for Normal Operations |
| Yellow KPI: Turn HVAC on to Comfortable Levels in Isolated Areas of the Building Where People are Present. |
| Red KPI: No People Present in the Building so Turn HVAC Off or to Minimum Levels to Avoid Heating or Cooling Empty Spaces. |
| Predictive: Turn the HVAC System On or Off Either Throughout the Building or in Specific Spaces Based on AI Occupancy Predictions. Coordinate with Real Time Symbolic AI to Verify Accuracy and Change as Needed. |
| Integration(s) |
|---|
| In Order to Make the Automations Possible, the IoT Platform must be Integrated with the Building HVAC System via BACnet and/or Modbus or Through a Building Management System. |
| In Scenarios Where Direct Integration is not Possible, the IoT Platform Must Notify Appropriate Personnel to Manually Make Changes to the HVAC System. |
| Outcome(s) |
|---|
| Reduced HVAC Usage Results in Reduced Energy Consumption and Therefore Reduced Greenhouse Gas Emissions. |
| 30% Less Energy Usage Results in Significant Cost Savings Which Makes Buildings More Profitable for Building Owners. |
Related
Discover more from Rob Tiffany
Subscribe to get the latest posts sent to your email.