NOMIA An AI is trained by KNX data

• Lighting: Switching , Dimming , Light Scenes , Timed Control , Automatic Occupancy Detection , Constant Light Control
• Heating, Ventilation & Air conditioning (HVAC): Central & Automatic Control , Automatic Occupancy Detection , Weather Dependent Control , Individual Room Control / Zone Control , CO2 / Humidity Measuring , Natural Ventilation , Boiler and burner Control
• Blind and Shutter Control: Individual Control , Group & Central Control
• Security and Safety: Smoke/Fire Detection , Faults (e.g. Leak Detection, Glass Break Detection, Window Contacts) , Access Control , Emergency Lighting , Supervision (e.g. Cameras)
• Operation and Visualisation: Switches / Push Buttons , PC Visualisation , Web Servers , Touch Panels & Display Panels , Tablets and smart phones
• Automation and Remote Access: Timed Functions , Logical Functions , System Supervision , Internet Access , Remote Programming , Messaging (e.g. SMS)
• Energy Management/Smart Metering: Metering , Data Logging , Visualisation , Peak Demand Monitoring , Load Shedding , Use of Renewable Energies
• Other: Audio Control , Irrigation Control

From smart to intelligent there is a giant leap. “Machine learning”. For this application an AI model is under continuous training targeting energy management and comfort. Room occupation is based on algorithms and machine learning not only on a presence detector. The hotel “knows” that there are guests in the rooms even if they are sleeping or resting at the balcony by combining the information from the corridor cameras, the room presence detector, the Wifi connections and the KNX traffic. The energy model predicts the energy demand and utilizes the renewable sources at the highest possible degree. Result of the efficient management is 20 suites with heated pools each, Restaurant, Bar, Kitchen, laundry, GYM, SPA, lighting and full A/C to operating with a 3X80A / 230-400V power source. To feed the AI model, KNX open API REST used and the naming of group addresses defined under certain rules. Once a day the model is deployed into the real working environment.