Where large amounts of data are continuously generated, a comprehensive analysis of this data and a fast reaction to special events is indispensable. A decisive advantage here is the use of artificial intelligence methods.
ATHION INTELLIENCE continuously monitors the entire energy system and detects anomalies and critical system states in real time using adaptive artificial intelligence methods.
For example, if ATHION INTELLIGENCE is used to measure conspicuous energy consumption or system values, the responsible employee or control room receives a real-time alarm of critical, abnormal states in the energy system.
ATHION INTELLIGENCE thus gives you the opportunity to react quickly and purposefully and to avoid uneconomical situations.
Your added values
- Continuously improving detection of anomalies and critical states thanks to self-learning algorithms
- Cost reduction by reducing plant downtime and increased efficiency of the energy system
- No detailed mapping of technical processes necessary
- Reduction of the complexity of the monitoring process
- Efficient troubleshooting through automated provision of context-sensitive alerts
- No investments necessary thanks to a flexible software-as-a-service model
- Dynamic monitoring of properties
- Detection of load curve anomalies
- Early detection of leaks (e.g. process water or compressed air)
- Consideration of product changes and different production processes (e.g. different number of production steps or product composition)
- Inclusion of calendar events (e.g. change in district heating consumption due to holiday periods)
- Monitoring of energy flows in relation to outdoor temperature
- Context-sensitive monitoring of the efficiency of generators and consumers as a function of power (partial load, full load)
The artificial neural networks behind ATHION INTELLIGENCE are individually trained to the conditions of the property to be monitored. The use of self-learning algorithms makes it possible to continuously improve the identification of anomalies.
Artificial intelligence methods are used, that do not rely on categorized data when training, but instead recognize hidden structures and patterns in the data.
This process is supplemented by the continuous recording of current information and systematic feedback on detected anomalies by the user or the occurrence of unusual events, thus continuously improving the results.