Learning device and inference device for state of air conditioning system

The learning device performs preliminary learning to create environment-specific threshold values before actual abnormality detection. The system pre-processes operation data from multiple operating environments to establish customized thresholds for each environment, so that when detection is needed, the already-optimized thresholds are ready for immediate use, improving precision without adding complexity to the detection process itself

The system changes the threshold parameter based on operating environment by learning different threshold values for different environments. Instead of using a fixed threshold, the system adapts the threshold parameter to match specific operating conditions (such as different outdoor temperatures, humidity levels, or usage patterns), thereby maintaining high measurement precision across varying conditions

The learning device and inference device are integrated into the air conditioning system itself, allowing the system to automatically learn and adapt to its own operating environment without external intervention. The system collects its own operation data, performs self-learning to establish environment-specific thresholds, and uses these thresholds for self-diagnosis, thereby improving reliability while minimizing the need for external complex detection systems

The learning device serves multiple functions: it collects operation data, learns environment-specific characteristics, generates customized threshold values, and supports abnormality detection. By making the learning device multi-functional, the patent reduces the need for separate dedicated components for each function, thereby improving reliability through comprehensive environmental adaptation while controlling overall device complexity