System for the energy saving pre-cooling/heating training of an air conditioner using deep reinforcement learning algorithm based on the user location, living climate condition and method thereof

摘要

The present invention relates to an air conditioner pre-cooling/pre-heating energy-saving learning system and a method of applying an in-depth reinforcement learning algorithm based on location information such as a user's commute, living conditions, and outdoor environment conditions. By learning various living conditions such as commuting location information, traffic conditions, and temperature and humidity conditions, the user learns when to properly heat and cool before reaching the office or residence, and learns how to save energy in the office or residence. It is about. In particular, in a simulated test environment, iterative learning is performed under various conditions (e.g., user location information, rate of change of location arrival due to traffic, setting temperature and humidity, temperature and humidity of the interior room, etc.) to stabilize the learning pattern within the target range. And if the energy saving pattern is shown, regardless of the air conditioner component mathematical model formula, the environmental model mathematical expression, and the air conditioner-environment interlocking mathematical model formula, whether the user is satisfied with the target range (temperature/humidity range and energy saving) Feedback), and the final learning result according to the stabilized learning pattern is stored in the neural network, and the user's location is based on the last learning result stored in the air conditioner located in the actual field. The present invention relates to an air conditioner pre-cooling and energy saving learning system and a method of applying an in-depth reinforcement learning algorithm that seeks user convenience and energy saving of an air conditioner by performing adaptive learning according to information and indoor and/or outdoor environments.