Abstract
This scientific article discusses the hardware and software implementation of an intelligent distributed system for predicting and controlling optimal heat distribution in a room. The system collects data from sensors and sends it to the server for further monitoring and training of the neural network. Prediction is based on a pre-trained neural network model. The principle of operation of modern air conditioners is based on maintaining the room temperature at a given temperature. The air conditioner generates a stream of air by pre-cooling or heating it. When the temperature reaches the required value, the air conditioner turns off (goes to sleep mode), continues to take the air temperature values, and when the temperature changes, it turns on again. The system uses the results of calculating a one-dimensional heat conduction problem to correct the trained neural model and makes a decision about switching on/off and changing the temperature regime of a certain air conditioner depending on the predicted data. In these methods, only the temperature close to the air conditioner is calculated using the temperature sensor of the air conditioner itself, and this does not give us a complete picture of the temperature distribution in the room. The temperature sensor located on the air conditioner itself does not reflect the general picture of the temperature distribution and, due to the characteristics of the room, the temperature in different areas can vary greatly. Also, the operation of the air conditioner does not take into account the operation of other air conditioners, the operation of each air conditioner is autonomous and is controlled only based on the readings of the associated temperature sensor.