NEURONETWORK CONTROL OF OFFICE ROOM VENTILATION SYSTEM

Abstract

The problem of air quality in enclosed spaces arises due to the operation of heating systems, the use of open fire sources, human respiration, the decomposition of synthetic materials, the ingress of dust from outside, and the presence of microorganisms. Exceeding permissible concentrations can adversely affect human well-being, making air quality control important. Traditional ventilation systems do not account for changing conditions and do not allow flexible air exchange regulation. The creation of modern automated ventilation systems using computational technologies and considering air parameter monitoring results poses a scientific challenge. Modern approaches to managing HVAC systems, which control the quality and comfort of indoor air, maintain air parameters according to sanitary standards, and allow for the creation of closed-loop control systems, have been analyzed. Control systems based on fuzzy logic theory and neural networks have been considered, which allow for effective air parameter management and ensure building energy efficiency. The aim of the article is to develop a neurocontroller for an office ventilation system that provides automated control considering the non-stationary behavior of the object by regulating the compressor motor speed of the ventilation cooling system. The proposed neural network control system uses the principle of inverse neural control for model training and ensures control by minimizing the error function. The neurocontroller is built as a multilayer neural network with sigmoid activation functions. To analyze the effectiveness of the proposed approach, a dataset in the form of time series containing measurements of temperature, humidity, and carbon dioxide concentration was used. Computational experiments with various neural network structures demonstrated the effectiveness of a network with a single hidden layer. It was established that the neurocontroller responds quickly to input signals, and the controlled ventilation system ensures the standard temperature in the room within an acceptable time and without overshoot. Comparison with P-regulator and PID-regulator indicates a significant advantage of the neurocontroller in ensuring control accuracy, response speed, and reducing energy consumption.