Developing a Neural Network Structure in Controlling Air Conditioning Systems

The air conditioning system is a stochastic, non-stationary, multi-connected, nonlinear industrial system. Using traditional management methods for such an object do not provide the required quality of regulation, as maintaining microclimate parameters (temperature, humidity in a certain space) with high precision. To control the multi-connected output parameters of the air conditioning system, an artificial neural network (ANN) is proposed. The type of neural network is Multi-Layer Perceptron (MLP). The neural network control algorithm is based on correction of weight coefficients using the gradient descent and back propagation algorithm.