Frequency-Variable Air Conditioner Controller Using Neural Network

It is common to control the frequency-variable air conditioner by using PED controller. However, the paper designs an algorithm for frequency-variable air conditioner by using neural networks. The neural network frame has the following characteristics: three-layer frame, two inputs, one output, while the hidden middle layer is designed with five nodes, input layer with linear signal, output layer with linear signal output. The neural network remembers the compressor track and running principle. The trained sample is used to control online. Experiments show that the scheme is capable of adapting to system variation. Even in poor condition the system can obtain the stable condition with current of 6.7A.