Predictive control model for variable air volume terminal valve opening based on backpropagation neural network

Abstract

To determine the valve opening for controlling air volume, the relationship among valve opening, static pressure and air volume is studied through experiment. A prediction model for valve opening based on back propagation neural network is proposed. Firstly, the appropriate number of hidden layers and neurons is determined by calculating the error of test set and training set. Secondly, the prediction effect of the model is tested under different conditions. The test result shows that the error between the predicted result and the expected result is less than 5%. Thirdly, the prediction results of the neural network model and the quadratic function model have high consistency. In contrast, neural networks have stronger nonlinear fitting capabilities and are more suitable for solving complex problems. Additionally, compared with the traditional feedback adjustment method, the approximation ability of the neural network model can be used to directly output the position of the VAV terminal demand valve, thus reducing the convergence time and stabilization time, and improving the energy-saving effect and indoor environment comfort. The proposed model can provide reference for the design and debugging of related air supply terminal products, not limited to ship VAV terminal.