Intelligent modelling to predict heat transfer coefficient of vacuum glass insulation based on thinking evolutionary neu
- PDF / 2,456,362 Bytes
- 22 Pages / 439.37 x 666.142 pts Page_size
- 50 Downloads / 192 Views
Intelligent modelling to predict heat transfer coefficient of vacuum glass insulation based on thinking evolutionary neural network Wang Lei1 · Omary Gastro1 · Yuanqi Wang2 · Nomenjanahary Homary Felicien3 · Li Hui1
© Springer Nature B.V. 2020
Abstract Vacuum glass is widely used in many construction applications, including single-family homes, as a proven energy-saving method with outstanding heat preservation characteristics. The thermal insulation performance of vacuum glass is closely related to its heat transfer coefficient. In this study, we applied neural network methods to predict the heat transfer coefficients of vacuum glass. Using MATLAB, a neural network intelligence model was established, and the traditional back-propagation neural network (BPNN) was optimised. First, a genetic algorithm was used to reduce the dimensions of the independent variable. Then, the Mind Evolutionary Computation algorithm was used to optimise the initial weight and threshold. Using the optimised BPNN intelligence model to predict the heat transfer coefficient of vacuum glass insulation, we derived an average absolute error of 0.0076. Keywords Neural network intelligence model · Back-propagation neural network · Genetic algorithm · Evolutionary algorithm · Vacuum glass insulation performance · Numerical simulation · Heat transfer coefficient Abbreviations BP Back-propagation NN Neural network IL Input layer HL Hidden layer OL Output layer
* Wang Lei [email protected] 1
College of Information and Communication Engineering, State Key Lab of Marine Resource Utilisation in South China Sea, Hainan University, Haikou 570228, China
2
Department of Industrial Engineering, University of Florence, 20133 Florence, Italy
3
College of Civil Engineering and Architecture, Hainan University, Haikou 570228, China
13
Vol.:(0123456789)
W. Lei et al.
1 Introduction The performance of vacuum glass gradually and nonlinearly decreases over time. This is concomitant with insulation performance deterioration, which also results in heat retention loss. Examples of degradation mechanisms include corrosion, cracking, and loosening of rim seals, which negatively affect thermal performance (Graves and Zarr 1997; Gross and Fricke 1997). The vacuum glass energy transfer process includes three components: input, output, and the system model. Input refers to the heat flow of the vacuum glass, and output refers to the temperature increase of a non-heated surface caused by the vacuum glass. For the overall system model, a neural grid can be used: first, the properties of vacuum glass with known performance parameters are derived, intelligent modelling is conducted through the neural grids, system characteristics are found, and the system is identified. This is termed the ‘black box’ principle: when building a model, only the system input and output data need be known. Through the relationship between the input and output, a reasonable approximation of the actual system can be obtained (Hou et al. 2004). In this study, we focus on a
Data Loading...
