ISHS Acta Horticulturae 440: International Symposium on Plant Production in Closed Ecosystems
NEURAL NETWORK INVERSE ANALYSIS FOR PLANT THERMAL BEHAVIOR |
| Authors: | . , H. Murase, N. Honami |
| Keywords: | culture vessel, finite element, heat equivalent mass, inverse technique, neural network |
| Abstract:
Development of finite element model to analyze such heat transfer problems can be helpful to study thermal system involving plants. However, the plant configuration is very complicated. Heat equivalent mass was introduced to represent a plant presence in the finite element model of the thermal system. The heat equivalent mass of plant can generate or absorb heat in the numerical system. Thermal characteristics of the heat equivalent mass were made similar to those of plants. Parameters that identify the thermal characteristics of the heat equivalent mass representing a plant are configuration of heat equivalent mass, thermal conductivity and amount of generated heat. The parameters of the heat equivalent mass were determined by a neural network-based finite element inverse technique. The inverse technique consisted of a finite element method and a neural network. Training data for the neural network inverse technique was generated by a finite element model. After the neural network training was completed, it became the inverse of the finite element model. The configuration of heat equivalent mass, thermal conductivity and generated heat from the neural network inverse technique agree with inspection data with the correlation coefficient of 0.9955, 0.8853 and 0.9777, respectively. This shows that the inverse technique can be used to determine the thermal characteristics of heat equivalent mass. | |
Download Adobe Acrobat Reader (free software to read PDF files) | |
