Application of genetic algorithm and artificial neural network on performance prediction of wood-plastics composite material

作者：Sun Jianping^*; Wang Fenghu; Li Peng; Cai Zhi

来源：Polymer Materials Science and Engineering, 2012, 28(1): 117-120.

• ANN
• Internal bonds
• Mechanical performance
• Modulus of rupture
• Performance prediction
• Technological parameters
• Thickness swelling
• Wood-plastics composite
• WPC

Genetic algorithm (GA) and artificial neural network (ANN) were employed to study on performance prediction of wood-plastics composite (WPC) material. Firstly, ANN was adopted to build the relationship between mechanical performance of WPC, which was internal bond strength (IB), modulus of rupture (MOR), modulus of elasticity (MOE) and thickness swelling (TS), and main technological parameters, which were hot-pression time (T), the recycled polypropylen (PP) and maleic anhydride (MA). Secondly, the weight and threshold of ANN model were optimized by GA. Finnally, mechanical performance of WPC was predicted by the optimized ANN models; and the proof generalization test of ANN was done. The results show the errors of MOE predicted by GA optimized model are 2%~15.5%, 9%~38% and 4%~70% respectively, which is less than those of non-optimized model. There is the same conclusion as for IB, MOR and TS.

• 出版日期2012
• 单位东北林业大学