Optimization design of panels made by crop straw based on genetic neural network

With the treatment of alkalescency, the crops rice straw particleboards were produced by the way of hot pressure. Effects of different addition amounts of isocyanate (MDI), urea formaldehyde resin (UF) and the fire retardant of FRW (fire retardant of wood) on the mechanical properties and flame-retardant performance were explored. In this study, the orthogonal experimental design and neural network were employed to construct the mapping model between the performances and the techniques. The genetic algorithm was adopted to optimize the weight and threshold of the model. The optimum design of the technique parameters were determined by the trained model and the given performances of the materials. With the proof generalization test, the optimized technological parameters of MDI, UF and FRW were 1.926%, 2.40% and 15.381%. The properties of rice straw particleboards hot-pressed by the optimized techniques were analyzed. Compared with errors of the non-optimization model, the error of modulus and rupture (MOR), Internal bond strength (IB) and heat release rate (HRR) was 11.7%, 20% and 8%, respectively. And after optimization by GA, the errors of performances were decreased 35.3%, 17.5% and 39%, respectively.