Study On The Softener Of The Glassfiber Binder And Optimization Based On Neural Network

Sound absorption performance of the glass-fiber mat is directly related to its flexible adhesive,domestic glass-fiber mat is very hard, and the sound absorption performance is very poor. It has agreat gap between our glass-fiber mat and the United States products. This paper based on referencesand previous experience, through a lot of experiments on modified phenolic resin binder, found thekey parameters affected the softness of the glass-fiber mat, select out the best formulation and process.Use the experimental data train the neural network, and then select the best process parameters: theamount of SS is30%, the amount of Wacker silicone is6%and the curing temperature is235℃. Withthis parameters to product the binder, and then spaying it from outside mixing sprinkler head to theglass-fiber, curing the mat to dry.According to the Japanese Industrial Standard JIS L1096:2010and BMS8-48standard, test thesoftness of the glass-fiber with hardness tester, test the hydrophobic properties of the glass-fiber withthe Static contact angle tester. Test results show that the softness of the glass-fiber reaches50mm,softening performance improved83%as compared with domestic products; contact angle is156°, theabsorption of per gram mat is not greater than1.4g, and it has a very good hydrophobic properties.The innovation of this paper is:(1) by adding SS and Wacker silicone into PF resin, collaboratemodified to get excellent glass-fiber mat.(2)by reduce the curing temperature and pressure to controlthe curing time, and increase the circulating hot air, to improve the performance of glass-fiber mat.(3)by design the neural networks, and train it using the factors of glass-fiber mat’s performance,forecast the best parameters, and get the parameters without experiment.