Research On Zeta Potential Of Glass Fiber Sus-Pensions And Neural Network Optimization Of Its Process Parameters

Glass fiber blanket made by the wet paper making technology is used as core material of VacuumInsulation Panel. The pore structure and its distribution is the key factor to reduce the thermal conduc-tivity and improve the effect of thermal insulation. But, the structure of the core material is under thecontrol of the Zeta potential of the glass fiber suspensions. Based on this, with the Zeta potential ofglass fiber suspensions as the research object, and the effect of glass fiber length, pH, the type andamount of dispersants on Zeta potential of glass fiber suspensions was investigated. Besides, usingMATLBA neural network work on glass fiber suspensions process parameters optimization, theoreti-cal simulation and forecast the best process parameters of dispersion.The results showed that firstly, with the glass fiber length reduced from38mm to10mm, theglass powder content in suspensions increased. With more anion adsorption around the glass fiber, thefiber surface charge density improved, which made the double electrode layer between the adjacentfibers expand and Zeta potential increased. Secondly, Zeta potential of glass fiber suspensions waslowest when pH value was3.5. When H+and OH-from glass fiber charge neutralization with chargefrom solution, the glass fiber suspensions reached it isopotential point at pH value was0.9. When pHvalue between1.5to3.5, the thermal conductivity of Vacuum Insulation Panel was lowest at2.322mW/(m K). Thirdly, the addition of HEC and HPMC made the Zeta potential of glass fibersuspensions decreased, and while the Zeta potential increased when the suspensions added with SHP.The thermal conductivity of Vacuum Insulation Panel made by glass fiber suspensions added SHP isbetter than that of the glass fiber suspension added HPMC and HEC.MATLBA neural network was used to optimize suspensions process parameters, neural networkmodels of3-4-1structural dispersant-Zeta potential, neural network models of1-3-2structuralPH-Zeta potential and thermal conductivity of VIP were build separately. After simulation tests on theobtained datas from the neural network,different Zeta potential correspondent to different content ofdispersants were obtained; Thus we indicate that the lowest thermal conductivity of the VIP is2.3155mW/(m K) when pH value is1.5, and the error is only0.28%.