Silica Composite Material At High Temperature Electromagnetic Properties Of The Neural Network Model

Quartz fibre-reinforced silicon dioxide ceramic (SiO₂/ SiO₂) composites have long been the primary ceramic material used for hypersonic missile radomes, owing to its outstanding thermal shock resistance, low and stable dielectric properties and its ease of fabrication. However, when Antenna window have both high surface temperatures (even ablation) and large temperature gradients in most reentry applications, it may produce unwanted signal attenuation. The magnitude of the attenuation could be calculated if the dielectric properties are known up to the ablation temperature. Unfortunately, such data was previously limited to about 1200℃. And in many cases, the change in the properties of composites with temperature variations cannot be described using a simple formula. Thus, it is crucial for the antenna window designer to extrapolate the high-temperature dielectric property and evolutionary behavior with a good model.This paper describes three key contributions of our research. First, the author performs data modeling and data presentation for experimental data of high-temperature dielectric property of ceramic matrix composites. It can provide decision support for the users of ceramic matrix composites and lay the root for the further research. Second, the author builds a RBF neural network model to simulate the high-temperature dielectric property of ceramic matrix composites. The result of simulation experiment with MATLAB indicates that the simulated result of loss tangent is more precise then BP neural network model. Third, a GA-RBF neural network model is advanced for simulating and forecasting the dielectric constant and loss tangent of ceramic matrix composites. The simulation experiment with MATLAB gets a satisfactory result. And the new model is excellent in the comparing with BP and RBF in precision and convergence rate.