基于神经网络的封接玻璃热膨胀系数研究

Study on the thermal expansion coefficient of artificial neural networks used in sealing glass

采用基于COM组件的混合编程技术编写了BP网络模型预测软件,并用该软件对封接玻璃SiO2-Al2O3-PbO-R2O系统的玻璃样品进行了热膨胀系数预测。预测结果表明,模型对给定组成玻璃热膨胀系数的预测值与实际测试值的相对误差在4%以内,该方法在对耐温耐压封接玻璃的配方设计中可起到重要作用。

A predicted software of BP artificial neural network model was by created using of hybrid programming based on COM builder. The thermal expansion coefficient of glass samples which belongs to the SiO2- Al2O3-PbO-R2O glass system had been predicted by the predicted software . It is found that the relative error between the experiment value and predicted value is less than 4 %. The method have an important effect on formulation design of high-temperature-high-pressure sealing glass.