The buildup factor calculations of concrete with different proportions of CRT based on a BP neural network by MCNP

Cathode ray tubes (CRTs) are glass materials that contain harmful heavy metal elements such as lead . CRTs have been widely used in electronic products sets for decades. At present, these discarded CRTs endanger the environment. Because the high-density heavy metal elements in CRTs have a good shielding effect for X/γ rays, these raw materials have become a good choice to produce radiation-proof concrete. In the selection and use of shielding materials, the buildup factor must be considered. Monte Carlo code MCNP is used to calculate the exposure buildup factor (EBF) of concrete with different proportions of CRT fragments in the photon energy range of 0.015–15 MeV for shielding thicknesses of up to 20 mean free paths (mfp). Back-propagation (BP) neural network is proposed to predict the EBF, and the prediction effect is evaluated, reliability of this method is verified; the average error is 3.4%, and the maximum error is 9.1%. Using this method, the EBF of concrete with an arbitrary proportion of CRTs, any energy level and any shielding thickness can be quickly obtained in the ranges of the parameters of the neural network training set (0.015–15 MeV and 0–20 mfp in this paper).

阴极射线管（Cathode ray tubes, CRTs）是一类含有铅等有害重金属元素的玻璃基材料。数十年来，CRTs已被广泛应用于各类电子整机产品中。当前，此类废弃阴极射线管已对环境构成威胁。由于CRTs内含的高密度重金属元素对X/γ射线具备优异的屏蔽性能，这类原料已成为制备防辐射混凝土的优质选择。在屏蔽材料的选型与应用过程中，必须考虑积累因子（buildup factor）。本研究采用蒙特卡罗程序MCNP，在光子能量范围0.015~15 MeV、屏蔽厚度最大达20个平均自由程（mean free paths, mfp）的条件下，计算了掺加不同比例CRTs碎料的混凝土的照射量积累因子（exposure buildup factor, EBF）。本研究提出采用反向传播（Back-propagation, BP）神经网络对EBF进行预测，并对预测效果开展评估，验证了该方法的可靠性：其平均误差为3.4%，最大误差为9.1%。借助该方法，可在本研究神经网络训练集的参数范围内（光子能量0.015~15 MeV、屏蔽厚度0~20 mfp），快速获取任意CRTs掺加比例、任意光子能量及任意屏蔽厚度下混凝土的EBF数值。