Faculty of Physics, Ho Chi Minh City University of Education, Ho Chi Minh City, Viet Nam; Faculty of Physics and Engineering Physics, University of Science, Ho Chi Minh City, Viet Nam; Vietnam National University, Ho Chi Minh City, Viet Nam.
Nuclear Technique Laboratory, University of Science, Ho Chi Minh City, Viet Nam; Vietnam National University, Ho Chi Minh City, Viet Nam.
Appl Radiat Isot. 2021 Mar;169:109570. doi: 10.1016/j.apradiso.2020.109570. Epub 2020 Dec 24.
The study presents a new ANN-based approach to determine the density of a liquid applied in the gamma transmission and gamma scattering techniques. This approach used the Monte Carlo simulation combined with an artificial intelligence technique and experimental data to estimate the density of liquids. Two advantages of the proposed approach: (1) it is able to determine the density of a liquid by only measuring the gamma spectrum (transmission spectrum or scattering spectrum) without knowing the composition of the liquid, and (2) it is able to determine the density of a liquid when it is contained in a tube of various diameters. The artificial neural network model was trained by data obtained from simulation and then was used to predict the density of seven liquids with density in the range of 0.6 g cm to 2.0 g cm for the purpose of validating the proposed approach. For the gamma transmission technique, there are 25/28 samples with relative deviations between reference and predicted densities of less than 5%. The remaining three samples have deviations in the range from 5.2% to 6.3%. For the gamma scattering technique, there are 17/21 samples with a relative deviation of less than 5%. The remaining four samples have a deviation in the range from 5.2% to 6.9%. The results proved that the artificial intelligence technique combined with Monte Carlo based on gamma transmission and gamma scattering techniques is an effective approach for estimating the density of a liquid.
本研究提出了一种新的基于 ANN 的方法，用于确定应用于伽马透射和伽马散射技术中的液体的密度。该方法结合了蒙特卡罗模拟和人工智能技术以及实验数据来估计液体的密度。所提出方法的两个优点：(1) 它能够通过仅测量伽马光谱（透射光谱或散射光谱）而无需知道液体的组成来确定液体的密度，以及 (2) 它能够在液体包含在各种直径的管中时确定液体的密度。人工神经网络模型通过从模拟中获得的数据进行训练，然后用于预测七种密度在 0.6 g cm 至 2.0 g cm 范围内的液体的密度，目的是验证所提出的方法。对于伽马透射技术，有 25/28 个样本的参考密度与预测密度之间的相对偏差小于 5%。其余三个样本的偏差在 5.2%到 6.3%之间。对于伽马散射技术，有 17/21 个样本的相对偏差小于 5%。其余四个样本的偏差在 5.2%到 6.9%之间。结果证明，基于伽马透射和伽马散射技术的结合蒙特卡罗的人工智能技术是估计液体密度的有效方法。
