Abstract The estimation of the irradiation date of a significant dose on a personal dosimeter can help to track the source of an exposure enhancing existing radiation protection programs. We reconstruct the irradiation date in the case of a single high dose exposure using thin-film LiF:Mg, Ti dosimeter in the TL-DOS project. In contrast to previous investigations on this topic, we include both pre-irradiation storing and post-irradiation fading time making a transfer to routine dosimetry possible. Using artificial neural networks, the irradiation date can be reconstructed with a total uncertainty of less than four days within a monitoring interval of one month. We compare these results with a classical approach using a single constructed variable showing lowered prediction uncertainties using neural networks throughout the given time interval with an improvement of up to the factor of four for irradiations shortly after annealing.

中文翻译：

---

使用人工神经网络从热释光剂量计的辉光曲线中获得与个人剂量监测相关的信息

摘要 在个人剂量计上估计显着剂量的辐照日期有助于跟踪照射源，从而增强现有的辐射防护计划。我们在 TL-DOS 项目中使用薄膜 LiF:Mg, Ti 剂量计重建单次高剂量暴露的照射日期。与之前关于该主题的研究相比，我们包括辐照前存储和辐照后衰落时间，从而可以转换为常规剂量测定。使用人工神经网络，可以在一个月的监测间隔内以少于四天的总不确定度重建辐照日期。