New Machine-Learning Method Helps Accurately Estimate Health Hazards of Nuclear Accidents
During nuclear accidents, the release of short-lived radionuclides poses acute health hazards to local populations. Obtaining accurate source-term information is crucial for nuclear emergency decision makers to determine appropriate protective measures. However, reliable monitoring instrument readings to estimate the source term based on core conditions, release routes, and release conditions are challenging to obtain.
To address this issue, researchers have been exploring various source-term inversion methods. In a recent study, a machine-learning method was used to estimate the release rates of four typical short-lived nuclides (Kr-88, Sr-91, Te-132, I-131) in two complex nuclear accident scenarios. The results revealed that the best estimation performance was achieved with the long short-term memory (LSTM) network.
The study found that the mean absolute percentage errors for the release rates of the four nuclides at 10 hours under the two nuclear accidents were as follows: 9.87% and 11.08% for Kr-88, 17.49% and 16.51% for Sr-91, 7.16% and 8.35% for Te-132, and 38.83% and 41.87% for I-131, respectively. Notably, Te-132 had the lowest mean absolute percentage errors among all the estimated nuclides.
Additionally, stability analysis revealed that the accuracy of the estimation was greatly influenced by the gamma dose rate. By considering these factors, the machine-learning method proved to be a valuable tool for estimating the release rates of short-lived nuclides during nuclear accidents.
This research demonstrates the potential of machine learning in providing accurate information to assist decision makers in emergency situations. The ability to estimate source terms more reliably contributes to enhanced emergency preparedness and the timely implementation of protective measures. By accurately assessing the health hazards caused by nuclear accidents, proactive steps can be taken to minimize the impact on local populations and the environment.
The findings from this study open doors to further advancements in the field of nuclear emergency management. Utilizing machine learning techniques can help mitigate the risks associated with nuclear accidents by providing critical information for decision-making processes. With ongoing research and continuous refinement of these methods, the accuracy and effectiveness of estimating health hazards related to nuclear accidents are expected to improve substantially.
This breakthrough in machine learning for estimating the health hazards of nuclear accidents brings us one step closer to better safeguarding our communities and the environment against the devastating effects of such incidents. As researchers continue to explore innovative approaches, decision makers will have more reliable tools at their disposal to protect public health and ensure effective emergency response strategies.
