http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Beam Energy Determination for a 10 kW LINAC through Monte Carlo Calculation
권호제 (사)한국방사선산업학회 2020 방사선산업학회지 Vol.14 No.4
Beam energy is one of the most important parameters in assessing the performanceof an electron accelerator. Consequently, many methods to estimate the beam energy have beendeveloped. Among them, the measurement strategy employing the depth-dose distribution curveis widely utilized. In the meantime, even though, recently, the 10 kW electron accelerator installedin Advanced Radiation Technology Institute (ARTI) in South Korea had been experiencedsome repairs including the replacement of the Klystron and some components constituting theModulator of the irradiator, there had been no trial to evaluate current performance of thisaccelerator. Therefore, as a part of evaluating the performance of the electron irradiator, beamenergy level was estimated by exploiting the depth-dose distribution curves in accordance withInternational Standard ISO/ASTM 51649. From the obtained results, it was possible to realizethat there is a difference by about 1 MeV between the initially designed beam energy level andcurrent energy value.
A Study on Sampling Methods and Algorithms for Variable Inferences of Diffusion Model
이윤종,권호제,김신애 (사)한국방사선산업학회 2020 방사선산업학회지 Vol.14 No.4
Atmospheric transport and dispersion is used to predict the diffusion of radioactivematerials to the atmosphere. Several variables are required as inputs to the model for accuratepredictions, but in the case of a sudden accident, the concentration, location and time of radioactiveemissions must be inferred from monitor measurements. In the inverse modeling method ofdetermining a parameter from a monitor measurement value, sampling is used to estimate apopulation by extracting some values. In this study, to enable sampling, a function in which theoutput value x becomes data, was defined, and algorithms for this were compared. Samplingtechniques were applied to exponential functions and normal distribution functions. In addition,algorithms for the inverse CDF method, reject sampling, compression reject sampling, adaptivereject sampling, and Importance Sampling were applied through simulation. The inverse CDFmethod is the most basic method and has the disadvantage of having to find the inverse function. Forcontinuous functions, sampling was difficult near - ∞ and ∞ during the sampling process. RejectionSampling can be easily sampled using the approximation function g(x) to find the target function,but when the c value increases, it takes much time for sampling. Adaptive rejection sampling (ARS)was economical because the area to be rejected was minimized. However, the complexity of theequation increases the computational complexity. In addition, when the dimensions increase, it isvery difficult to obtain a sample. Importance sampling must also have a cover function as close aspossible to the objective function. If the objective function is large and has a value only in a localarea of the data space, poor results may be obtained.