本篇論文假設標準模型對目標衰變的分支比預測正確,利用蒙地卡羅模擬探討了在 Belle II 實驗中尋找 B^+ → K^+νν 和 B^0 → K_S^0νν 衰變的可行性。為了辨別訊號事件,我們完整重建了 Υ(4S) → BB 事件中一顆衰變為強子的 B 介子,並限制除一 K 介子外,事件的剩餘部分中不得存在其他粒子。我們透過對抗式神經網路分類器來提升訊號背景比,並使用擬合方法來測量分支比。接著,我們利用艾西莫夫資料集和剖面概似比方法,以估算預期之訊號顯著性和分支比上限。 最後,本篇論文報告了在不同 Belle II 資料量下,所預期得到 90% 信心水準下之分支比上限與顯著性,以及在各 q^2 區間中, 90% 信心水準下之部分分支比上限。
In this thesis, we study the feasibility of searching the decays B^+ → K^+νν and B^0 → K_S^0νν at the Belle II experiment, based on the Monte-Carlo simulation data assuming that the Standard Model prediction of the branching ratios are correct. To recognize the signal events, we fully reconstruct the hadronic decay of one of the B mesons in each Υ(4S) → BB events and require that there are no other particles except one kaon in the rest of event. The signal-to-background ratios are improved by adversarial neural network classification, and then the branching fractions are measured by fitting method. The expected significance for discovery and the upper limit of branching fractions are obtained by the Asimov dataset and profile likelihood ratio method. As a result, we report the expected 90% C.L. upper limits of the branching fractions and the expected discovery significances that could be achieved with different amounts of Belle II data. The expected 90% C.L. upper limits of the partial branching ratio in different q^2 regions are also reported.