ABSTRACT

In response to the escalating tensions in the Taiwan Strait, the frequency of airspace threats by the People's Liberation Army (PLA) toward Taiwan has reached unprecedented levels. Between 2021 and 2022, Taiwan encountered over 500 PLA incursions, indicating a new level of frequency and threat intensity. This series of frontline aerial data confrontations poses a severe challenge to national security and presents a critical question for the national military: How can we identify the actions that pose the greatest threat to Taiwan's airspace security amidst this data deluge?
To address this challenge, this study introduces an innovative methodological framework, utilizing Gaussian Bayesian classification to analyze trends in PLA's incursions. By conducting a deep analysis of data captured by ground radar on PLA aircraft types and numbers, combined with Bayesian network models and prior probabilities, this research aims to predict which specific actions or models pose the greatest potential threat to Taiwan's airspace security. This approach not only reflects the strategic thinking of "more calculations lead to victory, fewer lead to defeat" from Sun Tzu's Art of War but also embodies the importance of utilizing data as a decision-support tool in the current wave of AI and big data technology.
The results of this study will provide the national military with a data-driven decision-making tool, enhancing the capability to predict and identify aerial threats, and optimizing response measures to potential incursions.

KEYWORDS: Gaussian naive Bayes classifier; C4ISR; Air Early Warning.