Abstract High-speed trains, as critical components of public transportation, have seen remarkable advancements globally in recent years. The evolution of communication technology has played a pivotal role in enhancing train performance and management efficiency, yet it has also rendered high-speed train systems increasingly susceptible to severe cybersecurity threats due to the widespread adoption of wireless communications, and Ethernet connectivity, among others. Consequently, railway departments and research institutions worldwide have devoted substantial efforts to tackle these issues. This paper delves into existing standards, guidelines, frameworks, and technologies, and, in conjunction with an analysis of high-speed train network architectures and potential vulnerabilities, devises a comprehensive security protection framework tailored to the characteristics of high-speed train networks. This framework aims to establish a robust defense system and implement resilient protective measures that effectively mitigate the impact of cyberattacks on overall operational functionality. To begin with, from the standpoint of preventing intrusion, it describes end-to-end secure communication mechanisms and interface designs encompassing both ground-train and intra-train communications. Subsequently, to minimize the effect of attacks on system functionality, it introduces a risk-mitigation-based attack tolerance architecture, and resilience recovery strategies tailored to the dynamic operation of high-speed trains. Lastly, by considering the future trends in high-speed train development, the paper explores emerging challenges, thereby guiding future research endeavors.

Federated Learning enables multiple clients to collaboratively train a model without sharing their individual data, thereby protecting local data privacy. However, attackers, such as untrusted servers, can still compromise the privacy of clients’ local training data through various inference attacks. One feasible approach to protect client privacy during training is the incorporation of differential privacy. Nevertheless, achieving an ideal level of privacy protection with differential privacy often degrades the model’s performance, significantly reducing its accuracy. To enhance model accuracy while minimizing the additional client heterogeneity introduced by differential privacy, this paper proposes a method that integrates personalized layers and data transformations, FusTP-FL. The core of our FusTP-FL is the incorporation of personalized layers and personalized data transformations within the client’s local training model, which further reduces client heterogeneity and improves model accuracy. We evaluated the model’s accuracy on six common datasets; experimental results demonstrate that the proposed FusTP-FL effectively enhances model accuracy across two different differential privacy modes (CDP and LDP), increasing it by up to 45%. Furthermore, we show that compared to PRIVATEFL, our method achieves lower client heterogeneity.

Summary Microperforated board is one of the important ways to absorb noise. However, the mechanism is still not well understood of the interaction between micropores, and the influence of the distance between the micropores on the sound absorption performance of microperforated plates is still not clarified. Professor Merlin introduced the influence of micropore interaction on the sound absorption performance of microperforated plates. Based on Maa and Melling theory, this paper proposed an absorption coefficient that considers the effect of perforation interactions. In addition, the formula was used to calculate the absorption coefficient and the resonance frequency and, subsequently, was tested by the conducted experiments. The results show that the structure parameters of the micro‐perforated panel affect its absorption coefficient, and the model can search the absorption coefficient more efficiently, with the calculated result close to the experimental value.