Large renewable penetration has been witnessed in power systems

resulting in reduced level of system inertia and increasing requirements for frequency response services. There have been plenty of studies developing frequency-constrained operation models for power system security. However

most existing literature only focuses on operational level rather than planning level. To fill this gap

this paper proposes a novel planning model for the optimal sizing problem of integrated power and gas systems

capturing both under and over frequency security requirements. A detailed unit commitment setup considering different ramping rates is incorporated into the planning model to accurately represent the scheduling behavior of each individual generator and accurate inertia calculation. The power importing and exporting behaviors of interconnectors are considered

which can influence the largest loss of generation and demand

accounting for under and over frequency security

respectively. Additionally

a deep learning-based clustering method featured by concurrent and integrated learning is introduced in the planning model to effectively generate representative days. Case studies have been conducted on a coupled 6-bus power and 7-node gas system as well as a 14-bus power and 14-node gas system to verify the effectiveness of the proposed planning model in accurate clustering performance and realistic investment decision making.