Variable renewable energy (VRE) integrated via non-synchronous inverters exhibits low inertia and fluctuating output, posing substantial frequency security challenges for future power systems. When frequency security constraints are omitted from generation planning, the resulting low-inertia generation portfolios often fail to meet critical frequency requirements. To address this issue, this paper proposes a novel frequency security constrained generation planning (FSCGP) model that leverages the frequency support potential of diverse power sources, including conventional thermal generators (CTGs), VRE units, concentrating solar power (CSP) units, and energy storage systems (ESSs). A physics-data hybrid-driven method is introduced to formulate frequency security constraints, enabling accurate representation of diverse frequency regulation characteristics, particularly the fast frequency support capabilities of inverter-based generators (IBGs). To further enhance the computational efficiency, several acceleration techniques are incorporated into the proposed FSCGP model. Case studies based on a modified IEEE RTS-79 system validate the effectiveness of the proposed FSCGP model. The numerical results identify the primary contributors to frequency security under different renewable energy penetration (REP) levels and demonstrate the cost-effectiveness of coordinating various frequency support sources, especially CSP units and IBGs, in mitigating challenges in low-inertia grids.