Frequency Constrained Coordinated Planning for Multi-type Inertia Resources with Deep Learning Assisted Linearization

With the increasing integration of renewable energy sources (RES) into power systems, the inertia declines and exhibits significant spatial distribution characteristics. To maintain the frequency stability and system economic operation, the optimal allocation of multi-type inertia resources poses a critical challenge. This paper proposes a frequency constrained coordinated planning strategy for multi-type inertia resources with deep learning assisted linearization. Frequency support capability for wind turbines (WTs), photovoltaic systems (PVs), and battery energy storage (BES) is established. A nodal inertia calculation method is proposed to characterize the spatial distribution of system inertia. Subsequently, we formulate a coordinated planning model for multi-type inertia resources that minimizes annual investment and operation costs while incorporating typical daily operation constraints and frequency security constraints. To handle the complex frequency security constraints efficiently, a deep learning assisted linearization method is introduced. The proposed approach is validated on a modified IEEE 39-bus test system.