Data-driven planning of mixed-generation power systems: Towards 100% RES-based grids

The increasing penetration of renewable energy resources in modern power systems introduces significant stability, reliability, and operational challenges. As inverter-based resources (IBRs) replace conventional synchronous generation, maintaining stability margins becomes increasingly complex, requiring innovative planning and analysis approaches. This paper presents a data-driven framework that automates parameterization, simulation, and stability assessment for mixed-generation systems. Leveraging a ternary coordinate system, the approach enables intuitive visualization of stability trends as a function of the generation mix, supporting planners in identifying stability boundaries and optimal resource compositions. Automated eigenvalue analysis and root cause identification are combined with a Convolutional-Recurrent Neural Network (CRNN) that predicts stability directly from time-domain data, offering a scalable alternative to conventional model-based methods. Applied to modified 9-bus and 39-bus systems, and an HVDC-interconnected network, the framework demonstrates its effectiveness in capturing stability trends and defining acceptable generation mix regions, highlighting its value as a planning tool for the transition towards reliable and fully renewable grids. • Scalable tool for simulation and stability analysis of mixed-generation grids. • Ternary mapping visualizes the impact of generation mix on system stability. • Hybrid CRNN enables predictive stability assessment from time-domain data. • Results guide design of stable, renewable-dominated and HVDC-linked systems.