Adaptive Hybrid Prediction-Correction with Trust-Region and Dynamic Line Search for AC/DC Load Flow

Hybrid AC/DC power systems are gaining importance due to the rising penetration of Distributed Energy Resources (DERs) and the widespread deployment of Voltage Source Converters (VSCs). Analyzing power flow in such systems remains challenging due to unbalanced operating conditions, strong nonlinearities, and strict operational limits. This paper presents an Adaptive Hybrid Prediction-Correction algorithm with Trust-Region and Dynamic Line Search (AHPC-TRDLS) for reliable and efficient load flow computation in large-scale hybrid systems. The framework couples a Newton-Raphson-based prediction stage with a Preconditioned Conjugate Gradient (PCG) correction, bounded adaptively by a trust-region radius. A dynamic line search mechanism is incorporated as a fallback whenever the PCG convergence stalls, ensuring numerical stability. Additional features include adaptive preconditioning, enforcement of voltage and reactive power limits, and the use of symmetrical components for unbalanced system modeling. The algorithm is implemented in MATLAB and evaluated on different hybrid AC/DC test systems of varying scales. Case studies confirm that the proposed approach offers improved convergence, robustness, and scalability under stressed, ill-conditioned, and unbalanced scenarios.