The fracture resistance of spent nuclear fuel (SNF) cladding containing reoriented hydrides is strongly affected by the distribution and morphology of radial hydrides. Previous studies demonstrated that fracture parameters in continuum damage mechanics (CDM) models can be calibrated using ring compression test (RCT) data by matching a single fracture indicator, typically the initial load drop point associated with crack initiation. While this approach enables accurate reproduction of the crack initiation point, it does not ensure consistency with other global structural responses observed during RCT. In the present study, the fracture parameter identification framework is extended to achieve a globally consistent reproduction of RCT behavior of SNF cladding. Four fracture-related parameters defined in the CDM model are treated as design variables, and their combined effects on RCT responses are systematically investigated. An Optimal Latin Hypercube Design (OLHD) is employed to efficiently sample the parameter space, and a series of RCT simulations is conducted to generate response data over a wide range of parameter combinations. Based on the generated dataset, Gaussian Process Regression (GPR) surrogate models are constructed to approximate the nonlinear relationships between fracture parameters and key structural responses, including the initial load drop point, peak load and corresponding displacement, and fracture energy. Using the developed surrogate models, a multi-response optimization problem is formulated to identify the fracture parameter sets that simultaneously match these experimentally observed quantities. The proposed approach moves beyond single-indicator calibration and enables fracture parameters to be identified based on the overall shape and energy characteristics of the RCT load–displacement response. By integrating design of experiments, surrogate modeling, and multi-response optimization, this study provides a more robust methodology for fracture parameter identification of SNF cladding with hydrides and improves the predictive reliability of image-based crack simulation models under pinch loading conditions.