A dynamic reliability assessment framework for integrated energy systems: A new methodology to address cascading failures

• Reliability models for energy transmission and supply equipment are established based on the relationship between failure probability and load rate, considering their operational status. • Event tree analysis (ETA) is used to reason about the mechanisms of cascading failures in IES, combined with sequential MC simulation to predict failures both within and across power subsystems, leading to a comprehensive reliability assessment methodology. • A data-driven minimum load curtailment model is developed using LSSVM, trained with mechanism model results, resulting in a joint data-model-driven load curtailment method. • The proposed method's rationality, accuracy, and speed are validated through system tests, sensitivity analysis of reliability factors, and identification of system weak points. As the energy internet and integrated energy systems develop, the interconnections among different systems increase operational risks, highlighting the need for urgent reliability research. Recent large-scale blackouts, often caused by cascading failures, reveal that current reliability assessments frequently overlook dynamic equipment conditions and the risk of such failures. Traditional model-driven methods for single energy systems are becoming inadequate due to rapid operational changes. To address these challenges, this study proposes a reliability assessment method for integrated energy systems that considers equipment operational states and cascading failures. It introduces an equipment reliability model for simulating cascading failures due to equipment overloads after initial failures. A hybrid data-model driven approach is proposed to improve the efficiency of load reduction calculations. Then the reliability evaluation is realized by combining the analysis of system energy flow state and index calculation. The modified model simulates more failure events than conventional model and the reliability level reflected by the calculated index is lower than that of the conventional model assessment by 25.39 % to 179.13 %. Evaluation time is reduced by 98.10 % while maintaining an average relative error within 6 %. The subsystem reliability level increases by 69.72 % and decreases by 2.25 % depending on the coupling degree. Failures of less than 20 % of all fault types contributed 43.34 % to 69.59 % of the load reduction. In summary, this model effectively simulates cascading failures from changes in operating states and provides a rapid, accurate reflection of system reliability.Based on this method, the reliability influencing factors can be analyzed and the weak link can be identified.