The current research on quantitative assessment of nuclear emergency response plans suffers from the lack of a systematic assessment framework and methodological limitations，which are manifested in the fragmentation of assessment dimensions，weak relevance of the indicators，and significant subjective cognitive bias，resulting in serious constraints on the comparability and reproducibility of the results of the assessment. The current assessment situation is difficult to meet the urgent needs of nuclear emergency rescue teams for the continuous improvement of plan quality， especially the lack of effective quantitative tools for key performance indicators，such as the timeliness of plan response and the rationality of resource allocation. To address the kennel problems of discrete index system and strength the subjectivity in the evaluation of nuclear emergency rescue teams’emergency plans，this study constructs a multi-level quantitative assessment system for emergency plan quality based on the theory of complex adaptive systems，and the fusion of heterogeneous data from multiple sources and optimisation strategy of model integration. At the level of theoretical construction，this study firstly deconstructs the professional characteristics and operation mechanism in the process of nuclear emergency plan preparation. Combined with the empirical data accumulated during the regular operation and maintenance of nuclear emergency rescue teams，a three-dimensional assessment framework was innovatively proposed. The framework systematically integrates the core assessment dimensions such as normative principles，structural integrity requirements and content coverage，forming a hierarchical assessment system oriented to continuous quality improvement. In terms of methodological innovation，this study adopted the deep coupling strategy of hierarchical analysis method (AHP) and fuzzy comprehensive evaluation method (FCE) to construct a hybrid assessment model with dynamic correction function. Through the structured indicator system design method，a stepwise evaluation system containing 6 first-level indicators (system architecture， response process，resource allocation，training and rehearsal，information management，and continuous improvement) and 23 second-level indicators (e.g.，completeness of command system，timeliness of emergency response，and rate of equipment configuration up to the standard，etc.) has been established. Among them，the AHP module completes the allocation of indicator weights by constructing a 1-9 scale judgement matrix，focusing on solving the problem of quantifying the structural relationship between multi-level indicators；while the FCE module applies the trapezoidal affiliation function to achieve the quantitative conversion of qualitative indicators，effectively reducing the bias of subjective judgement. The model validation process adopts a dual testing mechanism：the theoretical level is to test the structural validity of the indicator system through expert argumentation; the practical level is to select a national nuclear emergency rescue team to carry out empirical research. Example analyses show that the model can accurately identify the weak links of the preplanning system.The quantitative assessment system constructed in this study has gotten breakthroughs in three aspects：First，the establishment of a multi-dimensional evaluation framework of ‘standard compliance-process rationality-performance compliance. Second，the development of a hybrid AHP-FCE algorithm; and third，the innovation of a dynamic correction mechanism，so that the model can automatically optimise the indicator structure with the changes in the emergency environment. The study provides methodological support for the construction of China’s modern assessment system of nuclear emergency response capability，which has important theoretical value and practical guidance significance. Subsequent research will focus on expanding the applicability of the assessment model in the scenarios of nuclear emergency response collaboration and new reactor applications，and continue to improve the universality and accuracy of the model.