Measurement and analysis of regional water-energy-food nexus resilience with an improved hybrid kernel extreme learning machine model based on a dung beetle optimization algorithm

CONTEXT In traditional water-energy-food nexus (WEFN) system analysis, the impact of resilience is insufficiently considered, and the accuracy of resilience assessment is low, which can easily lead to system imbalance. Modeling measures can yield optimal evaluation results and provide decision makers with a scientific and efficient basis for management. OBJECTIVE The objective of this study was to provide a new research paradigm for exploring the sustainable development of WEFN systems by quantitatively and accurately assessing WEFN resilience and identifying its key driving factors. METHODS The connotation of WEFN resilience was established, and a human-centered conceptual framework for WEFN resilience was constructed. A WEFN resilience evaluation system was scientifically and completely constructed using qualitative and quantitative index screening methods. The hybrid kernel extreme learning machine (HKELM) model was improved based on the dung beetle optimization (DBO) algorithm to measure and analyze the WEFN resilience of the China Beidahuang Group. The CRiteria Importance Through Intercriteria Correlation (CRITIC) method was adopted to calculate the evaluation index weight of the WEFN system of the Beidahuang Group. Model performance verification methods, such as application of the theory of serial number summation, were used to evaluate the performance of the DBO-HKELM model. RESULTS AND CONCLUSIONS The results showed that over time, the Beidahuang Group's WEFN resilience exhibited a fluctuating upward trend. From 2006 to 2008, WEFN resilience declined to level II because of predatory agricultural production and management practices. From 2008 to 2018, the WEFN resilience increased to level IV due to the protection of uncultivated land, the optimization of irrigation methods, and the increasing maturity of the agricultural modernization development model. Likewise, from 2018 to 2020, the WEFN resilience stabilized at level IV because of the optimization of agricultural planting structures and the reduction in coal consumption. Spatially, the WEFN resilience of various branches of the Beidahuang Group exhibited a trend of “gradually decreasing from north to south.” Differences in certain variables, such as the proportion of area irrigated with electromechanical wells, were the most important factors related to changes in WEFN resilience over time. The DBO-HKELM model has notable advantages over the FA-HKELM and NGO-HKELM models in terms of fitting accuracy, optimization rate, reliability, rationality, and robustness. SIGNIFICANCE These research results could enrich quantitative assessment methods for determining WEFN system recovery strength and provide scientific support for the safe operation and sustainable development of agricultural economies and societies.