Parameter identification of condenser heat transfer model based on improved flower pollination algorithm

Condenser heat transfer parameters serve as critical indicators for assessing the operational status of condensers. However, these parameters cannot be directly obtained through measurement or inferred solely from condenser characteristic curves, which limits their practical applicability in performance monitoring. To overcome this limitation, we develop a parameter identification method for condenser heat transfer models based on an improved flower pollination algorithm (FPA). Specifically, a condenser heat transfer model is first formulated, and a fitness function is defined to guide the optimization process. On this basis, three enhancement strategies are introduced: a dynamic pollination probability mechanism, a chaotic initialization scheme, and an elite pollen mutation-crossover operator. These improvements are designed to address inherent shortcomings of the classical FPA, including its tendency to converge prematurely to local optima and its limited fitting performance. Comparative experiments conducted on real operational datasets demonstrate that the proposed improved FPA achieves significantly higher parameter identification accuracy and superior model fitting compared with the classical FPA.