A Dynamic Modeling Algorithm for Endangered Species Population Dynamics based on Dynamic Parameter Optimization and Bayesian Learning

Aiming at the problem that the fixed parameter assumption of the classical logistic model is difficult to portray the superlinear death effect and Allee effect, this paper proposes a population modeling algorithm that integrates dynamic parameter optimization and Bayesian learning. By constructing a time-varying parameter space, we establish the dual feedback correction equations of environmental capacity and growth rate, and design an adaptive sliding window mechanism to dynamically optimize the parameter sensitivity. The algorithm combines the improved Bayesian optimization module and introduces dynamic exploration factors to balance the global search and local convergence efficiency. Experiments using the Danding Crane population as an example show that the dynamic model shortens the response time lag by 53%, improves the recovery efficiency of manual intervention by 40%, and reduces the prediction error by 42% compared with the traditional method under sudden environmental changes. The framework provides a highly adaptive modeling tool for endangered species conservation, which can be extended to the dynamic management of complex ecosystems.