基于模拟退火算法的鱼类栖息地适宜性启发式优化建模

his paper presents a modeling framework called AnnHSI for fish HSI modeling and intelligent optimization. The AnnHSI model was constructed based on the projection of the space of logistic regression-based HSI to the space of simulated annealing as well as based on the objective function. The projection aims to minimize accumulative errors between the computed ground probabilities and observed probabilities converted from commercial fishing data. The proposed AnnHSI framework is composed of the construction of the problem to be solved and the annealing schedule. The validation and effectiveness of the AnnHSI framework have been proved consequently, by using simulation data, i.e. randomly generated normalized marine environmental factors and fishing ground probabilities range from 0 to 1. The research demonstrated that the AnnHSI framework is effective and efficient in retrieving and optimizing fish HSI parameters. The fish HSI parameters retrieved by the AnnHSI framework vary under different constraints, i.e. the bounds constraints on the HSI parameters. Compared with the results without constraints, those of the AnnHSI model with bounds constraints are more reasonable. In addition, the implementations of the AnnHSI modeling framework with 100, 1 000, 5 000 and 10 000 samples respectively, demonstrated a strong capability of the AnnHSI to process the mass data of fishing ground.