基于贝叶斯优化的SVR模型预测福建沿海海上养殖装备灾害损失

打开文本图片集

中图分类号：X43；P7 文献标志码：A 文章编号：1005—9857（2025）07-0028-08

Predicting the Losses of Marine Aquaculture Equipment Due to Disasters Along the Fujian Coastline Using a Bayesian-Optimized SVR Model

LI Xurui，DONG Dibo，GUO Qiaoying （Instituteof Smart Marine and Engineering，Fujian Universityof Technology，Fuzhou 35ol18，China）

Abstract： To predict the economic losses caused by disasters affecting marine aquaculture equipment and to ensure the healthy development of the marine aquaculture industry， this study analyzes and forecasts the losses of marine aquaculture equipment in the coastal areas of Fujian Province.First，25 variables related to marine aquaculture equipment in the Fujian coastal region were selected as the research subjects，and the Principal Component Analysis method was used for dimensionality reduction， extracting four principal components that significantly affect the loss rate. Next，a Bayesian-optimized Support Vector Regression （SVR） model was employed to predict the disaster losses of marine aquaculture equipment. The results were compared with those of the BP Neural Network model，Random Forest algorithm，and K-Nearest Neighbors algorithm.The results indicate that the SVR model performed excelently in predicting marine aquaculture equipment disaster losses， with an average relative error as low as （20号 0.116% . The model demonstrated good accuracy and robustness in addressing the nonlinear characteristics of economic loss data related to marine aquaculture equipment， providing a reference model for predicting disaster losses of marine vulnerable assets.

Keywords： Disaster loss， Marine aquaculture equipment， Principal Component Analysis （PCA），Bayesian optimization，Support Vector Regression（SVR）

0 引言

我国作为全球最大的水产养殖国家，其产量占据了全球水产养殖总产量的 70% 以上[1-2]，且沿海养殖设施南北分布差异显著。（剩余11301字）