摘要
基于DPSIR框架构建新型城镇化与乡村旅游耦合评价体系,通过LSTM神经网络揭示系统演化规律。选取常住人口城镇化率、城乡收入比等12项指标表征经济、社会、生态子系统交互作用,设计三层LSTM模型捕捉时序依赖特征。模型采用6步时间窗口输入与64神经元隐藏层结构,结合SHAP值解析驱动因子贡献度。研究发现,政策引导与生态约束构成系统演进的双重杠杆,乡村旅游示范县建设通过空间重构强化要素流动,而生态保护区覆盖率则形成发展阈值边界。模型验证显示其对协调度突变点具有灵敏响应,MAE低于0.03的预测精度为城乡协同发展提供动态监测工具。
This study constructs a coupling evaluation system for new urbanization and rural tourism based on the DPSIR framework,employing LSTM neural networks to reveal system evolution patterns.Twelve indicators including urbanization rate and urban-rural income ratio are selected to characterize interactions among economic,social,and ecological subsystems.A three-layer LSTM model with 6-step time window input and 64-neuron hidden layer is designed to capture temporal dependencies.SHAP value analysis identifies policy guidance and ecological constraints as dual leverage mechanisms,where rural tourism demonstration zones enhance factor mobility through spatial restructuring,while ecological reserve coverage establishes development thresholds.Validation shows the model's sensitivity to coordination degree turning points,with MAE below 0.03 providing a precision monitoring tool for urban-rural synergy.
作者
范平
陈萍萍
FAN Ping;CHEN Pingping(Tourism College of Zhejiang China,Hangzhou 311231,China;Woosong University,Daejeon 34514,Republic of korea)
出处
《佳木斯大学学报(自然科学版)》
2025年第7期152-154,99,共4页
Journal of Jiamusi University:Natural Science Edition
基金
2024年度浙江省教育厅高职教育十四五第二批教学改革项目(jg20240104)
2022年度浙江省教育科学规划教学创新专项课题(2022JCD005)。
关键词
LSTM神经网络
新型城镇化
乡村旅游
耦合协调度
熵值法
LSTM neural network
new urbanization
rural tourism
coupling coordination degree
entropy method
