摘要
为了项目法人在可行性研究阶段准确快速的估算出复杂的山区中小河流治理的成本且能有效的控制成本,建立了基于LM算法的BP神经网络模型对山区中小河流治理成本进行预测,并对治理成本的影响因素做出敏感性分析。结果表明,LM算法优化的BP神经网络模型比标准BP神经网络模型训练效果好,收敛快,且预测结果更精确,其预测结果可在可行性研究阶段作为一个有效的参考数据;在八个影响治理成本的因素中,护岸形式与生态措施是影响治理成本的关键,在根据预算做方案调整时可以主要针对这两个方面进行改善。
In this paper,the author,in order to make the project entities estimating management costs accurately and quickly for small rivers in complicated mountain areas,and controlling effectively the costs,established BP neural network model based on LM calculation method to predict the management costs for small rivers in mountain area,and to carry out sensitivity analysis for the effect factors of the costs.The results show that comparing with standard BP neural network model,there many advantages in optimized BP neural network model based on LM calculation method,including that training effect is better,convergence speed is faster,and prediction results are more accurate,in addition the results can be used as effect reference data in the stage of feasibility study.Among 8 factors effecting on management costs,both revetment types and ecological measures are the key factors effecting the costs,when adjusting the scheme on the basis of the budget,the improvement can be conducted mainly aiming at the two terms.
出处
《吉林水利》
2020年第5期27-31,共5页
Jilin Water Resources
关键词
山区中小河流
LM算法
神经网络
预测模型
敏感性分析
small rivers in mountain area
LM calculation method
neural network
prediction model
sensitivity analysis