辽宁省农田水利工程取费标准模型研究  

Modeling Study on Fee Collection Standards for Farmland Water Conservancy Projects in Liaoning Province

作　　者：韩坤 刘文合[1] 王铁良[1] 李春生[1] 郝聪明 刘顺国 HAN Kun;LIU Wen-he;WANG Tie-liang;LI Chun-sheng;HAO Cong-ming;LIU Shun-guo(College of Water Conservancy,Shenyang Agricultural University,Shenyang 110866,Liaoning Province,China;Engineering Construction Service Center of the Ministry of Agriculture and Rural Affairs,Beijing 100081,China;Liaoning Agricultural Development Service Center,Shenyang 110030,Liaoning Province,China)

机构地区：[1]沈阳农业大学水利学院,辽宁沈阳110866 [2]农业农村部工程建设服务中心,北京100081 [3]辽宁省农业发展服务中心,辽宁沈阳110030

出　　处：《节水灌溉》2025年第3期91-99,共9页Water Saving Irrigation

基　　金：辽宁省地方标准立项计划项目(2021186,2024223,2024224,2024225)。

摘　　要：国家尚未发布农田水利工程建设项目预算定额标准,在实践中各地区关于其工程取费标准值主要是借鉴水利、建筑、国土等行业依据,直接造成国家农田水利工程行业造价管理工作的混乱。为了合理确定和有效控制工程投资,科学制定人工预算单价、措施费和间接费取费标准值,该研究采用主成分分析法(PrincipalComponent Analysis,PCA)对农田水利工程人工预算单价影响因素进行筛选,并将此作为BP神经网络(Back-Propagation,BP)的输入层,同时采用蜣螂算法(Dung Beetle Optimizer,DBO)优化网络模型的权值和阈值,构建基于PCA-DBO-BP的人工预算单价预测模型;然后在考虑农田水利工程施工特点的基础上,结合模糊数学和灰色系统两种理论方法建立了农田水利工程措施费和间接费的测算模型。以辽宁省2004-2023年的数据样本进行了实例研究,结果表明:①与线性拟合、BP神经网络模型进行对比,PCA-DBO-BP模型的预测值与真实值最接近,平均相对误差只有0.74%,RMSE、MAE和决定系数R^(2)分别为1.676元、1.211元和0.978,具有更高的预测精度和泛化性。②通过灰色模糊模型确定农田水利工程措施费费率为3.90%,间接费费率为7.83%,相对误差分别为1.53%和2.02%,证明该模型准确合理、具有一定的理论和实践价值。该工程取费标准模型研究成果,完善了工程造价理论体系,为科学确定农田水利工程投资提供了科学依据和指导。The current state has not yet released the farmland water conservancy project construction project budget quota standard.In practice,regional project fee standards are primarily borrowed from those of construction,water conservancy,and land industries.This practice has directly caused confusion in the cost management of the national farmland water conservancy industry.this study employs Principal Component Analysis(PCA)to screen and downscale the influencing factors of such projects′labor budget unit prices.Additionally,the Dung Beetle Optimizer(DBO)is utilized to optimize the weights and thresholds of the BP neural network model,thereby constructing a labor budget unit price prediction model based on PCA-DBO-BP.Then,considering the construction characteristics of irrigation engineering for farmland,based on field survey and research data,we employed the relative comparison method in conjunction with three proximity measures:fuzzy proximity,Euclidean proximity,and gray correlation,to determine the comprehensive similarity.Subsequently,grounded in gray fuzzy theory,we establish a cost measurement model for both the engineering measures and overhead costs of irrigation engineering for farmland.An example study was conducted using data samples from Liaoning Province spanning from 2004 to 2023.The results indicated that:①Compared to linear fitting and BP neural network models,the predicted value of the PCA-DBO-BP model was closest to the actual value.The model evaluation indices,R^(2),RMSE and MAE,were 0.978,1.676,and 1.211,respectively,outperforming both the BP and linear fitting models,and achieving optimal values across all metrics.The screening of influencing factors,optimization of algorithms,and comparison of different models demonstrate that the PCA-DBO-BP model for predicting manual budget unit prices exhibits higher prediction accuracy and generalization ability.②Using the grey fuzzy model,the comprehensive similarity was calculated,with a maximum similarity of 0.853 and a minimum of 0.528.Based on the

关 键 词：农田水利工程 主成分分析 BP神经网络 蜣螂优化算法 灰色模糊理论 人工预算单价 措施费 间接费 

分 类 号：S277[农业科学—农业水土工程] F303[农业科学—农业工程]