遗传算法用于超远距离泵送混凝土造价控制技术分析  

Analysis of cost control technology of ultra long distance pumping concrete based on genetic algorithm

作　　者：马行耀[1] MA Xingyao(Department of Economics and Management,Zhejiang College of Construction, Hangzhou 311231, China)

机构地区：[1]浙江建设职业技术学院经管系,浙江杭州311231

出　　处：《混凝土》2018年第5期94-97,共4页Concrete

摘　　要：对超远距离泵送混凝土造价控制技术进行了研究和分析,利用遗传算法对初始阈值和权值进行了优化,所选择的研究对象为"超远泵送混凝土工程特征",工程特征涉及的内容和种类非常广泛,以建筑工程造价为基础,通过估算分析的方法研究主要因素,最终在工程特征类目当中选择了7个指标进行研究,墙体、混凝土、结构、层数、装饰、基础、平面组合,通过基于遗传网络的超远距离泵送混凝土建筑工程造价预测模型,训练大量该类型的样本,最终确定其单位立方米造价实际值和预测值,单位立方米造价预测值和实际值绝对误差最小为1.00元/m^2,最大误差为18.00元/m^2,相对误差最大为1.24%,最小为0.06%,均小于误差为5%的范畴。The cost control technology of ultra long distance pumping concrete is studied and analyzed based on genetic algorithm.The initial threshold and the weights are optimized by genetic algorithm,selection of ultra long distance pumping concrete engineering characteristics as the research object,because the project features category more,according to the construction project cost estimation,analysis of the main factors,selection of structure,foundation,concrete,floors,walls,decorative and graphic combination of the seven as the project feature category index,through the ultra long distance pump based on genetic network concrete construction project cost prediction model,the training of similar samples,decrease the cost of test samples obtained actual value and predictive value,cost prediction value and the actual value of the minimum absolute error is 1 yuan/m^2,the maximum error is 18 yuan/m^2,the maximum relative error is 1.24%,the minimum is 0.06%,less than 5% error categories.

关 键 词：造价控制 遗传算法 混凝土 单位立方米造价 

分 类 号：TU528.01[建筑科学—建筑技术科学]