学术论文

      基于GA-BP算法的混凝土抗压强度指标筛选

      Selection of concrete compressive strength index based on GA-BP algorithm

      摘要:
      利用遗传算法GA-BP神经网络筛选出最能表征混凝土抗压强度的因素指标,并以选取的因素指标作为输入变量,以抗压强度为输出变量,创建支持向量机(SVM)回归模型,克服了冗余因素指标对模型精度的影响.实例分析表明,采用筛选后的指标创建的SVM模型具有较高的精确度,对抗压强度的预测效果优于未经筛选的指标创建的SVM模型,简化了混凝土抗压强度评定的过程,只需考虑少量因素指标就能完成对抗压强度的评定.
      Abstract:
      Genetic algorithm GA-BP neural network was used to select the factors that best charac-terize the compressive strength of concrete. The SVM regression model was established by using the se-lected factor as input variable and compressive strength as output variables,to overcomes the influence of redundant factors on model accuracy. The example shows that the SVM model created by the selected in-dex has high accuracy, and the prediction effect of compressive strength is better than that SVM model without index selection, which simplifies the process of compressive strength evaluation of concrete, only need to consider a few factors to complete the assessment of the compressive strength.
      作者: 王江荣 白保琦
      Author: WANG Jiangrong BAI Baoqi
      作者单位: 兰州石化职业技术学院 信息处理与控制工程学院,兰州,730060
      年,卷(期): 2017, (6)
      分类号: TP18
      在线出版日期: 2018年1月4日
      基金项目: 兰州市科学技术局计划项目,兰州石化职业技术学院科技资助项目,甘肃省科技厅计划项目,甘肃省财政厅专项资金立项资助