基于BP网络的粉煤灰混凝土强度计算

Calculation of strength of fly-ash concrete based on BP network

为计算粉煤灰混凝土在一定条件下的强度,基于改进的BP算法,建立了6个粉煤灰混凝土28 d抗压强度BP网络训练及计算模型.模型均以水灰比为输入,分别以粉煤灰等量取代水泥5%卵石混凝土28 d抗压强度、粉煤灰等量取代水泥15%卵石混凝土28 d抗压强度、粉煤灰等量取代水泥25%卵石混凝土28 d抗压强度、粉煤等量取代水泥5%碎石混凝土28 d抗压强度、粉煤灰等量取代水泥15%碎石混凝土28 d抗压强度及粉煤灰等量取代水泥25%碎石混凝土28 d抗压强度为输出.该6个模型的计算结果为:模型1相对误差平均值为3.371%;模型2相对误差平均值为4.415%;模型3相对误差平均值为3.483%;模型4相对误差平均值为4.743%;模型5相对误差平均值为3.346%;模型6相对误差平均值为5.317%.由此可见,所建立的BP网络模型,训练及计算结果较为理想.

In order to calculate the strength of fly-ash concretes under controlled conditions, six optimized BP models were established to calculate compression strength （28d） of fly-ash concrete. All the input of networks were the W/C ratio. And the output of networks was the compression strength of cobble concrete（equal-substitution of fly ash for cement 5% , 15% and 25% ） , and the compression strength of broken stone concrete（ equal - substitution of fly ash for cement 5% , 15% and 25% ）, respectively. The computional results of these six models about the mean value of the relative error are shown as follows in order： 3. 371%,4. 415%,3. 483% ,4. 743% ,3. 346% and 5. 317%. Thus the results of training and calculation of the networks are proved to be very good.