基于遗传算法优化BP神经网络的沙漠砂混凝土强度预测

Strength prediction of desert sand concrete based on genetic algorithm optimized BP neural network

沙漠砂混凝土在工程建设中应用前需要做大量的试验进行验证,不仅会影响建设周期,还会消耗大量的建筑材料。针对沙漠砂混凝土强度受多种影响因素耦合作用,传统预测模型存在一定缺陷,借助全局搜索能力较强的遗传算法改进神经网络,输入层参数为水胶比、砂率、沙漠砂替代率、粉煤灰掺量、减水剂用量,建立遗传算法优化BP神经网络的沙漠砂混凝土强度预测模型。并通过实例验证,将BP神经网络预测的沙漠砂混凝土强度与遗传算法优化BP神经网络预测结果进行对比。结果表明:基于遗传算法优化BP神经网络的沙漠砂混凝土强度预测模型具有较好的操作性和预测精度,为提高沙漠砂混凝土强度预测精度开拓新的途径。

Desert sand concrete needs to do a large number of tests for verification before application in engineering construction,which not only affects the construction cycle,but also consumes a large amount of construction materials.For the desert sand concrete strength by a variety of influencing factors coupled role,the traditional prediction model has certain defects,this study with the help of global search ability of genetic algorithm to improve the neural network,the input layer parameters for the water-cement ratio,sand rate,desert sand substitution rate,the amount of fly ash mixing,the amount of water-reducing agent to establish genetic algorithm optimisation of the BP neural network of the desert sand concrete strength prediction model.And the strength of desert sand concrete predicted by BP neural network is compared with the prediction result of genetic algorithm optimised BP neural network through example verification.The results show that the desert sand concrete strength prediction model based on genetic algorithm optimised BP neural network has better operability and prediction accuracy,which opens up a new way to improve the prediction accuracy of desert sand concrete strength.