基于人工神经网络和Monte-Carlo法的混凝土配合比优化设计研究

结合BP人工神经网络和Monte Carlo随机试验法 ,建立混凝土配合比的直接优化设计模型 ,并开发出实用软件。该设计过程包括 :( 1 )首先建立配合比试验样本数据库 ,然后根据不同混凝土的设计要求 ,检索该数据库并动态建立网络模型 ;( 2 )以混凝土原材料和制作工艺作为输入单元 ,混凝土最终性能指标作为输出单元 ,训练、测试BP神经网络 ,并检验其可靠性 ;( 3)以建立的BP神经网络模型和其它配合比限制条件作为约束条件 ,混凝土单位成本作为优化目标建立Monte Carlo直接优化模型 ,设计出混凝土初步配合比 ;( 4 )按《普通混凝土配合比设计规程》对上述设计出的配合比进行试拌、调整 ,得到实际可使用配合比。

Optimization of concrete mixture on the basis of artificial neural network and Monte-Carlo techniques

An optimization method of concrete mix proportioning using Back-Propagation Neural Network program and Monte-Carlo techniques is described.The process includes:(1)After a database of mix proportioning experiment samples is built,the network model could be established by searching the database,on the basis of different design requests of concrete;(2)By taking the properties and use levels of raw materials,and mixing technology of the concrete as input units, while taking required properties of the concrete as output units,the B-P network could be trained and checked;(3)The Monte-Carlo optimization model could be built with constraints on the B-P network model and required mix proportioning,and with objective function on the unit cost of concrete;and (4) on the basis of the optimization results and the design specifications of JGJ 55-2000,the practical concrete mix proportioning could be tested,adjusted and determined.