Green High Performance Concrete And Its Prediction Neural Network Method

The green high performance concrete (GHPC) is a kind of green, the environmental protection , developed in recent years new building material conforming to our country s sustainable developmental strategy. Resting on the theory about " superimposition effect" of several thin mineral material's intensity .This article conducts researches on several factors affecting working performance of concrete newly mixed and mechanical performance of hardened concrete , determines the best adding proportions of several thin mineral material , has produced the C80 green high performance concrete recommendation mixture ratio, Using orthogonal experiment method. Using multivariable linear regression module in SPSS which is more popular statistics software package, This article analyses 5 factors affecting concrete performances ,give multivariable linear regression equation of concrete's 28d compressive strength, basing on hydro-gel ratio and silicon ash amount mixed.The artificial neural networks (ANN) are an extremely active interdisciplinary science developed in recent years. It has the capacities of distribution parallel processing, the non-linear mapping, the auto-adapted study, the Lu stick fault-tolerant and so on. This article uses the most popular BP nerve network model recently to carry on a more precise forecast to the GHPC 28d compressive strength. But the BP algorithm essence is the search algorithm which drops along the gradient, thus inevitably has the shortcomings, sensitive to the initial weight and prone to falling into the partial minimum. This article uses the highly parallel genetic algorithms , search in overall situation Avoiding falling into partial minimum and strong Robustness ,To carry on optimization to the Back-propagation network, Using respective special skill , solves the GHPC28d compressive strength forecast problem, obtains the good effect.