Application Research Based On Intelligent Ideas In Cement Performance Analysis

The cement material is not only widely used in industrial and civil architecture, but also applied extensively to traffic, civic building, country irrigation and ocean engineering. It is highly valued as one of the most important raw materials. According to forecast, cements and other cementing materials are still major building materials from this century to far.With the development of material science, profound changes of people's knowledge toward cement materials are taking place, which have characteristics and tendencies as following: 1)The knowledge about essence of cement material from macroscopy to microscopy is gradually deepened, and the relationship between the performance and the interior structure of cement materials is gradually revealed. Consequently, the theoretical basis is provided for developing new variety and enlarging application domain. 2)About the rule of cement material producing process and hydrating and hardening, people's knowledge is enhancing from experience to theory, from phenomenon to essence, so providing theoretical basis to effectively control the producing process of cement material and product as well as adopt new craft and technology.It needs a mass of data to correctly reflect a good many of complex factors and the performance of cement in the process of hydrating, researches of domestic and foreign are focus on data forecast by experiment, mathematical modal analysis and the Neural Network. Obviously, the way of experiment costs very large manpower and material resource, mathematical model analysis is mainly based on experiment data and experience to construct its mathematical model, subsequently make analysis, thus error is unavoidable. While the Neural Network is a technology processing information with intelligence, which tries to stimulate human's way of dealing with problem to understand and utilize information. By automatic adjustment for the network structure and the processing of weight value, the Neural Network can carry out partial functions of the non-living creature's nerve network system and deal with complex industrial process with high dimension and strong jamming. So it provides another feasible measure to stimulate the process of cement's hydrating and hardening under complex situation.But the main existing problem of the Neural Network technology is absent of ability for global search and easy to trap in local minimum. However the Genetic Algorithm is original from natural selection and the random search algorithm of natural inheritance, this capability of global research can optimize the structure and the learning rules of Neural Network. So the mixture of the Neural Network and the Genetic Algorithm can effectively improve the learning ability of Neural Network. But the traditional BP-GA exists disfigurement of many times of iteration. Thus this article proposes an improved BP-GA syncretic arithmetic, to resolve tranditional BP-GA problem of low constringency speed. It enhances the performance of network learning, efficiently extends application scope.Each algorithm has its applied limitation, another algorithm this article researched is the Particles Swarm Optimization Algorithm. It is an optimization algorithm based on the theory of swarm intelligence. Swarm intelligence instructed optimization research is produced by cooperation and competition among swarms in colony. It also is a global research strategy of colony. So the combination of the Particles Swarm Algorithm and the Neural Network is able to resolve the lower reliability in the study of the Neural Network. But tradional BP-PSO algorithm only trains network weight value and threshold value, exists limitation such as high degree redundance, slow constringency speed. By adopting a dynamic inertia gene and condensed network structure, this article improve BP-PSO algorithm, efficiently solve these problems.The computer intelligent algorithm is applied in this paper, it mainly combined Neural Network, Genetic Algorithm, Particle Swarm Optimization and other optimal algorithm, construct Neural Network learning model, based on analyzing the experimental data of the process of the cement hydrating and hardening, considering many complex factors to get the relationship among the ingredients, the particle size, additive and performance (expansible degree of cement, strength, time of coagulation, standard water demand), and putout this result by module to find a effective reason for producing of high performance cements.