Research On Reversed Modeling Method For Cement Hydration Simulation Using Cellular Automata

Cement is indispensable to building materials which has been widely used in the field of industry. The microstructure development affects the cement strength directly. Therefore, a better understanding of the microstructure development evolution process occurring at each day is necessary to increase the predictability of the cement performance.Due to the extreme complexity of cement hydration, people still could not understand the process of cement hydration fully. Traditional ways using physical and chemistry method to study the cement hydration process according to the experimental sample are complex and have low accuracy. With the development of computational materials science, cement hydration modeling using computer technology becomes a hot point in the development of materials science. By adopting the idea of reverse modeling, this paper establishes the three-dimensional cellular automaton model to simulate the microstructure evolution process of cement hydration. Using variety of ways to extract hydration rules from the observed data, this paper carried out the different simulation experiment in different model. The mainly research of this paper is described as follows:(1) Feature extraction and data preprocessingThis paper comes up with a feature extraction method of cement three-dimensional particle based on μCT image. The μCT image of cement can ensure the authenticity of the sample data because it won’t cause too much damage to the sample itself in the process of image scanning. At the same time, the gray value of μCT image can represent the average linear attenuation coefficient of the region which can guarantee the correlation between μCT image gray value and the actual cement material.(2) Multi Expression Programming and Particle Swarm OptimizationIn order to adapt to the cement hydration characteristics of the data, the paper puts forward a new hybrid programming algorithm using Multi Expression Programming(MEP) algorithm to optimize the structure of function and Particle Swarm Optimization(PSO) to optimize the function the coefficient. By testing the regression problems in UCI data sets, the result shows that the hybrid programming method in solving the problem of general function found has a good performance compared with the rest of the commonly used algorithms.(3) Analysis of reverse modeling method for cement hydrationThis paper use kinds of reverse modeling method to construct the microstructure development of the cement hydration with cellular automata model finally. The principle of the model structure and realization process is described in details. At the same time, the graphical representation of the evolution is established to represent the error of predicted value.