Research On Modeling Method For Cement Hydrationusing Markov Random Field And The Future Use

Cement is an important industrial material,and it has been widely used in all aspects of the fields of industry,so that it is important for people to study and explore the mechanisms of its hydration.But due to the extreme complexity of cement hydration,people still have not fully understood the process of the reaction,and the traditional methods of analysis need to consume a lot of time and do not behave the characteristic of time-continuity.With the development of computer science and technology,the cement hydration research has been went into the era of computing-materials,although the subsequent researches have also made some recognized results,such as Bentz's cellular automata model.However,this model is generally based on the basis of the pixels and features,before modeling must do the complex cement images registration and other operations,so that even a very small pixels bias may result in modeling foundered.And this model requires not only the precise equipment,but also should do a lot of computing works.Here thanks to the extensive application of Markov random field in the image,we put this applied to the field of cement hydration modeling.Due to the use of Markov random field,cement hydration modeling will only consider the cement micro-image neighborhood pixels' probability distributions,without must to do the complex images registration and feature selection first,so it can improve the accuracy and is easy to use.The mainly research of this article is described as follows:(1)The Markov Random Fields(MRF)Parameter Estimation Method that used in the Cement Microstructure PicturesThis article has proposed a new MRF parameter estimation method named weighted least squares fit method(WLS).The issues of do the MRF parameter estimations has always been a big challenge to the users,all of the common methods that used today have some problems.In order to make the MRF model can be successfully used to do cement hydration modeling,it need to use a parameter estimation method that behave with high accuracy,low time complexity and noise robustness.The WLS method we proposed is a complete system,which includes the main WLS parameter estimation method,the zero value processing method and the parameter fitness evaluation method.And the experiments show that our WLS method hashigher accuracy and is less sensitive to noise than the least squares method(LS).(2)Cement Hydration Modeling Based on MRFSince this is the first time the Markov model has been applied to the cement hydration modeling field,we have conducted a feasibility analysis.This paper presents a fitness function of cement hydration similarity based on sampled image,which solves the biggest difficulty of cement hydration modeling based on Markov field.And then combines it with particle swarm optimization(PSO)to establish the MRF Model of cement hydration process.(3)The Future use after the MRF Model of Cement Hydration Process Have been EstablishedThe application prospect and methods of do cement hydration research based on the established MRF model are given.The artificial neural network can be used to simulate the change of the parameters of cement hydration MRF model.And then could generate sampling images towards the change of model parameters to predict the change of cement microstructure and the cement hydration process.It also provides a new chance to explore the relationship between the MRF model parameters and the specific performance of cement.