Ordinary Portland Cement And Retarding Superplasticizer Compatibility Issues Based On Fuzzy Neural Network To Predict

The fuzzy neural network(FNN) is the research focus of rising in recent years, organic integration of deducing fuzzily and the neural network, the deduce fuzzily system do not possess from the learning ability, but the artificial neural network can't express deducing function of human brain with fuzzy language.Because of giving the possible explanation to power value of neural network and structurization in advance to neural network, make the neural network easy to be understanded by people, so the network have good adaptivity , strong generalization ability, better robustnes, so it is widely used to predict, intellectual control, the pattern-recognition, the disappearance of noise in signal etc field.Applying the admixture to the concrete craft extensively, the admixture has already become the fifth component in concrete, but compatibility issue of cement and admixture put on to people's face. The factor influencing compatibility is numerous, such as cement, admixture, and the construction method,etc., this paper has done more comprehensive summary to these influence factors. The method to measure compatibility has the flow degree of thick liquid cement net and Marsh tube, these methods are all to judge the compatibility between the two on the basis of testing. Though these test methods are simple and convenient and direct, the result received is not judgement on the influence factor based on compatibility and regular understanding, just to already there is a fact that is judgedding.This thesis sums up 8 main influence compatibility factors which on the basis of summarizing forefathers' research results, for fuzzy neural input of network, and make fuzzy comprehensive judge to compatibility of cement and admixture, obtain value of compatibility degree, is used in the output of the fuzzy neural network. Adopt the method based on cluster to build the fuzzy neural network, train the fuzzy neural network by mixing the algorithm of studying, find out the input parameter and output the non-linear relation of parameters from training the sample data, utilize such non-linear relation to predict the compatibility of admixture and cement , apply to the construction site. This thesis adopts MATLAB 7.0 programming, has developed the compatibility prediction system of common portland cement and set retarding and water reducing admixture, the system is make up of the training data entry system and compatibility prediction system. Importing the composition that the mineral of the cement makes up and admixture wanted to be predicted, the system can provide the value of compatibility degree that from 0 to1, according to the size of the value, can judge the compatibility between them. Because data collection is limited, the system can only predict four kinds of commonly used admixture.