Font Size: a A A

Improvement And Application Of Fuzzy Neural Network Prediction Algorithm

Posted on:2018-03-09Degree:MasterType:Thesis
Country:ChinaCandidate:J GaoFull Text:PDF
GTID:2348330515996654Subject:Engineering
Abstract/Summary:PDF Full Text Request
Fuzzy neural network is an important branch in the field of artificial neural network,and widely applied in the system of control and modeling,and have good achievement.But in the face of high dimension,low correlation data,because some results and data of one or more related,so the model forecast results are very poor,and not good to generate fuzzy rules.To solve corresponding problems,this article carries on the corresponding discussion and research,including: the selection of input vector method,a fuzzy neural network learning ability,the division of the fuzzy set,fuzzy rules generation and error.The work of this paper include:1,studied the present situation of the fuzzy neural network domestic and abroad,studies the education data mining technology at domestic and abroad,and points out some problems existing in the theory and application.2,to solve the existing problems,and compares a variety of models,the Adaptive Fuzzy neural reasoning System(the Adaptive Network--based Fuzzy Inference System),on the basis of Fuzzy neural Network model for the appropriate improvement,to make it more suitable for high dimension,low correlation data,you can find out the relationship among several large data of data items,describes the relationship,and generate the understanding of knowledge.3,using a modified adaptive neural fuzzy inference system for computer validation,generated a number of input variables,and some related to the output variable,part has nothing to do with the output variables,to find out the related to the output variable input and interpreted accordingly.4,the new model will be applied on the education of college students data,according to the correlation of subjects,make corresponding performance predictions.To predict with some future course in some subjects,to solve the traditional neural network problem,such as facing high dimension input,the training time is too long.The new method can find the relationship between the various subjects,and make the corresponding explanation,prediction accuracy is higher,smaller error,and has a strong explanatory.
Keywords/Search Tags:fuzzy neural network, adaptive fuzzy neural inference system, ANFIS, prediction to high dimensionality low association, data mining, Predicting result
PDF Full Text Request
Related items