A new meterage-data mining algorithm based on fuzzy neural network and genetic algorithm

Data Mining is a new direction for meterage data processing recently, which discoveries information from the large-scale meterage databases. The paper discusses the application of Extended TS Fuzzy Neural Network (FNN) and Genetic Algorithm (GA) in data mining. The Extended TS FNN is proposed by Takagi and Sugeno, which can be expressed with if X epsilon Rj, then y=gj(X), where Rj is expressed by a neural network, gj(X) is the output of the neural network. GA is proposed by Holland, which is a general and effective method to optimize problems. We proposes a new algorithm in this paper, which combines fuzzy neural network and genetic algorithm to process information. In the algorithm, the GA is used to cluster the input data(divide the space of input data). The system utilizes clustering data to adaptively construct and optimize FNN model. And at last, TS model (which is combined with n FNN) is used to forecasted the system's output by (j)Sigma(R)mu(j)g(j), where mu(j) is calculated with TS model. The TS model's output is gained with the information of clustering result by genetic algorithm. The combination of GA and Extended TS Fuzzy improves the effectiveness of the system: (1) higher precision, (2) able to process fuzzy information, (3) suitable to handle large-scale data, (4) owning good adaptability. The paper gives the instance of the algorithm, which validates the conclusion of the paper.