A Hybrid of Differential Evolution and Cultural Algorithm for Recurrent Functional Neural Fuzzy Networks and Its Applications

By applying the recurrent functional neural fuzzy network (RFNFN) and a novel evolutionary learning algorithm this study presents an evolutionary neural fuzzy network (NFN). The proposed new evolutionary learning algorithm is based on an effective combination of the modified differential evolution (MDE) and cultural algorithm, which is called the cultural-based MDE (CMDE) method. After the four individuals are randomly chosen from the population for mutation, the proposed CMDE method can search the whole solution space more efficiently. In addition, during the evolutionary process of the cultural algorithm, the performance enhancement is highly affected by the belief space extraction and use of the corresponding information. The proposed RFNFN model uses functional-link neural networks (FLNN) as the consequent part of the fuzzy rules, while in this study the orthogonal polynomials and linearly independent functions are adapted as in a functional expansion of the FLNNs. In summary, the RFNFN model is not only able to generate the consequent part as a nonlinear combination of input variables, but also capable of tackling complicated temporal problem through the added feedback connections. Experimental results in prediction and control applications have shown that the proposed RFNFN-CMDE mechanism outperforms other compared approaches, in terms of convergent speed and root-mean-squared error. Corresponding Author: Cheng-Jian Lin is with the Computer Science and Information Engineering Department, National Chin-Yi University of Technology, Taichung City, Taiwan 411, R.O.C. E-mail: [email protected] Hsueh-Yi Lin is with the Computer Science and Information Engineering Department, National Chin-Yi University of Technology, Taichung City, Taiwan 411, R.O.C. E-mail: [email protected] Chi-Feng Wu is with the Digital Content Application and Management Department, Wenzao Ursuline College of Languages, Kaohsiung City, Taiwan 811, R.O.C. E-mail: [email protected] Cheng-Hung Chen is with the Electrical Engineering Department, National Formosa University, Yunlin, Taiwan 632, R. O. C. E-mail: [email protected] Manuscript received 10 Jan. 2010; revised 28 March 2012; accepted 10 Dec. 2012.