A TSK-Type Neurofuzzy Network Approach to System Modeling Problems

Abstract

We develop a neurofuzzy network technique to extract TSK-type fuzzy rules from a given set of input-output data for system modeling problems. Fuzzy clusters are generated incrementally from the training dataset, and similar clusters are merged dynamically together through input-similarity, output-similarity, and output-variance tests. The associated membership functions are defined with statistical means and deviations. Each cluster corresponds to a fuzzy IF-THEN rule, and the obtained rules can be further refined by a fuzzy neural network with a hybrid learning algorithm which combines a recursive singular value decomposition-based least squares estimator and the gradient descent method. The proposed technique has several advantages. The information about input and output data subspaces is considered simultaneously for cluster generation and merging. Membership functions match closely with and describe properly the real distribution of the training data points. Redundant clusters are combined, and the sensitivity to the input order of training data is reduced. Besides, generation of the whole set of clusters from the scratch can be avoided when new training data are considered.

Data-driven fuzzy modeling for Takagi–Sugeno–Kang fuzzy system

“…All these offline algorithms have one advantage that they are independent of the input order of training instances. However, they often take a long time and need a large amount of memory [42]. Online methods, on the other hand, consider training data one at a time.…”

“…They may contain enough new information to form a new rule or to modify an existing one [2][3][4]. Online algorithms are able to run efficiently, but they suffer from the data presentation ordering problem, i.e., the performance of an online method may be greatly affected by the input order of training instances [42]. Angelov and Filev [4] proposed an approach to the online learning of TSK type models.…”

Esquema de controle adaptativo de tráfego de redes baseado em um algoritmo de predição fuzzy

“…Foi demonstrado que o modelo fuzzy TSK é capaz de descrever com eficiência o comportamento de processos reais com características não-lineares, como é o caso do sistema descrito pela equação (1) (Ouyang et al, 2005;Takagi and Sugeno, 1985). Além disso, sabe-se que sistemas de controle baseados em lógica fuzzy para aplicações em tempo real podem ser implementados utilizando por exemplo, processadores DSP (Digital Signal Processing) (Iliev et al, 2000) e circuitos programáveis FPGA (Field Programmable Gate Array) (Juang et al, 2004;Surmann et al, 2006 …”

“…Assim, um modelo de tráfego eficiente deve capturar fielmente as características do tráfego de redes. Muitos estudos mostram que modelos fuzzy possuem vantagens sobre os modelos lineares em descrever o comportamento não-linear e variante no tempo de processos reais desconhecidos, como é o caso dos fluxos de tráfego de redes (Ouyang et al, 2005;Chen et al, 2000). De fato, a modelagem fuzzy é capaz de representar um sistema complexo não-linear através da combinação de vários modelos locais lineares invariantes no tempo (Chen et al, 2007;Sugeno and Yasukawa, 1993).…”

Artificial wavelet neuro-fuzzy model based on parallel wavelet network and neural network

“…Learning pattern of SWNF model with all wavelet functions for Example Learning pattern of MWNF model with all wavelet functions for Example 1 performance index of MWNF model with Morlet activation function is J = 1.61 × 10 −7 which is better than others methods with J = 2.18 × 10 −6 inQuang et al (2005).…”