Elsevier

Applied Soft Computing

Volume 9, Issue 1, January 2009, Pages 305-307
Applied Soft Computing

A fuzzy filter for improving the quality of the signal in adaptive-network-based fuzzy inference systems (ANFIS)

https://doi.org/10.1016/j.asoc.2008.02.005Get rights and content

Abstract

This study is addressed to improve the quality of the signal of the Adaptive-Network based Fuzzy Inference System (ANFIS) reducing the level of fluctuations in the output due to periodical disturbances. The fuzzy filter computes the disturbances as periodic signals with two components, one at high frequency and another at low frequency. It was incorporated in the layer 0 of the network reducing iteratively effectively the heavy noise.

Introduction

Intelligent control emerged as a viable alternative to conventional model-based control schemes. This is because with fuzzy logic and neural networks issues such as uncertainty variations in plant parameters and structure can be dealt with more effectively hence improving the robustness of the control system. The application of the fuzzy contribution explores the basic adaptive noise logic to industrial level generally, is well accepted; however, no formal methods to identify the fuzzy inference cancellation of noise and periodical disturbances and rules exist. Several publications can be found in the literature. Terptra et al. [1] use an implicit fuzzy model (a fuzzy tuning the inference parameters improve the quality of the rule base) to analyze quantitative statements of the differences between the actual values and those predicted by quantitative models. Layne et al. [2] introduced a new control algorithm, which was developed from linguistic self-organizing. The algorithm has the advantage of improved performance feedback and more efficient knowledge-base modification over the linguistic. Li and Chiu [3] introduced a mathematical model for tuning the adaptation system in conventional controllers. Recently, Roy and Ganguli [4] designed a filter using neural network and genetic algorithm for processing the signals without noise, they obtained good numerical results using a helicopter rotor. On the other hand, Zaheeruddin and Garima [5] used Adaptive Neuro Fuzzy Inference System (ANFIS) for predicting the effects of the noise pollution in the human work. Also, Buragohain and Mahanta [6] used ANFIS for minimizing by application of an engineering statistical technique called full factorial design. However, in the literature, articles about the incorporation of an additional layer or modification in the ANFIS structure for reducing the noise in the signal-processing had not been found. The novel approach provided in this paper focuses on the pioneering Adaptive Neuro Fuzzy Inference System (ANFIS) [7] where the specific objectives are:

  • Construction of an adaptive-network-based fuzzy inference system.

  • Analysis of the possible periodic disturbances and information useful for decision-making.

  • Evaluation of the new ANFIS using a pilot plant.

Section snippets

Structure of the ANFIS and decomposition of the disturbance

The ANFIS is a hybrid neuro-fuzzy inference system that emulates a Sugeno controller [7]. ANFIS uses a feed-forward neural network with five layers that is adapted by a supervised learning algorithm, see Fig. 1. The network uses unweighted connections and works with different activation functions in each layer. The learning algorithm modifies the premise parameters of the fuzzy sets according to the present training data. The ANFIS architecture is as follows:

  • Layer 0: this layer consists of n

Discussion and results

The new filter-ANFIS system was tested on a Wood dryer pilot plant. A diagram of the dryer is depicted in Fig. 3. The dryer works between 373 and 464 K where ventilating fans inside are used for drying the wood. The dryer was operated using pine wood (the moisture changes with the type of the wood). The control system works the very simple manner: the controlled variable is temperature and the manipulated variables are the fan speed (airflow) and the electric power. Fig. 2, Fig. 3 depict two

Conclusion

In order to improve the quality of the signal in ANFIS, a simple fuzzy filter was created and incorporated in the layer 0 of the control structure of a wood dryer. The new ANFIS requires lower computational complexity and the results indicated that the new ANFIS gives satisfactory reduction of the noise to pilot plant scale.

References (10)

There are more references available in the full text version of this article.

Cited by (9)

  • Incorporation of expert variability into breast cancer treatment recommendation in designing clinical protocol guided fuzzy rule system models

    2012, Journal of Biomedical Informatics
    Citation Excerpt :

    The second step is to optimize rule parameters by using a hybrid method which combines gradient descent and a least squares approach. The ANFIS method has been widely used in various applications, such as [13–16]. However, these conventional fuzzy model tuning methods focus only on improving global model accuracy without consideration of preserving model interpretability, two conflicting modeling objectives [3,17].

  • Design and analysis of experiments in ANFIS modeling of a 3-DOF planner manipulator

    2018, 2018 Proceedings of the Japan-Africa Conference on Electronics, Communications, and Computations, JAC-ECC 2018
  • Systems modeling and intelligent control of meat drying process

    2015, 2015 10th System of Systems Engineering Conference, SoSE 2015
  • Design and analysis of experiments in ANFIS modeling for stock price prediction

    2011, International Journal of Industrial Engineering Computations
View all citing articles on Scopus
View full text