New Fuzzy Interpolative Reasoning Methods for Sparse Fuzzy Rule-Based Systems

• Main Authors: Yu-chaun Chang, 張昱銓
• Other Authors: Shyi-ming Chen
• Format: Others
• Language: en_US
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/2m6g6n

博士 === 國立臺灣科技大學 === 資訊工程系 === 100 === Fuzzy interpolative reasoning is an important research topic for sparse fuzzy rule-based systems. It not only can overcome the drawback of sparse fuzzy rule-based systems, but also can help to reduce the complexity of large fuzzy rule bases for fuzzy rule-based systems. In this dissertation, we present five new fuzzy interpolative reasoning methods for sparse fuzzy rule-based systems based on type-1 fuzzy sets and interval type-2 fuzzy sets, respectively. In the first method of our dissertation, we present a new fuzzy interpolative reasoning method for sparse fuzzy rule-based system based on the areas of fuzzy sets. The proposed method uses the weighted average method to infer the fuzzy interpolative reasoning results. In terms of the six evaluation indices, the experimental results show the proposed method performs more reasonably than the existing methods. In the second method of our dissertation, we present a new method for multi-variables fuzzy forecasting based on fuzzy clustering and fuzzy rule interpolation techniques. We apply the proposed method to the temperature prediction problem and the Taiwan Stock Exchange Capitalization Weighted Stock Index (TAIEX) data. The experimental results show that the proposed method produces better forecasting results than existing methods. In the third method of our dissertation, we present a new weighted fuzzy interpolative reasoning method for sparse fuzzy rule-based systems. It is based on genetic algorithm (GA)-based weight-learning techniques. The proposed method can deal with fuzzy rule interpolation with weighted antecedent variables. We also present a GA-based weight-learning algorithm to automatically learn the optimal weights of the antecedent variables of the fuzzy rules. We also apply the proposed weighted fuzzy interpolative reasoning method and the proposed GA-based weight-learning algorithm to deal with the truck backer-upper control problem, multivariate regression problems and time series prediction problems. Based on statistical analysis techniques, the experimental results show that the proposed weighted fuzzy interpolative reasoning method using the optimally learned weights obtained by the proposed GA-based weight-learning algorithm has statistically significantly smaller error rates than the existing methods. In the fourth method of our dissertation, we present a new method for fuzzy rule interpolation for sparse fuzzy rule-based systems based on the ratios of fuzziness of interval type-2 fuzzy sets. The proposed method can deal with fuzzy rule interpolation based on polygonal interval type-2 fuzzy sets and bell-shaped interval type-2 fuzzy sets. The experimental results show that the proposed method gets more reasonable results than the existing methods. In the fifth method of our dissertation, we present a new method for fuzzy rule interpolation with interval type-2 Gaussian fuzzy sets for sparse fuzzy rule-based systems. We also present a learning algorithm to learn the optimal interval type-2 Gaussian fuzzy sets for sparse fuzzy rule-based systems based on genetic algorithms. We also apply the proposed fuzzy rule interpolation method and the proposed learning algorithm to deal with multivariate regression problems and time series prediction problems. The experimental results show that the proposed fuzzy rule interpolation method using the optimally learned interval type-2 Gaussian fuzzy sets produces higher accuracy than the existing methods.