CAJ | 학술논문

在深入分析已有的基于相似度的模糊推理算法的基础上，针对其计算复杂、未考虑规则前件权重及规则激活方法可能导致结论不全面的问题，提出一种基于综合相似度的区间值模糊推理算法。引入综合相似度、规则阈值、规则可信度等概念，给出新的模糊规则激活方法及推理结论可信度的计算方法。通过算例说明了新方法在多重多维模糊推理的情况下具有还原性，且计算简单，适用于实际应用。
재심입분석이유적기우상사도적모호추리산법적기출상，침대기계산복잡、미고필규칙전건권중급규칙격활방법가능도치결론불전면적문제，제출일충기우종합상사도적구간치모호추리산법。인입종합상사도、규칙역치、규칙가신도등개념，급출신적모호규칙격활방법급추리결론가신도적계산방법。통과산례설명료신방법재다중다유모호추리적정황하구유환원성，차계산간단，괄용우실제응용。
Based on deep analysis on existing similarity-based fuzzy reasoning algorithms, we put forward an interval-valued fuzzy inference algorithm, which is based on comprehensive similarity, to tackle the problems of complicated calculation, neglecting the weight of all premises of fuzzy rules, and the possible incomplete conclusions led by rules activation method. This algorithm introduces the concepts of comprehensive similarity, rules thresholds and rules reliability degree, etc.Furthermore, it presents a new activation method for fuzzy rules and the calculation method of reliability degrees of reasoning conclusion.It is proved through numerical example that the new algorithm has the reductive property under the circumstance of multiple and multidimensional fuzzy inference, and its calculation is simple, thus it is suitable for practical applications.