【问题标题】:In fuzzy logic, how can I add membership functions for different perspectives?在模糊逻辑中,如何为不同的视角添加隶属函数?
【发布时间】:2015-01-29 11:34:59
【问题描述】:

我正在将 jFuzzyLogic 库用于 FL 项目。

我创建了一个 FCL 文件,其中包含三个变量,服务、房间和食物。

FUZZIFY food
    TERM bad := (0, 1) (4, 0) ; 
    TERM good := (1, 0) (4,1) (6,1) (9,0);
    TERM excellent := (6, 0) (9, 1);
END_FUZZIFY

FUZZIFY service
    TERM poor := (0, 1) (4, 0) ; 
    TERM good := (1, 0) (4,1) (6,1) (9,0);
    TERM excellent := (6, 0) (9, 1);
END_FUZZIFY

FUZZIFY room
    TERM poor := (0, 1) (4, 0) ; 
    TERM good := (1, 0) (4,1) (6,1) (9,0);
    TERM excellent := (6, 0) (9, 1);
END_FUZZIFY

我还有两条规则(这些规则并不详尽):

RULEBLOCK rules
    AND : MIN;          // Use 'min' for 'and' (also implicit use 'max' for 'or' to fulfill DeMorgan's Law)
    ACT : MIN;          // Use 'min' activation method
    ACCU : MAX;         // Use 'max' accumulation method

    RULE 1 : IF food IS bad OR service IS poor OR room IS poor THEN trustWeight IS less;
    RULE 2 : IF food IS excellent OR service IS excellent OR room IS excellent THEN trustWeight IS high;

END_RULEBLOCK

最终输出是以下集合:

DEFUZZIFY trustWeight
    TERM less := (0,0) (0.25,1) (0.5,0);
    TERM high := (0.5,0) (0.75,1) (1,0);
    METHOD : COG;       // Use 'Center Of Gravity' defuzzification method
    DEFAULT := 0;       // Default value is 0 (if no rule activates defuzzifier)
END_DEFUZZIFY

我的逻辑是,根据不同类型的人的输入,应该分配不同的权重。例如,当一个商人给一个房间打分(5/10)和一个家庭给一个房间打分(5/10)时,输出不应该是一样的 我需要能够拥有如下规则:

  • 如果是家庭式的,而且房间很差,那么 trustWeight 就更小了
  • 如果是商人,房间不好,那么信任度就高

根据人的类型(和其他固定因素),我应该能够得到不同的 TrustWeight 结果。

这可能吗?如果是这样,我该怎么做?

【问题讨论】:

    标签: java fuzzy-logic jfuzzylogic


    【解决方案1】:

    我会添加另一个输入人员:

    FUZZIFY person
      TERM family := 1;
      TERM business := 2;
    END_FUZZIFY
    

    以及必要的规则:

    RULE 3 : IF person IS business THEN trustWeight IS high;
    RULE 4 : IF person IS family THEN trustWeight IS less;
    

    【讨论】:

      猜你喜欢
      • 2010-12-19
      • 1970-01-01
      • 1970-01-01
      • 1970-01-01
      • 1970-01-01
      • 1970-01-01
      • 2013-02-18
      • 2011-10-16
      • 2011-03-30
      相关资源
      最近更新 更多