Introduction to Fuzzy Logic and Fuzzy Sets — Beginner-Friendly Notes

1. Why Fuzzy Logic Exists

Classical sets/logic are crisp: hot or not, true or false. Real language is vague: "warm", "tall", "high temperature" are matters of degree. Fuzzy logic formalises degrees of truth in [0,1].

Crisp: $x \in Set$ is 0 or 1. Fuzzy: $μ_Set(x) \in [0,1]$ is degree of membership.

2. Fuzzy Sets — Core Idea

A fuzzy set $A$ is characterised by membership function $μ_A(x)$.

Example — Temperature:

• Crisp: $Hot = \{x | x > 30°C\}$ → 30.1°C is hot, 29.9°C is not (brittle edge).
• Fuzzy: $μ_Hot(x)$ rises gradually: $μ_Hot(25)=0.2$, $μ_Hot(30)=0.6$, $μ_Hot(35)=1.0$ — no sharp boundary.

Common membership shapes: triangular, trapezoidal, Gaussian.

3. Operations (like sets, but on membership degrees)

For fuzzy sets $A,B$ at element $x$:

• Complement: $μ_{
  eg A}(x)=1-μ_A(x)$
• Intersection (AND): $μ_{A \land B}(x)=\min(μ_A(x), μ_B(x))$ (often $\min$)
• Union (OR): $μ_{A \lor B}(x)=\max(μ_A(x), μ_B(x))$ (often $\max$)
• Product alternative: $μ_{A \land B}=μ_A \cdot μ_B$ (also used)

4. Fuzzy Logic — Rules and Inference (concept)

Fuzzification: crisp input → fuzzy degrees (e.g., temp 32°C → Hot 0.7, Warm 0.3).

Fuzzy rules: IF $Temperature$ is Hot AND $Humidity$ is High THEN $Fan$ is Fast (linguistic rules).

Inference: combine rule premises via $\min/\max$, then aggregate consequents.

Defuzzification: fuzzy result → crisp action (e.g., centroid of output fuzzy set → fan speed 78%).

Pipeline:

Crisp inputs → Fuzzify → Rule Evaluation → Aggregate → Defuzzify → Crisp output

Example — AC controller: inputs: temp/membership, humidity/membership → rules like "if hot and humid then cool high" → defuzzify to compressor power.

5. Fuzzy vs Probability (exam must-distinguish)

• Probability: uncertainty about whether event occurs (randomness) — e.g., $P(rain)=0.6$.
• Fuzziness: degree to which property holds (vagueness) — e.g., "rain is heavy" to degree 0.6.
  Both use [0,1] but mean different things.

6. Where Used

• Control systems (washing machines, AC, cement kilns), early expert systems, Japanese trains, consumer electronics — where human-like vague rules suffice without precise models.

7. One-Line Summary

Fuzzy sets generalise crisp sets with membership $μ_A(x)\in[0,1]$ and fuzzy logic reasons with linguistic if-then rules via fuzzification, $\min/\max$ inference and defuzzification, modelling vagueness rather than randomness.

8. Quick Self-Check

1. Give $μ_Hot(28°C)$ for a triangular Hot that starts at 20, peaks at 40.
2. Why is "tall person = 180cm threshold" crisp but inadequate?