KomuraSoft LLC
Chapter 1

Introduction — why thresholds alone cause trouble

Before we step into fuzzy control, let's see with numbers how large a step threshold control produces at a boundary.

Difference between threshold control and fuzzy control

Even when the sensor reading wobbles just a little near the boundary, threshold control tends to switch abruptly, while fuzzy control can change smoothly by the amount of overlap.

Start by thinking of a fan with three speed levels

Throughout this course we work with a small controller that decides fan output from temperature and humidity. As a point of comparison, we first set up the following threshold controller.

ConditionOutputValue
Temperature < 23°CLow20
23°C ≤ Temperature < 27°C and Humidity < 45%Low20
23°C ≤ Temperature < 27°C and 45% ≤ Humidity < 70%Medium50
23°C ≤ Temperature < 27°C and Humidity ≥ 70%Medium50
Temperature ≥ 27°C and Humidity < 45%Medium50
Temperature ≥ 27°C and Humidity ≥ 45%High85

This table is for comparison only. From Chapter 2 onward we replace it with "labels that overlap with each other."

The problem is the jump "right at the boundary"

For example, keep humidity fixed at 69% and let the temperature go from 26.8°C to 27.1°C. To a human the sensation is at most "it got a little warmer," but the threshold controller jumps Medium 50 → High 85 the instant it crosses 27°C.

Sensor readings wobble
Real temperature and humidity sensors have fluctuations of 0.1–0.2°C and a few percent, so the output tends to bounce back and forth near a boundary.
Human perception is continuous
Phrases like "a bit warm" or "slightly muggy" describe a feeling that strengthens continuously, not in steps. Drawing only 0/1 boundaries there easily produces a controller that diverges from what people actually feel.

Check 1 — Trace the threshold jump by hand

Answer using only the threshold rules from this chapter. Outputs are low = 20, medium = 50, high = 85.

Q1. With a room temperature of 22.5°C and humidity of 40%, what is the threshold controller's output?

%

Q2. With a room temperature of 26.8°C and humidity of 69%, what is the threshold controller's output?

%

Q3. With a room temperature of 27.1°C and humidity of 69%, what is the threshold controller's output?

%

Q4. What is the output difference between the previous two questions, in points?

points

The intuition to take away from this chapter

The starting point of fuzzy control is not "erase ambiguity rigorously" — it is to avoid carrying steps straight into the controller. From Chapter 2 we look at how, instead of removing the boundaries, we let labels overlap slightly and handle them as numbers between 0 and 1.

For reference: the temperature boundaries 23°C and 27°C used in the threshold table above correspond roughly to the overlap regions (19–25°C and 25–29°C) of the temperature membership functions "cold / comfortable / hot" introduced in Chapter 2. The humidity boundaries 45% and 70% are likewise replaced by the overlap regions of the humidity labels "dry / normal / humid."