Despite its namesake, “fuzzy logic” isn’t as fuzzy as you probably think. In fact, fuzzy logic has already been hard-at-work for most of us, operating behind-the-scenes for over 20 years now in more places than the average expert will admit.
What is Fuzzy Logic?
To put it simply, fuzzy logic is a rule-based system. As a system, fuzzy logic is capable of relying on the practical experience it captures from an operator. It’s particularly useful when capturing such knowledge from an experienced operator.
Fuzzy logic didn’t just lay the groundwork for modern AI. Rather, fuzzy logic in itself is a form of AI. It’s a subset of modern Artificial Intelligence as we know it today.
Fuzzy logic is capable of performing a kind of “decision making” processes, and that means its loosely grouped into the AI toolkit as most software developers see it. When it comes to understanding fuzzy logic and its connection to modern AI, you’ll find that it’s much more powerful than most give it credit for.
How It Began
Considering that fuzzy logic was a huge milestone in the development of the AI technology we have today, it’s quite impressive to note that fuzzy logic dates back to the mid-1960s. However, it’s important to realize that it was not demonstrated in the form of a practical application until sometime in the 1970s.
Since the 1970s, the Japanese have been working tirelessly with fuzzy logic. In fact, they are now traditionally known as the largest producer of applications that run on fuzzy logic. Because of that, Japanese manufacturing has led to fuzzy logic’s inclusion in all sorts of applications where you would never have guessed its presence.
Most impressively, this includes washing machines, cameras, and even applications that help stock traders do their work better.
It has only been in the past decade or so that the United States has begun to catch on to just how useful fuzzy logic can be for various applications. Countless applications are using fuzzy logic at their core, but fail to inform about its usage, likely because “fuzzy logic” is known for having such a negative connotation.
What’s In A Name?
Fuzzy logic is a very diverse system and that has led to its increasing popularity for non-engineering applications. The stock trading applications that rely on it are good examples of that. However, fuzzy logic can also be found within handwriting recognition systems and even medical diagnosis systems, and that adds a great deal of credibility to this system.
While fuzzy logic is capable of functioning for just about any kind of system that features inputs and outputs, many engineers fear it simply because they lack understanding.
With that said, fuzzy logic doesn’t have to be difficult to grasp.
Although the math associated with fuzzy logic may seem intimidating, it’s very similar to binary logic where everything is either 0 or 1. With fuzzy logic, values are between 0 and 1.
As a simple example, you can think of fuzzy logic’s values ranging from 0% to 100%. In the world of fuzzy logic, consider the variable YOUNG. You might call age five 100% young, age eighteen 50% young, and age thirty 0% young. This is compared to binary logic which would make everything below eighteen 100% young and everything above eighteen 0% young.
That simple example can help people wrap their heads around fuzzy logic a bit better, and the design of such a system begins with membership functions, with a set for every input and a set for every output. A rule set is then applied to those functions and a “crisp” value is achieved as the output.
Obviously, this is a simple explanation of how the fuzzy logic operates. In a working system, many inputs and possibly many outputs would exist, and that would lead to a complex set of functions along with the need for many rules. Oftentimes, a fuzzy logic system will contain 40 rules or more. Regardless, the core principles remain the same.
Where Do PID Loops Come In?
PID stands for proportional–integral–derivative controller (also known as a “three term controller”). PID is a control loop feedback mechanism that can be found in many different industrial control systems. Countless other applications also use PID, especially those that require continuous, modulating control.
A PID controller works by constantly calculating an error value, which is the difference between the measurement process variable, or PV, and the desired set point, or SP. As this error value is calculated, the PID controller applies a correction based on the P, I, and D, which are the proportional, integral, and derivative terms. That’s how PID gets its name.
When it comes to practical applications, PID automatically acts accurately and responsively to correct a function. One example that most people use every day is the cruise control feature on their car. An external influence, such as a hill (gradient) will decrease the car’s speed. The PID algorithm will restore the car’s speed to the desired speed using in the most optimal way possible, without overshooting or delay by directly controlling the power output of the car’s engine.
Like fuzzy logic, PID got an early start. In fact, PID far pre-dates fuzzy logic with the first analysis (all theoretical, of course) dating back to the early 1920s. Practical applications were developed soon after and included automatic steering for large ships. Later on, it was used to control processes automatically across the manufacturing industry, largely being implanted in pneumatic controllers and soon being incorporated with electronic controllers.
Today, a universal usage for the PID concept exists for applications that require optimized and accurate control that can be delivered automatically without the need for outside assistance or interference.
PID and fuzzy logic are often integrated with one another, with PID controllers usually being at the base and fuzzy logic being used to schedule variable gains. In other words, the two together allows for even more diverse applications.
These two systems have led to the development of modern industrial AI technology almost single-handledly.