|
|
Easy to understand about Fuzzy control!
Fuzzy Control (Swing-up Inverted Pendulum)
It is Fuzzy control to translate the human rule of thumb into certain rules as inputs / outputs and use them for control. In the recent applications of Fuzzy control, however, it is attempted to control an object which has no human rule of thumb thus leaving it difficult to introduce such rules. Fuzzy control is a technology which can exert the best effect when the conventional control method is hard to apply to the object due to difficulty, time and cost to introduce a mathematical model or due to nonlinear model. Our Fuzzy control training kit for a swing-up inverted pendulum provides you the best mean to easily understand about Fuzzy control.
- Easy to understand about Fuzzy control
- Indifferent from the conventional mathematical logic, Fuzzy control adopts certain rules which is the translation of the human experience, sixth-sense and know-how into the membership functions. Accordingly, you can learn how to translate human rule of thumb into certain rules for Fuzzy control. Since the result of Fuzzy inference is sought through a load-center computation but a simple judgement of ON-OFF, you can also learn the steps how to rationally judge the problems in consideration with various aspects at a time as if a human being.
- Easy to realize "Swing-up Inverted Pendulum"
- The "Swing-up inverted pendulum" is different from the so-called "Inverted pendulum" which simply requires a single control not to topple down the pendulum. Three stages of controls are realized here with a minimum of inputs, outputs and rules (3 inputs, 1 output, 11 rules): namely, 1st stage to swing the tilted pendulum up toward its top-dead-point (TDP), 2nd stage to stop the pendulum at a proximity of its TDP and 3rd stage to stabilize the pendulum at around its TDP.
- Inference method based on human rule of thumb
- In Mamdani’s "Minimax Load-Center" method which is introductory applied at the earlier stage of Fuzzy control, the inference logic has problems such as: that the inference result is nonlinear or that there exist the rules which cause no effect. "Multiplication-Addition Load-Center" method is developed to make the inference result linear as well as each and every rule effective. We have adopted "Linear Fuzzy Inference" method which is an advanced one from the later method to make the input rules more suitable to human intuition.
- Easy rule creation with mouse
- The antecedent rules (input conditions) can be easily created by selecting a button with such shape (triangle, trapezoid, rectangle, slash, edge) of the membership functions with mouse and mouse-click a label position at the antecedent sector on the Rule-Setting window. While the consequent rules, which use a Singleton (Bar-shape) as its membership functions, can be also easily created by mouse-clicking at any position of 13 labels in the consequent sector thus enabling a fine tuning of the outputs. Since the all shapes of the membership functions are already fixed, it is easy to create the rules without introducing the shape anew.
Fuzzy Inference Specification
| Number of Rules (Max) | 128 |
| Number of Inputs / Outputs (Max) | 4-Inputs / 4-Outputs |
MF Shapes for Antecedent Sector
(MF= Membership Function) | Triangle, Trapezoid, Rectangle, Slash, Edge |
| MF for Consequent Sector | Singleton |
| Inference Logic | Linear Inference Logic |
| Deterministic Algorithm | Load-Center Method |
| Inference Device | Software |
| Resolution Power | 10-bit each for Antecedent / Consequent |
RealPlayer Sample Movie
Shows the moves of the pendulum. (from Accessory Video)
Click the button to download RealPlayer.
 (C) 2006 ADWIN Corp.
|
|
