Page 191 - The Mechatronics Handbook

E/H System Generic Fuzzy Control

Fuzzy control is an advanced control technology that can mimic a human’s operating strategy in con-
trolling complex systems and can handle systems with uncertainty and nonlinearity (Pedrycz, 1993). One
common feature of fuzzy controllers is that most such controllers are designed based on natural language
control laws. This feature makes it possible to design a generic controller for different plants if the control
of those plants can be described using the same natural language control laws (Zhang, 2001).
The speed control on a hydraulic cylinder actually is achieved by regulating the supplied flow rate to
the cylinder. In different hydraulic systems, the size of the cylinder and the capability of hydraulic system
are usually different, but the control principles are very similar. Representing cylinder speed control
operation, using natural language without loss of generality, the control laws are the same for all systems:
To have a fast motion, open the valve fully.
To make a slow motion, keep the valve open a_little.
To hold the cylinder at its current position, return the valve to the center.
To make a reverse motion, operate the valve to the other direction.
This natural language model represents the general roles in controlling the cylinder speed via an E/H
control valve on all hydraulic systems. The differences in system parameters on different systems can be
handled by redefining the domain of the fuzzy variable, such as fully, a_lot, and a_little, using fuzzy
membership functions (Passino and Yurkovich, 1998). This model provides the basis for designing a
generic fuzzy controller for E/H systems. The adoption of the generic controller on different systems can
be as easy as redefining the fuzzy membership function based on its system parameters.
Figure 10.6 shows the block diagram of a generic fuzzy controller consisting of two input variable
fuzzifiers, a control rule base, and a control command defuzzifier. The two input fuzzifiers were designed
to convert real-valued input variables into linguistic variables with appropriate fuzzy memberships. Each
fuzzifier consists of a set of fuzzy membership functions defining the domain for each linguistic input
variable. A real-valued input variable is normally converted into two linguistic values with associated
memberships. The definitions of these fuzzy values play a critical role in the design of generic fuzzy
controllers and are commonly defined based upon hydraulic system parameters. The fuzzy controller uses
fuzzy control rules to determine control actions according to typical behaviors in the speed control of
hydraulic cylinders. The control outputs are also linguistic values and associated with fuzzy memberships.
For example, if the demanding speed is negative_small (NS) and the error in speed was positive_small
(PS), the appropriate valve control action will be positive_small (PS).
The appropriate control actions were determined based on predefined control rules. Since each real-
valued variable commonly maps into two fuzzy values, the fuzzy inference engine fires at least two control
rules containing these fuzzy values to determine the appropriate control action. Therefore, at least two
appropriate fuzzy-valued control actions will be selected. However, the E/H controller can only implement
one specific real-value control command at a given time. It is necessary to convert multiple fuzzy-valued
control commands into one real-valued control signal in this fuzzy controller.

Commands
fuzzifier
Control Signal G (s)
Rules Defuzzifier H
Status
fuzzifier

H (y)
C

FIGURE 10.6 Block diagram of fuzzy E/H control system. The fuzzy controller consists of input variable fuzzifiers,
control rules, and a signal defuzzifier.

©2002 CRC Press LLC