Agriculture Reference
In-Depth Information
analysis and design. Classical control techniques are particularly good for single-
input/single-output systems. They also promote an intuitive understanding of com-
ponent and system behavior.
If the system is computer-controlled and the sampling time of the signal by the
computer is not short compared to the system dynamics, the effect of the time sam-
pling will affect the modeling of the system performance for prediction, analysis,
and design. If that computer relative slowness is the case, digital control theory must
be used instead of the classical control theory and the Laplace transforms replaced
with
z
-transforms. As computers have become very fast compared to the dynamics of
most agricultural systems, the use of digital control theory is less important.
The modern control theory was developed in the latter half of the twentieth century.
It operates in the time domain and uses state variables to describe the system. The
dynamics of the system being controlled are represented by a vector of first-order time
differential equations describing the changes in the state variables. Techniques from
linear algebra, such as eigenvalues, are used in system analysis and synthesis. The
modern control theory generally handles multiple-input/multiple-output systems bet-
ter than the classical control theory. It also is often convenient if a system needs to be
synthesized to achieve a certain level of performance. Of course, these comparisons
between classical and modern control theory are broad generalizations, and the selec-
tion of whether to use classical control theory or modern control theory to analyze
or design a system depends on user preferences and the particulars of a given situa-
tion. Historically, it appears that the classical control theory has been used much more
widely for agricultural automation systems than the modern control theory.
One important inherent assumption required for most analyses of both classical
and modern control is that the system is linear. For example, “linear” means that
doubling the setpoint will double the output. Because linearity greatly simplifies
analysis, many system components, such as sensors and actuators, are purposely
designed to be linear. Although some agricultural systems are inherently relatively
linear, especially over restricted ranges of operation, many are not. If possible, the
systems should be linearized to promote understanding and control. There are tech-
niques to analyze and control nonlinear systems. However, the techniques are rela-
tively difficult and have not been widely used in agricultural automation applications.
One type of nonlinear automatic control that has achieved some usage is the rule-
based controller. It may be an expert system or embody some other form of artificial
intelligence. For controlling some systems, there may be a lookup table or some
other form of control map. In these systems, based on what ranges the information
from the sensors are located in, the controller outputs are accordingly specified. If
boundaries between the ranges are fuzzy, this is fuzzy control.
Controllers may be mechanical. One of the early agricultural automation exam-
ples was the flyball governor on steam engines. Based on Watt's pioneering work,
the governor utilized the centrifugal force generated by spinning balls connected
to the engine's output speed. When the speed was not correct, the changing force
would move a valve, thereby correcting the steam flow to the engine and ultimately
the engine speed. Another famous agricultural example is Harry S. Ferguson's draft
control system. This hydromechanical system on tractors automatically raised or
lowered the implement to maintain a near-constant draft force on the tractor.
