2 Tutorial

Simulate Fuzzy Inference Systems in Simulink

You can simulate a fuzzy inference system (FIS) in Simulink using either the Fuzzy Logic
Controller or Fuzzy Logic Controller with Ruleviewer blocks. Alternatively, you can
evaluate fuzzy systems at the command line using evalfis.

For more information on creating fuzzy inference systems, see “Build Mamdani Systems
Using Fuzzy Logic Designer” on page 2-32 and “Build Mamdani Systems at the Command
Line” on page 2-71.

Simulate Fuzzy Inference System

Once you have implemented a fuzzy inference system using Fuzzy Logic Designer, using
Neuro-Fuzzy Designer, or at the command line, you can simulate the system in
Simulink.

For this example, you control the level of water in a tank using a fuzzy inference system
implemented using a Fuzzy Logic Controller block. Open the sltank model.

open_system('sltank')

2-84
 Simulate Fuzzy Inference Systems in Simulink

For this system, you control the water that flows into the tank using a valve. The outflow
rate depends on the diameter of the output pipe, which is constant, and the pressure in
the tank, which varies with water level. Therefore, the system has nonlinear
characteristics.

The two inputs to the fuzzy system are the water level error, level, and the rate of
change of the water level, rate. The output of the fuzzy system is the rate at which the
control valve is opening or closing, valve.

To implement a fuzzy inference system, specify the FIS name parameter of the Fuzzy
Logic Controller block as the name of a FIS structure in the MATLAB® workspace. In this
example, the block uses the FIS structure tank.

For more information on this system, see “Water Level Control in a Tank” on page 2-96.

As a first attempt to control the water level, set the following rules in the FIS. These rules
adjust the valve based on only the water level error.

2-85
2 Tutorial

• If the water level is okay, then do not adjust the valve.

• If the water level is low, then open the valve quickly.
• If the water level is high, then close the valve quickly.

rule1 = "If level is okay then valve is no_change";

rule2 = "If level is low then valve is open_fast";
rule3 = "If level is high then valve is close_fast";
rules = [rule1 rule2 rule3];
tank = parsrule(tank,rules);

Simulate the model and view the water level.

open_system('sltank/Comparison')
sim('sltank',100)

2-86
Simulate Fuzzy Inference Systems in Simulink

2-87
2 Tutorial

These rules are insufficient for controlling the system, since the water level oscillates
around the setpoint.

To reduce the oscillations, add two more rules to the system. These rules adjust the valve
based on the rate of change of the water level when the water level is near the setpoint.

• If the water level is okay and increasing, then close the valve slowly.
• If the water level is okay and decreasing, then open the valve slowly.

rule4 = "If level is okay and rate is positive then valve is close_slow";
rule5 = "If level is okay and rate is negative then valve is open_slow";
rules = [rule1 rule2 rule3 rule4 rule5];
tank = parsrule(tank,rules);

Simulate the model.

sim('sltank',100)

2-88
Simulate Fuzzy Inference Systems in Simulink

2-89
2 Tutorial

The water level now tracks the setpoint without oscillating.

You can also simulate fuzzy systems using the Fuzzy Logic Controller with Ruleviewer
block. The sltankrule model is the same as the sltank model, except that it uses the
Fuzzy Logic Controller with Ruleviewer block.

open_system('sltankrule')

2-90
 Simulate Fuzzy Inference Systems in Simulink

During simulation, this block displays the Rule Viewer from the Fuzzy Logic Designer
app.

sim('sltankrule',100)

2-91
2 Tutorial

2-92
 Simulate Fuzzy Inference Systems in Simulink

If you pause the simulation, you can examine the FIS behavior by manually adjusting the
input variable values in the Rule Viewer, and observing the inference process and output.

You can also access the Fuzzy Logic Designer editors from the Rule Viewer. From the
Rule Viewer, you can then adjust the parameters of your fuzzy system using these editors,
and export the updated system to the MATLAB workspace. To simulate the updated FIS,
restart the simulation. For more information on using these editors, see “Build Mamdani
Systems Using Fuzzy Logic Designer” on page 2-32.

Access Intermediate Fuzzy Inference Results

You can access intermediate fuzzy inference results using the Fuzzy Logic Controller
block. You can use this data to visualize the fuzzy inference process or troubleshoot the
performance of your FIS. To access this data, enable the corresponding parameters in the
block, and connect signals to the corresponding output ports.

2-93
2 Tutorial

Block Parameter Description Output Port

Fuzzified Inputs Fuzzified input values, obtained by evaluating fi
the input membership functions of each rule at
the current input values.
Rule firing Rule firing strengths, obtained by evaluating the rfs
strengths antecedent of each rule.
Rule outputs Rule outputs, obtained by evaluating the ro
consequent of each rule.
Aggregated Aggregate output for each output variable, ao
outputs obtained by combining the corresponding
outputs from all the rules.

For more information see, Fuzzy Logic Controller.

Simulation Modes
The Fuzzy Logic Controller block has the following two simulation modes:

• Interpreted execution — Simulate fuzzy systems using precompiled MEX files.

Using this option reduces the initial compilation time of the model.
• Code generation — Simulate fuzzy system without precompiled MEX files. Use this
option when simulating fuzzy systems for code generation applications. Doing so
simulates your system using the same code path used for generated code.

To select a simulation mode, set the Simulate using parameter of the block. By default,
the block uses Interpreted execution mode for simulation.

Map Command-Line Functionality to Fuzzy Logic Controller

Block
The parameters and ports of the Fuzzy Logic Controller block map to the input and output
arguments of evalfis or the properties of evalfisOptions. The following table shows
the block parameters and ports that map to evalfis arguments.

evalfisArgument Description Block Parameter or Port

fis Fuzzy inference system FIS name

2-94
 See Also

evalfisArgument Description Block Parameter or Port

input, when a single row Input variable values in
output, when a single row Output variable values out
fuzzifiedInputs Fuzzified inputs fi
ruleOutputs Rule outputs ro
aggregateOutput Aggregated outputs ao

The following table shows the block parameters that map to evalfisOptions
properties.

evalfisOptions Property Description Block Parameter or Port

NumSamplePoints Number of points in output Number of samples for
fuzzy sets output discretization
OutOfRangeInputValueM Diagnostic message Out of range input value
essage behavior when an input is
out of range
NoRuleFiredMessage Diagnostic message No rule fired
behavior when no rules fire
EmptyOutputFuzzySetMe Diagnostic message Empty output fuzzy set
ssage behavior when an output
fuzzy set is empty

The remaining parameters of the Fuzzy Logic Controller block do not map to arguments
of evalfis.

See Also
Blocks
Fuzzy Logic Controller | Fuzzy Logic Controller with Ruleviewer

More About
• “Temperature Control in a Shower” on page 2-104
• “Water Level Control in a Tank” on page 2-96

2-95
2 Tutorial

Water Level Control in a Tank

This model shows how to implement a fuzzy inference system (FIS) in a Simulink® model.

Simulink Model

This model controls the level of water in a tank using a fuzzy inference system
implemented using a Fuzzy Logic Controller block. Open the sltank model.

open_system('sltank')

For this system, you control the water that flows into the tank using a valve. The outflow
rate depends on the diameter of the output pipe, which is constant, and the pressure in
the tank, which varies with water level. Therefore, the system has nonlinear
characteristics.

2-96
 Water Level Control in a Tank

Fuzzy Inference System

The fuzzy system is defined in a FIS structure, tank, in the MATLAB® workspace. For
more information on how to specify a FIS in a Fuzzy Logic Controller block, see Fuzzy
Logic Controller.

The two inputs to the fuzzy system are the water level error, level, and the rate of
change of the water level, rate. Each input has three membership functions.

figure
plotmf(tank,'input',1)
figure
plotmf(tank,'input',2)

2-97
2 Tutorial

The output of the fuzzy system is the rate at which the control valve is opening or closing,
valve, which has five membership functions.

plotmf(tank,'output',1)

2-98
 Water Level Control in a Tank

Due to the diameter of the outflow pipe, the water tank in this system empties more
slowly than it fills up. To compensate for this imbalance, the close_slow and
open_slow valve membership functions are not symmetrical. A PID controller does not
support such asymmetry.

The fuzzy system has five rules. The first three rules adjust the valve based on only the
water level error.

• If the water level is okay, then do not adjust the valve.

• If the water level is low, then open the valve quickly.
• If the water level is high, then close the valve quickly.

2-99
2 Tutorial

The other two rules adjust the valve based on the rate of change of the water level when
the water level is near the setpoint.

• If the water level is okay and increasing, then close the valve slowly.
• If the water level is okay and decreasing, then open the valve slowly.

showrule(tank)

ans =

5x75 char array

'1. If (level is okay) then (valve is no_change) (1) '

'2. If (level is low) then (valve is open_fast) (1) '
'3. If (level is high) then (valve is close_fast) (1) '
'4. If (level is okay) and (rate is positive) then (valve is close_slow) (1)'
'5. If (level is okay) and (rate is negative) then (valve is open_slow) (1) '

In this model, you can also control the water level using a PID controller. To switch to the
PID controller, set the const block to a value greater than or equal to zero.

Simulation

The model simulates the controller with periodic changes in the setpoint of the water
level. Run the simulation.

sim('sltank',100)
open_system('sltank/Comparison')

2-100
Water Level Control in a Tank

2-101
2 Tutorial

The water level tracks the setpoint well. You can adjust the performance of the controller
by modifying the rules of the tank FIS. For example, if you remove the last two rules,
which are analogous to a derivative control action, the controller performs poorly, with
large oscillations in the water level.

See Also
Blocks
Fuzzy Logic Controller | Fuzzy Logic Controller with Ruleviewer

2-102
 See Also

More About
• “Simulate Fuzzy Inference Systems in Simulink” on page 2-84
• “Temperature Control in a Shower” on page 2-104

2-103