PIDTune
PIDTune[lsys]
gives a feedback PID controller for the linear timeinvariant system lsys.
PIDTune[lsys,"carch"]
gives a controller of architecture "carch" ("P", "PI", "PID", etc).
PIDTune[lsys,{"carch","trule"}]
gives a controller using the tuning rule "trule".
PIDTune[lsys,…,"prop"]
returns the value for the property "prop".
Details and Options
 PIDTune will produce a PID controller g_{fb} such that the closedloop system csys is good at rejecting disturbances as well as follow changes in the reference .
 The ideal PID controller effectively computes the control signal through with proportional, integral, and derivative terms.
 By default, the transfer function representing g_{fb} is returned.
 The system lsys can be a continuoustime linear SISO system, such as TransferFunctionModel, StateSpaceModel, etc.
 The following controller architectures "carch" can be used:

"P" proportional "PI" proportional integral "PD" proportional derivative "PID" proportional integral derivative "PFD" "PD" with filtered derivative part "PIFD" "PID" with filtered derivative part  By default, a "PI" controller is chosen.
 The tuning rule "trule" is chosen automatically, unless explicitly specified, based on stability properties of the system lsys.
 The following tuning rules "trule" can be used:

"AMIGO" approximate constrained integral gain optimization "AMIGOFrequency" "AMIGO" from frequency response "ChienHronesReswick" uses rules based on step response "CohenCoon" dominant pole placement for amplitude decay ratio 0.25 "DisturbanceRejection" minimizes infinity norm of disturbance response "ErrorIntegral" minimizes an integral of error "KappaTau" dominant pole placement "KappaTauFrequency" dominant pole placement from frequency response "LambdaTuning" uses internal model control "LoopShaping" exact "AMIGO" "SkogestadIMC" uses internal model control based on halfrule reduction "TyreusLuyben" uses rules based on frequency response "ZieglerNichols" uses rules based on step response "ZieglerNicholsFrequency" uses rules based on frequency response  PIDTune[lsys,…,"PIDData"] returns a PIDData object pid that can be used to extract additional properties using the form pid["property"].
 PIDTune[lsys,…,"property"] can be used to directly give the value of pid["property"].
 PIDTune can optionally compute a feedforward filter g_{ff} to improve tracking of the reference independently of disturbance rejection capabilities of the feedback controller g_{fb}.
 The feedforward weights b and c affect what fraction of the reference signal is provided to the different terms of the PID controller. The resulting controller is effectively given by , but provided as a separate feedforward filter g_{ff} having same effect when combined with the feedback controller g_{fb}.
 Properties related to different transfer functions include:

"Feedback" feedback g_{fb} "Feedforward" feedforward g_{ff} "OpenLoop" series connection of g_{fb} and lsys "DisturbanceOutput" transfer function from to "DisturbanceControl" transfer function from to "ReferenceOutput" transfer function from to "ReferenceControl" transfer function from to "SensorOutput" transfer function from to "SensorControl" transfer function from to "ISA" transfer function from and to  Properties related to parametrizations of transfer functions:

"FeedbackIdealParameters" {k_{p},t_{i},t_{d}} in "FeedbackSeriesParameters" {k_{ps},t_{is},t_{ds}} in "FeedbackParallelParameters" {k_{pp},k_{ip},k_{dp}} in "FeedforwardParameters" {b,c} in the g_{ff} "DerivativeFilterParameter" n controlling derivative filters in g_{ff} and g_{fb}  The derivative filter replaces the direct derivative with its filtered version , effectively series connecting a lowpass filter with the derivative.
 Properties related to tuning rule and internal tuning model for lsys:

"TuningRule" chosen tuning rule "TuningModel" model and parameters {"tmodel",{par_{1},…}}  The possible tuning models and parameters:

{"FOPTD2",{k_{v},l}} first order plus time delay {"FOPTD3",{k_{s},l,t_{c}} first order plus time delay {"Frequency2",{k_{u},t_{u}}} ultimate gain and ultimate period {"Frequency3",{k_{u},t_{u},k_{s}} ultimate gain, ultimate period, and static gain {"SOPTD",{k_{s},l,t_{c1},t_{c2}}} secondorder plus time delay  The parameters are static gain , velocity constant , delay , time constants , ultimate gain and ultimate period .
 PIDTune also accepts a nonlinear system specified by AffineStateSpaceModel or NonlinearStateSpaceModel, for which the controller computations use the approximate Taylor linearization and the simulations use the nonlinear system.
 The following options can be given:

PIDFeedforward None reference weights for g_{ff} PIDDerivativeFilter None filtered derivative Method Automatic methods to use  The settings for Method control the parameter estimation method used to derive the tuning models.
 Possible settings for Method include:

Automatic automatically choose the estimation method "CharacteristicArea" use the characteristic area of the step response "GainMargins" use the phase crossover frequency and gain margin "InflectionPoint" use the inflection point of the step response "MethodOfMoments" match moments "TwoPointSampling" use the 28% and 63% points of the step response
Examples
open allclose allBasic Examples (2)
Scope (17)
Basic Uses (4)
Find a PI controller for plant in state space form:
Or for the plant as a transfer function:
Specify the type of controller:
Proportional integral and derivative:
Evaluate the performance of the resulting controller:
Get the PID parameters in different standard forms suitable for different implementation technologies:
Properties (4)
Obtain the PIDData object and extract a property from it:
Find the computed controllers, PID feedback controller, and feedforward filter:
The feedback controller transfer function and the PID controller:
By default, there is no feedforward filter:
Find closed loop transfer functions from reference, process disturbance, and sensor noise to output:
The reference to output transfer function measures the ability to follow reference changes:
The disturbance to output transfer function measures the ability to reject process disturbances:
The sensor noise to output transfer function measures the ability to reject measurement noise:
Find transfer functions from reference, process disturbance, and sensor noise to control output:
The reference to control transfer function measures the control effort to follow reference:
The disturbance to output transfer function measures the control effort to reject the disturbance:
The sensor noise to output transfer function measures the control effort to reject sensor noise:
Controller Architectures (4)
By default, the controller architecture is a proportional integrating controller:
Specify the controller architecture:
Proportional ("P") controller:
Proportional integral ("PI") controller:
Proportional derivative ("PD") controller:
Proportional integral derivative ("PID") controller:
Filtered PD ("PFD") controller:
Filtered PID ("PIFD") controller:
Use integral action to eliminate steadystate reference following error:
Derivative action may result in a faster reference response:
Tuning Rules (5)
The tuning rule is automatically determined:
The property "TuningRule" gives the tuning rule that was used:
Obtain the automatically selected tuning rule for a given lsys and "carch":
The "CohenCoon" tuning rule allows for PD and PFD architectures:
Controllers designed with the "LoopShaping" rule result in a maximum sensitivity of about 1.4:
The Nyquist plot of the openloop transfer function lies outside the sensitivity circle of radius 1/1.4:
The "TyreusLuyben" rule may give a stabilizing controller for an unstable system:
The "ZieglerNicholsFrequency" rule may also stabilize the system:
Options (4)
PIDFeedforward (2)
The default feedforward transfer function is unity:
This may result in a large overshoot to step reference inputs:
Use the PIDFeedforward option to improve the tracking performance:
Automatically compute the reference weights for the feedforward filter:
PIDDerivativeFilter (1)
Applications (4)
Process Control (2)
For a system with three cascaded water tanks, find a PID controller for a constant water level in tank 3:
The transfer function from the feed rate in tank 1 to the liquid level in tank 3:
Compute a liquidlevel PID controller for the system:
The reference response from the input to the first tank to the water level in the third tank:
The PID controller also improves the gain and phase margins:
An isothermal continuously stirred tank reactor (CSTR):
The transfer function from the dilution rate to the product concentration:
Compute a PID controller that controls the product concentration:
The control effort for a setpoint change of 0.1 gmol/liter in the product concentration:
The achieved product concentration:
Use a tuning rule, which gives a faster response:
A faster response is accompanied by an increased peak control effort:
Electrical Motor Control (1)
Nonlinear System Control (1)
In a continuous stirredtank reactor where the reaction occurs, the objective is to cause the temperature to track a desired trajectory by using the fluid temperature as the control input:»
The nonlinear model of the system:
The reference trajectory is modeled as the response of a firstorder system:
The openloop system does not track the desired trajectory:
Design a PID controller and compute the response of the nonlinear system with the controller:
Properties & Relations (2)
Tuning rules that give a good reference response may not give a good disturbance response:
Reference responses to a step reference change:
Disturbance responses to a step disturbance change:
A PI controller introduces phase lag at low frequencies: