11/1/2017 Convert System Model from System Identification to Control Design National Instruments
Convert System Model from System Identification to Control Design
Publish Date: dic 24, 2009
Overview
This document introduces how to convert a model you created in the LabVIEW System Identification Toolkit into a control
design model that you can use with the LabVIEW Control Design and Simulation Module.
Table of Contents
1. System and Model
2. Model Type Defined in Different LabVIEW Toolkits
3. Example – Transfer Function Model
4. Example – ARX Model
5. Summary
1. System and Model
The mathematical model for a dynamic system describes how the outputs of the system respond to inputs, as the following
diagram shows.
The outputs of a dynamic system contain responses to inputs and noise, including system noise and measurement noise. A
general mathematical model consists of the deterministic part and the stochastic part. The deterministic part describes how
outputs respond to inputs. The stochastic part describes the noise characteristics, which is the way that the noise disturbs the
outputs.
A general linear model is defined as the following equation:
Where G defines the model of the deterministic part and H defines the model of the stochastic part.
2. Model Type Defined in Different LabVIEW Toolkits
Models defined in the LabVIEW System Identification Toolkit include determinant and stochastic parts.
The determinant part comes from the measured input u(t) and the stochastic part comes from the noise e(t). Models defined in
the control design field treat the stochastic part as regular input channels. If the estimated models defined in the System
Identification Toolkit are to be used for the control design purpose, these models must be converted to control design models.
The System Identification Toolkit provides the SI Convert to Models of CDT VI to convert a model you obtained in the
LabVIEW System Identification Toolkit. You can convert the model into a transfer function, zeropolegain, or statespace
model that you can use with the LabVIEW Control Design Toolkit.
The measured only? input of the VI specifies whether to ignores the stochastic part e(t) in the conversion.
If the measured only? input is FALSE, this VI treats the noise e(t) as an input of the system.
When the VI converts the stochastic part e(t) to regular input channels, the VI normalizes the noise channels. For a general
linear model is defined as y(t)=Gu(t)+He(t), the noise covariance is defined by cov(e)=LLT, where L is a lower triangular matrix
obtained from the Cholesky factorization of cov(e). The modified general linear model is defined as y(t)=Gu(t)+HLv(t). Its noise
input v(t) is normalized noise as a unit variance, which is cov(v)=I.
The new converted linear model becomes y(t)=[G, HL][u(t), v(t)]T with normalized noise v(t) as a new input channel. During
model simulation, the input noise signal v(t) can be generated using the Gaussian White Noise VI on the Signal Generation
palette with its variance input being set to 1.
3. Example – Transfer Function Model
http://www.ni.com/whitepaper/10764/en/#toc4 1/3
11/1/2017 Convert System Model from System Identification to Control Design National Instruments
The following figure shows the block diagram of a VI that identifies continuoustime transfer function model and converts the
estimated model to the defined model for control design in the LabVIEW Control Design and Simulation Module The identified
continuoustime transfer function model does not contain stochastic part, so the measured only? input is specified as TRUE
in the SI Convert to Models of CDT VI.
The following figure shows the transfer function model identified in the System Identification Toolkit.
The following figure shows the converted transfer function model for control design in the LabVIEW Control Design and
Simulation Module.
4. Example – ARX Model
The following figure shows the block diagram of a VI that identifies ARX model, and converts the estimated model to the
defined model control design in the LabVIEW Control Design and Simulation Module .Identified ARX model contains the
stochastic part. Users can specify the value of the measured only? input of the SI Convert to Models of CDT VI to include or
exclude the stochastic part.
The A order, B order and delay of the ARX model are specified as 2, 1 and 1 respectively. The ARX model identified in system
identification toolkit is displayed as the following figure.
When the measured only? input is set to TRUE, this VI converts the transfer function model without the stochastic part. The
converted model for control design in the LabVIEW Control Design and Simulation Module is displayed as the following figure.
When measured only? input is set to FALSE, this VI converts the transfer function model with the stochastic part. The
converted model for control design in the LabVIEW Control Design and Simulation Module is displayed as following figure. The
stochastic part of the ARX model is treated as a new input channel, so the converted transfer function model becomes a MISO
system with two inputs and one output.
http://www.ni.com/whitepaper/10764/en/#toc4 2/3
11/1/2017 Convert System Model from System Identification to Control Design National Instruments
5. Summary
With the LabVIEW System Identification Toolkit, you can estimate the model of a dynamic system from the inputoutput data.
You then can use the SI Convert to Models of CDT VI to convert a model you obtained in the System Identification Toolkit
into a model that you can use with the LabVIEW Control Design and Simulation Module.
http://www.ni.com/whitepaper/10764/en/#toc4 3/3