wavenet

Create a wavelet network nonlinearity estimator object

collapse all in page

Syntax

NL = wavenet
NL = wavenet(Name,Value)

Description

NL = wavenet creates a default wavelet network nonlinearity estimator object for estimating nonlinear ARX and Hammerstein-Wiener models. Use dot notation to customize the object properties, if needed.

NL = wavenet(Name,Value) creates a wavelet network nonlinearity estimator object with properties specified by one or more Name,Value pair arguments. The properties that you do not specify retain their default value.

Object Description

wavenet is an object that stores the wavelet network nonlinearity estimator for estimating nonlinear ARX and Hammerstein-Wiener models.

Use wavenet to define a nonlinear function y=F(x,θ), where y is scalar, x is an m-dimensional row vector of regressors, and θ represent the parameters in wavelet expansion. The wavelet network function is based on the following function expansion:

F(x,θ)=(x−r)PL+as_1f(bs_1((x−r)Q−cs_1))+…                          +as_nsf(bs_ns((x−r)Q−cs_ns))                          +aw_1g(bw_1((x−r)Q−cw_1))+…                          +aw_nwg(bw_nw((x−r)Q−cw_nw))+d

f(z)=e-0.5zzT

g(z) = (m-zzT)e-0.5zzT

Here,

  • f(z) is a radial function called the scaling function, and z is the input to the scaling function. z is a 1-by-q row vector. q is the number of components of x used in the scaling and wavelet functions.

  • g(z) is a radial function called the wavelet function, and z is the input to the wavelet function.

  • θ represents the following parameters of the nonlinearity estimator:

    • P and Q — Projection matrices of size m-by-p and m-by-q, respectively.

      P and Q are determined by principal component analysis of estimation data. Usually, p = m. If the components of x in the estimation data are linearly dependent, then p<m. The number of columns of Q is q. q is the number of components of x used in the scaling and wavelet function.

      When used in a nonlinear ARX model, q is equal to the size of the NonlinearRegressors property of the idnlarx object. When used in a Hammerstein-Wiener model, m=q=1 and Q is a scalar.

    • r – Mean value of the regressor vector computed from estimation data, specified as a 1-by-m vector.

    • as, bs, aw, and bw — Scaling and wavelet parameters, specified as scalars. Parameters with the s subscript are scaling parameters, and parameters with the w subscript are wavelet parameters.

    • L — Specified as a p-by-1 vector.

    • cs and cw – Specified as a 1-by-q vectors.

    • d — Output offset, specified as a scalar.

The value F(x) is computed by evaluate(NL,x), where NL is the wavenet object.

For wavenet object properties, see Properties.

Examples

collapse all

Open Live Script
NL = wavenet;

Exclude the linear term from the wavelet expansion.

NL.LinearTerm = 'off';
Open Live Script

Load the estimation data.

load twotankdata;

Create an iddata object from the estimation data.

z = iddata(y,u,0.2);

Create a wavelet network nonlinearity estimator with 5 units.

NL = wavenet('NumberOfUnits',5);

Estimate the nonlinear ARX model.

sys = nlarx(z,[4 4 1],NL);
Open Live Script

Load the estimation data.

load motorizedcamera;

Create an iddata object.

z = iddata(y,u,0.02,'Name','Motorized Camera','TimeUnit','s');

z is an iddata object with 6 inputs and 2 outputs.

Specify the model orders.

Orders = [ones(2,2),2*ones(2,6),ones(2,6)];

Specify different nonlinearity estimators for each output channel.

NL = [wavenet('NumberOfUnits',2),linear];

Estimate the nonlinear ARX model.

sys = nlarx(z,Orders,NL);
Open Live Script

Load the estimation data.

load motorizedcamera;

Create an iddata object.

z = iddata(y,u,0.02,'Name','Motorized Camera','TimeUnit','s');

z is an iddata object with 6 inputs and 2 outputs.

Specify the model orders and delays.

Orders = [ones(2,6),ones(2,6),ones(2,6)];

Specify the same nonlinearity estimator for each input channel.

InputNL = saturation;

Specify different nonlinearity estimators for each output channel.

 OutputNL = [deadzone,wavenet];

Estimate the Hammerstein-Wiener model.

sys = nlhw(z,Orders,InputNL,OutputNL);

To see the shape of the estimated input and output nonlinearities, plot the nonlinearities.

plot(sys)

Click on the input and output nonlinearity blocks on the top of the plot to see the nonlinearities.

Input Arguments

Name-Value Pair Arguments

Specify optional comma-separated pairs of Name,Value arguments. Name is the argument name and Value is the corresponding value. Name must appear inside quotes. You can specify several name and value pair arguments in any order as Name1,Value1,...,NameN,ValueN.

Use Name,Value arguments to specify additional properties of wavenet nonlinearity. For example, NL= wavenet('NumberofUnits',5) creates a wavelet nonlinearity estimator object with five nonlinearity units in wavelet expansion.

Properties

NumberofUnits

Number of nonlinearity units in wavelet expansion, specified as a positive integer or one of the following values:

  • 'Auto' — The number of units are automatically determined from estimation data.

  • 'Interactive' — The number of units are determined during model estimation.

Default: 'Auto'

LinearTerm

Inclusion of linear term, specified as one of the following values:

  • 'on' — The linear term, (x−r)PL, is included in the wavelet expansion.

  • 'off' — The linear term in not included in the wavelet expansion.

Default: 'on'

Parameters

Parameters in wavelet expansion, specified as a structure with the following fields:

Field NameDescriptionDefault
RegressorMean

Means of the regressors, r, specified as a 1-by-m vector. m is the number of regressors, x.

For Hammerstein-Wiener models, m = 1.

[]
NonLinearSubspace

Projection matrix, Q, specified as an m-by-q matrix.

[]
LinearSubspace

Projection matrix, P, specified as an m-by-p matrix.

[]
LinearCoef

Linear coefficients, L, specified as a p-by-1 vector.

[]
ScalingDilation

Scaling function dilation, bs_ns, specified as an ns-by-1 matrix.

[]
WaveletDilation

Wavelet function dilation, bw_nw, specified as an nw-by-1 matrix.

[]
ScalingTranslation

Scaling function translation, cs_ns, specified as an ns-by-q matrix.

[]
WaveletTranslation

Wavelet function translation, cw_nw, specified as an nw-by-q matrix.

[]
ScalingCoef

Scaling function coefficients, as_ns, specified as an ns-by-1 vector.

[]
WaveletCoef

Wavelet function coefficients, aw_nw, specified as an nw-by-1 vector.

[]
OutputOffset

Output offset, d, specified as a scalar.

[]

The parameters are typically not assigned directly. They are estimated by the identification algorithm (nlarx or nlhw) when wavenet is used in a Nonlinear ARX (idnlarx) or Hammerstein-Wiener (idnlhw) model.

Options

Options specifying the initial wavelet nonlinearity structure, specified as a structure with the following fields:

Field NameDescriptionDefault
FinestCell

Minimum number of data points in the smallest cell, specified as an integer or character vector. A cell is the area covered by the significantly nonzero portion of a wavelet.

'auto' — Compute the value from the data.
MinCells

Minimum number of cells in the partition, specified as an integer.

16
MaxCells

Maximum number of cells in the partition, specified as an integer.

128
MaxLevels

Maximum number of wavelet levels, specified as an integer.

10
DilationStep

Dilation step size, specified as real scalar.

2
TranslationStep

Translation step size, specified as real scalar.

1

Output Arguments

collapse all

Wavelet nonlinearity estimator object, returned as a wavenet object.

Algorithms

wavenet can be used in both Nonlinear ARX and Hammerstein-Wiener models.

  • When used in a Nonlinear ARX model:

    • If the Focus estimation option (see, nlarxOptions) is 'prediction', wavenet uses a fast, noniterative technique for estimating parameters [1]. Successive refinements after the first estimation use an iterative algorithm.

    • If Focus is 'simulation', wavenet uses an iterative technique for estimating parameters.

    To always use a noniterative or iterative algorithm, specify the IterativeWavenet option of nlarxOptions.

  • When used in a Hammerstein-Wiener model, wavenet parameters are determined by iterative minimization.

References

[1] Zhang, Q. “Using wavelet network in nonparametric estimation.” IEEE Trans. on Neural Networks, Vol. 8, Number 2, March 1997, pp. 227-236.

See Also

|

Introduced in R2007a

System Identification Toolbox Documentation

Support