Mercurial > hg > camir-aes2014

annotate toolboxes/FullBNT-1.0.7/bnt/examples/dynamic/ho1.m @ 0:e9a9cd732c1e tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
first hg version after svn
author wolffd
date Tue, 10 Feb 2015 15:05:51 +0000
parents
children
rev   line source
wolffd@0 1 function ho1()
wolffd@0 2
wolffd@0 3 % Example of how to create a higher order DBN
wolffd@0 4 % Written by Rainer Deventer <deventer@informatik.uni-erlangen.de> 3/28/03
wolffd@0 5
wolffd@0 6 bnet = createBNetNL();
wolffd@0 7
wolffd@0 8 %%%%%%%%%%%%
wolffd@0 9
wolffd@0 10
wolffd@0 11 function bnet = createBNetNL(varargin)
wolffd@0 12 % Generate a Bayesian network, which is able to model nonlinearities at
wolffd@0 13 % the input. The only input is the order of the dynamic system. If this
wolffd@0 14 % parameter is missing, the an order of two is assumed
wolffd@0 15 if nargin > 0
wolffd@0 16 order = varargin{1}
wolffd@0 17 else
wolffd@0 18 order = 2;
wolffd@0 19 end
wolffd@0 20
wolffd@0 21 ss = 6; % For each time slice the following nodes are modeled
wolffd@0 22 % ud(t_k) Discrete node, which decides whether saturation is reached.
wolffd@0 23 % Node number 2
wolffd@0 24 % uv(t_k) Visible input node with node number 2
wolffd@0 25 % uh(t_k) Hidden input node with node number 3
wolffd@0 26 % y(t_k) Modeled output, Number 4
wolffd@0 27 % z(t_k) Disturbing variable, number 5
wolffd@0 28 % q(t_k), number6 6
wolffd@0 29
wolffd@0 30 intra = zeros(ss,ss);
wolffd@0 31 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0 32 % Within each timeslice ud(t_k) is connected with uv(t_k) and uh(t_k) %
wolffd@0 33 % This part is used to model saturation %
wolffd@0 34 % A connection from uv(t_k) to uh(t_k) is omitted %
wolffd@0 35 % Additionally y(t_k) is connected with q(t_k). To model the disturbing%
wolffd@0 36 % value z(t_k) is connected with q(t_k). %
wolffd@0 37 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0 38 intra(1,2:3) = 1; % Connections ud(t_k) -> uv(t_k) and ud(t_k) -> uh(t_k)
wolffd@0 39 intra(4:5,6) = 1; % Connectios y(t_k) -> q(t_k) and z(t_k) -> q(t_k)
wolffd@0 40
wolffd@0 41
wolffd@0 42
wolffd@0 43 inter = zeros(ss,ss,order);
wolffd@0 44 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0 45 % The Markov assumption is not met as connections from time slice t to t+2 %
wolffd@0 46 % exist. %
wolffd@0 47 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0 48 for i = 1:order
wolffd@0 49 if i == 1
wolffd@0 50 inter(1,1,i) = 1; %Connect the discrete nodes. This is necessary to improve
wolffd@0 51 %the disturbing reaction
wolffd@0 52 inter(3,4,i) = 1; %Connect uh(t_{k-1}) with y(t_k)
wolffd@0 53 inter(4,4,i) = 1; %Connect y(t_{k-1}) with y(t_k)
wolffd@0 54 inter(5,5,i) = 1; %Connect z(t_{k-1}) with z(t_k)
wolffd@0 55 else
wolffd@0 56 inter(3,4,i) = 1; %Connect uh(t_{k-i}) with y(t_k)
wolffd@0 57 inter(4,4,i) = 1; %Connect y(t_{k-i}) with y(t_k)
wolffd@0 58 end
wolffd@0 59 end
wolffd@0 60
wolffd@0 61 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0 62 % Define the dimensions of the discrete nodes. Node 1 has two states %
wolffd@0 63 % 1 = lower saturation reached %
wolffd@0 64 % 2 = Upper saturation reached %
wolffd@0 65 % Values in between are model by probabilities between 0 and 1 %
wolffd@0 66 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0 67 node_sizes = ones(1,ss);
wolffd@0 68 node_sizes(1) = 2;
wolffd@0 69 dnodes = [1];
wolffd@0 70
wolffd@0 71 eclass = [1:6;7 2:3 8 9 6;7 2:3 10 11 6];
wolffd@0 72 bnet = mk_higher_order_dbn(intra,inter,node_sizes,...
wolffd@0 73 'discrete',dnodes,...
wolffd@0 74 'eclass',eclass);
wolffd@0 75
wolffd@0 76 cov_high = 400;
wolffd@0 77 cov_low = 0.01;
wolffd@0 78 weight1 = randn(1,1);
wolffd@0 79 weight2 = randn(1,1);
wolffd@0 80 weight3 = randn(1,1);
wolffd@0 81 weight4 = randn(1,1);
wolffd@0 82
wolffd@0 83 numOfNodes = 5 + order;
wolffd@0 84 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0 85 % Nodes of the first time-slice %
wolffd@0 86 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0 87 % Discrete input node,
wolffd@0 88 bnet.CPD{1} = tabular_CPD(bnet,1,'CPT',[1/2 1/2],'adjustable',0);
wolffd@0 89
wolffd@0 90
wolffd@0 91 % Modeled visible input
wolffd@0 92 bnet.CPD{2} = gaussian_CPD(bnet,2,'mean',[0 10],'clamp_mean',1,...
wolffd@0 93 'cov',[10 10],'clamp_cov',1);
wolffd@0 94
wolffd@0 95 % Modeled hidden input
wolffd@0 96 bnet.CPD{3} = gaussian_CPD(bnet,3,'mean',[0, 10],'clamp_mean',1,...
wolffd@0 97 'cov',[0.1 0.1],'clamp_cov',1);
wolffd@0 98
wolffd@0 99 % Modeled output in the first timeslice, thus there are no parents
wolffd@0 100 % Usuallz the output nodes get a low covariance. But in the first
wolffd@0 101 % time-slice a prediction of the output is not possible due to
wolffd@0 102 % missing information
wolffd@0 103 bnet.CPD{4} = gaussian_CPD(bnet,4,'mean',0,'clamp_mean',1,...
wolffd@0 104 'cov',cov_high,'clamp_cov',1);
wolffd@0 105
wolffd@0 106 %Disturbance
wolffd@0 107 bnet.CPD{5} = gaussian_CPD(bnet,5,'mean',0,...
wolffd@0 108 'cov',[4],...
wolffd@0 109 'clamp_mean',1,...
wolffd@0 110 'clamp_cov',1);
wolffd@0 111
wolffd@0 112 %Observed output.
wolffd@0 113 bnet.CPD{6} = gaussian_CPD(bnet,6,'mean',0,...
wolffd@0 114 'clamp_mean',1,...
wolffd@0 115 'cov',cov_low,'clamp_cov',1,...
wolffd@0 116 'weights',[1 1],'clamp_weights',1);
wolffd@0 117
wolffd@0 118 % Discrete node at second time slice
wolffd@0 119 bnet.CPD{7} = tabular_CPD(bnet,7,'CPT',[0.6 0.4 0.4 0.6],'adjustable',0);
wolffd@0 120 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0 121 % Node for the model output %
wolffd@0 122 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0 123 bnet.CPD{8} = gaussian_CPD(bnet,10,'mean',0,...
wolffd@0 124 'cov',cov_high,...
wolffd@0 125 'clamp_mean',1,...
wolffd@0 126 'clamp_cov',1);
wolffd@0 127 % 'weights',[0.0791 0.9578]);
wolffd@0 128
wolffd@0 129
wolffd@0 130 %%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0 131 % Node for the disturbance %
wolffd@0 132 %%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0 133 bnet.CPD{9} = gaussian_CPD(bnet,11,'mean',0,'clamp_mean',1,...
wolffd@0 134 'cov',[4],'clamp_cov',1,...
wolffd@0 135 'weights',[1],'clamp_weights',1);
wolffd@0 136
wolffd@0 137 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0 138 % Node for the model output %
wolffd@0 139 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0 140 bnet.CPD{10} = gaussian_CPD(bnet,16,'mean',0,'clamp_mean',1,...
wolffd@0 141 'cov',cov_low,'clamp_cov',1);
wolffd@0 142 % 'weights',[0.0188 -0.0067 0.0791 0.9578]);
wolffd@0 143
wolffd@0 144
wolffd@0 145
wolffd@0 146
wolffd@0 147 %%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0 148 % Node for the disturbance %
wolffd@0 149 %%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0 150 bnet.CPD{11} = gaussian_CPD(bnet,17,'mean',0,'clamp_mean',1,...
wolffd@0 151 'cov',[0.2],'clamp_cov',1,...
wolffd@0 152 'weights',[1],'clamp_weights',1);
wolffd@0 153
wolffd@0 154
wolffd@0 155
wolffd@0 156