Initial State Dynamical System LSTM Network

4 次查看（过去 30 天）

Michael Hesse 2020-11-18

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/652753-initial-state-dynamical-system-lstm-network

评论： Michael Hesse 2020-11-19

%% - cleanup
clear;
close all;
clc;
%% - data
t = linspace(0, 5, 1000);
odefcn = @(t, x) [x(2, :); 10*sin(x(1, :))-x(2, :)];
x0 = [pi/2, 0]';
[~, x] = ode45(odefcn, t, x0);
x = x';
X = x(:, 1:end-1);
Y = x(:, 2:end);
%% - define and train lstm network
numFeatures = 2;
numResponses = 2;
numHiddenUnits = 200;
layers = [sequenceInputLayer(numFeatures);
          lstmLayer(numHiddenUnits);
          fullyConnectedLayer(numResponses);
          regressionLayer];
opts = trainingOptions('adam', 'MaxEpochs', 100, 'Plots', 'training-progress');
net = trainNetwork(X, Y, layers, opts);
%% - prediction
net = resetState(net);
xpred = x0;
for i = 1 : length(t)-1
    [net, xpred(:, i+1)] = predictAndUpdateState(net, xpred(:, i));
end
%% - plotting
figure(1);
plot(t, x);
hold on;
grid on;
plot(t, xpred, '--');

This is an example code where I want to predict the trajectory of a pendulum via LSTM neural network. How can I provide the initial state x0 into the network? If you look at the figure the second state directly jumps from x0 to the state [0, 0]'. Why does this happen?

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Michael Hesse 2020-11-19

在 MATLAB Online 中打开

Here is a possible workaround. Instead of learning the next state, one can learn the difference to the next state.

%% - cleanup
clear;
close all;
clc;
%% - data
t = linspace(0, 5, 1000);
odefcn = @(t, x) [x(2, :); 10*sin(x(1, :))-x(2, :)];
x0 = [pi/2, 0]';
[~, x] = ode45(odefcn, t, x0);
x = x';
X = x(:, 1:end-1);
Y = x(:, 2:end) - x(:, 1:end-1);
%% - define and train lstm network
numFeatures = 2;
numResponses = 2;
numHiddenUnits = 200;
layers = [sequenceInputLayer(numFeatures, 'Normalization', 'zscore');
          lstmLayer(numHiddenUnits);
          fullyConnectedLayer(numResponses);
          regressionLayer];
opts = trainingOptions('adam', 'MaxEpochs', 100, 'Plots', 'training-progress');
net = trainNetwork(X, Y, layers, opts);
%% - prediction
xpred = x0;
for i = 1 : length(t)-1
    [net, dxpred] = predictAndUpdateState(net, xpred(:, i));
    xpred(:, i+1) = xpred(:, i) + dxpred;
end
%% - plotting
figure(1);
plot(t, x);
hold on;
grid on;
plot(t, xpred, '--');