Neural Network: fill missing data in time series

Question

Tyler Roman 2016-1-12

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/263500-neural-network-fill-missing-data-in-time-series

编辑： KSSV 2018-2-22

I'm trying to use neural network to fill in missing data values in a time series of methane fluxes. I've used an input series with 6 parameters to train the network with the non-missing flux data. I then want to use the network to find the values of the missing flux data, for which I have input data for the original 6 parameters used to train the network. I'm getting an error message saying the 'Number of inputs does not match net.numInputs' which is confusing because both the original and new input data 6 parameters. Here is the section of code I am using:

% set up the data
x = tonndata(input1,false,false);
x2 = tonndata(input2,false,false);
t = tonndata(target1,false,false);
t2=tonndata(target2,false,false);
% design network
net = narxnet;
[X,Xi,Ai,T] = preparets(net,x,{},t);
net = train(net,X,T,Xi,Ai);
% view(net)
% Simulate network
[Y,Xf,Af] = sim(net,X,Xi,Ai);
% Closed Loop Network
netc = closeloop(net,Xf,Af);
% view(netc)
%setting up the second set of input data
[X2,Xi2,Ai2,T2] = preparets(net,x2,{},t2);
[y2,xf,af] = netc(X2,Xi2,Ai2);