I want to train a dataset for detection of CT saturation using Neural Networks/ SVMS.
Check this link, it is a very nice tutorial
http://lab.fs.uni-lj.si/lasin/wp/IMIT_files/neural/NN-examples.pdf
@S.Anbazhagan....I know about the tools.Question is:how can i fit my data in this tool?e.g i have multiple signals as input which are trained to predict the required output signal from test data.
First of you create training input matrix, each column represents your input signals.
Next, you create training target matrix associated with your input signal matrix
Similarly
@S.Anbazhagan.Thank you for your reply.
Is there any code for it? or Neural Network tool will do this for me?
the code is very simple, just read each line, and add it to your X.
X=[];
for i=1:num_datapoints
%%% load data line here, and the truth labels (0/1) for binary classification
X=[X,data(:,num_datapoints)];
Y=[Y,labels(num_datapoints)];
end
if you specify how your data is stored, u can modify the above code accordingly
also make sure you split your dataset into training, validation (for trying different parameters such as learning rate) and test, on which you check your accuracy.
Theoretically
1.Convergence based on back-propagation learning
2. Simulated annealing (in matlab has the function too) with bounds given.
https://www.mathworks.com/discovery/simulated-annealing.html
@S. Anbazhagan ..when i change CT parameters for getting different training patterns i-e Changing following conditions at each simulation: type of fault, fault resistance,
fault location,. Then i obtained different current and voltage wave forms at each simulation.Now how can i inputs all these different current and voltage wave forms in Matrix form to feed as input and target in ANN?
- Badges
- Science topic