How do i normalize my data for logsig transfer function?

I have 3 inputs in a 3 row matrix 1 output in a single row matrix

% code
InputData =
 1.0e+03 *
  0.0254    0.0254    0.0254    0.0254    0.0253    0.0253    0.0253    0.0252    0.0252
  1.0081    1.0080    1.0080    1.0080    1.0080    1.0080    1.0079    1.0079    1.0079
  0.0843    0.0843    0.0843    0.0841    0.0843    0.0844    0.0844    0.0844    0.0846
OutputData =
-76.9479  -76.8323  -76.7923  -76.7183  -76.6507  -76.6531  -76.6235  -76.6654  -76.7230
end

How do I normalize my input and output data to suit 3 different pairs of activation functions (purelin & logsig, logsig & logsig, tansig & logsig)? In each pair, the first activation function is for the hidden layer and the second is for the output layer.