I am trying to obtain the response model for resistance spot welding.I have used TANSIG function for hidden layer and PURELIN function for output layer to obtain weight and bias. Now I need to form an equation of the output obtained from ANN.
You can obtain the equation ( a long one!) when you have the weights and biases. I suggest that you read the manual of MATLAB neural network and go through through the examples to get a good grasp of ANN. Here is the part from manual:
For multiple-layer networks the layer number determines the superscript on the weight matrix. The appropriate notation is used in the two-layer tansig/purelin network shown next.
The figures from the manuals demonstrate the sort of equations you will obtain.
Technically speaking, computer science researchers usually adopt Genetic Programming approaches to obtain the explicit equation of an response model. In the case of neural networks, however, you can do so.
You can obtain the equation ( a long one!) when you have the weights and biases. I suggest that you read the manual of MATLAB neural network and go through through the examples to get a good grasp of ANN. Here is the part from manual:
For multiple-layer networks the layer number determines the superscript on the weight matrix. The appropriate notation is used in the two-layer tansig/purelin network shown next.
The figures from the manuals demonstrate the sort of equations you will obtain.
1- if you need the equation out of MATLAB environment, for multi-layer perception you can extract the equation. but for other networks, i do not know.
2- if you need the equation inside of MATLAB environment:
In software like MATLAB, there is a function to calculate outputs from your rained network and there is no need to know weights and biases or other details, to calculate outputs. you should be aware of normalize and denormalize functions.
If you need outputs inside of MATLAB as an input for another calculations, easily use net(input) to calculate outputs.