Predicting values from neural network fitting

Using the Neural Net Fitting application I have trained, tested, and validated an ANN that fits my data very well. This neural network has 2 hidden layers. My end goal is to have mathematical model in MATLAB that can represent the results of the neural network. My attempt so far is to use the resultant bias and weights in the following equation:

h = tansig(IW1_1*x + b1)
y = purelin (LW2_1*h + b2)

I would like to input a known value of x from the original data and have the model return the appropriate y value. My understanding of this concept breaks down between the neural network world of biases and weights and how that correlates to the typical mathematical world of y=f(x).