Dear Adel, as far as I understand, you want to use opiimization methods to solve inverse problems for Volterra eq. in Heat conduction.
I am not familiar with the problem, but as a general rule, the choice of algorithm depends on the proof of uniqueness (needs regularization? which data you have? Temper. of thermal gradient?), and the well-posedness of the problem (mostly, on its sensitivity to noise and uncertainties).
If I should attack this problem, I would use some kind of neural network, maybe incorporating fuzzy operator to deal with uncertainty.