Hybrid neural network (HNN) has received great attention especially for modeling a nonlinear system. Hence, HNN for modeling and optimization for determining optimum temperature profiles with maximal biogas production was studied basing on mathematical models initially describing the anaerobic digestion in biogas production process. The experiment of a 30-day process was conducted using sediments from shrimp ponds to simulate the daily performance of the HNN consisting of two-hidden layers with 7 and 9 nodes and the optimization of ambient temperature with satisfactory results obtained: a significant higher yield at 3.30 times compared with conventional methods.