Artificial neural network combined with genetic algorithm were used to model and optimize the preparation process that could orientate the preparation of chickpea protein Maillard Reaction flavor peptide. Chickpea protein hydrolysates （CPHs） with different degrees of hydrolysis （DHs） were prepared with the method of double-enzyme digestion technology. After having done the Maillard Reaction with hydrolysates， Maillard peptide produced was evaluated by descriptive sensory analysis. The proposed parameters were used to build Maillard Reaction prediction model. After the all program of the GA-BP networks， the optimized parameters were： the hydrolysis degree 24%， the initial reaction temperature 85 ℃， reaction time 31 min， initial pH 9.7 and D（+）-xylose 1.2 g. Under these optimal parameters the MRPs’ sensory score was approximately 5.0 and the sensory evaluation was excellent. The further experiment confirmed that those MRPs presented an obvious meaty and umami taste with a long duration. Thus the optimized reaction conditions could be considered as a thermal optimum reaction conditions to produce the MRPs. And the research results could provide theoretical basis and method reference for the development of high quality vegetable protein fresher seasoning.