Optimization of goal programming algorithm of liquor blending based on neural network

As to liquor blending process, the existed blending technique based on goal programming algorithm was difficult to determine the weighting coefficients (priority factors) . So artificial neural network was utilized to revise the goal programming algorithm and a three-layered forward BP neural network structure was adopted, then the structure was trained through appropriate training samples which were between physicochemical index vector and priority factor vector. As a consequence, the optimal priority factors were gained, and then were applied to the formula model, resulting in liquor blending optimal formula solution. Last, simulation results showed that the neural network optimization algorithm was fast, convergent and feasible, which could get a less 5% cost and more precise formula whose physicochemical index curve was more close to the target one. Therefore, the optimization algorithm based on neural network could be effectively applied to liquor blending process, and be able to meet the multi-objective optimal formula.