In this paper,an extreme optimization method of neural network genetic algorithm(NN-GA)is used to design a model of cell voltage and optimize it,and so as to reduce the production cost of electrolytic aluminum.First,a method of kernel principal component is proposed to analysis the parameters in the process of electrolytic aluminum.And then,the method of Extreme Learning Machine Genetic Algorithm(ELM-GA)and Back Propagation neural network Genetic Algorithm(BP-GA)is applied to model and optimize cell voltage.Through the practical data to qualify,the optimal production cell voltage and the corresponding production conditions were obtained.The results show the optimization effect on ELM-GA is better than that of BP-GA.By using the ELM-GA to optimize cell voltage,the small cell voltage is obtained and the purpose of energy saving can be achieved.