Abstract: This article has established one model of electrobath' s output-rate based on BP neural networks by taking working current, working voltage and unit time' s weight of adding raw-material as networks' input, the ratio of output and input of the metal as networks' output and large amount of the working data on the spot as the training sample. The research indicates this model can not only accurately forecast the relation between various quantities and the output-rate, but also provide new argument for increasing the electrolysis product and enhancing the production efficiency.