Abstract: In this paper,we use several methods like the gray system model and linear regression model to predict each evaluation of the manufacturing industrial security,then combine the predictive value with historical data to establish a three-ply BP neural network. On this basis,the future status of our manufacturing industrial security system is alarmed in advance.Early warning results of the BP neural network show that:our manufacturing industrial security will keep in good condition in future,we only need to guard against it mildly.