Research on modeling method of coke oven´s blast blower suction based on improved BP neural network

For the complicated mechanism and difficult traditional modeling of the blast blower cooling system of coke oven, a new dynamic mathematic model for the fore end suction of blast blower cooling system has been firstly established. This model avoids the complex process of analysis on blast blower cooling system, makes full use of the existing field data, and traverses to all sections of the production process. Then artificial bee colony (ABC) algorithm is used to optimize the BP neural network´s structure parameters and an analysis is conducted on different type of input variables. Finally, a validation for the established model is performed with the test samples that collected from different process sections, and simulation results show that the obtained model has a faster convergence and higher approximation accuracy.