An intelligent cross-process prediction model integrating the mechanistic model and LSTM

The strip shape control during the temper rolling process after continuous annealing is affected by various factors such as the upstream flatness and annealing temperature. The control accuracy of the traditional mechanistic model fails to meet the requirements. In response to this, this paper established a cross-process strip shape presetting method that integrated a mechanistic model and an intelligent model. Based on the finite element simulation technology, the core factors influencing the strip shape are simulated and calculated. Furthermore, the Long Short-Term Memory (LSTM) algorithm is adopted to establish a prediction model between the upstream processes and roll bending, which makes up for the deficiencies of the traditional mechanistic model.