Optimization Research On Rough Rolling Process Of H-Beam Based On ANN And GA

H-beam is one kind of sectional steel with optimized area, reasonable ratio of weight to strength, and has widespread application in national economic construction because of its strong anti-bending ability, simple construction, low cost and small quality. In recent years, the demand of H-beam is increasing, and the quality requirement is improving. In order to improve quality and production efficiency, reduce production energy consumption, and gain greater economic benefits, it is essential to optimize rolling process parameters.This research chooses the topic of optimizing rough rolling process parameters of H-beam. Neural network algorithm is used to build mathematical modeling and genetic algorithm is used to optimize process parameters. And then, rolling energy consumption, quality and production efficiency after optimization were compared with the results before optimization.First, using traditional method to forecasting rolling energy consumption and product quality has some shortages, so neural network algorithm is used to establish forecasting models. On the basis of that method, two neural network models were established, with the research parameters which are the beginning rolling temperature, rolling speed, interval time, and press-down quantity and the output parameters which were rolling power, austenitic grain diameter. The training samples are obtained from the finite element software simulation, and in order to reduce the quantity of training samples and improve the accuracy of the models, process parameters are designed according to mix orthogonal test. The corresponding computer programs are wrote with MATLAB and trained many times to obtain the neural network model with excellent performance. After that, it proved that using neural network algorithm to build mathematical models is feasible and accurate after comparing the simulation value with FEM and the calculation value with neural network.Based on these models, genetic algorithm is used to optimize the rolling process, and there are three optimizing targets, rolling energy consumption, product quality, and production efficiency, which need to be changed into single target for the purpose of looking for the non-inferior solution. How the mutation rate, the crossover rate and the number of population influence the optimizing results is analyzed. From the results contrast, it can be seen that the process parameters after optimization is more reasonable, because the rolling energy is reduced, and the quality of H-beam is improved.Finally, Visual C++ software is chosen as the development platform to get the interactive interface which includes models establishing and optimization calculation. Optimal rolling parameters can be gained automatically and be showed in the interface.This research has important theoretical and realistic significance in improving the existing rolling process of H-beam.