The Process Parameter Optimization Of Nitric Acid Production

Off-line steady optimization of nitric acid process unit is discussed in this thesis. In the research process, the author first penetrates this factory understanding production craft, has carried on mass data gathering and the analysis, and unified this factory actual production experience, in view of the nitric acid production installment movement condition, had determined the main controllable independent variable collection and needs to optimize objective function. And the determination entire optimization process divides into the modeling and seeks the superior two parts to complete.The nitric acid production process optimization question modeling uses the artificial neural network to complete. According to the actual situation, after has chosen the appropriate network architecture and the algorithm, the use gathers 50 groups of actual production data, train a BP neural network, after the network training had ended, uses other 50 groups of production data to carry on the test to this network model, and performs to revise, finally obtains the nitric acid production process parameter optimization question the model, and has guaranteed this network model accuracy.After finishing modeling, in order to choose appropriate optimization algorithm, a mass of research work on different optimization algorithm was done, in addition, aiming at eliminating the noise influence in the optimization, a new performance rule was put forward in this paper, that is, average performance rule of adjacent region in parameter optimization. If combine optimization algorithm with this performance rule, the filter and optimization search can be accomplished at the same time. To testing the feasibility of this method, a mass of simulation data was choose to finish actual calculate. The results indicate that the method is efficient. According to the simulation results, the improved PSO (particle swarm optimization) was selected to combine with average performance rule to solve the optimization problem.Finally, guaranteed the neural network model can accurately reflect the system movement operating mode, the PSO was combined with average performance rule tocomplete computation which optimizes to this nitric acid production process parameter Comparison with the actual production outcome, the optimization result is satisfied.