Computational Intelligent Method Investigation And Application In The Process Industry.

Process simulation and optimization are two important and high associated issues in petroleum and chemical engineering enterprise. They can bring tremendous economic benefit with their improvement. In the multi variable, strong coupling large-scale, complication and modernization of process industry such as petroleum and chemical engineering, taking single control method and optimization means can't meet production requirement. With multi subjects crossover and penetrating each other, the requirement of efficient optimization means and intelligent computing is urgent progressive. Since industrial process objective is a complicable, restricted, nonlinear, multioptimum, modeling difficulty, searching feasible intelligence algorithm becomes the main investigable targets. In this paper, some intelligent algorithms are investigated aiming at the problem of process simulation and optimization.Since ANN has the disadvantages on convergence speed, a fuzzy Particle Swarm Optimization (PSO) and Artificial Neural Network model is investigated. The improved approach is applied successfully to the foundation of the soft measurement model in cracking furnace and PTA solventdehydration. From both theoretical computing and practical application, the validity and reliability of proposed algorithm are verified by referrence case and actrul industrial case. Aiming at optimization problem with more than one competing objective, multi particle swarm optimization based on fuzzy Pareto is investigated. Multi PSO based on fuzzy Pareto is easy to be implemented. The algorithm has strong general search ability and robust capability. In this algorithm, Pareto solutions distribute uniformly, approach to Pareto front, and extend the range of Pareto solutions. The performance of Pareto solution is improved. The posterior fuzzy evaluation is applied to obtain objective preference fuzzy information and satisfying solutions. Fuzzy multi-objective particle swarm optimization based on Pareto is applied to the optimization of the ethylene cracking furnace.Ant colony optimization (ACO) algorithm' parameter based on evolutionary mechanism is studied in this paper to improve the performance of the solution of ACO. In this approach, parameters of the traditional ACO are taken as attributes of every single ant, and the population of ant is chosen by evolutionary operator. The results show that the algorithm improves solution performance.To decrease the loss of acetic acid in PTA solvent dehydration system, the dynamic characteristic of azeotropic distillation is investigated. Dynamic model of industrial Purified Terephthalic Acid (called PTA) solvent dehydration process is investigated to realize and characterize the process in this work. The dynamic behavior of an azeotropic distillation column separating acetic acid and water using n-butyl acetate as the entrainer is extensively studied using this model. Responses of the column to feed flow rate and aqueous reflux rate are simulated. The movement temperature front is simulated. The comparison results between industrial value and simulation value make clear that the model and algorithm are particularly effective. On the basis of simulation and analysis, optimization strategy can be implemented effectively. Ant colony based on evolutionary and fuzzy multi-objective particle swarm optimization based on Pareto are applied to investigation of the loss of acetate acid in PTA solvent dehydration industry. The results make clear that ant colony based on evolutionary offer the optimization value to industritial process and fuzzy multi-objective particle swarm optimization offer the optional range of the decision variable. It is obvious that the improved algorithm has significant reference to solve the loss problem of the acetate acid in PTA solvent dehydration industry. It is noted from results that intelligence algorithm has a promising application in the process industry.