| As a prosperously developing country,China makes products and consumes energy in top several countries of the world.It is also an important problem of the polluted and even destroyed environment in China.In China,the industry sector takes up the most energy consumption.Meanwhile,the iron and steel,the chemical and petrochemical,and the non-metallic mineral sub-sectors consume the most energy.Ethylene cracking process in the chemical and petrochemical industry is an important production in national economy.Ethylene is one of the most produced chemical products and consumes more than half of the energy in the industry.It is essential for it to improve the production efficiency,energy saving and emission reduction.Single objective optimization is no longer enough for needs of various developments.The multi-objective optimization tries to coordinate multiple objectives,which are often contradicted with each other.As usual,the optimization of one objective may lead to the weakening of another objective or other objectives.Therefore,there is need for multi-objective optimization to find methods for coordinating objectives without transferring them into one single objective.The main research contents and finding are summarized as follows:1)Concerning the multi-objective optimization problem,in order to improve the exploration and exploitation abilities of intelligent optimization algorithms,find global optimization solutions,researches about the membrane algorithm are made.Making use of the distributed structure and maximally parallel computation feasibilities,the membrane computing incorporates the selection,crossover and mutation rules for evolving objects.The competing communication process is proposed for objects communication between hierarchical membranes.Communicated objects are archived in the skin membrane as global elitists,and are selected again and communicated back into inner membranes,serving as guidance for the next iteration evolution.The evolutionary membrane algorithm based on competing communication is verified by several classical test functions,with the found solutions converged to the real Pareto front and distributed uniformly as close as possible to the real ones.2)Concerning the many-objective optimization problem with more than three objectives,as the increasing number of objectives makes the number of non-dominated solutions increase exponentially,the algorithm selection pressure is reduced.Thus,the algorithm may not be able to find the solutions and the difficulty of solving the problem is increased.The selection strategy based on the reference point,instead of the crowding distance,is introduced into the membrane algorithm,as well as the simulated binary crossover and polynomial mutation.On the other hand,the domination mechanism in the competing communication for deleting dominated solutions are adopted as the ?-dominance mechanism.Conducting experiments of the proposed many-objective optimization membrane algorithm on the extensible test functions with different numbers of objectives and decision variables,the optimization results demonstrate that the solutions in the objective space are converged to each real objective value,covering regions of real objective values uniformly and distributed well for different objectives.All the extreme values of each objective can be reached but not reached simultaneously for all objectives,due to objectives'conflicts.3)Further to help the decision making process,a multi-atrribute decision making method based on the grey correlation analysis and information gain is applied to make the final decision from the solution set.Wherein,the attributes'weights are decided by the difference mean square method,based on the difference between the data of different solutions and different attributes.The solution set is sorted by the relative closeness,calculated by the grey correlation degree and information gain between each solution and the desired solution,and the undesired solution.The experiment results of test cases prove the effectiveness and rationality of the proposed decision making method.4)Concerning the production optimization of the ethylene cracking process,the proposed algorithm is applied to solve the constrained multi-objective optimization of the ethylene cracking process for maximization of ethylene yields,propylene yields and the cracking period.The constraints are under strict control for deleting candidate solutions that are not met with.The operation variables in the cracking process that are important for production are selected as decision variables.The multi-objective optimization results demonstrate a set of solutions distributed uniformly and converged in the reasonable ranges of the real-world problem.Decision makers are able to make selection based on preference ordering or previous experience.The optimization results also reduce the use of the feed oil as well as the emission of carbon dioxide.Applying the membrane algorithm based on the reference point selection to the many-objective optimization of the ethylene cracking process,in order to realize the production,energy efficiency and environment protection goals.Based on the multi-objective optimization model,the maximization of the butadiene yield is added.The optimization result shows that the algorithm is able to find solutions located in the reasonable ranges of realistic production,and the Pareo front in the objective space covered by the solutions are converged and distributed uniformly for all objectives.The many-objective optimization results of the ethylene cracking process are able to achieve simultaneous optimization of multiple product yields,as well as realizing energy saving and emission reduction.Applying the decision making method to the multi-objective and many-objective optimization results of the ethylene cracking process,the solutions selected from the closessness sorting demonstrate good trade-off between optimized objectives.The amount of energy saving and emission reduction,as well as the yields of products of ethylene,propylene,or with butadiene,are all above the average values of the whole solution set,serving as valuable guidance for decision making.The trends of product yields of the decision making solution along with the running time(determined by the cracking period),or along the tube are basically in accordance with the trends under the initial condition.It demonstrates again that the optimal solution obtained by the proposed optimization algorithm,and decided by the decision making method are effective and reasonable for the ethylene cracking process.Moreover,the solution produces optimal yields of ethylene,propylene and butadiene,as well as realizes energy saving and emission reduction. |
PDF Full Text Request