Multi-objective Scheduling Optimization Of Short-term Production Plan For Crude Oil Operations

As a non-renewable energy source,crude oil is of great significance to the development of modern economy.Production with limited resources and energy is a problem that refining companies need to solve urgently.The production scheduling plan needs to be rationally arranged,which can not only improve the profits of refining enterprises,but also greatly reduce the emission of greenhouse gases,and help refining enterprises become environmentally friendly.In general,crude oil short-term scheduling is divided into three levels: production planning,production scheduling and process control,which respectively implement the decision-making on different time scales.Among them,the primary processing of crude oil in the production scheduling layer usually has a huge production scale.At the same time,it contains a lot of constraints and complex production process.At present,there is still no mature software or tools,detaild short-term production plans need to be manually generated by experienced decision makers.As a result,it often takes a long time to make a production plan,and requires high professional skills for decision makers.Moreover,after the brain drain,refining companies need to spend a lot of time to train new decision makers.In order to solve this problem,many scholars use mathematical programming methods to model it.By such methods,the main issue is how to model the time and it can be represented as either discrete one or continuous one.However,as the scale of the problem becomes huge,these methods are difficult to solve because of the large number of decision variables.Therefore,in order to simplify the problem,some scholars divide the entire system into two levels from the perspective of control theory.In the upper level,a linear programming model was proposed to maximize the production efficiency and the refining plan of each distiller is obtained.In the lower level,a detailed production scheduling is generated to achieve these refining plans.At the same time,in order to obtain the schedulable conditions of the system in different states and the feasibility of the solution,the problem is modeled and analyzed using Petri net.In order to reduce the multiple costs incurred during the primary processing of crude oil,based on the above two-level solution model,intelligent search algorithms are used to solve the problem.First of all,in order to optimize the four objectives,including the cost of oil tankers to use ports,the cost of tankers switching oil tanks,the cost of using oil tanks in the port,and the crude oil mixing cost at the bottom of the oil tanks,the crude oil unloading decision is made by assigning the storage tank in the portto meet the pipline task queue.At the same time,the multi-objective genetic algorithm with elite strategy is used to search the solution space to obtain an oprimized non-dominated solution set.An industrial example is used for experimental analysis and good results are obtained;Secondly,in order to find a better non-dominated solution set,a dual-population co-evolutionary algorithm based on the improved backbone particle swarm algorithm and the multi-objective genetic algorithm with elite strategy is proposed.The communication of the two populations is controlled by Pareto difference entropy,which improves the diversity and convergence performance of the algorithm.As a result,the four objectives were oprimized,including the cost of the fuel supply tank,the switching cost of the fuel supply tank,the cost of crude oil mixing in the pipeline,and the blending cost of the fuel tank bottom.Finally,through an industrial example,the algorithm is compared with several representative multi-objective evolutionary algorithms to verify the feasibility and effectiveness of the algorithm.