Optimization method of ethylene cracking furnace production objectives using multi-objective particle swarm optimization algorithm

Abstract

The present invention discloses a method for optimizing the production objectives of an ethylene cracking furnace using a multi-objective particle swarm optimization algorithm. First, the positions of particles are updated through quantum behavior, thereby reducing the possibility of the algorithm falling into a local optimal solution and improving the convergence speed of the algorithm. Then, in order to prevent the algorithm from converging too quickly and thus missing certain optimal solutions, a search strategy of dynamically switching between global search and local search is introduced to further improve the distribution of the algorithm. The algorithm is then verified using a standard test function to achieve optimization of the production objectives of the ethylene cracking furnace, solving the problem of inaccuracy in the optimization of traditional multi-objective optimization algorithms. The present invention effectively optimizes the ethylene yield and propylene yield targets in the ethylene cracking furnace, achieves energy conservation and emission reduction of the ethylene cracking furnace, and thus helps to achieve the "dual carbon" goals.

Description

Technical Field

[0001] The present invention relates to the technical field of ethylene production, and in particular to a method for optimizing production targets of an ethylene cracking furnace using a multi-objective particle swarm optimization algorithm. Background Art

[0002] The petrochemical industry (abbreviated as petrochemical) is an industry that uses petroleum and natural gas resources to produce chemical products through various chemical and physical processes. The main production processes of petrochemicals include cracking, gasification, separation, synthesis and polymerization. Among them, cracking and separation are the most basic production processes for producing basic raw materials such as ethylene. Petrochemicals are an important raw material industry and a basic industry of the national economy. It provides a large amount of chemical raw materials for industry, agriculture, transportation and national defense construction, and is directly related to the development of the entire national economy. Ethylene is the most important basic raw material for petrochemicals. In 2021, Sinopec's ethylene production and carbon emission reduction were 13.38 million tons and 2.38 million respectively. Figure 1 This is a schematic diagram of the traditional cracking furnace structure. Figure 2 Figure 2 is a schematic diagram of a traditional reaction tube structure. Figure 1 and Figure 2 As shown, the ethylene cracking furnace is the key equipment in the entire production process and also the equipment that consumes the most energy in the entire production process. Therefore, in order to improve the production efficiency of ethylene while reducing the impact on the environment, it is particularly important to carry out technological innovation and operation optimization of the ethylene cracking furnace.

[0003] Domestic experts have proposed various approaches to optimizing ethylene cracking furnaces, including using data-driven mixed combustion optimization strategies to address combustion optimization in ethylene cracking furnaces; and optimizing feedstock or individual components by adjusting the feed composition and optimizing the unit's feeding pattern. However, these approaches are complex and limited in their general applicability. Others have utilized swarm intelligence optimization algorithms, including the NSGA-II algorithm and the Grey Wolf Optimization Algorithm. However, these algorithms could be further improved in terms of convergence speed, convergence performance, and distribution. To better optimize ethylene cracking furnace production targets, this paper proposes a multi-objective particle swarm optimization algorithm based on quantum behavior. Particle position updates rely solely on position information and a constructed potential well, significantly improving the swarm's convergence speed and preventing particles from falling into local optima. Furthermore, to prevent excessive convergence from missing optimal solutions, a dynamic switching strategy between global and local search is introduced. For example, global search is primarily performed in the early stages of algorithm iterations, while local search is primarily performed in the later stages. Finally, the algorithm was used to optimize the production of an ethylene cracking furnace, with ethylene yield and propylene yield as optimization targets. Experimental results showed that compared with the original operating conditions, fixed-cycle adjustment of operating conditions achieved very good results in optimizing the targets.

[0004] In recent years, my country's ethylene industry has experienced rapid growth, making it the world's second-largest ethylene producer after the United States. By the end of 2021, my country had 61 ethylene production companies and 79 operational ethylene plants, with a combined annual production capacity of 41.68 million tons, accounting for approximately 18% of the global total. These included 41 steam cracking ethylene plants (including heavy oil catalytic thermal cracking) with a production capacity of 29.48 million tons; 27 coal / methanol to olefins (CTO / MTO) plants with a production capacity of 7.15 million tons; and 6 ethane cracking ethylene plants (including mixed paraffin cracking) with a production capacity of 4.9 million tons.

[0005] In the ethylene production process, ethylene cracking accounts for approximately 50% of energy consumption. Numerous scholars at home and abroad have conducted extensive research on ethylene cracking furnaces. Ethylene cracking is a complex chemical process involving multiple reactions and the formation of a variety of products. The operational performance of an ethylene cracking furnace is directly related to the yields of various products, and thus directly affects the economic benefits of a cracking unit. Summary of the Invention

[0006] To address the limitations and defects of the prior art, the present invention provides a method for optimizing the production objectives of an ethylene cracking furnace using a multi-objective particle swarm optimization algorithm, comprising:

[0007] Step S1, initialize the population and individual positions, filter the non-dominated particles into external storage, divide the target space, and assign a grid index to each particle in the external storage;

[0008] Step S2: Calculate the dynamic control parameter ω and the contraction-expansion coefficient α. The calculation formulas for the dynamic control parameter ω and the contraction-expansion coefficient α are as follows:

[0009] ω=e (1-it) / MaxIt *rand (I)

[0010] α=0.5·(MaxIt-it) / MaxIt+0.5 (2)

[0011] Among them, ω represents the dynamic control parameter, it represents the current number of iterations, MaxIt represents the maximum number of iterations, rand represents a random number between (0, 1), and α represents the contraction and expansion coefficient;

[0012] Step S3: Select the global optimal particle according to the grid congestion;

[0013] Step S4, calculating the particle potential well in quantum behavior;

[0014] Step S5: determining whether the dynamic control parameter ω is greater than a preset threshold;

[0015] If the judgment result is that the dynamic control parameter ω is greater than the preset threshold, step S6 is executed; if the judgment result is that the dynamic control parameter ω is less than or equal to the preset threshold, step S7 is executed;

[0016] Step S6, determining a global search for particles according to the dynamic control parameter ω;

[0017] Step S7, determining to perform local search on particles according to the dynamic control parameter ω;

[0018] Step S8, controlling the limit range of the particle position;

[0019] Step S9, randomly mutating the particle positions;

[0020] Step S10: Update the optimal solution of each individual in the particle swarm;

[0021] Step S11: Taking the maximum ethylene yield and the maximum propylene yield as optimization targets, the operation cycle of the ethylene cracking furnace is 63 days, the feed gas hydrocarbon ratio and the outlet temperature adjusted every 7 days as model parameters, and optimizing the production target of the ethylene cracking furnace using a multi-objective particle swarm optimization algorithm based on quantum behavior improvement. The expression of the optimization target of the ethylene cracking furnace production target optimization model is as follows:

[0022]

[0023]

[0024] The constraint conditions of the ethylene cracking furnace production target optimization model are expressed as follows:

[0025] DSR min ≤DSR≤DSR max (5)

[0026] Q min ≤Q naphtha ≤Q max (6)

[0027] COT min ≤COT≤COT max (7)

[0028]

[0029]

[0030]

[0031] T w ≤T max (11)

[0032] Among them, DSR is the feed gas hydrocarbon ratio, DSR min and DSR max are the upper and lower limits of the feed gas hydrocarbon ratio, Q naphtha is the naphtha flow rate, Q min and Q max are the upper and lower limits of the naphtha flow rate, COT is the outlet temperature, and COT min and COT max are the upper and lower limits of the outlet temperature, respectively. crack is the number of days to cleavage, and are the upper and lower limits of the cracking days, P ethy is the daily ethylene production rate, is the lower limit of daily ethylene production rate, P prop is the daily propylene production rate, is the lower limit of daily propylene production rate, T w is the tube wall temperature, T max is the upper limit of the tube wall temperature.

[0033] Optionally, step S9 includes:

[0034] According to quantum behavior theory and an iterative formula, the position information of the particle is randomly updated according to probability. The iterative formula is as follows:

[0035] X i，j (t+1)=p i，j (t)±α·|C j (t)-X i，j (t)|·ln[1 / u i，j (t)] (12)

[0036]

[0037]

[0038] Among them, X i，j (t) represents the current position of the particle, u i，j (t) is a random number uniformly distributed in the interval (0, 1), C j (t) represents the average of the best individual position of the jth attribute of each particle in the tth iteration, p i，j (t) represents the potential well of the jth property of the ith particle in the tth iteration, G j (t) represents the jth attribute value at the global optimal position.

[0039] Optionally, the step S10 includes:

[0040] Step 11: Release the optimal individual in the updated particle swarm into the external storage space;

[0041] Step 12: Determine whether the external storage space is full;

[0042] If the judgment result is that the external storage space is full, execute step 13; if the judgment result is that the external storage space is not full, execute step 14;

[0043] Step 13: select individuals with a congestion degree greater than a preset standard congestion degree according to the grid congestion degree and delete them;

[0044] After completing step S13, proceed to step S14;

[0045] Step 14: Compare the number of iterations with the preset maximum number of iterations;

[0046] If the number of iterations is less than the preset maximum number of iterations, step 15 is executed; if the number of iterations is greater than or equal to the preset maximum number of iterations, the process ends;

[0047] Step 15: Increase the current number of iterations by 1;

[0048] After step S15 is completed, step S2 is executed.

[0049] The present invention has the following beneficial effects:

[0050] To address the problems of traditional multi-objective particle swarm optimization algorithms being prone to falling into local optimal solutions and having slow convergence speeds, the present invention uses a multi-objective particle swarm optimization algorithm based on quantum behavior to optimize the production targets of ethylene cracking furnaces. First, the positions of the particles are updated through quantum behavior, thereby reducing the possibility of the algorithm falling into local optimal solutions and improving the convergence speed of the algorithm. Then, to prevent the algorithm from converging too quickly and thus missing certain optimal solutions, a search strategy that dynamically switches between global search and local search is introduced to further improve the distributive nature of the algorithm. The algorithm is then verified using standard test functions ZDT1, ZDT2, ZDT3, ZDT4, and ZDT6, thereby optimizing the production targets of the ethylene cracking furnace and solving the inaccuracy problem of traditional multi-objective optimization algorithms. The multi-objective particle swarm optimization algorithm based on quantum behavior provided by the present invention effectively optimizes the ethylene yield and propylene yield targets in the ethylene cracking furnace by integrating quantum behavior into the multi-objective particle swarm optimization algorithm, helping enterprises achieve optimized operations and improve their economic efficiency, while also achieving energy conservation and emission reduction in the ethylene cracking furnace, thereby contributing to the achievement of the "dual carbon" goals. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 Schematic diagram of the traditional cracking furnace structure.

[0052] Figure 2 Schematic diagram of the traditional reaction tube structure.

[0053] Figure 3 This is a flowchart of a multi-objective particle swarm optimization algorithm based on quantum behavior provided in Example 1 of the present invention.

[0054] Figure 4 This is a Pareto distribution diagram of ethylene yield and propylene yield provided in Example 1 of the present invention.

[0055] Figure 5 Schematic diagram of the outlet temperatures of the individual with fixed-cycle temperature regulation and the individual with constant temperature provided in the first embodiment of the present invention.

[0056] Figure 6 Schematic diagram of the decrease rate of ethylene yield of an individual with fixed-cycle temperature regulation and an individual with constant temperature provided in Example 1 of the present invention.

[0057] Figures 7a-7e This is the Pareto result distribution diagram of the IMOQPSO algorithm, MOPSO algorithm, PESA-II algorithm, NSGA-II algorithm, and MOEA / D algorithm provided in Example 1 of the present invention.

[0058] Figures 8a-8e This is the Pareto result distribution diagram of the IMOQPSO algorithm, MOPSO algorithm, PESA-II algorithm, NSGA-II algorithm, and MOEA / D algorithm provided in Example 1 of the present invention.

[0059] Figures 9a-9e This is the Pareto result distribution diagram of the IMOQPSO algorithm, MOPSO algorithm, PESA-II algorithm, NSGA-II algorithm, and MOEA / D algorithm provided in Example 1 of the present invention.

[0060] Figures 10a-10e This is the Pareto result distribution diagram of the IMOQPSO algorithm, MOPSO algorithm, PESA-II algorithm, NSGA-II algorithm, and MOEA / D algorithm provided in Example 1 of the present invention.

[0061] Figures 11a-11e This is the Pareto result distribution diagram of the IMOQPSO algorithm, MOPSO algorithm, PESA-II algorithm, NSGA-II algorithm, and MOEA / D algorithm provided in Example 1 of the present invention. DETAILED DESCRIPTION

[0062] To enable those skilled in the art to better understand the technical solution of the present invention, the optimization method for the production target of an ethylene cracking furnace using the multi-objective particle swarm optimization algorithm provided by the present invention is described in detail below with reference to the accompanying drawings.

[0063] Example 1

[0064] This embodiment provides a multi-objective particle swarm optimization algorithm based on quantum behavior for optimizing production targets in ethylene cracking furnaces. To address the problem that traditional multi-objective particle swarm optimization algorithms are prone to getting stuck in local optimal solutions and experiencing slow convergence when optimizing ethylene cracking furnaces, this algorithm first updates particle positions using quantum behavior, thereby reducing the likelihood of the algorithm getting stuck in local optimal solutions and improving its convergence speed. To prevent the algorithm from converging too quickly and missing certain optimal solutions, a search strategy that dynamically switches between global and local searches is introduced, further improving the algorithm's distributive nature. This optimizes the production targets of the ethylene cracking furnace, thereby achieving energy conservation and emission reduction in the ethylene cracking furnace and contributing to the "dual carbon" goals.

[0065] In this example, under the same output and conditions, a multi-objective particle swarm optimization algorithm improved by quantum behavior is applied to the optimization of ethylene cracking furnace operation, with ethylene yield and propylene yield as optimization targets. The change in ethylene and propylene yields is mainly achieved by temperature regulation, which is measured by the outlet temperature (COT).

[0066] This embodiment provides a novel multi-objective optimization algorithm. The algorithm first updates the position information of particles based on quantum behavior, and then introduces a dynamic search strategy to optimize the production target of the ethylene cracking furnace. The flow chart of the algorithm is shown in FIG. Figure 3 shown.

[0067] This example first initializes the population and individual positions, selects non-dominated freed particles for external storage, partitions the target space, and assigns a grid index to each particle in the external storage. Next, it performs an iterative update. This iterative process requires sequentially calculating the dynamic control parameter ω and the contraction and expansion coefficient α, selecting the globally optimal particle based on grid congestion, calculating the potential well in quantum behavior, and determining an update strategy based on the dynamic control parameter ω to update the particle's position information.

[0068] The calculation formulas for the dynamic control parameter ω and the contraction and expansion coefficient α are shown in (1) and (2). The purpose is to prevent quantum behavior from causing the convergence speed to be too fast and miss some local optimal solutions. Subsequently, the dynamic control parameter ω will be used to determine whether to perform a global search or a local search for particles.

[0069] ω=e (1-it) / MaxIt *rand (1)

[0070] α=0.5·(MaxIt-it) / MaxIt+0.5 (2)

[0071] When updating the particle position information, we use quantum behavior knowledge and obtain it randomly according to probability through an iterative formula. The purpose is to reduce the possibility of particles falling into local optimality and speed up the convergence of the algorithm. The iterative formula is as follows:

[0072] X i，j (t+1)=p i，j (t)±α·|C j (t)-X i，j (t)|·ln[1 / u i，j (t)] (12)

[0073]

[0074]

[0075] Among them, X i，j (t) represents the current position of the particle, u i，j (t) is a random number uniformly distributed in the interval (0, 1), C j (t) represents the average of the best individual position of the jth attribute of each particle in the tth iteration, p i，j (t) represents the potential well of the jth property of the ith particle in the tth iteration, G j (t) represents the jth attribute value at the global optimal position.

[0076] Then the optimal solution of each individual in the particle swarm is updated and placed in the external storage space. When the external storage space is full, the individuals with higher congestion are selected for deletion based on the grid congestion. This strategy takes individual diversity into consideration.

[0077] Finally, this improved multi-objective optimization algorithm is applied to the ethylene cracking furnace to achieve the optimization of ethylene yield and propylene yield. The Pareto distribution of ethylene yield and propylene yield is as follows: Figure 4 As shown in Figure 2. Since heat conduction can also cause changes in ethylene and propylene yields, in order to reduce the impact of pipeline coking on ethylene yield while ensuring that propylene yield remains essentially unchanged, the outlet temperature needs to be periodically increased. Randomly select an individual with a fixed period of temperature adjustment and an individual with a constant temperature from the Pareto result set. The comparison of their outlet temperatures is shown in Figure 2. Figure 5 As shown, the rate of decrease of ethylene yield when the outlet temperature is adjusted at a fixed period and the outlet temperature remains unchanged is as follows: Figure 6 This proves that the algorithm proposed in this embodiment can effectively optimize the ethylene cracking furnace.

[0078] In this example, the maximum ethylene yield and the maximum propylene yield are used as optimization targets. The operation cycle of the ethylene cracking furnace is 63 days. The feed gas hydrocarbon ratio and the outlet temperature adjusted every 7 days are used as model parameters. The multi-objective particle swarm optimization algorithm based on quantum behavior improvement is used to optimize the production target of the ethylene cracking furnace. The expression of the optimization target of the ethylene cracking furnace production target optimization model is as follows:

[0079]

[0080]

[0081] The constraint conditions of the ethylene cracking furnace production target optimization model are expressed as follows:

[0082] DSR min ≤DSR≤DSR max (5)

[0083] Q min ≤Q naphtha ≤Q max (6)

[0084] COT min ≤COT≤COT max (7)

[0085]

[0086]

[0087]

[0088] T w ≤T max (11)

[0089] Among them, DSR is the feed gas hydrocarbon ratio, DSR min and DSR max are the upper and lower limits of the feed gas hydrocarbon ratio, Q naphtha is the naphtha flow rate, Q min and Q max are the upper and lower limits of the naphtha flow rate, COT is the outlet temperature, and COT min and COT max are the upper and lower limits of the outlet temperature, respectively. crack is the number of days to cleavage, and are the upper and lower limits of the cracking days, P ethy is the daily ethylene production rate, is the lower limit of daily ethylene production rate, P prop is the daily propylene production rate, is the lower limit of daily propylene production rate, T w is the tube wall temperature, T max is the upper limit of the tube wall temperature.

[0090] In order to verify the effectiveness of the multi-objective optimization particle swarm algorithm based on quantum behavior improvement, it is necessary to first test it with a standard data set. We selected ZDT1, ZDT2, ZDT3, ZDT4, and ZDT6, as detailed in Table 1:

[0091] Table 1 Test functions

[0092]

[0093] Figures 7a-7e This is the Pareto result distribution diagram of the IMOQPSO algorithm, MOPSO algorithm, PESA-II algorithm, NSGA-II algorithm, and MOEA / D algorithm provided in Example 1 of the present invention. Figures 8a-8e This is the Pareto result distribution diagram of the IMOQPSO algorithm, MOPSO algorithm, PESA-II algorithm, NSGA-II algorithm, and MOEA / D algorithm provided in Example 1 of the present invention. Figures 9a-9e This is the Pareto result distribution diagram of the IMOQPSO algorithm, MOPSO algorithm, PESA-II algorithm, NSGA-II algorithm, and MOEA / D algorithm provided in Example 1 of the present invention. Figures 10a-10e This is the Pareto result distribution diagram of the IMOQPSO algorithm, MOPSO algorithm, PESA-II algorithm, NSGA-II algorithm, and MOEA / D algorithm provided in Example 1 of the present invention. Figures 11a-11e The Pareto result distribution diagrams of the IMOQPSO algorithm, MOPSO algorithm, PESA-II algorithm, NSGA-II algorithm, and MOEA / D algorithm provided in Example 1 of the present invention are shown. This example obtains their Pareto result distribution diagrams. Pareto is the optimal result set found by the optimization algorithm after iteration. By observing the result diagram, it can be seen that the IMOQPSO algorithm is significantly superior to other algorithms in terms of convergence, convergence speed, and distribution. The evaluation indicators of the particle swarm optimization algorithm have two aspects: convergence and distribution. Convergence represents the distance from the true Pareto solution. The closer the distance, the better the convergence. Distribution represents the uniformity of the distribution on the true Pareto. Convergence and distribution are generally evaluated by the inversion generation distance, or IGD. The smaller the mean (Mean) and standard deviation (Std) of IGD, the better the convergence and distribution of the algorithm. The IGD of each algorithm on the test function is shown in Table 2.

[0094] Table 2 IGD values ​​of the algorithms on the test function

[0095]

[0096] Table 2 shows that the convergence and distribution of the IMOQPSO algorithm on the test function are better than other algorithms in most cases, thus verifying the effectiveness of the IMOQPSO algorithm.

[0097] This example demonstrates the feasibility and effectiveness of a multi-objective particle swarm optimization algorithm based on quantum-behavioral optimization for optimizing ethylene and propylene yields in an ethylene cracking furnace. Furthermore, this model, when applied to temperature regulation in an ethylene cracking furnace, can reduce the rate of ethylene degradation, improve ethylene utilization, conserve energy, and increase the company's economic benefits.

[0098] It will be understood that the above embodiments are merely exemplary embodiments for illustrating the principles of the present invention, and the present invention is not limited thereto. Those skilled in the art will appreciate that various modifications and improvements can be made without departing from the spirit and substance of the present invention, and such modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A method for optimizing the production target of an ethylene cracking furnace using a multi-objective particle swarm optimization algorithm, characterized in that: include: Step S1, initialize the population and individual positions, filter the non-dominated particles into external storage, divide the target space, and assign a grid index to each particle in the external storage; Step S2: Calculate the dynamic control parameter ω and the contraction-expansion coefficient α. The calculation formulas for the dynamic control parameter ω and the contraction-expansion coefficient α are as follows: ω=e (1-it) / MaxIt *row (1) α=0.5·(MaxIt-it) / MaxIt+0.5 (2) Among them, ω represents the dynamic control parameter, it represents the current number of iterations, MaxIt represents the maximum number of iterations, rand represents a random number between (0,1), and α represents the contraction and expansion coefficient; Step S3: Select the global optimal particle according to the grid congestion; Step S4, calculating the particle potential well in quantum behavior; Step S5: determining whether the dynamic control parameter ω is greater than a preset threshold; If the judgment result is that the dynamic control parameter ω is greater than the preset threshold, step S6 is executed; if the judgment result is that the dynamic control parameter ω is less than or equal to the preset threshold, step S7 is executed; Step S6, determining a global search for particles according to the dynamic control parameter ω; Step S7, determining to perform local search on particles according to the dynamic control parameter ω; Step S8, controlling the limit range of the particle position; Step S9, randomly mutating the particle positions; Step S10: Update the optimal solution of each individual in the particle swarm; Step S11: Taking the maximum ethylene yield and the maximum propylene yield as optimization targets, the operation cycle of the ethylene cracking furnace is 63 days, the feed gas hydrocarbon ratio and the outlet temperature adjusted every 7 days as model parameters, and optimizing the production target of the ethylene cracking furnace using a multi-objective particle swarm optimization algorithm based on quantum behavior improvement. The expression of the optimization target of the ethylene cracking furnace production target optimization model is as follows: The constraint conditions of the ethylene cracking furnace production target optimization model are expressed as follows: DSR min ≤DSR≤DSR max (5) Q min ≤Q naphtha ≤Q max (6) COT min ≤COT≤COT max (7) T w ≤T max (11) Among them, DSR is the feed gas hydrocarbon ratio, DSR min and DSR max are the upper and lower limits of the feed gas hydrocarbon ratio, Q naphtha is the naphtha flow rate, Q min and Q max are the upper and lower limits of the naphtha flow rate, COT is the outlet temperature, and COT min and COT max are the upper and lower limits of the outlet temperature, respectively. crac+ is the number of days to cleavage, and are the upper and lower limits of the cracking days, P ethy is the daily ethylene production rate, is the lower limit of daily ethylene production rate, P prop is the daily propylene production rate, is the lower limit of daily propylene production rate, T w is the tube wall temperature, T max is the upper limit of the tube wall temperature.

2. The method for optimizing the production target of an ethylene cracking furnace using a multi-objective particle swarm optimization algorithm according to claim 1, characterized in that: The step S9 includes: According to quantum behavior theory and an iterative formula, the position information of the particle is randomly updated according to probability. The iterative formula is as follows: X i，j (t+1)=p i，j (t)±α@|C j (t)-X i，j (t)|·ln[1 / u i，j (t)] (12) Among them, X i,j (t) represents the current position of the particle, u i,j (t) is a random number uniformly distributed in the interval (0,1), C j (t) represents the average of the best individual position of the jth attribute of each particle in the tth iteration, p i,j (t) represents the potential well of the jth property of the ith particle in the tth iteration, G j (t) represents the jth attribute value at the global optimal position.

3. The method for optimizing the production target of an ethylene cracking furnace using a multi-objective particle swarm optimization algorithm according to claim 2, characterized in that: The step S10 then includes: Step 11: Liberate the optimal individual in the updated particle swarm into the external storage space; Step 12: Determine whether the external storage space is full; If the judgment result is that the external storage space is full, execute step 13; if the judgment result is that the external storage space is not full, execute step 14; Step 13: select individuals with a congestion degree greater than a preset standard congestion degree according to the grid congestion degree and delete them; After completing step S13, proceed to step S14; Step 14: Compare the number of iterations with the preset maximum number of iterations; If the number of iterations is less than the preset maximum number of iterations, step 15 is executed; if the number of iterations is greater than or equal to the preset maximum number of iterations, the process ends; Step 15: Increase the current number of iterations by 1; After step S15 is completed, step S2 is executed.

Images