Improved artificial bee colony optimization method for energy-saving scheduling of ethylene cracking furnace cluster

Abstract

This invention discloses an improved artificial bee colony optimization method for energy-saving scheduling of ethylene cracking furnace groups. The method includes: proposing a mixed-integer nonlinear programming scheduling model for the energy-saving scheduling problem of ethylene cracking furnace groups, with the optimization objective of maximizing the daily average net profit within an adjustable scheduling range; and solving the mixed-integer nonlinear programming scheduling model based on an improved artificial bee colony algorithm to obtain the scheduling optimization result. This invention proposes a new mixed-integer nonlinear programming scheduling model considering emission constraints and uses an improved artificial bee colony algorithm to solve the model. The proposed algorithm has advantages such as simple structure and ease of implementation. It can not only search a wide solution space in a short time, but its inherent integrated weighted optimization mechanism can also drive the algorithm to reach a better search region, thereby obtaining a satisfactory solution to the problem and maximizing daily profit while taking environmental benefits into account.

Description

Technical Field

[0001] This invention relates to an improved artificial bee colony optimization method for energy-saving scheduling of ethylene cracking furnace groups, belonging to the field of ethylene production model optimization. Background Technology

[0002] Cracking furnaces are the core equipment in ethylene production. Industrial ethylene production typically employs multiple cracking furnaces operating in parallel to break down hydrocarbon feedstocks into smaller molecule hydrocarbon compounds such as ethylene. During operation, coking inevitably occurs on the inner walls of the furnace tubes, leading to a decrease in ethylene yield. Therefore, periodic shutdowns for coking removal are necessary. When multiple cracking furnaces simultaneously process various types of feedstocks, and operating costs and product yields are constantly changing, optimizing the scheduling of the cracking furnace group to maximize production profits is of great significance.

[0003] On the other hand, ethylene cracking furnaces are also major industrial energy consumers and key targets for environmental protection, especially during the coking stage, which emits large amounts of carbon dioxide, carbon monoxide, and dust. In today's increasingly severe situation of energy conservation and emission reduction, how to use new methods and technologies to meet emission reduction requirements is an important issue in the furnace group scheduling process.

[0004] In addition, with increasing global awareness of environmental protection, China is also emphasizing the pollution and carbon emissions from industry. The ethylene industry is a high-energy-consuming and high-carbon-emission process. Therefore, while maximizing profits, we must also take into account environmental benefits and adopt new models and methods to achieve the goal of energy conservation and emission reduction. Summary of the Invention

[0005] This invention provides an improved artificial bee colony optimization method for energy-saving scheduling of ethylene cracking furnace groups, which is used to optimize the scheduling of ethylene cracking furnace groups and maximize daily profits while taking into account environmental benefits.

[0006] The technical solution of this invention is: an improved artificial bee colony optimization method for energy-saving scheduling of ethylene cracking furnace groups, comprising:

[0007] A mixed-integer nonlinear programming scheduling model is proposed for the energy-saving scheduling problem of ethylene cracking furnace groups, with the optimization objective of maximizing the daily average net profit within an adjustable scheduling range.

[0008] The energy-saving scheduling problem of ethylene cracking furnace group was solved by a mixed-integer nonlinear programming scheduling model based on an improved artificial bee colony algorithm, and the scheduling optimization results were obtained.

[0009] The mixed-integer nonlinear programming scheduling model for the energy-saving scheduling problem of the ethylene cracking furnace group includes: an objective function and constraints; wherein, the objective function is to maximize the average daily net profit within an adjustable scheduling range; the constraints include: material constraints, integer constraints, time constraints, boundary conditions, and non-simultaneous coking constraints.

[0010] The objective function consists of three parts: the first is the sales revenue of various products, the second is the purchase cost of raw materials, and the third is the cost of descaling; the specific function can be expressed by equation (1):

[0011]

[0012] Wherein, NF, NC, NB and NP represent the number of feed types, the number of cracking furnaces, the number of batches and the number of products, respectively; This is an exponential decay equation used to describe the dynamic change in the yield of product l relative to time during the pyrolysis of feed i in furnace j in a batch operation; P l It is the unit price of product l; l i,j,k D represents the product generated in the kth batch of cracking furnace j from feed i. i,j It is the feed flow rate of feed i in pyrolysis furnace j; It is the profit of all products in a batch of operations; Cr i D i,j t i,j,k The unit price of feed i is Cr i The feed flow rate D of feed i in cracking furnace j i,j The batch processing time t of feed i in the kth batch of cracking furnace j i,j,k The product of these terms represents the material cost incurred by feed i in the kth batch of furnace j; Cv i,j D i,j t i,j,k The unit operating cost Cv of feed i in cracking furnace j i,j With feed flow rate D i,j Batch processing time t i,j,k The product of these terms represents the operating cost incurred by feed i in the kth batch of cracking furnace j; Cs i,j y i,j,k The cost of decoking feed i in cracking furnace j, Cs i,j And by the binary variable y i,j,kControl is defined as follows: if feed i is processed in the kth batch of pyrolysis furnace j, it is assigned as 1; otherwise, it is assigned as 0. T represents the total cycle time; t represents the batch processing time. In addition, the model is based on the following assumptions: each batch operation in any pyrolysis furnace can only process one type of feed; the model accuracy of the pyrolysis furnace meets the requirements; the emission curve during the decoking process roughly follows a normal distribution, with the emission peak located at the midpoint of the decoking duration; multiple furnaces in the pyrolysis furnace group system cannot decoke simultaneously, and only one pyrolysis furnace performs decoking operation at the same time.

[0013] The material constraint: G i The actual feed flow rate minus the minimum upstream feed flow rate (Flo) i The actual feed flow rate is expressed as Flo. i +G i Its definition is shown in equation (2), and equation (3) gives G. i Boundary values;

[0014]

[0015] in, This represents the sum of the feed flow rates consumed in all batches across all cracking furnaces;

[0016]

[0017] Among them, Fup i Flo is the highest feed rate upstream. i This is the minimum feed flow rate upstream.

[0018] The integer constraint:

[0019]

[0020]

[0021]

[0022] Wherein, variable y i,j,k It is an integer variable with a value of 0 or 1, expressed by equations (4)-(5); equation (4) indicates that all feeds should be processed in one cycle operation; equation (5) indicates that for each cracking furnace, the first batch is forced to be the feed; equation (6) indicates that a single batch can be used to crack a maximum of one feedstock.

[0023] The time constraint mainly limits the batch processing time t. i,j,k Batch start time S j,k and end time E j,k These three time variables; Equation (7) represents t i,j,kIt should be within a certain range, where tlo i,j Indicates the shortest processing time, tup i,j This indicates the maximum processing time; the upper and lower limits of this time are specified by the operating characteristics of the cracking furnace or industrial experience.

[0024]

[0025]

[0026]

[0027]

[0028]

[0029]

[0030]

[0031]

[0032]

[0033] Equations (8) to (11) are used to represent the start time S of each batch on each furnace. j,k and end time E j,k A binary variable z was introduced. j If the start time of the current cycle of the first batch in furnace j is greater than the end time of the last batch in the previous cycle, then z j z is 1 if it is positive and 0 otherwise; j The logical constraints are represented by equations (12) and (13), where M represents the theoretical upper limit of the total time spent; equation (14) indicates that the total cycle time of each cracking furnace is the same, equal to the processing time of all batches and its decoking time τ. i,j The sum of; Equation (15) indicates that the start time of the first batch should be less than or equal to the total cycle time T.

[0034] The boundary conditions are as follows:

[0035]

[0036] E j,k ,S j,k ,t i,j,k ,T≤M (17)

[0037]

[0038] Equations (16) to (18) indicate that the lower bound of all continuous variables is zero, and all start times Sj,k End time E j,k Batch processing time t i,j,k The total cycle time T should be less than the upper limit M; G i The actual feed rate minus the minimum upstream feed rate; variable z j ,y i,j,k ,x j,k,j',k' Defined as a binary variable.

[0039] The non-simultaneous coking constraint: The following two formulas indicate that multiple pyrolysis furnaces cannot be shut down for coking at the same time within a certain period of time;

[0040]

[0041]

[0042] In this regard, the decoking times between different pyrolysis furnaces should not overlap, which introduces another binary variable x. j,k,j',k' When its value is 1, the k-th coking in furnace j is after the k'-th coking in furnace j' and they are not coking simultaneously; if the k-th coking in furnace j is before the k'-th coking in furnace j' and they are not coking simultaneously, then it is 0; M represents the theoretical upper limit of the total time spent; S is the start time of the batch processing on the furnace. j',k'+1 S j,k+1 End time E j,k E j',k' .

[0043] The specific steps of the scheduling model solution algorithm based on the improved artificial bee colony algorithm are as follows:

[0044] Step 1, Encoding and Decoding Method: An integer encoding and decoding strategy based on the arrangement of input raw materials is adopted;

[0045] Step 2, Population Initialization: Use reverse learning to generate SN uniformly distributed initial populations;

[0046] Step 3, Hired Bee Stage: Place hired bees on nectar sources, find a nectar source for each hired bee and evaluate the quality of the nectar source; hired bees use integrated optimal guidance to find new nectar sources near the assigned nectar source;

[0047] Step 4, Observation Bee Stage: After all the hired bees have completed their search for new nectar sources, the observation bees calculate the probability P of each nectar source being selected using formulas (24) and (25) based on the nectar source information found by the hired bees. p Then, a roulette wheel method is used to randomly select nectar sources;

[0048]

[0049]

[0050] In equation (24), P p Represents the probability that nectar source p is selected, fitness p Let f represent the fitness value of the p-th nectar source, which is calculated as shown in formula (25), where f p This represents the objective function value of the p-th honey source;

[0051] Step 5, Scout Bee Phase: After all the observation bees have completed their search for new nectar sources, they evaluate and decide which nectar source to keep. If a nectar source is not updated after a limit number of iterations, that nectar source X is discarded. p The food source will be discarded, and the corresponding mercenary bee will be transformed into a scout bee; the scout bee will then generate a new food source to replace it based on reverse learning.

[0052] Step 6, Termination Condition: Set the termination condition to the maximum number of iterations. If the condition is met, output "Optimal Scheduling Strategy". Otherwise, repeat Step 3, Step 4, and Step 5 until the termination condition is met.

[0053] The integrated optimal bootloader includes:

[0054] The population is sorted by fitness value, and a new individual V is generated by a weighted combination of all individuals that are better than the current population individuals. p,q Guided updates for current population individuals:

[0055]

[0056] Among them, X p,q θ represents the scheduling optimization result of the q-th pyrolysis furnace among individuals with fitness ranking p in the current population; p,q This represents the correlation coefficient; q∈{1…NC}, where NC is the number of pyrolysis furnaces. u represents the current individual's ranking number, W p The weight corresponding to the individual with fitness ranking p is calculated using the following formula:

[0057]

[0058] Among them, WF p For X p Balance fitness, X p The scheduling optimization result for the individual with fitness rank p in the current population is defined by the following formula:

[0059]

[0060] Among them, F p For X p The degree of adaptability.

[0061] The beneficial effects of this invention are as follows: This invention discloses an improved artificial bee colony optimization method for energy-saving scheduling of ethylene cracking furnace groups. Firstly, this invention models the conventional ethylene cracking furnace group scheduling problem. Then, considering environmental issues, it schedules coke removal emissions at night to reduce the impact on the environment and human health. By adding time window constraints, pollutant emissions are reduced, achieving the goal of energy conservation and emission reduction. Finally, a new mixed-integer nonlinear programming (MINLP) scheduling model considering emission constraints is proposed, and the improved artificial bee colony algorithm is used to solve the MINLP problem. The proposed algorithm has advantages such as simple structure and ease of implementation. It can not only search a wide solution space in a short time, but its inherent integrated weighted optimization mechanism can also drive the algorithm to reach a better search region, thereby obtaining a satisfactory solution to the problem. This achieves the goal of maximizing daily profits while taking environmental benefits into account. The new scheduling model and method maximize profits while also considering environmental benefits, achieving the goal of energy conservation and emission reduction. Attached Figure Description

[0062] Figure 1 This is a detailed flowchart of the overall algorithm of the present invention;

[0063] Figure 2 This is a schematic diagram of the scheduling results of the present invention. Figure 1 ;

[0064] Figure 3 This is a schematic diagram of the scheduling results of the present invention. Figure 2 ;

[0065] Figure 4 This is a schematic diagram of the integrated optimal boot operation of the present invention;

[0066] Figure 5 This is a schematic diagram of the pyrolysis furnace system of the present invention. Detailed Implementation

[0067] The invention will be further described below with reference to the accompanying drawings and embodiments, but the scope of the invention is not limited to the description.

[0068] like Figure 1-5 As shown, an improved artificial bee colony optimization method for energy-saving scheduling of ethylene cracking furnace groups includes: proposing a mixed-integer nonlinear programming scheduling model for the new energy-saving scheduling problem of ethylene cracking furnace groups considering emission constraints, with the optimization objective of maximizing the daily average net profit within an adjustable scheduling range; solving the mixed-integer nonlinear programming scheduling model for the energy-saving scheduling problem of ethylene cracking furnace groups based on the improved artificial bee colony algorithm to obtain the scheduling optimization result, thus maximizing the daily profit while taking into account environmental benefits. Figure 5This is a schematic diagram of a cracking furnace system. The types of feed for each cracking furnace may be different. For example, the feed for furnace No. 4 is HVGO, LPG and NAP, while the feed for furnace No. 2 is LPG and NAP.

[0069] Furthermore, the mixed-integer nonlinear programming scheduling model for the energy-saving scheduling problem of the ethylene cracking furnace group includes: an objective function and constraints; wherein, the objective function is to maximize the average daily net profit within an adjustable scheduling range; the constraints include: material constraints, integer constraints, time constraints, boundary conditions, and non-simultaneous coking constraints.

[0070] Furthermore, before building the model, it is necessary to determine the known conditions and the information to be solved.

[0071] The known conditions include: average feedstock flow rate; types of feedstock that can be processed in each furnace; yield variations of pyrolysis products for different types of feedstock in different furnaces; decoking time for each pyrolysis furnace; upper and lower limits of batch processing time; decoking cost; feedstock purchase cost and product sales price; and the customized planned time range. The information to be determined includes: the number of batches allocated to each pyrolysis furnace; the type of feedstock processed in each batch operation; the start and end times of each batch operation; and the decoking sequence of the entire furnace system.

[0072] The information to be determined is the optimal scheduling scheme for the pyrolysis furnace group, which is obtained by solving the improved artificial bee colony algorithm. Based on the above assumptions, the pyrolysis furnace group scheduling problem is modeled, and the specific modeling steps are as follows:

[0073] The model is based on the following assumptions: each batch operation in any pyrolysis furnace can only handle one type of feed; the model accuracy of the pyrolysis furnace meets the requirements; the emission curve during the decoking process roughly follows a normal distribution, with the emission peak located at the midpoint of the decoking duration; multiple furnaces in the pyrolysis furnace group system cannot decoke simultaneously, and only one pyrolysis furnace can perform decoking operation at the same time.

[0074] The objective function of the scheduling model is to maximize the average daily net profit within an adjustable scheduling range. This involves three parts: the first is the sales revenue of various products, the second is the purchase cost of raw materials, and the third is the cost of descaling. The specific function can be expressed by equation (1):

[0075]

[0076] Wherein, NF, NC, NB and NP represent the number of feed types, the number of cracking furnaces, the number of batches and the number of products, respectively; This is an exponential decay equation used to describe the dynamic change of product l yield relative to time during the pyrolysis of feed i in furnace j in a batch operation. i,j,l bi,j c i,j,l All are constants, determined by the feed rate, outlet temperature, and steam-to-hydrogen ratio of the cracking furnace, respectively; typically, the dynamic change of product output relative to time is non-linear. Therefore, describing the dynamic product output using an exponential formula is closer to reality than using a linear function; P l It is the unit price of product l; l i,j,k D represents the product generated in the kth batch of cracking furnace j from feed i. i,j It is the feed flow rate of feed i in pyrolysis furnace j; It is the profit of all products in a batch of operations; Cr i D i,j t i,j,k The unit price of feed i is Cr i The feed flow rate D of feed i in cracking furnace j i,j The batch processing time t of feed i in the kth batch of cracking furnace j i,j,k The product of these terms represents the material cost incurred by feed i in the kth batch of furnace j; Cv i,j D i,j t i,j,k The unit operating cost Cv of feed i in cracking furnace j i,j With feed flow rate D i,j Batch processing time t i,j,k The product of these terms represents the operating cost incurred by feed i in the kth batch of cracking furnace j; Cs i,j y i,j,k The cost of decoking feed i in cracking furnace j, Cs i,j And by the binary variable y i,j,k Control, if feed i is processed in the kth batch of cracking furnace j, it is specified as 1, otherwise it is specified as 0; T represents the total cycle time; t represents the batch processing time; to calculate the net profit of a batch operation, the material, operating and decoking costs mentioned above should be subtracted; the numerator of equation (1) is the total net profit of the furnace group system in a cycle operation, which is divided by the total cycle time, indicating that it represents the average net profit of the cracking furnace system per day.

[0077] Furthermore, the material constraints are as follows: based on actual production conditions, the total amount of each raw material consumed by the pyrolysis furnace is represented as the actual feed flow rate; G i The actual feed flow rate minus the minimum upstream feed flow rate (Flo) i The actual feed flow rate is expressed as Flo. i +G i Its definition is shown in equation (2), and equation (3) gives G. i Boundary values;

[0078]

[0079] in, This represents the sum of the feed flow rates consumed in all batches across all cracking furnaces;

[0080]

[0081] Among them, Fup i Flo is the highest feed rate upstream. i This represents the minimum feed rate upstream.

[0082] Furthermore, the integer constraint: variable y i,j,k It is an integer variable with a value of 0 or 1, expressed by equations (4)-(5); before determining the scheduling optimization result, it is unknown how many batches of NB need to be run in each cracking furnace in a cycle, and it needs to be determined according to the actual scheduling optimization result; equation (4) indicates that all feed should be processed in one cycle operation; equation (5) indicates that for each cracking furnace, the first batch is forced to be a feed; equation (6) indicates that a single batch can be used to crack a maximum of one raw material;

[0083]

[0084]

[0085]

[0086] Furthermore, the time constraint primarily limits the batch processing time t. i,j,k Batch start time S j,k and end time E j,k These three time variables; Equation (7) represents t i,j,k It should be within a certain range, where tlo i,j Indicates the shortest processing time, tup i,j This indicates the maximum processing time; the upper and lower limits of this time are specified by the operating characteristics of the cracking furnace or industrial experience.

[0087]

[0088]

[0089]

[0090]

[0091]

[0092]

[0093]

[0094]

[0095]

[0096] Equations (8) to (11) are used to represent the start time S of each batch on each furnace. j,k and end time E j,k Because the model is a cyclic scheduling problem, the first batch may start from the previous cycle or the current cycle; therefore, a binary variable z is introduced. j If the start time of the current cycle of the first batch in furnace j is greater than the end time of the last batch in the previous cycle, then z j z is 1 if it is positive and 0 otherwise; j The logical constraints are represented by equations (12) and (13), where M represents the theoretical upper limit of the total time spent; equation (14) indicates that the total cycle time of each cracking furnace is the same, equal to the processing time of all batches and its decoking time τ. i,j The sum of; Equation (15) indicates that the start time of the first batch should be less than or equal to the total cycle time T.

[0097] Furthermore, the boundary conditions are as follows:

[0098]

[0099] E j,k ,S j,k ,t i,j,k ,T≤M (17)

[0100]

[0101] Equations (16) to (18) indicate that the lower bound of all continuous variables is zero, and all start times S j,k End time E j,k Batch processing time t i,j,k The total loop time T should be less than the upper limit M; variable z j ,y i,j,k ,x j,k,j',k' Defined as a binary variable.

[0102] Furthermore, the non-simultaneous coking constraint: the following two formulas indicate that multiple pyrolysis furnaces cannot be shut down for coking at the same time within a certain period of time;

[0103]

[0104]

[0105] In this regard, the decoking times between different pyrolysis furnaces should not overlap, which introduces another binary variable x.j,k,j',k' When its value is 1, the kth coking in furnace j is after the k'th coking in furnace j' and there is no simultaneous coking; if the kth coking in furnace j is before the k'th coking in furnace j' and there is no simultaneous coking, then it is 0.

[0106] In summary, with equation (1) as the objective function and equations (2)-(20) as constraints, a basic energy-saving scheduling model for ethylene cracking furnace groups can be established. This model includes a target of daily average profit, material constraints, integer constraints, time constraints, boundary conditions, and non-simultaneous coking constraints determined by actual industrial production conditions, and basically describes the energy-saving scheduling problem of ethylene cracking furnace groups.

[0107] Furthermore, the specific steps of the scheduling model solution algorithm based on the improved artificial bee colony algorithm are as follows:

[0108] Step 1, Encoding and Decoding Method: An integer encoding and decoding strategy based on the arrangement of input raw materials is adopted; For the gen-th generation individual r, based on the raw materials, the general expression is X. r NC represents the number of ethylene cracking furnaces. The raw material input arrangement for furnace No. 1 is given, and the arrangement length is the total number of batches for that furnace. Assuming there are 6 ethylene cracking furnaces and 5 types of input raw materials, let the code for individual r be... The number of furnaces is the length of the list. [2,2,3] indicates that the feeding order of furnace 1 is: two batches of raw material 2, one batch of raw material 3, and feeding in 3 batches.

[0109] Step 2, Population Initialization: Use reverse learning to generate SN evenly distributed initial populations, i.e. SN nectar sources;

[0110] Step 3, Hired Bee Stage: Place hired bees on nectar sources, find a nectar source for each hired bee and evaluate the quality of the nectar source; the minimum number of hired bees is SN, which is selected in this embodiment. Assign one hired bee to each nectar source and evaluate the quality of the nectar source; the hired bees use integrated optimal guidance to find new nectar sources near the assigned nectar source;

[0111] Furthermore, such as Figure 4 As shown, the integrated optimal guidance:

[0112] The population is sorted by fitness value, and a new individual V is generated by a weighted combination of all individuals that are better than the current population individuals. p,q Guided updates for current population individuals:

[0113]

[0114] Among them, X p,qθ represents the scheduling optimization result of the q-th pyrolysis furnace among individuals with fitness ranking p in the current population; p,q The correlation coefficient is a random number following a uniform distribution (0, 0.5); q∈{1…NC}, where NC is the number of pyrolysis furnaces. u represents the current individual's ranking number, W p The weight corresponding to the individual with fitness ranking p is calculated using the following formula:

[0115]

[0116] Among them, WF p For X p Balance fitness, X p The scheduling optimization result for the individual with fitness rank p in the current population is defined by the following formula:

[0117]

[0118] Among them, F p For X p The fitness of;

[0119] Using a balanced fitness calculation weight can prevent a slightly better ranking than X when u is too large. p The individual weights tend to 0 and are diluted during integration; WF p The formula is defined as being derived from experimental simulation results;

[0120] Step 4, Observation Bee Stage: After all the hired bees have completed their search for new nectar sources, the observation bees calculate the probability P of each nectar source being selected using formulas (24) and (25) based on the nectar source information found by the hired bees. p Then, a roulette wheel method is used to randomly select nectar sources;

[0121]

[0122]

[0123] In equation (24), P p Represents the probability that nectar source p is selected, fitness p Let f represent the fitness value of the p-th nectar source, which is calculated as shown in formula (25), where f p This represents the objective function value of the p-th honey source;

[0124] Step 5, Scout Bee Phase: After all the observation bees have completed their search for new nectar sources, they evaluate and decide which nectar source to keep. If a nectar source is not updated after a limit number of iterations, that nectar source X is discarded. pThe food source will be discarded, and the corresponding mercenary bee will be transformed into a scout bee; the scout bee will then generate a new food source to replace it based on reverse learning.

[0125] Step 6, Termination Condition: Set the termination condition to the maximum number of iterations. If the condition is met, output "Optimal Scheduling Strategy". Otherwise, repeat Step 3, Step 4, and Step 5 until the termination condition is met.

[0126] To further verify the effectiveness of this invention, production data from an ethylene cracking unit in a petrochemical enterprise in Gansu Province were used for testing. This unit is controlled by an Emerson system, and key production process data such as flow rate, outlet temperature, and steam-to-hydrogen ratio, as well as the cracking furnace's own parameters, are collected in real time by field intelligent instruments (sensors). The collected data is then transmitted to the Emerson configuration database via the HART bus. This invention exports the production data of a complete operating cycle under normal conditions from the database, then establishes a mixed-integer nonlinear programming (MINLP) furnace group scheduling model based on this production data, and solves the proposed MINLP scheduling model using an improved artificial bee colony algorithm. An example of the obtained scheduling optimization results is shown below. Figure 2 , 3 As shown. The total cycle time T is set to 150, and the theoretical upper limit of M is set to 360, both in days. The population size is set to SN = 50, and the number of times not updated is Limit = 8; the number of hired bees and the number of observation bees are both set to SN.

[0127] Figure 2 , Figure 3 The diagram illustrates an optimized scheduling scheme for a single cracking furnace. Figure 2 This indicates that there are two processing batches in cracking furnace No. 1. Batch 1 is fed with Fpd, and the batch processing starts on day 0 and ends on day 20. Batch 2 starts on day 20 and ends on day 113. Figure 3 This indicates that there are 3 processing batches in cracking furnace No. 2. Batch 1 is fed with Fb, and the batch processing starts on day 0 and ends on day 12; batch 2 starts on day 12 and ends on day 60; batch 3 starts on day 60 and ends on day 131. The yield of the ethylene cracking furnace gradually decreases as the feed reactants are consumed.

[0128] According to another aspect of the present invention, an improved artificial bee colony optimization system for energy-saving scheduling of ethylene cracking furnace groups is provided, comprising: a first module for proposing a mixed-integer nonlinear programming scheduling model for the energy-saving scheduling problem of ethylene cracking furnace groups, with the optimization objective of maximizing the daily average net profit within an adjustable scheduling range; and a second module for solving the mixed-integer nonlinear programming scheduling model for the energy-saving scheduling problem of ethylene cracking furnace groups based on an improved artificial bee colony algorithm to obtain the scheduling optimization result.

[0129] The specific embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention.

Claims

1. An improved artificial bee colony optimization method for energy-saving scheduling of ethylene cracking furnace groups, characterized in that: include: A mixed-integer nonlinear programming scheduling model is proposed for the energy-saving scheduling problem of ethylene cracking furnace groups, with the optimization objective of maximizing the daily average net profit within an adjustable scheduling range. The scheduling optimization results were obtained by solving the mixed-integer nonlinear programming scheduling model of the energy-saving scheduling problem of ethylene cracking furnace group based on the improved artificial bee colony algorithm. The specific steps of the scheduling model solution algorithm based on the improved artificial bee colony algorithm are as follows: Step 1, Encoding and Decoding Method: An integer encoding and decoding strategy based on the arrangement of input raw materials is adopted; Step 2, Population Initialization: Generate population using reverse learning. An initial population with a uniform distribution; Step 3, Hired Bee Stage: Place hired bees on nectar sources, find a nectar source for each hired bee and evaluate the quality of the nectar source; hired bees use integrated optimal guidance to find new nectar sources near the assigned nectar source; Step 4, observation stage: when all the employed bees complete the search of new honey source, the observation bees calculate the probability P of each honey source selected by formula (24) (25) according to the information of the honey source searched by the employed bees p Then the roulette method is used to randomly select the honey source; (24) (25) In equation (24) This represents the probability that nectar source p is selected. Let represent the fitness value of the p-th nectar source, which is calculated as shown in formula (25), where This represents the objective function value of the p-th honey source; Step 5, Scout Bee Phase: After all the observation bees have completed their search for new nectar sources, they make a decision on whether to keep or discard any of the nectar sources. If a certain nectar source has passed through... If the honey source is not updated after the next cycle, it is still not updated. The bee that was hired to use this nectar source will be abandoned and transformed into a scout bee; the scout bee will then generate a new nectar source to replace it based on reverse learning. Step 6, Termination Condition: Set the termination condition to the maximum number of iterations. If the condition is met, output "Optimal Scheduling Strategy". Otherwise, repeat Step 3, Step 4, and Step 5 until the termination condition is met. The integrated optimal bootloader includes: The population is sorted according to fitness values, consisting of all individuals superior to the current population. New individuals generated by weighted combination of other individuals Guided updates for current population individuals: (21) in, This represents the scheduling optimization result of the q-th pyrolysis furnace among the individuals with fitness ranking p in the current population; Represents the correlation coefficient; NC represents the number of pyrolysis furnaces. , where u is the current individual's ranking number. The weight corresponding to the individual with fitness ranking p is calculated using the following formula: (22) in, for Balance fitness The scheduling optimization result for the individual with fitness rank p in the current population is defined by the following formula: (23) in, for The degree of adaptability.

2. The improved artificial bee colony optimization method for energy-saving scheduling of ethylene cracking furnace groups according to claim 1, characterized in that: The mixed-integer nonlinear programming scheduling model for the energy-saving scheduling problem of the ethylene cracking furnace group includes: an objective function and constraints; wherein, the objective function is to maximize the average daily net profit within an adjustable scheduling range; the constraints include: material constraints, integer constraints, time constraints, boundary conditions, and non-simultaneous coking constraints.

3. The improved artificial bee colony optimization method for energy-saving scheduling of ethylene cracking furnace groups according to claim 2, characterized in that: The objective function consists of three parts: the first is the sales revenue of various products, the second is the purchase cost of raw materials, and the third is the cost of descaling; the specific function can be expressed by equation (1): (1) in, , , and These represent the types and quantities of feed materials, the number of pyrolysis furnaces, the number of batches, and the quantity of products, respectively. It is an exponential decay formula used to describe the feeding in batch operations. In the furnace Products during medium pyrolysis The dynamic changes in output relative to time; It is a product The unit price; Indicates feed In the pyrolysis furnace No. Products generated in batch, It is the feed In the pyrolysis furnace The feed flow rate in the middle; It is the profit of all products in a batch of operations; For feeding unit price With feed In the pyrolysis furnace Feed flow rate in Feeding In the pyrolysis furnace The Batch processing time in batch The product of the feed In the furnace The The material costs incurred in the batch; For feeding In the pyrolysis furnace Unit operating cost With feed flow rate Batch processing time The product of the feed In the pyrolysis furnace The Operating costs incurred during the batch; For feeding In the pyrolysis furnace The cost of descaling And by binary variables Control, if feeding In the pyrolysis furnace The If processing is performed in a batch, it is specified as 1; otherwise, it is specified as 0. T represents the total cycle time; t represents the batch processing time. In addition, the model is based on the following assumptions: each batch operation in any pyrolysis furnace can only process one type of feed; the model accuracy of the pyrolysis furnace meets the requirements; the emission curve during the decoking process roughly follows a normal distribution, and the emission peak is located at the midpoint of the decoking duration; multiple furnaces in the pyrolysis furnace group system cannot decoke simultaneously, and only one pyrolysis furnace performs decoking operation at the same time.

4. The improved artificial bee colony optimization method for energy-saving scheduling of ethylene cracking furnace groups according to claim 3, characterized in that: The material constraints: The actual feed flow rate minus the minimum upstream feed flow rate The actual feed flow rate is expressed as Its definition is shown in equation (2), and equation (3) gives Boundary values; (2) in, This represents the total feed consumed in all batches across all cracking furnaces. The sum of flows; (3) in, This represents the highest feed flow rate upstream. This is the minimum feed flow rate upstream.

5. The improved artificial bee colony optimization method for energy-saving scheduling of ethylene cracking furnace groups according to claim 3, characterized in that: The integer constraint: (4) (5) (6) Among them, variables It is an integer variable with a value of 0 or 1, expressed by equations (4)-(5); equation (4) indicates that all feeds should be processed in one cycle operation; equation (5) indicates that for each cracking furnace, the first batch is forced to be the feed; equation (6) indicates that a single batch can be used to crack a maximum of one feedstock.

6. The improved artificial bee colony optimization method for energy-saving scheduling of ethylene cracking furnace groups according to claim 3, characterized in that: The time constraint: The time constraint mainly limits the batch processing time. Batch start time and end time These three time variables; Equation (7) represents It should be within a certain range, where, Indicates the shortest processing time. This indicates the maximum processing time; the upper and lower limits of this time are specified by the operating characteristics of the cracking furnace or industrial experience. (7) (8) (9) (10) (11) (12) (13) (14) (15) Equations (8) to (11) are used to represent the start time of each batch on each furnace. and end time Binary variables were introduced. If the start time of the current cycle of the first batch in furnace j is greater than the end time of the last batch in the previous cycle, then =1, otherwise =0; The logical constraints are represented by equations (12) and (13), where M represents the theoretical upper limit of the total time spent; equation (14) indicates that the total cycle time of each cracking furnace is the same, equal to the processing time of all batches and its decoking time. The sum; Equation (15) indicates that the start time of the first batch should be less than or equal to the total cycle time. .

7. The improved artificial bee colony optimization method for energy-saving scheduling of ethylene cracking furnace groups according to claim 3, characterized in that: The boundary conditions are as follows: (16) (17) (18) Equations (16) to (18) indicate that the lower bound of all continuous variables is zero, and all start times are zero. End time Batch processing time Total cycle time It should be less than the upper limit. ; The actual feed rate minus the minimum upstream feed rate; variable Defined as a binary variable.

8. The improved artificial bee colony optimization method for energy-saving scheduling of ethylene cracking furnace groups according to claim 3, characterized in that: The non-simultaneous coking constraint: The following two formulas indicate that multiple pyrolysis furnaces cannot be shut down for coking at the same time within a certain period of time; (19) (20) In this regard, the decoking times between different pyrolysis furnaces should not overlap, which introduces another binary variable. When its value is 1, the furnace The kth coke removal in the furnace The first in If the furnace is not cleaned simultaneously after the second coking process; The kth coking time in the furnace The first in If no coking occurs before the next decoking and there is no simultaneous decoking, then the value is 0; M represents the theoretical upper limit of the total time spent; the start time of the batch processing on the furnace. , End time , .

Images