Composite rectification sequence optimization method, equipment and storage medium

Abstract

The invention relates to the field of chemical rectification, and discloses a composite rectification sequence optimization method which comprises the following steps: acquiring an initial mixture to be rectified and a group of composite rectification sequences for rectifying the initial mixture; based on a genetic programming algorithm, converting the composite rectification sequence into an initial tree species group; performing iterative optimization on the initial tree species group on the basis of a tournament algorithm to obtain an optimized tree species group, and performing iterative optimization on nodes of the initial tree species group on the basis of an exhaustion method or an elite genetic algorithm to obtain a first optimal node; based on an exhaustion method or an elite genetic algorithm, iteratively optimizing nodes of the optimized tree species group to obtain a second optimal node; updating the composite rectification sequence into an optimized rectification sequence; and updating the initial rectification unit of the composite rectification sequence to the optimized rectification unit. According to the method provided by the embodiment of the invention, the rectification sequence optimization efficiency is improved, and the global optimality of the rectification unit is ensured.

Description

Technical Field

[0001] This invention relates to the field of chemical distillation, and in particular to a method, equipment and storage medium for optimizing a complex distillation sequence. Background Technology

[0002] When separating multi-component mixtures by distillation, the conventional distillation sequence formed by connecting simple distillation columns in series has excessive energy consumption and low distillation efficiency. In order to reduce the distillation cost of multi-component mixtures, the existing technology introduces a distillation column with a partition wall as a distillation unit, and combines a simple distillation column with a partition wall distillation column to form a composite distillation sequence.

[0003] However, existing complex distillation sequences have two problems:

[0004] First, the potential organizational structures among the various distillation units in a complex distillation sequence are diverse. As the number of components in a multi-component mixture increases, the number of potential distillation sequence structures will grow exponentially, leading to a "combinatorial explosion" problem.

[0005] Second, compared with simple distillation columns, isolation wall distillation columns have more design decision variables and more complex relationships between them, which greatly increases the optimization difficulty of the distillation unit.

[0006] Without addressing the aforementioned issues, existing technologies present excessive challenges in optimizing complex distillation sequences, resulting in significant errors in the optimization outcomes and making it difficult to obtain satisfactory results. Summary of the Invention

[0007] This invention provides a method, apparatus, and storage medium for optimizing a complex distillation sequence to solve the aforementioned technical problems.

[0008] In a first aspect, embodiments of the present invention provide a method for optimizing a compound distillation sequence based on a genetic programming algorithm. The method is applied to a compound distillation sequence comprising multiple simple distillation columns and a partition wall distillation column. The method includes: obtaining an initial mixture to be distilled and a set of compound distillation sequences for distilling the initial mixture; converting the compound distillation sequences into an initial tree population based on a genetic programming algorithm; iteratively optimizing the initial tree population based on a tournament algorithm to obtain an optimized tree population, and simultaneously iteratively optimizing the nodes of the initial tree population based on an exhaustive search method or an elite genetic algorithm to obtain a first optimal node; iteratively optimizing the nodes of the optimized tree population based on an exhaustive search method or an elite genetic algorithm to obtain a second optimal node; comparing the historical minimum annual total cost of the initial tree population and the optimized tree population, and updating the compound distillation sequence to the optimized distillation sequence corresponding to the one with the smaller historical minimum annual total cost; comparing the historical minimum annual total cost of the first optimal node and the second optimal node, and updating the initial distillation unit of the compound distillation sequence to the optimized distillation unit corresponding to the one with the smaller historical minimum annual total cost.

[0009] The compound distillation sequence optimization method based on genetic programming algorithm provided in this invention decomposes the optimization problem of compound distillation sequence into two parts: sequence optimization process and distillation unit optimization process. Through genetic programming algorithm, the optimization solution of compound distillation sequence and distillation unit can be solved simultaneously. The compound distillation sequence is updated to the optimized distillation sequence, and the initial distillation unit is updated to the optimized distillation unit. This realizes synchronous iterative optimization of compound distillation sequence and initial distillation unit, improves the optimization efficiency of distillation sequence, reduces distillation cost, solves the "combinatorial explosion" problem in the compound distillation sequence optimization process, improves the accuracy of optimization calculation results, and thus ensures the global optimality of distillation unit.

[0010] Optionally, the steps of transforming the complex distillation sequence into an initial tree population based on the genetic programming algorithm include: obtaining all components in the initial mixture based on the initial mixture; obtaining the terminal set corresponding to each component and all function sets corresponding to the simple distillation column and the isolation wall distillation column based on all components; and generating multiple initial tree individuals based on the genetic programming algorithm, the function sets, and the terminal sets to form an initial tree population.

[0011] Optionally, the number of functions is The number of terminal sets is the number of all component types in the initial mixture, where N is the number of all component types in the initial mixture.

[0012] Optionally, the steps of iteratively optimizing the initial tree population based on the tournament algorithm to obtain the optimized tree population include: selecting parent tree populations from the initial tree population based on the tournament algorithm; controlling the crossover of different parent individuals in the parent tree population to generate offspring individuals; controlling the mutation of offspring individuals to form multiple mutated offspring individuals, so as to obtain an optimized tree population including multiple mutated offspring individuals.

[0013] Optionally, the steps of selecting parent tree populations from the initial tree population based on the tournament algorithm include: splitting the initial tree population into multiple tree groups; selecting the tree individual with the lowest annual total cost in each tree group as the parent individual based on the tournament algorithm; and combining all parent individuals to form the parent tree population.

[0014] Optionally, the feed components are the same between the parent individuals that are cross-crossed.

[0015] Optionally, the step of controlling the crossover of different parent individuals in the parent tree population to generate offspring individuals includes: obtaining each feed component of different parent individuals in the parent tree population; controlling the crossover between parent individuals with the same feed component to generate offspring individuals.

[0016] Optionally, the step of controlling the offspring individuals to mutate and form multiple mutated offspring individuals to obtain an optimized tree population including multiple mutated offspring individuals includes: selecting any feed component; obtaining the subtree of the offspring individual corresponding to the selected feed component; controlling the growth of the subtree to form multiple mutated offspring individuals to obtain an optimized tree population.

[0017] Optionally, the step of comparing the historical minimum total annual cost of the initial tree population and the optimized tree population, and updating the compound distillation sequence to the optimized distillation sequence corresponding to the one with the smaller historical minimum total annual cost, includes: calculating the first initial total annual cost of each initial tree in the initial tree population and the first optimized total annual cost of each optimized tree in the optimized tree population; comparing the first initial total annual cost with the first optimized total annual cost, and if the first optimized total annual cost is less than or equal to the first initial total annual cost, then updating the initial tree population to the optimized tree population; and updating the compound distillation sequence to the optimized distillation sequence corresponding to the optimized tree with the historical minimum total annual cost based on the optimized tree population.

[0018] Optionally, after updating the compound distillation sequence to the optimized distillation sequence corresponding to the optimized tree population, the method further includes: if the total cost of the first optimized year is greater than the total cost of the first initial year, then detecting the current iteration number of the optimized tree population; obtaining the maximum iteration number of the optimized tree population; comparing the current iteration number of the optimized tree population with the maximum iteration number of the optimized tree population; if the current iteration number is less than the maximum iteration number, then the optimized tree population continues to iterate and optimize until the current iteration number is greater than or equal to the maximum iteration number.

[0019] Optionally, the nodes corresponding to the simple distillation column are iteratively optimized based on the exhaustive method; the nodes corresponding to the isolation wall distillation column are iteratively optimized based on the elite genetic algorithm.

[0020] Optionally, the steps of iteratively optimizing the nodes of the initial tree population to obtain the first optimal node include: selecting each node individual from the nodes of the initial tree population based on the elite retention strategy and roulette wheel method; controlling the crossover of different node individuals to generate node offspring individuals; controlling the mutation of node offspring individuals to form multiple mutated node individuals, so as to obtain the first optimal node including multiple mutated node individuals.

[0021] Optionally, the feed components are the same between the individual nodes that are crossing over.

[0022] Optionally, the steps for selecting each individual node from the nodes of the initial tree population based on the elite retention strategy and the roulette wheel method include: splitting the nodes of the initial tree population into multiple node groups; and selecting the individual with the lowest annual total cost in each node group as the node individual based on the elite retention strategy and the roulette wheel method.

[0023] Optionally, the step of controlling different node individuals to cross over and generate node offspring individuals includes: obtaining multiple feed components based on the initial mixture; controlling node individuals with the same feed components to cross over and generate node offspring individuals.

[0024] Optionally, the step of controlling the mutation of the child individuals of the controlled node to form multiple mutated node individuals to obtain a first optimal node including multiple mutated node individuals includes: selecting any feed component; obtaining the subtree corresponding to the selected feed component; controlling the growth of the subtree to form multiple mutated node individuals to obtain a first optimal node including multiple mutated node individuals.

[0025] Optionally, the step of comparing the historical minimum total annual cost of the first optimal node and the second optimal node, and updating the initial distillation unit of the compound distillation sequence to the optimized distillation unit corresponding to the one with the smaller historical minimum total annual cost, includes: calculating the second initial total annual cost of each individual in the first optimal node and the second optimized total annual cost of each individual in the second optimal node; comparing the second initial total annual cost with the second optimized total annual cost, and if the second optimized total annual cost is less than or equal to the second initial total annual cost, then updating the initial distillation unit to the optimized distillation unit corresponding to the second optimal node.

[0026] Optionally, after updating the initial distillation unit of the compound distillation sequence to the optimized distillation unit corresponding to the one with the smaller historical minimum annual total cost, the method further includes: if the second optimized annual total cost is greater than the second initial annual total cost, then detecting the current iteration number of the nodes in the optimized tree population; obtaining the maximum iteration number of the nodes in the optimized tree population; comparing the current iteration number with the maximum iteration number, and if the current iteration number is less than the maximum iteration number, then the nodes in the optimized tree population continue to iterate and optimize until the current iteration number is greater than or equal to the maximum iteration number.

[0027] In a second aspect, embodiments of the present invention provide a compound distillation sequence optimization device, the device comprising: a processor and a memory, the memory storing instructions; the processor calls the instructions in the memory, the processor calling the instructions in the memory so that the processor executes the compound distillation sequence optimization method based on genetic programming algorithm according to any of the foregoing embodiments of the first aspect of the present invention.

[0028] The processor of the compound distillation sequence optimization device provided in this embodiment of the invention executes the compound distillation sequence optimization method based on genetic programming algorithm according to any of the foregoing embodiments of the first aspect of the invention by calling instructions in the memory. It can decompose the optimization problem of compound distillation sequence into two parts: sequence optimization process and distillation unit optimization process. Through genetic programming algorithm, it can optimize the compound distillation sequence and the distillation unit at the same time. It updates the compound distillation sequence to the optimized distillation sequence and the initial distillation unit to the optimized distillation unit, realizing synchronous iterative optimization of compound distillation sequence and initial distillation unit. This improves the optimization efficiency of distillation sequence, reduces distillation cost, solves the "combinatorial explosion" problem in the compound distillation sequence optimization process, improves the accuracy of optimization calculation results, and thus ensures the global optimality of distillation unit.

[0029] Thirdly, embodiments of the present invention provide a machine-readable storage medium storing instructions that, when executed by a machine, implement the compound distillation sequence optimization method based on genetic programming algorithm according to any of the foregoing embodiments of the first aspect of the present invention.

[0030] The instructions stored in the machine-readable storage medium provided in this embodiment of the invention can be called by a processor and executed by the machine-readable storage medium of any of the foregoing embodiments of the first aspect of the invention to decompose the optimization problem of the compound distillation sequence into two parts: a sequence optimization process and a distillation unit optimization process. Through the genetic programming algorithm, the optimization solution of the compound distillation sequence can be obtained simultaneously with the optimization solution of the distillation unit. The compound distillation sequence is updated to the optimized distillation sequence, and the initial distillation unit is updated to the optimized distillation unit. This achieves synchronous iterative optimization of the compound distillation sequence and the initial distillation unit, improves the optimization efficiency of the distillation sequence, reduces the distillation cost, solves the "combinatorial explosion" problem in the optimization process of the compound distillation sequence, improves the accuracy of the optimization calculation results, and thus ensures the global optimality of the distillation unit. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0032] Figure 1 This is a flowchart of an embodiment of the compound distillation sequence optimization method based on genetic programming algorithm of the present invention;

[0033] Figure 2 This is a flowchart of step S120 in one embodiment of the compound distillation sequence optimization method based on genetic programming algorithm of the present invention;

[0034] Figure 3 This is a flowchart of step S130 in one embodiment of the compound distillation sequence optimization method based on genetic programming algorithm of the present invention;

[0035] Figure 4 This is a flowchart of step S140 in one embodiment of the compound distillation sequence optimization method based on genetic programming algorithm of the present invention;

[0036] Figure 5 This is a flowchart of step S150 in one embodiment of the compound distillation sequence optimization method based on genetic programming algorithm of the present invention.

[0037] Figure 6 This is a flowchart of step S160 in one embodiment of the compound distillation sequence optimization method based on genetic programming algorithm of the present invention;

[0038] Figure 7This is a flowchart illustrating the optimization of the distillation sequence and the overall optimization of the distillation unit in one embodiment of the compound distillation sequence optimization method based on genetic programming algorithm of the present invention.

[0039] Figure 8 This is a schematic diagram illustrating the transformation of a compound distillation sequence into a tree structure in one embodiment of the compound distillation sequence optimization method based on genetic programming algorithm of the present invention.

[0040] Figure 9 This is a schematic diagram illustrating iterative optimization in one embodiment of the compound distillation sequence optimization method based on genetic programming algorithm of the present invention;

[0041] Figure 10 This is a schematic diagram of a simple distillation column in one embodiment of the compound distillation sequence optimization method based on genetic programming algorithm of the present invention;

[0042] Figure 11 This is a schematic diagram of the structure of a partition wall distillation column in one embodiment of the compound distillation sequence optimization method based on genetic programming algorithm of the present invention;

[0043] Figure 12 This is a structural block diagram of the performance estimation tool in one embodiment of the compound distillation sequence optimization method based on genetic programming algorithm of the present invention;

[0044] Figure 13 This is a line graph showing the annual total cost of a compound distillation sequence after multiple iterations of optimization, as shown in one embodiment of the compound distillation sequence optimization method based on genetic programming algorithm of the present invention.

[0045] Figure 14 This is a schematic diagram of the optimal composite distillation sequence structure in one embodiment of the composite distillation sequence optimization method based on genetic programming algorithm of the present invention;

[0046] Figure 15 This is a structural block diagram of one embodiment of the compound distillation sequence optimization device of the present invention.

[0047] Explanation of icon numbers:

[0048] 100 - Performance estimation tools; 110 - Classification models; 120 - Decision modules; 130 - Predictive models;

[0049] 210 - Processor; 220 - Memory; 230 - Communication interface; 240 - Bus. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0051] It should be noted that all directional indications in the embodiments of the present invention, such as up, down, left, right, front, back, etc., are only used to explain the relative positional relationship and movement of the components in a specific posture as shown in the attached figure. If the specific posture changes, the directional indication will also change accordingly.

[0052] Furthermore, the use of terms such as "first" and "second" in this invention is for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature. Additionally, the technical solutions of the various embodiments can be combined with each other, but only on the basis of being achievable by those skilled in the art. When the combination of technical solutions is contradictory or impossible to implement, such a combination of technical solutions should be considered non-existent and not within the scope of protection claimed by this invention.

[0053] For ease of understanding, the following describes the compound distillation sequence optimization method based on genetic programming algorithm according to embodiments of the present invention. This method is applied to compound distillation sequences including multiple simple distillation columns and isolation wall distillation columns. For example... Figures 1 to 9 As shown, the compound distillation sequence optimization method based on genetic programming algorithm in this embodiment of the invention includes steps S110 to S160.

[0054] In step S110, the initial mixture to be distilled and a set of complex distillation sequences for distilling the initial mixture are obtained.

[0055] In this embodiment, the distillation unit of the complex distillation sequence is a simple distillation column (SDC) or a dividing-wall column (DWC). The initial mixture includes various compounds, such as alkanes, alkenes, butane, pentane, cyclohexane, benzene, toluene, and o-xylene. In this embodiment, the complex distillation sequence is a randomly selected set of sequences. The initial mixture can be distilled through this sequence to sequentially separate the various compounds in the initial mixture, thereby completing the distillation process.

[0056] In step S120, the compound distillation sequence is transformed into an initial tree population based on a genetic programming algorithm.

[0057] like Figure 8 As shown, in this embodiment, the composite distillation sequence is first optimized by using a genetic programming algorithm, and the composite distillation sequence is uniquely represented as a tree. That is, the composite distillation sequence in this embodiment is transformed into a tree structure. The distillation units in the composite distillation sequence, i.e., distillation columns, correspond to the nodes in the tree structure. The high-purity products obtained after the initial mixture is distilled by each distillation unit form the leaf nodes in the tree structure.

[0058] In this embodiment, the genetic programming algorithm, as an evolutionary algorithm, uses trees to encode candidate results, where each node represents an operation or function, and each leaf node represents an input variable or constant.

[0059] The tree structure of the compound distillation sequence obtained by the genetic programming algorithm can be optimized through multiple evolutions, thereby optimizing the compound distillation sequence, improving the distillation efficiency, and reducing the distillation cost.

[0060] like Figure 2 As shown, in some optional embodiments, step S120 includes steps S121 to S123.

[0061] In step S121, all components in the initial mixture are obtained based on the initial mixture.

[0062] In step S122, based on all components, the terminal set corresponding to each component and the function set corresponding to the simple distillation column and the isolation wall distillation column are obtained.

[0063] In step S123, based on the genetic programming algorithm, function set, and terminal set, multiple initial tree individuals are generated to form an initial tree population.

[0064] In this embodiment, the compounds of all components in the initial mixture are first obtained. Based on the number of compounds and the genetic programming algorithm, all possible function sets corresponding to nodes and all terminal sets corresponding to leaf nodes are calculated. The function sets and terminal sets are used for subsequent iterative optimization of the initial tree population.

[0065] Furthermore, the number of functions is The number of terminal sets is the number of all component types in the initial mixture, where N is the number of all component types in the initial mixture.

[0066] like Figure 3 , Figure 7 and Figure 9As shown, in step S130, the initial tree population is iteratively optimized based on the tournament algorithm to obtain the optimized tree population. At the same time, the nodes of the initial tree population are iteratively optimized based on the exhaustive method or the elite genetic algorithm to obtain the first optimal node.

[0067] The iterative optimization of the initial tree population includes four steps: growth, selection, crossover, and mutation. The growth step involves obtaining the function set and terminal set in step S120, and then growing multiple tree structures based on these sets to form the initial tree population. The growth of the tree structures continues until all types of compound components are separated. The nodes of the initial tree population are then optimized to obtain the first optimal node, thus optimizing the nodes of the tree population before iterative optimization.

[0068] like Figure 3 As shown, the steps for iteratively optimizing the initial tree population based on the tournament algorithm to obtain the optimized tree population include steps S131 to S133.

[0069] In step S131, parent tree populations are selected from the initial tree population based on the tournament algorithm. Step S131 is a selection step, and further, step S131 includes steps S1311 to S1313.

[0070] In step S1311, the initial tree population is split into multiple tree groups.

[0071] In step S1312, based on the tournament algorithm, the tree individual with the lowest annual total cost in each tree group is selected as the parent individual.

[0072] In step S1313, all parent individuals are combined to form a parent tree population.

[0073] In this embodiment, the initial tree population includes multiple individual trees. The multiple individual trees in the initial tree population are divided into multiple tree groups, with each tree group having the same number of individual trees. The individual tree with the lowest annual total cost in each tree group is then selected. Finally, the individual tree with the lowest annual total cost in each tree group is used as the parent individual to form the parent tree population, which is used for subsequent crossover and mutation steps.

[0074] In step S132, different parent individuals in the parent tree population are crossbred to produce offspring individuals.

[0075] Step S132 is an interleaved step. Further, step S132 includes steps S1321 to S1322.

[0076] In step S1321, each feed component of different parent individuals in the parent tree population is obtained.

[0077] In step S1322, crossover occurs between parent individuals with the same feed composition to generate offspring individuals.

[0078] Specifically, the parent individuals undergoing crossover have the same feed components. In this embodiment, by obtaining each feed component from different parent individuals in the parent tree population, parent individuals with the same feed components are obtained, and crossover between parent individuals with the same feed components is controlled to ensure the effectiveness of the crossover operation and reduce the probability of generating unreasonable offspring individuals.

[0079] In step S133, the offspring individuals are controlled to mutate, forming multiple mutated offspring individuals, to obtain an optimized tree population including multiple mutated offspring individuals. Step S133 is a mutation step, and further, step S133 includes steps S1331 to S1333.

[0080] In step S1331, any feed component is selected.

[0081] In step S1332, the subtree of the offspring individual corresponding to the selected feed component is obtained.

[0082] In step S1333, the growth of the subtree is controlled to form multiple mutated offspring individuals in order to obtain an optimized tree population.

[0083] In this embodiment, the mutation step is similar to the growth step. A feed component of a child individual is randomly selected for mutation. The growth of the subtree corresponding to the selected feed component is controlled to form a new child individual, forming a mutated child individual. Multiple mutated child individuals are combined to form an optimized tree population.

[0084] In the step of iteratively optimizing the nodes of the initial tree population using either exhaustive search or an elite genetic algorithm to obtain the first optimal node, exhaustive search is used to iteratively optimize the nodes corresponding to the simple distillation column. Elite genetic algorithm is used to iteratively optimize the nodes corresponding to the isolation wall distillation column.

[0085] Since the simple distillation column has a relatively simple structure and requires less computation, the nodes corresponding to the simple distillation column can be iteratively optimized by exhaustive search. However, the isolation wall distillation column is more complex. The elite genetic algorithm is used to iteratively optimize the nodes corresponding to the isolation wall distillation column, thereby iteratively optimizing the nodes in the initial tree population corresponding to the simple distillation column or the isolation wall distillation column, and thus obtaining the corresponding first optimal node.

[0086] like Figure 3 As shown, in this embodiment, the steps for iterative optimization of the distillation unit are similar to those for iterative optimization of the composite distillation sequence. The specific steps for iterative optimization of the distillation unit are as follows.

[0087] Furthermore, the steps of iteratively optimizing the nodes of the initial tree population to obtain the first optimal node include steps S134 to S136.

[0088] In step S134, each node individual is selected from the nodes of the initial tree population based on the elite retention strategy and roulette wheel method.

[0089] In this embodiment, step S134 includes steps S1341 to S1342.

[0090] In step S1341, the nodes of the initial tree population are split into multiple node groups.

[0091] In step S1342, based on the elite retention strategy and roulette wheel method, the individual with the lowest annual total cost in each node group is selected as the node individual.

[0092] In this embodiment, by selecting an elite retention strategy and a roulette wheel method, both the randomness and efficiency of the selection operation are ensured, while the best individuals in the population are preserved. The nodes of the initial tree population are divided into multiple node groups, each with the same number of nodes. The individual with the lowest annual total cost in each node group is selected and used as the node for subsequent crossover and mutation steps.

[0093] In step S135, different node individuals are controlled to cross over, generating offspring node individuals. The feed components are the same among the node individuals that cross over.

[0094] In this embodiment, step S135 includes steps S1351 to S1352.

[0095] In step S1351, multiple feed components are obtained based on the initial mixture.

[0096] In step S1352, crossover occurs between nodes with the same feed composition to generate offspring nodes. By obtaining each feed composition of a node, nodes with the same feed composition are obtained, and crossover occurs between nodes with the same feed composition to ensure the effectiveness of the crossover operation and reduce the probability of generating unreasonable offspring nodes.

[0097] In step S136, the child individuals of the control node undergo mutation to form multiple mutated node individuals, so as to obtain the first optimal node including multiple mutated node individuals.

[0098] In this embodiment, step S136 includes steps S1351 to S1352.

[0099] In step S1361, any feed component is selected.

[0100] In step S1362, the subtree corresponding to the selected feed component is obtained.

[0101] In step S1363, the growth of the subtree is controlled to form multiple mutated node individuals, so as to obtain the first optimal node including multiple mutated node individuals.

[0102] In this embodiment, the mutation step is similar to the growth step. The feed component of a node's offspring individual is randomly selected for mutation. The subtree of the node's offspring individual corresponding to the selected feed component is controlled to grow to form a new node individual, thus forming a mutated node individual. Multiple mutated node individuals are combined to form the first optimal node.

[0103] In step S140, the nodes of the optimized tree population are iteratively optimized using either exhaustive search or an elite genetic algorithm to obtain the second optimal node. After iteratively optimizing the initial tree population to obtain the optimized tree population, the nodes of the optimized tree population need to be iteratively optimized again. The step of iteratively optimizing the nodes of the optimized tree population to obtain the second optimal node is the same as the step of iteratively optimizing the nodes of the initial tree population to obtain the first optimal node.

[0104] like Figure 4 As shown, in this embodiment, step S140 includes steps S141 to S143.

[0105] In step S141, based on the elite retention strategy and roulette wheel method, each node individual is selected from the nodes of the optimized tree population.

[0106] In this embodiment, step S141 includes steps S1411 to S1412.

[0107] In step S1411, the nodes of the optimized tree population are split into multiple node groups.

[0108] In step S1412, based on the elite retention strategy and roulette wheel method, the individual with the lowest annual total cost in each node group is selected as the node individual.

[0109] In this embodiment, by selecting an elite retention strategy and a roulette wheel method, both the randomness and efficiency of the selection operation are ensured, while the best individuals in the population are preserved. The optimized tree population is divided into multiple node groups, each with the same number of nodes. The individual with the lowest annual total cost in each node group is selected and used as the node for subsequent crossover and mutation steps.

[0110] In step S142, different node individuals are controlled to cross over, generating offspring node individuals. The feed components are the same among the node individuals undergoing the crossover.

[0111] In this embodiment, step S142 includes steps S1421 to S1422.

[0112] In step S1421, multiple feed components are obtained based on the initial mixture.

[0113] In step S1422, crossover occurs between nodes with the same feed composition to generate offspring nodes. By obtaining each feed composition of a node, nodes with the same feed composition are obtained, and crossover occurs between nodes with the same feed composition to ensure the effectiveness of the crossover operation and reduce the probability of generating unreasonable offspring nodes.

[0114] In step S143, the child individuals of the control node undergo mutation to form multiple mutated node individuals, so as to obtain a second optimal node including multiple mutated node individuals.

[0115] In this embodiment, step S143 includes steps S1431 to S1433.

[0116] In step S1431, any feed component is selected.

[0117] In step S1432, the subtree corresponding to the selected feed component is obtained.

[0118] In step S1433, the growth of the subtree is controlled to form multiple mutated node individuals, so as to obtain a second optimal node including multiple mutated node individuals.

[0119] In this embodiment, the mutation step is similar to the growth step. The feed component of a node's offspring individual is randomly selected for mutation. The subtree of the node's offspring individual corresponding to the selected feed component is controlled to grow to form a new node individual, thus forming a mutated node individual. Multiple mutated node individuals are combined to form a second optimal node.

[0120] In this embodiment, the nodes corresponding to the simple distillation column are iteratively optimized using an exhaustive search method. The nodes corresponding to the isolation wall distillation column are iteratively optimized using an elite genetic algorithm.

[0121] like Figure 7As shown, SDC is a simple distillation column, and DWC is a wall-separated distillation column. In this embodiment, given the specified purity of the distilled products at the top and bottom of the distillation column, the simple distillation column, due to its simple structure and only two decision variables (the number of trays in the rectifying and stripping sections), requires less computation and can be easily and quickly obtained using an exhaustive search method. However, the wall-separated distillation column has eight design decision variables, including the number of trays in the six sections and the liquid and gas flow rates from the main column to the pre-fractionation column. Therefore, the wall-separated distillation column has more decision variables, and these variables are complexly coupled, making it difficult to optimize using simple exhaustive or sequential search methods. Elite genetic algorithms are more suitable for optimizing wall-separated distillation columns. Figure 7 As shown, the optimization process of the isolation wall distillation column using the elite genetic algorithm is roughly the same as the optimization process of the complex distillation sequence, while the simple distillation column can be optimized directly using the exhaustive method.

[0122] In step S150, the historical minimum annual total cost of the initial tree population and the optimized tree population is compared, and the composite distillation sequence is updated to the optimized distillation sequence corresponding to the one with the smaller historical minimum annual total cost.

[0123] like Figure 5 As shown, in some optional embodiments, step S150 includes steps S151 to S153.

[0124] In step S151, the first initial year total cost of each initial tree in the initial tree population and the first optimized year total cost of each optimized tree in the optimized tree population are calculated.

[0125] In step S152, the first initial annual total cost is compared with the first optimized annual total cost. If the first optimized annual total cost is less than or equal to the first initial annual total cost, the initial tree population is updated to the optimized tree population.

[0126] In step S153, based on the optimized tree population, the composite distillation sequence is updated to the optimized distillation sequence corresponding to the optimized tree individual with the lowest historical annual total cost.

[0127] In this embodiment, after obtaining the optimized tree population, the first initial annual total cost of each initial tree in the initial tree population and the first optimized annual total cost of each optimized tree in the optimized tree population are calculated. By comparing the calculation results, the fitness of the optimized tree population is obtained. Fitness corresponds to the annual total cost of each tree. The smaller the annual total cost of a tree, the higher the survival rate during iterative optimization in the genetic programming algorithm. The annual total cost of the optimized tree is calculated to compare and determine whether the distillation cost has been reduced after optimization iteration. If the optimized tree population has reduced the distillation cost, the composite distillation sequence is updated to the optimized distillation sequence corresponding to the optimized tree with the smallest annual total cost, thus completing the iterative optimization of the composite distillation sequence.

[0128] like Figure 5 As shown, after step S153, the compound distillation sequence optimization method based on genetic programming algorithm of this embodiment of the invention further includes steps S154 to S156.

[0129] In step S154, if the total cost of the first optimized year is greater than the total cost of the first initial year, the current iteration number of the optimized tree population is detected.

[0130] In step S155, the maximum number of iterations for the optimized tree population is obtained.

[0131] In step S156, the current iteration number of the optimized tree population is compared with the maximum iteration number of the optimized tree population. If the current iteration number is less than the maximum iteration number, the optimized tree population continues to iterate and optimize until the current iteration number is greater than or equal to the maximum iteration number.

[0132] In this embodiment, after completing one iteration of optimization, it is checked whether the current iteration count of the optimized tree population has reached the maximum iteration count. If the maximum iteration count has not yet been reached, the optimization iteration continues until the maximum iteration count is reached. The maximum iteration count can be preset, for example, it can be 1000 times, 2000 times, 3000 times or any other arbitrary number.

[0133] In step S160, the historical minimum annual total cost of the first optimal node and the second optimal node is compared, and the initial distillation unit of the compound distillation sequence is updated to the optimized distillation unit corresponding to the one with the smaller historical minimum annual total cost.

[0134] like Figure 6 As shown, in some optional embodiments, step S160 includes steps S161 to S163.

[0135] In step S161, the second initial year total cost of each individual in the first optimal node and the second optimized year total cost of each individual in the second optimal node are calculated.

[0136] In step S162, the second initial annual total cost is compared with the second optimized annual total cost. If the second optimized annual total cost is less than or equal to the second initial annual total cost, the initial distillation unit is updated to the optimized distillation unit corresponding to the second optimal node.

[0137] In this embodiment, based on the second initial total annual cost of each individual in the first optimal node, the individual with the lowest historical total annual cost in the first optimal node is selected, i.e., the individual with the lowest total annual cost. Similarly, based on the second optimized total annual cost of each individual in the second optimal node, the individual with the lowest historical total annual cost in the second optimal node is selected, i.e., the individual with the lowest total annual cost. The total annual cost of the individual with the lowest historical total annual cost in the first optimal node is compared with that of the individual with the lowest historical total annual cost in the second optimal node. If the total annual cost of the individual in the second optimal node is smaller, then based on the individual with the lowest total annual cost, the initial distillation unit is updated to the optimized distillation unit, thereby optimizing the distillation unit.

[0138] like Figure 6 As shown, after step S162, the compound distillation sequence optimization method based on genetic programming algorithm of this embodiment of the invention further includes steps S163 to S165.

[0139] In step S163, if the total cost of the second optimized year is greater than the total cost of the second initial year, the current iteration number of the nodes of the optimized tree population is detected.

[0140] In step S164, the maximum number of iterations for the nodes of the optimized tree population is obtained.

[0141] In step S165, the current iteration number is compared with the maximum iteration number. If the current iteration number is less than the maximum iteration number, the nodes of the optimized tree population continue to be iterated and optimized until the current iteration number is greater than or equal to the maximum iteration number.

[0142] In this embodiment, after completing one iteration of optimization, it is checked whether the current iteration count of the nodes in the optimized tree population has reached the maximum iteration count. If the maximum iteration count has not yet been reached, the optimization iteration continues until the maximum iteration count is reached. The maximum iteration count can be preset, for example, it can be 1000 times, 2000 times, or 3000 times, etc.

[0143] The compound distillation sequence optimization method based on genetic programming algorithm of this invention includes: obtaining an initial mixture to be distilled and a compound distillation sequence for distilling the initial mixture; converting the compound distillation sequence into an initial tree population based on genetic programming algorithm; iteratively optimizing the initial tree population to obtain an optimized tree population; updating the compound distillation sequence to the optimized distillation sequence corresponding to the optimized tree population; obtaining the initial nodes of the optimized tree individuals in the optimized tree population; iteratively optimizing the initial nodes to obtain optimized nodes; and updating the initial distillation unit of the optimized distillation sequence to the optimized distillation unit corresponding to the optimized node.

[0144] The compound distillation sequence optimization method based on genetic programming algorithm provided in this invention decomposes the optimization problem of compound distillation sequence into two parts: sequence optimization process and distillation unit optimization process. Using genetic programming algorithm, the compound distillation sequence and distillation units are optimized simultaneously. In the process of optimizing the compound distillation sequence, the nodes of the initial tree population are iteratively optimized first, and then the nodes of the optimized tree population are iteratively optimized again. This updates the compound distillation sequence to the optimized sequence corresponding to the one with the smaller historical minimum annual total cost, and updates the initial distillation units of the compound distillation sequence to the optimized distillation units corresponding to the one with the smaller historical minimum annual total cost. This achieves synchronous iterative optimization of the compound distillation sequence and the initial distillation units, improving the optimization efficiency of the distillation sequence, reducing distillation costs, solving the "combinatorial explosion" problem in the compound distillation sequence optimization process, improving the accuracy of the optimization calculation results, and thus ensuring the global optimality of the distillation units.

[0145] For ease of understanding, the present invention provides specific embodiments of the compound distillation sequence optimization method based on genetic programming algorithm as described below.

[0146] like Figure 8 and Figure 9 As shown, in this embodiment of the invention, the initial mixture is a hexa-component mixture, including six compound components: A, B, C, D, E, and F. Therefore, based on the initial mixture, a total of 15 function sets and 6 terminal sets can be obtained. Among them, the 15 function sets include 5 function sets belonging to SDC and 10 function sets belonging to DWC.

[0147] The SDC function set includes: A / BCDEF, AB / CDEF, ABC / DEF, ABCD / EF, and ABCDE / F.

[0148] The DWC function set includes: A / B / CDEF, A / BC / DEF, A / BCD / EF, A / BCDE / F, AB / C / DEF, AB / CD / EF, AB / CDE / F, ABC / D / EF, ABC / DE / F, and ABCD / E / F.

[0149] The six terminal sets include: A, B, C, D, E, and F, which correspond to six compound components, respectively.

[0150] In the embodiments of the present invention, A, B, C, D, E, and F are n-butane, n-pentane, benzene, toluene, o-xylene, and 2,4-methylbenzene, respectively. The boiling points of these six compound components under normal pressure are: n-butane (-0.5℃), n-pentane (36.07℃), benzene (80.09℃), toluene (110.63℃), o-xylene (144.43℃), and 2,4-methylbenzene (169.38℃).

[0151] In this embodiment, the feed flow rate L of the compound distillation sequence F =400 kmol / h, feed component ratio x F，A :x F，B :x F，C :x F，D :x F，E :x F，F =25:26:27:28:29:30, the product specification of each distillation unit in the compound distillation sequence is uniformly set to 99 mol%, and the distillation column corresponding to each distillation unit is an atmospheric pressure column with a pressure drop of 0.0068 atm.

[0152] like Figure 10 and Figure 11 As shown, in this embodiment, the SDC tower segment is divided into N S and N R DWC's tower sections are divided into N S1 N S2 N S3 N S4 N S5 and N S6 .

[0153] Where, N S For the rectification section of SDC, N R For the stripping section of SDC, N S1 N S2 N S3 N S4 N S5 N S6 These are tower sections 1 to 6 of DWC, L MP V represents the liquid flow rate from the main column to the pre-fractionation column in the DWC system. MP The vapor flow rate from the main column to the pre-fractionation column, TAC (Total Annual Cost) is the total annual cost, L F L is the feed flow rate. D L B L SThese represent the flow rates of the corresponding types of distillation products that flow out.

[0154] When optimizing DWC based on the genetic programming algorithm, the real-number encoding of a tree individual is: [N] S3 +N S4 +N S5 +N S6 N S1 +N S2 N S3 N S1 N S3 +N S4 L MP V MP The fitness value of a tree individual is equal to the TAC value of the DWC represented by that tree individual.

[0155] like Figure 12 As shown in Tables 1 to 3, the compound distillation sequence optimization method based on genetic programming algorithm of the above embodiments of the present invention is implemented by the classification model 110 and prediction model 130 of the performance estimation tool 100 to optimize the compound distillation sequence and distillation unit.

[0156] In this embodiment, 15 function sets correspond to 15 options. In order to ensure that the steady-state performance of the compound distillation sequence can be accurately evaluated regardless of the distillation operation involved, a performance estimation tool 100 based on an Artificial Neural Network (ANN) needs to be established for each of the 15 options before implementing the compound distillation sequence optimization.

[0157] like Figure 12 As shown, the performance estimation tool 100 includes a classification model 110, a judgment module 120, and a prediction model 130.

[0158] Classification model 110 is used to evaluate the feasibility of given model input variables. If a given design variable is not feasible, the infeasible input variable is eliminated.

[0159] The judgment module 120 is used to determine whether the given model input variables, i.e. the distillation unit parameters, can converge. If they can converge, the prediction model 130 is called to start calculating the model output variables, i.e. the predicted values ​​calculated by the prediction model 130.

[0160] For a simple distillation column, the model input variables for the performance estimation tool 100 are the feed flow rate, the types of compounds, and the number of trays in the rectifying and stripping sections. The model output variables are the flow rates and types of compounds of the TAC, top and bottom products.

[0161] For a wall-separated distillation column, the model input variables for the performance estimation tool 100 are the feed flow rate, composition, number of trays in the six sections of the wall-separated distillation column, and gas-liquid phase separation ratio. The model output variables are TAC, the flow rates of the top, side stream, and bottom products, and the types of compounds.

[0162] To establish a performance estimation tool 100 based on an artificial neural network, sufficient modeling data must first be collected. Before starting data collection, the input and output variables of the classification model 110 and the prediction model 130 in the performance estimation tool 100 need to be determined.

[0163] Considering that SDC and DWC need to be optimized under different feed flow rates and different compound compositions, the feed conditions and decision variables of SDC and DWC are defined as the input variables of the corresponding classification model 110 and prediction model 130.

[0164] The input variables for SDC's classification model 110 and prediction model 130 are defined as follows:

[0165] [x F，A x F，B x F，C x F，D x F，E X F，F L F N S +N R N R ];

[0166] The input variables for DWC's classification model 110 and prediction model 130 are defined as follows:

[0167] [x F，A x F，B x F,C x F，D x F，E x F，F N S3 +N S4 +N S5 +N S6 N S1 +N S2 N S3 N S1 N S3 +N S4 L MP V MP ].

[0168] Table 1 shows the output variables of the prediction model 130 for SDC and DWC, where SDC is a simple distillation column and DWC is a wall-separated distillation column.

[0169]

[0170] Table 1

[0171] The outputs of both SDC and DWC classification models 110 are Boolean variables. The collection of modeling data can be automatically performed by process simulation software. The feed flow rate ranges from 75 kmol / h to 510 kmol / h, and the relative deviation between the highest and lowest mole fractions in the feed components does not exceed 20%.

[0172]

[0173] Table 2

[0174] Table 2 shows the number of feasible and infeasible solutions in the modeling samples collected for the separation operations corresponding to the 15 options. Among them, the proportion of infeasible samples in SDC is 0.07%-3.11% of the total samples, and the proportion of infeasible samples in DWC is 4.01%-5.43% of the total samples.

[0175] Table 3 shows the R-values ​​of SDC and DWC classification model 110 and prediction model 130. 2 , where R 2 R represents the goodness of fit of the model. 2 The closer the value is to 1, the higher the accuracy of the model.

[0176]

[0177] Table 3

[0178] After collecting sufficient modeling samples, the classification model 110 and prediction model 130 in the performance estimation tool 100 are iteratively trained. All classification models 110 and prediction models 130 are configured as neural networks with two hidden layers, each containing 40 neurons. The ratio of training set to test set in the modeling samples is 4:1. The learning rate can be adaptively adjusted during training. The maximum number of iterations in this embodiment is 3000. Table 3 shows the training results of the performance estimation tool 100 for the separation operation corresponding to 15 options. It can be seen that the R-values ​​of the output variables of most models are... 2 The values ​​are all around 0.99, indicating that the iterative training of classification model 110 and prediction model 130 is excellent.

[0179] Based on the above embodiments of the present invention, a program for optimizing compound distillation sequences based on a genetic programming algorithm can be developed. In the embodiments of the present invention, the parameters of the genetic programming algorithm are preset as follows: population size = 60, maximum generation = 20, tournament size = 5, crossover rate = 0.7, and mutation rate = 0.3.

[0180] The parameters for the genetic programming algorithm used to optimize DWC are set as follows: population size = 200, maximum number of generations = 150, crossover rate = 0.7, mutation rate = 0.3.

[0181] Figure 13 The optimization process of the complex distillation sequence is shown. It can be seen that after 5 generations, the evolutionary process tends to stabilize.

[0182] like Figure 14 The optimal combined distillation sequence is shown. To verify the reliability of the optimization results, simulations were performed based on the optimal combined distillation sequence obtained by performance estimation tool 100. According to the simulation results, the predicted and actual TAC values ​​for column 1 in the optimal combined distillation sequence are 1026237 and 1025163, respectively; for column 2, they are 695772 and 728186, respectively; and for column 3, they are 1313585 and 1442575, respectively. The predicted and actual TAC values ​​for the entire combined distillation sequence are 3035596 and 3195924, respectively, with a relative prediction error of -5.02%. This prediction error is close to the prediction error result of performance estimation tool 100 and is within the acceptable error range for practical applications, meeting the accuracy requirements for the optimized results of the combined distillation sequence.

[0183] Based on the prediction results, for the six-component mixture feed in the embodiments of the present invention, the TAC of the optimal direct distillation sequence is $3,647,984 / year, and the TAC of the optimal indirect distillation sequence is $6,361,462 / year. The TAC of the optimal compound distillation sequence is reduced by 12.39% and 49.76% compared with the optimal direct distillation sequence and the optimal indirect distillation sequence, respectively.

[0184] As can be seen, the composite distillation sequence optimization method proposed in this embodiment simplifies the optimization difficulty, solves the "combinatorial explosion" problem in the composite distillation sequence optimization process, overcomes the convergence problem of training models for simple distillation columns and isolation wall distillation columns and their impact on sequence optimization, realizes effective iterative optimization of the optimal simple distillation column-isolation wall distillation column composite distillation sequence design, ensures the global optimality of the distillation unit, and also improves distillation efficiency and reduces distillation cost.

[0185] In addition to the above method embodiments, the present invention also provides, for example, Figure 15 The apparatus shown includes a compound distillation sequence optimization device, comprising a processor 210 and a memory 220, wherein the memory 220 stores instructions; the processor 210 calls the instructions in the memory 220, thereby causing the processor 210 to execute the compound distillation sequence optimization method based on genetic programming algorithm according to any of the foregoing embodiments of the present invention.

[0186] The aforementioned method for optimizing compound distillation sequences based on genetic programming algorithms of the present invention includes: obtaining an initial mixture to be distilled and a set of compound distillation sequences for distilling the initial mixture; converting the compound distillation sequences into an initial tree population based on a genetic programming algorithm; iteratively optimizing the initial tree population based on a tournament algorithm to obtain an optimized tree population, and simultaneously iteratively optimizing the nodes of the initial tree population based on an exhaustive search method or an elite genetic algorithm to obtain a first optimal node; iteratively optimizing the nodes of the optimized tree population based on an exhaustive search method or an elite genetic algorithm to obtain a second optimal node; comparing the historical minimum annual total cost of the initial tree population and the optimized tree population, and updating the compound distillation sequence to the optimized distillation sequence corresponding to the one with the smaller historical minimum annual total cost; comparing the historical minimum annual total cost of the first optimal node and the second optimal node, and updating the initial distillation unit of the compound distillation sequence to the optimized distillation unit corresponding to the one with the smaller historical minimum annual total cost.

[0187] The processor 210 of the compound distillation sequence optimization device provided in this embodiment of the invention executes the compound distillation sequence optimization method based on genetic programming algorithm according to any of the foregoing embodiments of the invention by calling instructions in memory 220. It can decompose the optimization problem of compound distillation sequence into two parts: sequence optimization process and distillation unit optimization process. Through genetic programming algorithm, it can optimize the compound distillation sequence and the distillation unit at the same time. It updates the compound distillation sequence to the optimized distillation sequence and the initial distillation unit to the optimized distillation unit, realizing synchronous iterative optimization of compound distillation sequence and initial distillation unit. This improves the optimization efficiency of distillation sequence, reduces distillation cost, solves the "combinatorial explosion" problem in the compound distillation sequence optimization process, improves the accuracy of optimization calculation results, and thus ensures the global optimality of distillation unit.

[0188] Furthermore, the compound distillation sequence optimization device provided in this embodiment of the invention may also include a communication interface 230 and a bus 240, with the processor 210, memory 220 and communication interface 230 electrically connected via the bus 240.

[0189] The memory 220 may include high-speed random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 230 (which can be wired or wireless), such as the Internet, wide area network, local area network, or metropolitan area network. The bus 240 may be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 15 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0190] Processor 210 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of processor 210 or by instructions in software form. The processor 210 can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this invention. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly manifested as execution by a hardware decoding processor, or execution by a combination of hardware and software modules in the decoding processor. The software module can reside in a mature storage medium in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory 220. The processor 210 reads information from memory 220 and, in conjunction with its hardware, completes the steps of the method described in the foregoing embodiments.

[0191] This invention also provides a machine-readable storage medium, which can be a non-volatile machine-readable storage medium or a volatile machine-readable storage medium. The machine-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the steps of the above-described composite distillation sequence optimization method based on genetic programming algorithm.

[0192] The machine-readable storage medium provided in this embodiment of the invention stores data and computer-executable instructions of the above method. The method includes: acquiring an initial mixture to be distilled and a set of compound distillation sequences for distilling the initial mixture; converting the compound distillation sequences into an initial tree population based on a genetic programming algorithm; iteratively optimizing the initial tree population based on a tournament algorithm to obtain an optimized tree population, and simultaneously iteratively optimizing the nodes of the initial tree population based on an exhaustive search method or an elite genetic algorithm to obtain a first optimal node; iteratively optimizing the nodes of the optimized tree population based on an exhaustive search method or an elite genetic algorithm to obtain a second optimal node; comparing the historical minimum annual total cost of the initial tree population and the optimized tree population, and updating the compound distillation sequence to the optimized distillation sequence corresponding to the one with the smaller historical minimum annual total cost; comparing the historical minimum annual total cost of the first optimal node and the second optimal node, and updating the initial distillation unit of the compound distillation sequence to the optimized distillation unit corresponding to the one with the smaller historical minimum annual total cost.

[0193] The instructions stored in the machine-readable storage medium provided in this embodiment of the invention can be called by a processor and executed by the machine-readable storage medium of any of the foregoing embodiments of the invention to decompose the optimization problem of the compound distillation sequence into two parts: a sequence optimization process and a distillation unit optimization process. Through the genetic programming algorithm, the optimization solution of the compound distillation sequence can be obtained simultaneously with the optimization solution of the distillation unit. The compound distillation sequence is updated to the optimized distillation sequence, and the initial distillation unit is updated to the optimized distillation unit. This achieves synchronous iterative optimization of the compound distillation sequence and the initial distillation unit, improves the optimization efficiency of the distillation sequence, reduces the distillation cost, solves the "combinatorial explosion" problem in the optimization process of the compound distillation sequence, improves the accuracy of the optimization calculation results, and thus ensures the global optimality of the distillation unit.

[0194] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0195] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0196] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing a compound distillation sequence based on a genetic programming algorithm, characterized in that, The method is applied to a complex distillation sequence comprising multiple simple distillation columns and a partition wall distillation column, the method comprising: Obtain an initial mixture to be distilled and a set of complex distillation sequences for distilling the initial mixture; Based on the genetic programming algorithm, the composite distillation sequence is transformed into an initial tree population; Based on the tournament algorithm, the initial tree population is iteratively optimized to obtain an optimized tree population. At the same time, based on the exhaustive method or the elite genetic algorithm, the nodes of the initial tree population are iteratively optimized to obtain the first optimal node. Based on the exhaustive search method or the elite genetic algorithm, the nodes of the optimized tree population are iteratively optimized to obtain the second optimal node; Compare the historical minimum total annual cost of the initial tree population and the optimized tree population, and update the composite distillation sequence to the optimized distillation sequence corresponding to the one with the smaller historical minimum total annual cost; The historical minimum annual total cost of the first optimal node and the second optimal node is compared, and the initial distillation unit of the composite distillation sequence is updated to the optimized distillation unit corresponding to the one with the smaller historical minimum annual total cost.

2. The method for optimizing compound distillation sequences based on genetic programming algorithm according to claim 1, characterized in that, The step of converting the composite distillation sequence into an initial tree population based on the genetic programming algorithm includes: Based on the initial mixture, obtain all components in the initial mixture; Based on all the components, obtain the terminal set corresponding to each component, and the set of all functions corresponding to the simple distillation column and the isolation wall distillation column; Based on the genetic programming algorithm, the function set, and the terminal set, multiple initial tree individuals are generated to form the initial tree population.

3. The method for optimizing compound distillation sequences based on genetic programming algorithm according to claim 2, characterized in that, The number of functions is The number of terminal sets is the number of all types of components in the initial mixture, where N is the number of all types of components in the initial mixture.

4. The method for optimizing compound distillation sequences based on genetic programming algorithm according to claim 1, characterized in that, The step of iteratively optimizing the initial tree population based on the tournament algorithm to obtain the optimized tree population includes: Based on the tournament algorithm, a parent tree population is selected from the initial tree population; By controlling the crossover of different parent individuals in the parent tree population, offspring individuals are generated; The offspring individuals are controlled to mutate, forming multiple mutated offspring individuals, to obtain an optimized tree population including multiple mutated offspring individuals.

5. The method for optimizing compound distillation sequences based on genetic programming algorithm according to claim 4, characterized in that, The step of selecting the parent tree population from the initial tree population based on the tournament algorithm includes: The initial tree population is divided into multiple tree groups; Based on the tournament algorithm, the tree individual with the lowest annual total cost in each tree group is selected as the parent individual; All the parent individuals are combined to form the parent tree population.

6. The method for optimizing compound distillation sequences based on genetic programming algorithm according to claim 4, characterized in that, The feed components are the same between the parent individuals that are cross-crossed.

7. The method for optimizing compound distillation sequences based on genetic programming algorithm according to claim 4, characterized in that, The step of controlling the crossover of different parent individuals in the parent tree population to generate offspring individuals includes: Obtain each feed component from different parent individuals in the parent tree population; The offspring individuals are generated by cross-pollinating between parent individuals with the same feed composition.

8. The method for optimizing compound distillation sequences based on genetic programming algorithm according to claim 7, characterized in that, The step of controlling the offspring individuals to mutate, forming multiple mutated offspring individuals, to obtain an optimized tree population including multiple mutated offspring individuals includes: Select any of the feed components described above; Obtain the subtree of the offspring individual corresponding to the selected feed component; The growth of the subtrees is controlled to form multiple mutant offspring individuals to obtain the optimized tree population.

9. The method for optimizing compound distillation sequences based on genetic programming algorithm according to claim 1, characterized in that, The step of comparing the historical minimum total annual cost of the initial tree population and the optimized tree population, and updating the composite distillation sequence to the optimized distillation sequence corresponding to the one with the smaller historical minimum total annual cost, includes: Calculate the first initial year total cost for each initial tree in the initial tree population, and the first optimized year total cost for each optimized tree in the optimized tree population; The first initial annual total cost is compared with the first optimized annual total cost. If the first optimized annual total cost is less than or equal to the first initial annual total cost, the initial tree population is updated to the optimized tree population. Based on the optimized tree population, the composite distillation sequence is updated to the optimized distillation sequence corresponding to the optimized tree individual with the lowest historical annual total cost.

10. The method for optimizing compound distillation sequences based on genetic programming algorithm according to claim 9, characterized in that, After the step of updating the composite distillation sequence to the optimized distillation sequence corresponding to the optimized tree population, the method further includes: If the first optimized annual total cost is greater than the first initial annual total cost, then the current iteration number of the optimized tree population is detected; Obtain the maximum number of iterations for the optimized tree population; The current iteration number of the optimized tree population is compared with the maximum iteration number of the optimized tree population. If the current iteration number is less than the maximum iteration number, the optimized tree population continues to iterate and optimize until the current iteration number is greater than or equal to the maximum iteration number.

11. The method for optimizing compound distillation sequences based on genetic programming algorithm according to claim 1, characterized in that, Based on the exhaustive method, the nodes corresponding to the simple distillation column are iteratively optimized. Based on the elite genetic algorithm, the nodes corresponding to the isolation wall distillation column are iteratively optimized.

12. The method for optimizing compound distillation sequences based on genetic programming algorithm according to claim 1, characterized in that, The step of iteratively optimizing the nodes of the initial tree population to obtain the first optimal node includes: Based on the elite retention strategy and roulette wheel method, each node individual is selected from the nodes of the initial tree population; Control the different individual nodes to cross over and generate child nodes; The node's offspring individuals are controlled to mutate, forming multiple mutated node individuals, in order to obtain the first optimal node including multiple mutated node individuals.

13. The method for optimizing compound distillation sequences based on genetic programming algorithm according to claim 12, characterized in that, The feed components are the same between the individual nodes that are crossing each other.

14. The method for optimizing compound distillation sequences based on genetic programming algorithm according to claim 12, characterized in that, The step of selecting each individual node from the nodes of the initial tree population based on the elite retention strategy and roulette wheel method includes: The nodes of the initial tree population are split into multiple node groups; Based on the elite retention strategy and roulette wheel method, the individual with the lowest annual total cost in each node group is selected as the node individual.

15. The method for optimizing compound distillation sequences based on genetic programming algorithm according to claim 12, characterized in that, The step of controlling the crossover of different individual nodes to generate offspring nodes includes: Based on the initial mixture, multiple feed components are obtained; The nodes with the same feed composition are controlled to cross each other to generate the node offspring.

16. The method for optimizing compound distillation sequences based on genetic programming algorithm according to claim 15, characterized in that, The step of controlling the mutation of the node's offspring individuals to form multiple mutated node individuals, in order to obtain the first optimal node including the multiple mutated node individuals, includes: Select any of the feed components described above; Obtain the subtree corresponding to the selected feed component; Controlling the growth of the subtree to form multiple mutated node individuals, to obtain the first optimal node including multiple mutated node individuals.

17. The method for optimizing compound distillation sequences based on genetic programming algorithm according to claim 1, characterized in that, The step of comparing the historical minimum total annual cost of the first optimal node and the second optimal node, and updating the initial distillation unit of the composite distillation sequence to the optimized distillation unit corresponding to the one with the smaller historical minimum total annual cost, includes: Calculate the second initial year total cost for each individual in the first optimal node, and the second optimized year total cost for each individual in the second optimal node; The second initial annual total cost is compared with the second optimized annual total cost. If the second optimized annual total cost is less than or equal to the second initial annual total cost, the initial distillation unit is updated to the optimized distillation unit corresponding to the second optimal node.

18. The method for optimizing compound distillation sequences based on genetic programming algorithm according to claim 17, characterized in that, The step of updating the initial distillation unit of the composite distillation sequence to an optimized distillation unit corresponding to the one with the smaller historical minimum annual total cost further includes: If the second optimized annual total cost is greater than the second initial annual total cost, then the current iteration number of the nodes in the optimized tree population is detected; Obtain the maximum number of iterations for the nodes of the optimized tree population; The current iteration count is compared with the maximum iteration count. If the current iteration count is less than the maximum iteration count, the nodes of the optimized tree population continue to be iterated and optimized until the current iteration count is greater than or equal to the maximum iteration count.

19. A composite distillation sequence optimization device, characterized in that, The device includes: a processor and a memory, wherein the memory stores instructions; The processor invokes the instructions in the memory to cause the device to implement the compound distillation sequence optimization method based on genetic programming algorithm as described in any one of claims 1 to 18.

20. A machine-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the machine, they implement the compound distillation sequence optimization method based on the genetic programming algorithm as described in any one of claims 1 to 18.

Images