Gasoline blending rescheduling method and system based on real-time optimization relaxation feedback
Through the gasoline blending and rescheduling method based on real-time optimization of relaxation feedback, combined with the NSGA-II algorithm and attribute relaxation feedback mechanism, the problems of fluctuations in the attributes and flows of refined oil in the oil refinery are solved, and the stability and efficiency of production plans are improved. .
Patent Information
- Application Number
- CN202410957028.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-17
- Publication Date
- 2024-11-15
AI Technical Summary
The fluctuations in crude oil supply and uncertainties in the production process in the refinery cause fluctuations in the attributes and flow of refined oil, resulting in the non-compliance of quality standards. The existing real-time optimization technology is limited when adjusting the mixing ratio, which affects the production efficiency and scheduling plan. stability.
Using the gasoline blending and rescheduling method based on real-time optimization of relaxation feedback, the gasoline blending and scheduling is optimized by building a multi-objective model, combining the NSGA-II algorithm and attribute relaxation feedback mechanism, to optimize the gasoline blending and scheduling, to ensure the balance between refined oil quality and production plan, and to improve the production process efficiency.
It achieves a balance between ensuring the quality of refined oils meets standards and maintaining the stability of production scheduling plans, improves the efficiency of the entire production process, and reduces operating costs and unqualified products in the production process.
Smart Images

Figure CN118966626A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of industrial device modeling, and in particular to a gasoline blending and rescheduling method based on real-time optimization relaxation feedback, a gasoline blending and rescheduling system based on real-time optimization relaxation feedback, and a computer-readable storage medium. Background Art
[0002] In the production process of the refinery, the properties and flow of the component oils often fluctuate due to uncertainties such as fluctuations in crude oil supply, errors in process and operation, and environmental factors. These fluctuations may cause the final finished oil to fail to meet the established quality standards, resulting in unqualified products. When unqualified finished oil appears, in order to make it meet the standards, these oils usually need to be re-blended. This process not only increases operating costs, but may also affect gasoline production efficiency and delivery time.
[0003] In order to reduce the impact of component oil properties and flow fluctuations, many refineries have adopted pipeline blending real-time optimization technology. This technology can implement rolling optimization during the component oil blending process, thereby adjusting the mixing ratio of the component oils in real time during the dynamic process to minimize the impact of fluctuations and ensure the quality stability of the finished oil. However, the real-time optimization process is subject to certain restrictions during adjustment. Since the production scheduling plan of the refinery needs to be executed within a certain range, the real-time optimization process has limited adjustment range for the component oil mixing ratio. If the adjustment range is too large, it may affect the execution of the subsequent production scheduling plan, resulting in chaos and inefficiency in the entire production process.
[0004] In order to overcome the above-mentioned defects of the prior art, there is an urgent need in the art for a gasoline blending rescheduling technology based on real-time optimized relaxation feedback, which is used to find a balance between ensuring that the quality of refined oil meets the standards and maintaining the stability of the production scheduling plan, thereby improving the efficiency of the entire production process. Summary of the invention
[0005] A brief summary of one or more aspects is given below to provide a basic understanding of these aspects. This summary is not an exhaustive overview of all conceived aspects, and is neither intended to identify the key or decisive elements of all aspects nor to define the scope of any or all aspects. Its only purpose is to give some concepts of one or more aspects in a simplified form as a prelude to a more detailed description that will be given later.
[0006] In order to overcome the above-mentioned defects in the prior art, the present invention provides a gasoline blending rescheduling method based on real-time optimization relaxation feedback, a gasoline blending rescheduling system based on real-time optimization relaxation feedback, and a computer-readable storage medium, which are used to find a balance between ensuring that the quality of refined oil meets the standards and maintaining the stability of the production scheduling plan, thereby improving the efficiency of the entire production process.
[0007] Specifically, the gasoline blending rescheduling method based on real-time optimization relaxation feedback provided in accordance with the first aspect of the present invention includes the following steps: constructing a multi-objective model of gasoline blending scheduling with the goal of maximizing the output and pickup volume of refined oil and minimizing the excess octane number, and determining its corresponding first constraint condition; obtaining sample data of gasoline blending scheduling, and solving the multi-objective model accordingly to determine the first optimal scheduling scheme; constructing a real-time optimization model considering attribute relaxation, and determining its corresponding optimization objective function and second constraint condition, wherein the real-time optimization model includes a cumulative blending model and a research octane number blending effect model; using attribute relaxation feedback to correct the objective function and the first constraint condition of the multi-objective model to construct a gasoline blending rescheduling model based on real-time optimization relaxation feedback; and solving the gasoline blending rescheduling model according to the sample data to determine the second optimal scheduling scheme based on real-time optimization relaxation feedback.
[0008] Further, in some embodiments of the present invention, the first objective function of maximizing the refined oil production and the pickup volume is expressed as: The second objective function of minimizing the excess octane number is expressed as: Among them, F t,c,n is the oil product movement flow rate of the nth storage tank of component oil c in period t; g t,p,m LP is a binary variable indicating whether the m-th storage tank of finished oil p in period t receives the finished oil. A value of 0 indicates that the m-th storage tank of finished oil p in period t does not receive the finished oil, and a value of 1 indicates that the m-th storage tank of finished oil p in period t receives the finished oil. t,p,m is the delivery quantity of the mth storage tank of refined oil p in period t; Q c,ron is the research octane property of component oil c; It is the lower limit of the research octane number property of finished oil.
[0009] Further, in some embodiments of the present invention, the first constraint condition includes: component tank material balance constraint: CS t,c,n =CS t-1,c,n +Cr t,c,n ·CP t,c -F t,c,n H, where CS t,c,nis the inventory of component oil in the nth storage tank of component oil c in period t. t-1,c,n Cr is the inventory of component oil in the nth storage tank of component oil c in period t-1. t,c,n Indicates whether the nth storage tank of component oil c in period t is used to receive the component oil planned to be produced in period t. CP t,c is the planned production of component oil c in period t. And / or component oil tank capacity constraint: in, is the lower limit of the inventory of the nth storage tank of component oil c; The inventory limit of the nth tank of component oil c. And / or blending formula constraints: in, Indicates the lower limit of the formula of component oil c when blending finished oil p; Indicates the upper limit of the formulation of component oil c when blending finished oil p. And / or finished product tank inventory material balance constraint: PS t,p,m =PS t-1,p,m +F t,p,m ·H-LP t,p,m , where PS t,p,m represents the inventory of the mth storage tank of finished oil p at the end of period t; PS t-1,p,m Represents the inventory of the mth tank of finished oil p at the end of period t-1. And / or finished oil tank capacity constraint: in, Indicates the lower limit of the inventory capacity of the mth storage tank of refined oil p; Indicates the upper limit of the inventory capacity of the mth storage tank of refined oil p. And / or the refined oil delivery quantity constraint: in, It represents the lower limit of the single pick-up quantity of the m-th storage tank of refined oil p; The upper limit of the single pick-up quantity of the m-th storage tank of refined oil p. And / or refined oil demand constraints: in, It represents the lower limit of demand for refined oil product p; Indicates the upper limit of demand for refined oil p. And / or refined oil attribute constraints: in, is the lower limit of the attribute of the finished oil product p; is the upper limit of the first attribute of the refined oil product p; Q c,j is the attribute value of the jth attribute of component oil c. And / or component tank parameter adjustment constraint: (1-x t,c,n )·F t,c,n =0 Among them, x t,c,n Binary variable for component oil tank selection, x t,c,n =1 means the nth tank parameter adjustment of component oil c in period t, xt,c,n =0 means no adjustment. And / or component oil receiving constraints: Among them, Cr t,c,n The storage tank of component oil c is used to receive component oil c planned to be produced in period t, and C represents the number of types of component oil to be adjusted. And / or finished product tank adjustment constraint: (1-g t,p,m )·F t,p,m =0, Among them, g t,p,m A binary variable representing the number of finished oil products that the mth storage tank receives in period t, F t,p,m It represents the flow rate of the product p received by the mth tank in the period t. And / or the flow balance constraint of the blending head: Among them, F t,c,n represents the flow rate of oil product movement in the nth tank of component oil c within period t, F t,p,m Represents the flow rate of finished oil p received by the mth storage tank in period t. And / or finished product tank blending and delivery constraints: LP t,p,m ·(1-g t,p,m )=0, where LP t,p,m represents the pickup quantity of the mth storage tank of refined oil p in period t, g t,p,m A binary variable indicating that the mth storage tank of finished oil p receives finished oil in period t.
[0010] Furthermore, in some embodiments of the present invention, the step of solving the multi-objective model includes: using the NSGA-II algorithm to solve the multi-objective model to obtain a corresponding non-dominated solution set; and selecting a knee-point-based global optimal solution from the non-dominated solution set as the first optimal scheduling solution.
[0011] Further, in some embodiments of the present invention, the step of using the NSGA-II algorithm to solve the multi-objective model to obtain the corresponding non-dominated solution set includes: constructing an initial population of size N; and evolving the initial population for a preset number of rounds based on crossover mutation until a non-dominated solution set after a specified number of evolutions is obtained, wherein for the t-th evolution, the parent population P of size N is replaced by t With a population Q of size N t Merge to get a merged population R of size 2N t , using the fast non-dominated sorting method, the merged population R t Sort multiple individuals in the , and divide the multiple individuals into l levels {F1, F2, ... F l}, and then create an empty offspring population P t+1 , and add the individuals of each level to the offspring population P in order from small to large t+1 To continue the t+1th evolution.
[0012] Further, in some embodiments of the present invention, the step of constructing an initial population of size N includes: creating binary variables of flow, pick-up quantity, and component oil reception in the gasoline blending scheduling plan as individuals in the population, wherein the binary variables include 0 / 1 variables and real number variables in encoding form; randomly generating a flow value as the component oil flow according to the selected component tank, and expressing the finished oil flow as the sum of the component oil flows to construct individual chromosomes; randomly selecting a finished product tank for pick-up and assigning it a random pick-up quantity, wherein pick-up is allowed from multiple finished product tanks in each cycle; randomly selecting the component tank to receive the component oil, wherein only one component tank is allowed to be selected for each component oil in each cycle; and in response to completing the construction of each individual, calculating the objective function value and constraint violation degree of the individual, and filling each individual into the empty target population in sequence according to the objective function value and the constraint violation degree until the target population is filled up to serve as the initial population.
[0013] Furthermore, in some embodiments of the present invention, the individuals of each level are added to the offspring population P in order from small to large. t+1 The steps in include: determining whether to add the i-th level F i After all individuals in the offspring population P t+1 Will the number of individuals be greater than N? In response to the added offspring population P t+1 The number of individuals is less than or equal to N, and the level F i All individuals in are added to the offspring population P t+1 and the progeny population P in response to the addition t+1 The number of individuals is greater than N, and the crowding operator is used to i The individuals in the above-mentioned are sorted based on the crowding distance, and the level F is i The remaining individuals with the smallest crowding distance are added to the offspring population P t+1 Until the progeny population P t+1 The number of individuals in reaches N.
[0014] Furthermore, in some embodiments of the present invention, the step of constructing a real-time optimization model that takes into account property relaxation includes: converting the property index range of the full tank oil product into the index range of the blending head, and optimizing the blending process by adjusting the property of the finished oil at the blending head:
[0015]
[0016] in, and They represent the upper and lower limits of the attribute j of the entire tank of oil, and They represent the upper and lower limits of the blending head attribute index that make the upper and lower limits of the attribute j of the entire tank oil product meet the specified range, VPro tol,j VPro represents the cumulative blending volume attribute product of the entire tank of oil attribute j. Heel,j Represents the attribute product of the tank bottom oil volume of the full tank oil attribute j, VolH represents the tank bottom oil volume, VolS represents the blending target volume, VolT represents the blended volume, and VolR represents the unblended volume.
[0017] Further, in some embodiments of the present invention, the step of constructing a real-time optimization model that takes into account property relaxation includes: expressing the research octane number blending effect model as: in, Represents the octane value of component oil i after compensation for blending effect, RON i Indicates the octane value of component oil i, pb i Indicates the blending effect compensation value of component oil i, RON b represents the predicted value of the octane number of the blended product, r i represents the mass formula of component oil i, ρ i Represents the density of component oil i.
[0018] Furthermore, in some embodiments of the present invention, the second constraint condition includes: attribute constraint condition: in, represents the property j of component oil i, and the relaxation parameter δ j >0 means that the attribute j of the refined oil does not satisfy the corresponding lower limit constraint, and the relaxation parameter δ j <0 means that the attribute j of the refined oil does not satisfy the corresponding upper limit constraint, and the relaxation parameter δ j = 0 means that the attribute j of the finished oil does not violate the attribute constraint. And / or the formula constraint: r i min ≤r i ≤r i max i=1,2,…,n, Among them, r i min represents the lower limit of the formulation of component oil i, r i max Indicates the upper limit of the formulation of component oil i. And / or the formulation change step constraint: d ≥ r i_last -r i , d ≥ r i -r i_last, where d represents the maximum step size of a single change, r i_last Indicates the blending formula of component oil i calculated by the last optimization. And / or component oil flow constraint: r c ·total_fiow<Power c , where total_fiow represents the total blending flow, Power c Represents the flow rate parameter of the oil pump in the movement path of component oil c.
[0019] Further, in some embodiments of the present invention, the step of modifying the objective function and the first constraint of the multi-objective model by adopting the attribute relaxation feedback method to construct a gasoline blending rescheduling model based on real-time optimization relaxation feedback includes: responding to the relaxation parameter δ i >0, increase the lower limit of the finished oil property constraint: In response to the relaxation parameter δ i <0, lower the upper limit of the refined oil property constraint: and a relaxation parameter δ corresponding to the octane value attribute ron > 0, add the relaxation parameter δ to the second objective function of minimizing the excess octane number ron :
[0020] In addition, the gasoline blending rescheduling system based on real-time optimized relaxation feedback provided according to the second aspect of the present invention includes: a memory on which computer instructions are stored; and a processor connected to the memory and configured to execute the computer instructions stored on the memory to implement the gasoline blending rescheduling method based on real-time optimized relaxation feedback as described in any one of the first aspect of the present invention.
[0021] In addition, the computer-readable storage medium provided according to the third aspect of the present invention stores computer instructions, which, when executed by a processor, implement the gasoline blending rescheduling method based on real-time optimized relaxation feedback as described in any one of the first aspects of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] The above features and advantages of the present invention can be better understood after reading the detailed description of the embodiments of the present disclosure in conjunction with the following drawings. In the drawings, the components are not necessarily drawn to scale, and components with similar related properties or features may have the same or similar reference numerals.
[0023] Figure 1 A schematic diagram of a gasoline blending device provided according to some embodiments of the present invention is shown.
[0024] Figure 2A flow chart of a gasoline blending rescheduling method based on real-time optimized relaxation feedback according to some embodiments of the present invention is shown.
[0025] Figure 3 A refined oil pickup quantity scheduling scheme provided according to some embodiments of the present invention is shown.
[0026] Figure 4 A component oil receiving scheduling scheme provided according to some embodiments of the present invention is shown.
[0027] Figure 5 Flow scheduling schemes for various storage tanks provided according to some embodiments of the present invention are shown.
[0028] Figure 6 The gasoline blending process tank volume allocation according to some embodiments of the present invention is shown.
[0029] Figure 7 A flowchart of real-time rolling optimization of gasoline blending provided according to some embodiments of the present invention is shown.
[0030] Figure 8 A gasoline blending rescheduling flow chart based on real-time optimized relaxation feedback according to some embodiments of the present invention is shown.
[0031] Fig. 9 A component oil receiving scheduling scheme provided according to some embodiments of the present invention is shown.
[0032] Fig.10 Flow scheduling schemes for various storage tanks provided according to some embodiments of the present invention are shown.
[0033] Fig.11 A refined oil pickup quantity scheduling scheme provided according to some embodiments of the present invention is shown. DETAILED DESCRIPTION
[0034] The following specific embodiments illustrate the implementation of the present invention, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. Although the description of the present invention will be introduced in conjunction with the preferred embodiment, this does not mean that the features of this invention are limited to this implementation. On the contrary, the purpose of introducing the invention in conjunction with the implementation is to cover other options or modifications that may be extended based on the claims of the present invention. In order to provide a deep understanding of the present invention, the following description will include many specific details. The present invention can also be implemented without using these details. In addition, in order to avoid confusion or blurring the focus of the present invention, some specific details will be omitted in the description.
[0035] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0036] In addition, the terms "upper", "lower", "left", "right", "top", "bottom", "horizontal" and "vertical" used in the following description should be understood as the directions shown in the paragraph and the related drawings. Such relative terms are only used for the convenience of description and do not mean that the device described therein must be manufactured or operated in a specific direction, and therefore should not be understood as limiting the present invention.
[0037] It is understood that although the terms "first", "second", "third", etc. may be used herein to describe various components, regions, layers and / or parts, these components, regions, layers and / or parts should not be limited by these terms, and these terms are only used to distinguish different components, regions, layers and / or parts. Therefore, the first component, region, layer and / or part discussed below may be referred to as a second component, region, layer and / or part without departing from some embodiments of the present invention.
[0038] As mentioned above, in order to reduce the impact of component oil properties and flow fluctuations, many refineries have adopted pipeline blending real-time optimization technology. This technology can implement rolling optimization during the component oil blending process, thereby adjusting the mixing ratio of the component oils in real time during the dynamic process to minimize the impact of fluctuations and ensure the quality stability of the finished oil. However, the real-time optimization process is subject to certain restrictions during adjustment. Since the production scheduling plan of the refinery needs to be executed within a certain range, the real-time optimization process has limited adjustment range for the component oil mixing ratio. If the adjustment range is too large, it may affect the execution of the subsequent production scheduling plan, resulting in confusion and reduced efficiency of the entire production process.
[0039] In order to overcome the above-mentioned defects of the prior art, the present invention provides a gasoline blending rescheduling technology based on real-time optimization relaxation feedback, which is used to find a balance between ensuring that the quality of finished oil meets the standards and maintaining the stability of the production scheduling plan, thereby improving the efficiency of the entire production process.
[0040] In some non-limiting embodiments, the gasoline blending rescheduling method based on real-time optimized relaxation feedback provided in the first aspect of the present invention can be implemented based on the gasoline blending rescheduling system based on real-time optimized relaxation feedback provided in the second aspect of the present invention. Specifically, the gasoline blending rescheduling system based on real-time optimized relaxation feedback is configured with a memory and a processor. The memory includes but is not limited to the above-mentioned computer-readable storage medium provided in the third aspect of the present invention, on which computer instructions are stored. The processor is connected to the memory and is configured to execute the computer instructions stored on the memory to implement the above-mentioned gasoline blending rescheduling method based on real-time optimized relaxation feedback provided in the first aspect of the present invention.
[0041] The following will describe the working principle of the above-mentioned gasoline blending rescheduling system based on real-time optimization and relaxation feedback in combination with some embodiments of the gasoline blending rescheduling method based on real-time optimization and relaxation feedback. Those skilled in the art can understand that these embodiments of the gasoline blending rescheduling method based on real-time optimization and relaxation feedback are only some non-restrictive implementation methods provided by the present invention, which are intended to clearly demonstrate the main concept of the present invention and provide some specific solutions that are convenient for the public to implement, rather than to limit all functions or all working modes of the gasoline blending rescheduling system based on real-time optimization and relaxation feedback. Similarly, the gasoline blending rescheduling system based on real-time optimization and relaxation feedback is also only some restrictive implementation methods provided by the present invention, and does not constitute a limitation on the execution subject or execution order of each step in these gasoline blending rescheduling methods based on real-time optimization and relaxation feedback.
[0042] Please refer to Figure 1 , Figure 1 A schematic diagram of a gasoline blending device provided according to some embodiments of the present invention is shown.
[0043] like Figure 1 As shown, Figure 1The figure is a schematic diagram of a gasoline blending device. The gasoline blending system mainly includes component tanks, finished product tanks, blending heads, pipelines, oil pumps and other equipment. During the production process of the refinery, crude oil processing continues, and the various fractions produced are processed by secondary devices to form component oils with specific properties. These components are continuously output to the blending head through pipelines, mixed into finished oils of different grades in the static blender and stored in the corresponding finished product tanks, and finally picked up from the finished product tanks according to customer needs. Among them, each component oil and finished oil can be stored in multiple tanks, but at any time, each tank can only store one oil product. A blending head can blend multiple finished oils, but at any time, the blending head can only produce one finished oil. Once blending begins, the blending head must run for a period of time before switching to another finished oil. After the blending head switches the finished oil, it is necessary to adjust the blending component oil formula and the finished oil quality index. In order to reduce operating costs, the number of finished oil switches should be minimized. Each blending head can only load finished oil tanks into one product tank at any time, and at any time, each finished product tank can only store one finished oil. Finished tanks can be received and delivered, but not simultaneously.
[0044] Please refer to Figure 2 , Figure 2 A flow chart of a gasoline blending rescheduling method based on real-time optimized relaxation feedback according to some embodiments of the present invention is shown.
[0045] like Figure 2 As shown, the gasoline blending rescheduling method based on real-time optimization relaxation feedback can first perform functional step S1: constructing a refinery gasoline blending scheduling model.
[0046] Specifically, the model is modeled using a discrete time model. The time span of each scheduling cycle is equal, and all events within the scheduling time range occur at the start and end time points of the cycle. The gasoline blending scheduling problem contains multiple decision-making objectives. The model mainly considers two objective functions, namely, the maximum output and pickup volume of finished oil and the minimum excess of research octane number attributes. NSGA-II is used to solve it, and the knee point is selected as the optimal solution to avoid the problem that linear weighting easily falls into local optimality. Taking into account the limitations of production processes, operating rules, product quality indicators, etc. in the gasoline blending process, formulas for various constraints are constructed in the model. In this model, it is assumed that there are C types of blending component oils, and component oil c (c = 1, 2, ..., C) has N c There are M storage tanks, P products, and product p (p = 1, 2, ..., P) has M p storage tanks, T scheduling cycles, the length of each cycle is H, and there are J attributes.
[0047] The first objective function of maximizing the production and pickup of refined oil is expressed as:
[0048]
[0049] The second objective function of minimizing the octane excess is expressed as:
[0050]
[0051] Among them, F t,c,n is the flow rate of oil product movement in the nth tank of component oil c in period t; g t,p,m LP is a binary variable indicating whether the m-th storage tank of finished oil p receives finished oil in period t. A value of 0 indicates that the m-th storage tank of finished oil p does not receive finished oil in period t, and a value of 1 indicates that the m-th storage tank of finished oil p receives finished oil in period t. t,p,m is the delivery volume of the mth storage tank of refined oil p in period t; Q c,ron Research octane number properties for component oil c; It is the lower limit of the octane number property of the finished oil by the research method.
[0052] The first constraints include:
[0053] a. Material balance constraints for component tanks are as follows:
[0054] CS t,c,n =CS t-1,c,n +Cr t,c,n ·CP t,c -F t,c,n ·H
[0055] Among them, CS t,c,n is the inventory of component oil in the nth storage tank of component oil c in period t; CS t-1,c,n is the inventory of component oil in the nth storage tank of component oil c in period t-1; Cr t,c,n Indicates whether the nth storage tank of component oil c in period t is used to receive the component oil planned to be produced in period t; CP t,c is the planned production of component oil c in period t.
[0056] Here, the constraint indicates that the inventory of the component oil storage tank at the end of period t is equal to the inventory at the end of period t-1 plus the component oil volume received in that period, minus the component oil volume consumed in period t.
[0057] b. Constraints on component tank capacity are as follows:
[0058]
[0059] in, is the lower limit of the inventory of the nth storage tank of component oil c; is the inventory upper limit of the nth storage tank of component oil c.
[0060] Here, the constraint means that the inventory level of the component tank must be kept within the upper and lower limits allowed by the process.
[0061] c. The blending formula constraints are as follows:
[0062]
[0063] in, Indicates the lower limit of the formula of component oil c when blending finished oil p; Indicates the upper limit of the formulation of component oil c when blending finished oil p.
[0064] Here, the constraint represents the component oil formula according to the proportion of the component oil flow in the total flow, and the component oil formula should meet the upper and lower limit constraints of the formula. Since the component oil formula is calculated by flow normalization, it is ensured that the component oil formula always meets the constraint that the sum is 1.
[0065] d. The material balance constraint of finished product tank inventory is as follows:
[0066] PS t,p,m =PS t-1,p,m +F t,p,m ·H-LP t,p,m
[0067] Among them, PS t,p,m represents the inventory of the mth storage tank of finished oil p at the end of period t; PS t-1,p,m Represents the inventory of the mth storage tank of finished oil p at the end of period t-1.
[0068] Here, the inventory of the constrained finished oil storage tank at the end of period t is equal to the inventory at the end of period t-1 plus the amount of finished oil received in period t, and minus the amount of finished oil delivered in period t.
[0069] e. The capacity constraint of finished oil tank is as follows:
[0070]
[0071] in, Indicates the lower limit of the inventory capacity of the mth storage tank of refined oil p; Represents the upper limit of the inventory capacity of the mth storage tank of refined oil p.
[0072] Here, the constraint means that the finished product tank inventory must remain within the upper and lower limits allowed by the process.
[0073] f. The constraints on the quantity of refined oil products to be picked up are as follows:
[0074]
[0075] in, The lower limit of the single pick-up quantity of the mth storage tank of refined oil p; The upper limit of the single pick-up quantity of refined oil p from the mth storage tank.
[0076] Here, the constraint means that in the mth storage tank of the refined oil product p, the pick-up quantity should satisfy the upper and lower limit constraints of the pick-up quantity.
[0077] g. The demand constraints for refined oil products are as follows:
[0078]
[0079] in, It represents the lower limit of demand for refined oil product p; Indicates the upper limit of demand for refined oil product p.
[0080] Here, the constraint means that within T scheduling cycles, the total amount of refined oil received by each storage tank of refined oil p should meet the upper and lower limits of the refined oil demand.
[0081] h. The finished oil attribute constraints are as follows:
[0082]
[0083] in, is the lower limit of the j-th attribute of the refined oil product p; is the upper limit of the jth attribute of the refined oil product p; Q c,j is the attribute value of the jth attribute of component oil c.
[0084] Here, the constraint indicates that the finished oil produced in each cycle should meet the upper and lower limits of the attribute indicators of the target product.
[0085] i. Constraints on component tank parameters are as follows:
[0086] (1-x t,c,n )·F t,c,n =0
[0087]
[0088] Among them, x t,c,n Binary variable for component oil tank selection, x t,c,n =1 means the nth tank parameter adjustment of component oil c in period t, x t,c,n =0 means no adjustment.
[0089] Here, the constraint means that for any period t, at most one storage tank of oil distribution c participates in production.
[0090] j. The component oil receiving constraints are as follows:
[0091]
[0092] Among them, Crt,c,n The nth storage tank representing component oil c is used to receive component oil c planned to be produced within cycle t, and C represents the number of types of component oils to be adjusted.
[0093] Here, the constraint means that within any period t, each component oil produced by the component oil processing device can only be received by one storage tank of the component oil.
[0094] k. The parameters and constraints of the finished product tank are as follows:
[0095] (1-g t,p,m )·F t,p,m =0
[0096]
[0097] Among them, g t,p,m A binary variable representing the number of finished oil products that the mth storage tank receives in period t, F t,p,m It represents the flow rate of finished oil p received by the mth storage tank within period t.
[0098] Here, g t,p,m A value of 0 means that finished oil is not received, and a value of 1 means that finished oil is received. This constraint means that in any cycle t, only one of all the finished oil storage tanks can be selected to receive the finished oil produced in this cycle.
[0099] 1. The flow balance constraint of the blending head is as follows:
[0100]
[0101] Among them, F t,c,n represents the flow rate of oil product movement in the nth tank of component oil c within period t, F t,p,m It represents the flow rate of finished oil p received by the mth storage tank within period t.
[0102] Here, the constraint means that within any period t, the total flow rate of all the blending component oils should be equal to the finished oil flow rate, that is, the input flow rate and output flow rate of the blending head are balanced.
[0103] m. The finished product tank blending and delivery constraints are as follows:
[0104] LP t,p,m ·(1-g t,p,m )=0
[0105] Among them, LP t,p,m represents the pickup quantity of the mth storage tank of refined oil p in period t, g t,p,m A binary variable indicating that the pth storage tank receives finished oil in period t.
[0106] Here, g t,p,mA value of 0 indicates that finished oil is not received, and a value of 1 indicates that finished oil is received. This constraint indicates that within any period t, the finished product tank cannot simultaneously receive and deliver finished oil.
[0107] Furthermore, the gasoline blending rescheduling system based on real-time optimization relaxation feedback can obtain sample data of gasoline blending scheduling, and solve the multi-objective model accordingly to determine the first optimal scheduling plan. Specifically, the steps of solving the multi-objective model include: solving the multi-objective model using the NSGA-II algorithm to obtain the corresponding non-dominated solution set; and selecting the global optimal solution based on the knee point from the non-dominated solution set as the first optimal scheduling plan. Here, since the NSGA-II algorithm searches out a set of non-dominated solution sets, it is usually difficult for decision makers to select the optimal solution from them. At the same time, the high complexity of the gasoline blending scheduling optimization problem and the large number of feasible solutions make it difficult to express decision preferences clearly. Therefore, the knee point is introduced to select the global optimal solution from the non-dominated solution set. The optimal solution selected by the knee point is the optimal scheduling plan.
[0108] The NSGA-II algorithm is used to solve the multi-objective model. First, an initial population of size N can be constructed. Binary variables of flow, pick-up quantity, and component oil reception in the gasoline blending scheduling scheme are created as individuals in the population, where the binary variables include 0 / 1 variables and real number variables in encoding form; a flow value is randomly generated as the component oil flow according to the selected component tank, and the finished oil flow is expressed as the sum of the component oil flow to construct individual chromosomes; a finished product tank is randomly selected for pick-up and a random pick-up quantity is assigned to it, where pick-up is allowed in multiple finished product tanks in each cycle; a component tank is randomly selected to receive the component oil, where only one component tank is allowed to be selected for each component oil in each cycle; and in response to the completion of the construction of each individual, the objective function value and constraint violation degree of the individual are calculated, and each individual is sequentially filled into the empty target population according to the objective function value and constraint violation degree until the target population is filled up as the initial population.
[0109] Specifically, the NSGA-Ⅱ algorithm is a multi-objective evolutionary algorithm improved on the basis of NSGA. Since the NSGA algorithm has a high complexity in calculating non-dominated sorting, the NSGA-Ⅱ algorithm uses fast non-dominated sorting technology to reduce the algorithm's computational complexity from O(MN3) to O(MN2). In addition, NSGA-Ⅱ also introduces an elite retention strategy to accelerate the evolution process by merging offspring and parents and then selecting them. In order to ensure the diversity of solutions, the NSGA algorithm uses a shared function to calculate the proximity between population members. However, this method relies on the selection of shared parameters, and the complexity of calculating the distance between all individuals in the population is high. Therefore, the NSGA-Ⅱ algorithm proposes a crowding operator to calculate the density estimate of the individual, and uses the density estimate as the judgment condition for the dominance relationship.
[0110] Furthermore, the fast non-dominated sorting method sets two attributes for any individual i in the population. The first attribute is the dominance number n i , which represents the number of individuals that dominate individual i in the population, and another attribute is the dominated set S i , which stores the individuals in the population that are dominated by individual i. When sorting, put the individuals with i=0 in the latest layer, then traverse all the individuals in this layer, and then sort each individual’s S i Traverse and convert all the individuals n i Subtract 1. After the traversal is completed, the number of sorting layers is increased by 1, and the next layer of individuals is assigned until all individuals are assigned a layer. In the sorted population, the smaller the number of layers corresponding to the individuals, the closer it is to the Pareto frontier.
[0111] Furthermore, the crowding operator is a dominance operator. When the individual objective function values are the same, the dominance relationship between individuals is judged by the density estimation value, and the individuals are guided to be evenly distributed on the Pareto frontier. Density estimation refers to the density around an individual in a population, which is represented by the average distance between two points along each target on both sides of the individual. The nearest neighbors on both sides of the individual are used as vertices to form a rectangle, and the perimeter of the rectangle is called the crowding distance. The larger the crowding distance value, the smaller the density around the individual. Before calculating the crowding distance, the individuals in the population are sorted according to the target value, and the individuals corresponding to each target boundary value are assigned an infinite crowding distance value. During the calculation process, the function value of the middle individual is set to the normalized difference of the function value. The crowding operator determines the dominance relationship based on the non-dominated level and the crowding distance. For individuals x and y in a population, if the following two conditions are met, then individual x is said to dominate individual y.
[0112] a. When the non-dominated rank of x is less than the non-dominated rank of y.
[0113] The nondominated ranks of bx and y are equal and the crowding distance of x is greater than that of y.
[0114] Then, the initial population is subjected to a preset number of rounds of evolution based on crossover mutation until a non-dominated solution set is obtained after a specified number of evolutions.
[0115] a. For the tth evolution, the parent population P of size N is t With a descendant population Q of size N t Merge to get a population R of size 2N t Here, the binary tournament method is used to select the parent population P. t Select the parent individuals, perform crossover, mutation, and repair operations, and generate a child population Q of size N t
[0116] b. Use the fast non-dominated sorting method to sort the population R t Sort and divide the individuals in the population into l levels {F1, F2, ... F l}.
[0117] c. Create an empty offspring population P t+1 , add individuals of each level to the population P in order from small to large t+1 middle.
[0118] d. If we add the i-th level F i After the individuals, the offspring population P t+1 If the number of individuals is less than or equal to N, then F i All individuals are added to the population P t+1 Otherwise, the crowding operator is used to i Sort the individuals in the cluster by the crowding distance from large to small, and sort the sorted F i Individuals are added to the population P one by one t+1 until the number of individuals in the population reaches N.
[0119] Further, it is determined that the i-th level F is added i After all individuals in the offspring population P t+1 Will the number of individuals be greater than N? In response to the added offspring population P t+1 If the number of individuals is less than or equal to N, the level F i All individuals in are added to the offspring population P t+1 In response to the added progeny population P t+1 The number of individuals is greater than N, and the crowding operator is used to i The individuals in the cluster are sorted based on the crowding distance, and the level F is assigned to each i The remaining individuals with the smallest crowding distance are added to the offspring population P t+1 Until the offspring population P t+1 The number of individuals in reaches N.
[0120] Specifically, in the NSGA-Ⅱ algorithm, the gasoline blending scheduling plan is represented as an individual in the population. The plan contains binary variables of flow, pick-up quantity, and component oil reception, which involve two encoding forms: 0-1 variables and real number variables. Therefore, individual genes are represented by multi-chromosomes. When constructing individual chromosomes, a flow value is randomly generated as the component oil flow according to the selected component tank, and the finished oil flow is equal to the sum of the component oil flow. Next, a finished product tank is randomly selected for pick-up and a random pick-up quantity is assigned. Pick-up can be performed in multiple finished product tanks in each cycle. Finally, a component tank is randomly selected to receive the component oil. Only one tank can be selected for each component oil in each cycle. After each individual is constructed, the objective function value and constraint violation degree of the individual are calculated. Construct an initial population of size, generate individuals in the above manner and fill in the initial population until the population is full.
[0121] The algorithm uses multiple chromosomes to represent individuals, which contain binary variables and continuous variables. Different crossover operations need to be performed for different chromosomes. Therefore, the algorithm uses a combination of multiple crossover operators to perform crossover operations. During the execution of the crossover operation, different crossover operators are selected with the same probability.
[0122] Similar to the crossover method, this algorithm uses a combination of multiple different mutation operators to perform mutations on individual genes. In order to make the mutation more random, each combination of mutation operations is random, and each gene point will be selected with the same probability whether to mutate.
[0123] The algorithm introduces random sorting to modify the dominance criterion, forcing individuals to converge from the infeasible domain to the feasible domain. Random sorting combines the feasibility rule and the Pareto dominance rule, and controls the judgment of the dominance relationship through the probability parameter pf. When the random probability is less than pf, the Pareto dominance method is used to determine the dominance relationship of the individual; otherwise, the feasibility rule is used to determine the dominance relationship of the individual. The random sorting method retains some individuals with better target values but lower feasibility with a certain probability during the evolution process, improves the diversity of the population, and prevents the algorithm from falling into the local optimum.
[0124] Here, since the algorithm searches for a set of non-dominated solutions, it is usually difficult for decision makers to select the optimal solution from them. At the same time, the high complexity of the gasoline blending scheduling optimization problem and the large number of feasible solutions make it difficult to express decision preferences clearly. Therefore, this section introduces the knee point to select the global optimal solution from the non-dominated solution set. The knee point is considered to be the best candidate point on the Pareto frontier.
[0125] Based on the above steps and methods, those skilled in the art can take the actual production process of a domestic refinery as an example, set the scheduling time range to 7 days, the scheduling cycle is in days, and the scheduling result is a gasoline blending production plan for 7 cycles. The component oils used for gasoline blending in this refinery are: reformed gasoline, MBTE, non-aromatic gasoline, catalytic gasoline, C5 (light naphtha), and alkylate. The finished oil products are: 92# gasoline (National VIB), 95# gasoline (National VIB), and 98# gasoline (National VIB). The NSGA-II algorithm is used to solve the scheduling model, and the algorithm parameters are set as follows: the population size is 100, the number of iterations is 30,000 times, the crossover probability is 0.9, the mutation probability is 0.2, and the random sorting method probability threshold pf is set to 0.2.
[0126] Please refer to Tables 1 and 2. Table 1 shows the upper and lower limits of the component oil formula. Table 2 shows the properties of the component oils involved in the blending.
[0127] As shown in Table 1 and Table 2, loose formulation constraints are used in the scheduling optimization of Table 1, and 92# gasoline, 95# gasoline and 98# gasoline use the same constraint range. Table 2 has a total of 12 attribute indicators, and each attribute value takes the average attribute value of the component oil.
[0128] Table 1 Upper and lower limits of component oil formulation
[0129] Constraints Reformed gasoline MBTE Non-aromatic gasoline Catalytic gasoline C5 Alkylate Recipe Cap / % 50 15 39 85 8 81 Formula lower limit / % 0 0 0 0 0 0
[0130] Table 2 Properties of the component oils involved in the blending
[0131]
[0132] Please refer to Tables 3 to 5. Table 3 shows the planned production of component oils. Table 4 shows the lower limits of the finished oil attribute indicators. Table 5 shows the upper limits of the finished oil attribute indicators.
[0133] As shown in Table 3, the experiment assumes that the component oil is produced at a uniform rate, and the data in the table represent the output of the component oil in the corresponding cycle. Tables 4 and 5 are the lower and upper limits of the finished oil attribute indicators, respectively. In the national standard for gasoline, attributes usually have only upper or lower limits. In order to stabilize the attributes of finished oil, upper and lower limits are set for all attributes in actual production.
[0134] Table 3 Component oil planned production
[0135] Cycle / day Reforming gasoline / t MTBE / t Non-aromatic gasoline / t Catalytic gasoline / t C5 / t Alkylate oil / t 1 2000 750 980 5800 360 575 2 2000 750 980 5800 360 575 3 2000 750 980 5800 360 575 4 2000 750 980 5800 360 575 5 2000 750 980 5800 360 575 6 2000 750 980 5800 360 575 7 2000 750 980 5800 360 575
[0136] Table 4 Lower limits of finished oil property indicators
[0137]
[0138] Table 5 Upper limits of finished oil property indicators
[0139]
[0140] Please refer to Tables 6 to 8. Table 6 shows the inventory information of component tanks, Table 7 shows the inventory information of finished product tanks, and Table 8 shows the constraints on finished oil production and pick-up quantity.
[0141] As shown in Tables 6 and 7, the tables contain the upper and lower limits of each tank's inventory and the inventory of each tank at the initial time of scheduling. Among them, G315 and G316 are MTBE tanks, G311 and G312 are non-aromatic gasoline tanks, G701 and G702 are catalytic gasoline tanks, G804 and G806 are reforming gasoline tanks, G305 and G307 are alkylate tanks, and G709 is a C5 tank. G902, G905, and G939 are 92# gasoline tanks, G903, G908, and G940 are 95# gasoline tanks, and G703 and G704 are 98# gasoline tanks.
[0142] Table 9 Group tank inventory information
[0143] Storage tank G804 G806 G315 G316 G311 G312 G701 G702 G709 G310 Initial inventory / t 1600 1600 1200 1200 500 500 2500 2500 400 1200 Inventory lower limit / t 1600 1600 1200 1200 500 500 2500 2500 400 1200 Inventory limit / t 6500 6500 6200 6200 3000 3000 12000 12000 1300 6000
[0144] Table 10 Finished Can Inventory Information
[0145] Storage tank G902 G903 G905 G908 G939 G940 G703 G704 Initial inventory / t 3000 3000 4800 4800 2400 2400 1250 1250 Inventory lower limit / t 3000 3000 4800 4800 2400 2400 1250 1250 Inventory limit / t 40000 40000 40000 40000 20000 20000 7200 7200
[0146] As shown in Table 8, the table contains the upper and lower limits of the single pickup quantity of each brand of refined oil and the upper and lower limits of the total production of each brand of refined oil within the scheduling time range.
[0147] Table 11 Constraints on refined oil production and delivery volume
[0148] Pickup 92# Single delivery quantity 95# Single delivery quantity 98# Single delivery quantity 92# Total output 95# Total output 98# Total output Lower limit / t 200 200 100 20000 20000 5000 Upper limit / t 20000 20000 10000 40000 30000 10000
[0149] Please refer to the optimal scheduling solution selected by the knee point Figure 3 to Figure 5 and Table 9, Figure 3 A refined oil pickup quantity scheduling scheme provided according to some embodiments of the present invention is shown. Figure 4 A component oil receiving scheduling scheme provided according to some embodiments of the present invention is shown. Figure 5 The flow scheduling schemes of various storage tanks provided according to some embodiments of the present invention are shown. Table 9 shows the predicted values of the properties of the blended products in each period.
[0150] Table 12 Predicted values of properties of blended products in each period
[0151]
[0152] like Figure 3 As shown in the figure, the horizontal axis is the cycle, the vertical axis is the component oil and finished oil tank number, and different tanks are distinguished by different colors. The bar chart in the figure indicates whether the tank has picked up goods in the corresponding period, and the numbers in the bar chart indicate the amount of goods picked up.
[0153] like Figure 4 As shown, the bar graph in the figure indicates that the component tank receives the component oil in the corresponding period, and the numbers in the bar graph indicate the amount of component oil received.
[0154] like Figure 5 As shown, the bar graph in the figure indicates that the storage tank participates in blending in the corresponding period, and the numbers in the bar graph indicate the pipeline flow rate of the oil product.
[0155] As shown in Table 9, the property values of the refined oil meet the corresponding property indicators, and the excess octane number is small.
[0156] As can be seen from the above optimization results, the multi-objective optimization method for gasoline blending scheduling can meet the production needs of gasoline blending and formulate a reasonable and feasible scheduling plan. In the scheduling plan, the properties, output, and delivery volume of finished oil all meet the corresponding constraints, basically achieving the goal of a small total excess of research octane number and maximizing the output and delivery volume of finished oil.
[0157] Afterwards, the gasoline blending rescheduling method based on real-time optimization relaxation feedback may execute step S2: constructing a gasoline blending real-time optimization model considering attribute relaxation.
[0158] Please refer to Figure 6 , Figure 6 The gasoline blending process tank volume allocation according to some embodiments of the present invention is shown.
[0159] like Figure 6 The ultimate goal of gasoline blending is to make the gasoline in the finished tank meet various quality indicators and strictly control the key attributes. To achieve this goal, in the real-time optimization process, it is necessary to calculate the properties of the finished oil at the blending head based on the volume and properties of the oil in the finished tank and the unblended volume to compensate for the properties of the oil in the tank.
[0160] First, the basic idea of cumulative blending is to calculate the property value of the unblended volume of finished oil according to the deviation between the properties and indicators of the tank bottom oil and the blended volume, so as to ensure that the properties of the entire finished tank are qualified. Therefore, it is necessary to convert the property index range of the entire tank of oil products into the index range of the blending head, and optimize the blending process by adjusting the properties of the finished oil at the blending head. Taking any property as an example, the conversion formula of the blending head property index is as follows:
[0161]
[0162]
[0163] in, and It means that the upper and lower limits of the attribute j index of the whole tank oil product meet the specified range, and the upper and lower limits of the index that the attribute of the unblended volume should meet, that is, the upper and lower limits of the attribute index of the blending head; and VPro is the upper and lower limits of the attribute j index of the brand of oil in the whole tank, which is equal to the lower limit of the brand index + internal control margin and the upper limit of the brand index - internal control margin respectively; tol,j VPro represents the cumulative blending volume attribute product of attribute j; Heel,j Represents the attribute product of the tank bottom oil volume of attribute j; VolH represents the tank bottom oil volume; VolS represents the blending target volume; VolT represents the blended volume; VolR represents the unblended volume.
[0164] Secondly, a research octane number blending effect model is constructed. In the gasoline blending process, some property values of the finished oil can be obtained by linearly adding the properties of the component oils, while the superposition of some properties is nonlinear. Octane number is a typical quality indicator that does not have a linear additive property among gasoline-related quality indicators. It is necessary to compensate the octane number of each component oil before calculating the octane number of the finished oil using a linear method. In order to accurately predict the octane number properties of the finished oil, an online updated blending effect model is introduced, and the recursive least squares method is used to update the model parameters online in real time. The specific form of the research octane number blending effect model is as follows:
[0165]
[0166]
[0167] in, Represents the octane value of component oil i after compensation for blending effect; RON i pb represents the octane value of component oil i; i Represents the blending effect compensation value of component oil i; RON b Represents the predicted value of the octane number attribute of the blended product; r i represents the mass formula of component oil i; ρ i Represents the density of component oil i.
[0168] Here, octane number is the most important quality characteristic of gasoline, which is mainly used to measure the anti-knock performance of gasoline as a fuel. Generally, the higher the octane number, the better the anti-knock performance of the oil, which can reduce the frequency of knocking during gasoline combustion and reduce damage to the engine. The octane number is divided into research octane number (RON) and motor octane number (MON), which are determined on a standard CFR engine according to ASTM protocols D-2699 and D2700, respectively. RON represents the anti-knock performance of the engine under low speed and frequent acceleration conditions, while MON represents the performance of the engine under more severe high-speed conditions. In both octane number determination experiments, highly active n-heptane and highly stable isooctane are used as alternative fuels, and their octane numbers range from 0 to 100.
[0169] Finally, determine the objective function and the second constraint. In the model, the objectives are to minimize the cost, research octane number property excess and slack, and multiple objectives are handled by weighted sum. The objective function is expressed as follows:
[0170]
[0171] Among them, W m Represents the excess octane number weight; RON min Indicates the lower limit of the octane number of the finished oil; j indicates the number of attributes; δ j represents the relaxation of attribute j; α j represents a relaxation switch, when α j 1 means that the relaxation parameter is enabled. Among the various attribute indicators of gasoline, some key attributes need to strictly comply with the constraints. j Set to 0 to ensure that the attribute value does not violate the indicator constraint. j Indicates the relaxation weight coefficient of attribute j. In the optimization calculation, a larger relaxation parameter weight is set to avoid the deviation of the refined oil attribute from the normal range due to the relaxation parameter being too large. In addition, different weight coefficients can be set for each attribute to distinguish the importance of the attribute.
[0172] The second constraint includes:
[0173] Identify the constraints:
[0174] a. Attribute constraints: For attributes based on quality formula, their attribute values should satisfy the following:
[0175]
[0176] in, Represents the property j of the component oil i. Relaxation parameter δ j > 0, indicating that the property j of the refined oil does not satisfy the lower limit constraint, and the relaxation parameter δ j<0 means that the attribute j of the refined oil does not meet the upper limit constraint, and the relaxation parameter δ j =0 means that the attribute j of the refined oil product does not violate the attribute constraint.
[0177] And the attributes based on volume formula, whose attribute values should meet the following constraints:
[0178]
[0179] b. Formula constraints: During the gasoline blending process, in order to ensure that the properties of the finished oil are qualified, the proportion of each component oil in the finished oil should meet the corresponding constraints. At the same time, the sum of the proportions of each component oil is 1. The constraints are as follows:
[0180] r i min ≤r i ≤r i max i=1,2,…,n…
[0181]
[0182] Among them, r i min Indicates the lower limit of the formulation of component oil i; r i max Indicates the upper limit of the formulation of component oil i.
[0183] Here, the upper and lower limits of the recipe are to add or subtract a margin value on the basis of the initial recipe of the scheduling plan, so as to limit the real-time optimization range and prevent a large deviation between the optimized recipe and the initial recipe.
[0184] Since there is a situation where the optimization formula is not unique, when the target values of two solutions are the same, as long as and remain unchanged, they are both optimal solutions, which will cause the solution to jump between multiple optimal solutions to a certain extent, causing oscillation. In order to ensure the smooth operation of the system, a constraint is added to move closer to the previous formula, and the step size constraint of the formula change is as follows:
[0185] d≥r i_last -r i
[0186] d≥r i -r i_last
[0187] Where d represents the maximum step length of a single change; r i_last Indicates the blending formula of component oil i calculated by the last optimization.
[0188] c. Flow constraint conditions: During the gasoline blending process, each component oil is transported to the blending head through the oil pump for blending, and the component oil flow rate is controlled by adjusting the pump. Therefore, the component oil flow rate delivered by the oil pump should not exceed its rated flow parameter, otherwise the component oil flow rate cannot meet the blending requirements. The component oil flow constraints are as follows:
[0189] r c ·total_fiow<Power c
[0190] Among them, total_fiow represents the total blending flow; Power c Represents the flow parameter of the oil pump in the movement path of component oil c.
[0191] Please refer to Figure 7 , Figure 7 A flowchart of real-time rolling optimization of gasoline blending provided according to some embodiments of the present invention is shown.
[0192] like Figure 7 As shown in the figure, the specific steps of real-time rolling optimization of gasoline blending scheduling are as follows:
[0193] a. At the beginning of the scheduling cycle, the initial blending plan of the current cycle is obtained, including data such as component oil formula, total blending flow, parameter storage tank, tank bottom oil information, oil product movement path and finished oil index, and blending is performed according to the plan.
[0194] b. Obtain real-time measurement data from the online analysis system and calculate the blending head attribute index and blending effect value based on the real-time data.
[0195] c. Calculate the current optimal blending formula based on the gasoline blending real-time optimization model considering attribute relaxation and real-time data.
[0196] d. Determine whether there are attribute relaxations that are not 0 in the optimization calculation results. If the relaxation parameters are all 0, the current optimization result is the current optimal blending solution; otherwise, the attribute relaxation parameters are fed back to the blending scheduling layer, and the previous round of optimization results are retained as the current optimal blending solution.
[0197] e. Send the current optimal blending plan to the blending control system, which will execute the plan.
[0198] f. Determine whether the current cycle has ended. If the current cycle has not ended, execute the next round of optimization calculation; otherwise, end the real-time optimization calculation and wait for the next cycle scheduling plan.
[0199] Thereafter, the gasoline blending rescheduling method based on real-time optimization of relaxation feedback may execute step S3: constructing a gasoline blending rescheduling model based on real-time optimization of relaxation feedback.
[0200] The attribute constraints and objective function of the gasoline blending scheduling model in S1 are modified by using attribute relaxation feedback, and then the scheduling optimization algorithm is re-executed to calculate the best scheduling plan after the feedback time point.
[0201] When the relaxation parameter δ i When >0, it means that the finished oil attribute j violates the lower limit constraint. The attribute constraint condition corresponding to equation 74 in the scheduling model is modified to increase the lower limit constraint of the finished oil attribute. The attribute constraint modification is as follows:
[0202]
[0203] When the relaxation parameter δ i When <0, it means that the finished oil attribute j violates the upper limit constraint. The attribute constraint condition corresponding to equation 74 in the scheduling model is modified to reduce the upper limit constraint of the finished oil attribute. The attribute constraint modification is as follows:
[0204]
[0205] When the relaxation parameter δ corresponding to the octane attribute ron When >0, it means that the octane number attribute of the finished oil violates the lower limit constraint. At this time, the lower limit of the octane number attribute will cause the finished oil attribute to be unqualified. It is necessary to correct the objective function of the minimum excess octane number attribute of 66 gasoline in the scheduling model, and add a relaxation parameter to the lower limit constraint of the octane number attribute. The objective function is corrected as follows:
[0206]
[0207] Please refer to Figure 8 , Figure 8 A gasoline blending rescheduling flow chart based on real-time optimized relaxation feedback according to some embodiments of the present invention is shown.
[0208] like Figure 8 As shown in FIG. 1 , the specific steps of gasoline blending rescheduling based on real-time optimization relaxation feedback are as follows:
[0209] a. Set the initial data for gasoline blending and scheduling tasks, including component oil properties, component tank inventory, component oil output, finished product tank inventory, finished oil quality indicators, formula restrictions, pick-up quantity restrictions, gasoline tank area flow chart and equipment parameters, etc.
[0210] c. Based on the initial data blended with the scheduling model in step S1, the NSGA-II algorithm is used for optimization calculation to generate a multi-period gasoline blending scheduling plan.
[0211] d. Execute the blending plan for each cycle in cyclic order.
[0212] e. Execute the gasoline blending real-time rolling optimization program to optimize the gasoline blending process in real time.
[0213] f. Determine whether there are slack parameters that are not 0 during the real-time optimization process. If so, modify the attribute constraints and objective function in the scheduling model according to the slack parameters, update the tank inventory, pickup quantity, product demand and other data, and use the remaining time of the current cycle as the time slot length of the first scheduling cycle, then jump to a to recalculate the scheduling plan; otherwise, execute the next cycle blending plan.
[0214] Finally, the gasoline blending rescheduling method based on real-time optimization relaxation feedback can execute step four S4: verifying the gasoline blending rescheduling strategy using a gasoline blending simulation case.
[0215] Please refer to Tables 10 to 13. Tables 10 and 11 record the current component oil tank inventory and finished product tank inventory respectively. Table 12 is the real-time optimization data of component oil, including price, parameter switch, pump flow, initial formula upper limit, formula lower limit and real-time formula, where the upper and lower limits of the formula are based on the initial formula plus or minus 5, which is used to limit the change range of the optimized formula. Table 13 is the real-time data of oil properties, including real-time measurement data of component oil properties, real-time measurement data of blending head properties, tank bottom oil volume attribute product and cumulative blending volume attribute product.
[0216] The gasoline blending rescheduling system based on real-time optimization relaxation feedback can use the optimization result of step S1 as the initial gasoline blending scheduling plan, and feed back the attribute relaxation that is not zero at a certain moment in the real-time optimization process to the scheduling layer for correction.
[0217] Assume that the fourth cycle blending plan is currently being executed, the execution time is 8 hours, the current product brand is 92#, the target output is 6668.94t, the finished product tank is G903, the initial tank bottom oil is 3000t, and the blended volume is 2000t. The output and pick-up volume of 92# gasoline are 8917.6t and 4673.6t respectively, the output and pick-up volume of 95# gasoline are 12243.12t and 1078.54t respectively, and the output and pick-up volume of 98# gasoline are 6867.72t and 4392.75t respectively.
[0218] Table 13 Component tank inventory
[0219] Component tank G804 G806 G315 G316 G311 G312 G701 G702 G709 G305 G307 Inventory volume / t 1224.76 968.62 1217.91 750 1441.59 1480 3162.34 1463.7 1040.21 511.69 741.44
[0220] Table 14 Finished Cans Inventory
[0221] Finished cans G902 G903 G940 G905 G908 G939 G703 G704 Inventory volume / t 6068.09 6315.33 2000 4000 16243.02 2000 7549.52 1000
[0222] Table 15 Real-time optimization parameters
[0223] Oil price Parameter adjustment switch Pump flow Initial recipe Recipe Cap Recipe lower limit Real-time recipe Reformed gasoline 7000 1 280 22.37 27.37 17.37 23.52 MTBE 6500 0 0 0 0 0 0 Non-aromatic gasoline 7000 1 200 12.93 17.93 7.93 12.56 Catalytic gasoline 8000 1 444 64.7 69.7 59.7 63.92 C5 5300 0 0 0 0 0 0 Alkylate 8200 0 0 0 0 0 0
[0224] Table 16 Real-time data of oil properties
[0225]
[0226] Combining the above real-time data with the gasoline blending real-time optimization model considering attribute relaxation, the Gurobi solver is used to solve the problem. The Gurobi solver version used is 10.0.1. In the objective function of the real-time optimization model, the blending cost weight and the research octane number attribute excess weight are set to 1 and 100 respectively, all attribute relaxation weights are set to 100000, and the attribute relaxation switch is set to 1.
[0227] Please refer to Table 14, which shows the obtained relaxation parameter values of the properties. Among them, the relaxation parameters of the research octane number and vapor pressure properties are greater than 0, indicating that the predicted research octane number and vapor pressure of the refined oil in the real-time optimization calculation results do not meet the lower limit of the index.
[0228] Table 17 Attribute relaxation parameters
[0229]
[0230] Please refer to Figures 9 to 11 . Fig. 9 A component oil receiving scheduling scheme provided according to some embodiments of the present invention is shown. Fig.10 Flow scheduling schemes for various storage tanks provided according to some embodiments of the present invention are shown. Fig.11 A refined oil pickup quantity scheduling scheme provided according to some embodiments of the present invention is shown.
[0231] like Figures 9 to 11 As shown in FIG. 1 , the attribute relaxation parameter is fed back to the gasoline blending scheduling model to correct the gasoline blending scheduling model. The left side of the dotted line represents the scheduling plan before rescheduling, and the right side of the dotted line represents the scheduling plan after rescheduling.
[0232] Compared with the experimental results of step S1, it can be seen that after the 4th cycle, the production of refined oil, the flow rate of each storage tank, the component oil receiving tank and the refined oil delivery volume are adjusted on the basis of the original scheduling plan.
[0233] Since the property relaxation of the research octane number and vapor pressure in the real-time optimization results is greater than 0, it is necessary to adjust the lower limit constraints of the research octane number and vapor pressure and the objective function of the minimum excess of the research octane number according to the relaxation parameters. After the model is corrected, only when the predicted value of the finished oil property is higher than the lower limit of the index superimposed with the property relaxation can the final product meet the finished oil index.
[0234] Please refer to Table 15, which shows the predicted values of finished oil properties in each cycle.
[0235] As shown in Table 15, the research octane number and vapor pressure properties of the finished oil in the 4th cycle and subsequent cycles in the table are all higher than the corrected lower limit, which proves the feasibility of model correction through rescheduling strategy.
[0236] Table 18 Prediction of finished oil properties in each cycle
[0237]
[0238] The above case analysis shows that the proposed gasoline blending rescheduling strategy based on real-time optimization relaxation feedback can coordinate real-time optimization and scheduling optimization. When uncertain factors have a negative impact on gasoline blending, the scheduling model can be corrected through attribute relaxation parameters, so that a new scheduling plan can be formulated in time to avoid unqualified finished oil in the finished product tank. The scheduling plan obtained through rescheduling optimization meets the gasoline blending process requirements and operating rules, and can ensure the stable operation of gasoline blending. Experiments have proved that the proposed rescheduling strategy can improve the robustness of the gasoline blending optimization process, reduce the participation of scheduling personnel in the gasoline blending process, and achieve the improvement of the economic benefits of the refinery.
[0239] In summary, the gasoline blending rescheduling method based on real-time optimization of relaxation feedback, the gasoline blending rescheduling system based on real-time optimization of relaxation feedback, and the storage medium provided by the present invention can be used to find a balance between ensuring that the quality of refined oil meets the standards and maintaining the stability of the production scheduling plan, thereby improving the efficiency of the entire production process.
[0240] Although the above methods are illustrated and described as a series of actions for simplicity of explanation, it should be understood and appreciated that these methods are not limited by the order of the actions, because according to one or more embodiments, some actions may occur in a different order and / or concurrently with other actions from those illustrated and described herein or not illustrated and described herein but understandable to those skilled in the art.
[0241] The previous description of the disclosure is provided to enable any person skilled in the art to make or use the disclosure. Various modifications to the disclosure will be apparent to those skilled in the art, and the general principles defined herein may be applied to other variations without departing from the spirit or scope of the disclosure. Thus, the disclosure is not intended to be limited to the examples and designs described herein, but should be granted the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A gasoline blending rescheduling method based on real-time optimization relaxation feedback, characterized in that: The following steps are involved: With the goal of maximizing the output and delivery volume of refined oil products and minimizing the excess octane number, a multi-objective model for gasoline blending scheduling is constructed, and its corresponding first constraint condition is determined; Obtaining sample data of gasoline blending scheduling, and solving the multi-objective model based on the sample data to determine a first optimal scheduling solution; Constructing a real-time optimization model that takes attribute relaxation into consideration, and determining the corresponding optimization objective function and the second constraint condition, wherein the real-time optimization model includes a cumulative blending model and a research octane number blending effect model; Adopting the attribute relaxation feedback method, modifying the objective function and the first constraint condition of the multi-objective model to construct a gasoline blending rescheduling model based on real-time optimization relaxation feedback; and The gasoline blending rescheduling model is solved according to the sample data to determine a second optimal scheduling solution based on real-time optimization relaxation feedback.
2. The gasoline blending and rescheduling method according to claim 1, characterized in that: The first objective function of maximizing the oil product output and the pickup volume is expressed as: The second objective function of minimizing the excess octane number is expressed as: Among them, F t,c,n is the oil product movement flow rate of the nth storage tank of component oil c in period t; g t,p,m LP is a binary variable indicating whether the m-th storage tank of finished oil p in period t receives the finished oil. A value of 0 indicates that the m-th storage tank of finished oil p in period t does not receive the finished oil, and a value of 1 indicates that the m-th storage tank of finished oil p in period t receives the finished oil. t,p,m is the delivery quantity of the mth storage tank of refined oil p in period t; Q c,ron is the research octane property of component oil c; It is the lower limit of the research octane number property of finished oil.
3. The gasoline blending and rescheduling method according to claim 2, characterized in that: The first constraint condition includes: Component tank material balance constraints: CS t,c,n =CS t-1,c,n +Cr t,c,n ·CP t,c -F t,c,n ·H Among them, CS t,c,n is the inventory of component oil in the nth storage tank of component oil c in period t; CS t-1,c,n is the inventory of component oil in the nth storage tank of component oil c in period t-1; Cr t,c,n Indicates whether the nth storage tank of component oil c in period t is used to receive the component oil planned to be produced in period t; CP t,c is the planned production of component oil c in period t; and / or Component tank capacity constraints: in, is the inventory lower limit of the nth storage tank of component oil c; is the upper limit of the inventory of the nth storage tank of component oil c; and / or Blending recipe constraints: in, Indicates the lower limit of the formula of component oil c when blending finished oil p; Indicates the upper limit of the formulation of component oil c when blending finished oil p; and / or Finished product tank inventory material balance constraints: PS t,p,m =PS t-1,p,m +F t,p,m ·H-LP t,p,m Among them, PS t,p,m represents the inventory of the mth storage tank of finished oil p at the end of period t; PS t-1,p,m represents the inventory of the mth storage tank of refined oil p at the end of period t-1; and / or Product oil tank capacity constraints: in, Indicates the lower limit of the inventory capacity of the mth storage tank of refined oil p; represents the upper limit of the inventory capacity of the mth storage tank of refined oil product p; and / or Constraints on the quantity of refined oil products to be picked up: in, It represents the lower limit of the single pick-up quantity of the m-th storage tank of refined oil p; represents the upper limit of the single pick-up quantity of the mth storage tank of refined oil product p; and / or Constraints on demand for refined oil products: in, It represents the lower limit of demand for refined oil product p; represents the upper limit of the demand for refined oil product p; and / or Constraints on refined oil properties: in, is the lower limit of the j-th attribute of the refined oil product p; is the upper limit of the jth attribute of the refined oil product p; Q c,j is the attribute value of the jth attribute of component oil c; and / or Component tank parameter adjustment constraints: (1-x t,c,n )·F t,c,n =0 Among them, x t,c,n Binary variable for component oil tank selection, x t,c,n =1 means the nth tank parameter adjustment of component oil C in period t, x t,c,n =0 means no adjustment; and / or Component oil receiving constraints: Among them, Cr t,c,n The nth storage tank representing component oil c is used to receive component oil c planned to be produced within period t, and C represents the number of types of the reference component oil; and / or Finished product tank parameter adjustment constraints: (1-g t,p,m )·F t,p,m =0 Among them, g t,p,m A binary variable representing the number of finished oil products that the m-th storage tank receives in period t, F t,p,m represents the flow rate of the finished oil p received by the mth storage tank in period t; and / or Blending head flow balance constraints: Among them, F t,c,n represents the flow rate of oil product movement in the nth tank of component oil c within period t, F t,p,m represents the flow rate of the finished oil p received by the mth storage tank in period t; and / or Finished product tank blending and delivery constraints: LP t,p,m ·(1-g t,p,m )=0 Among them, LP t,p,m represents the delivery volume of the mth storage tank of refined oil p in period t, g t,p,m A binary variable indicating that the mth storage tank of finished oil p receives finished oil in period t.
4. The gasoline blending and rescheduling method according to claim 3, characterized in that: The steps of solving the multi-objective model include: Solving the multi-objective model using the NSGA-II algorithm to obtain a corresponding non-dominated solution set; and A global optimal solution based on a knee point is selected from the non-dominated solution set as the first optimal scheduling solution.
5. The gasoline blending and rescheduling method according to claim 4, characterized in that: The step of using the NSGA-II algorithm to solve the multi-objective model to obtain a corresponding non-dominated solution set includes: Construct an initial population of size N; and The initial population is subjected to a preset number of rounds of evolution based on crossover mutation until a non-dominated solution set after a specified number of evolutions is obtained, wherein for the t-th evolution, the parent population P of size N is replaced by t With a population Q of size N t Merge to get a merged population R of size 2N t , using the fast non-dominated sorting method, the merged population R t Sort multiple individuals in the , and divide the multiple individuals into l levels {F1, F2, ... F l }, and then create an empty offspring population P t+1 , and add the individuals of each level to the offspring population P in order from small to large t+1 To continue the t+1th evolution.
6. The gasoline blending and rescheduling method according to claim 5, characterized in that: The step of constructing an initial population of size N comprises: Creating binary variables of flow, pickup, and component oil received in the gasoline blending scheduling plan as individuals in the population, wherein the binary variables include 0 / 1 variables and encoding forms of real number variables; A flow value is randomly generated according to the selected component tank as the component oil flow, and the finished oil flow is expressed as the sum of the component oil flows to construct the individual chromosome; Randomly select finished cans for picking up and assign them random picking quantities, wherein picking up is allowed from a plurality of the finished cans in each cycle; Randomly selecting the component tank to receive the component oil, wherein only one component tank is allowed to be selected to receive each component oil in each cycle; and In response to completing the construction of each individual, the objective function value and constraint violation degree of the individual are calculated, and each individual is sequentially filled into the empty target population according to the objective function value and the constraint violation degree until the target population is filled to serve as the initial population.
7. The gasoline blending and rescheduling method according to claim 5, characterized in that: The individuals of each level are added to the offspring population P in order from small to large. t+1 The steps include: Determine the addition of the i-th level F i After all individuals in the offspring population P t+1 Whether the number of individuals will be greater than N; In response to the added offspring population P t+1 The number of individuals is less than or equal to N, and the level F i All individuals in are added to the offspring population P t+1 in; and In response to the added offspring population P t+1 The number of individuals is greater than N, and the crowding operator is used to i The individuals in the above-mentioned are sorted based on the crowding distance, and the level F is i The remaining individuals with the smallest crowding distance are added to the offspring population P t+1 Until the progeny population P t+1 The number of individuals in reaches N.
8. The gasoline blending and rescheduling method according to claim 3, characterized in that: The steps of constructing a real-time optimization model considering attribute relaxation include: Convert the property index range of the full tank oil into the index range of the blending head, and optimize the blending process by adjusting the properties of the finished oil at the blending head: in, and They represent the upper and lower limits of the attribute j of the entire tank of oil, and They represent the upper and lower limits of the blending head attribute index that make the upper and lower limits of the attribute j of the entire tank oil product meet the specified range, VPro tol,j VPro represents the cumulative blending volume attribute product of the entire tank of oil attribute j. Heel,j Represents the attribute product of the tank bottom oil volume of the full tank oil attribute j, VolH represents the tank bottom oil volume, VolS represents the blending target volume, VolT represents the blended volume, and VolR represents the unblended volume.
9. The gasoline blending and rescheduling method according to claim 8, characterized in that: The step of constructing a real-time optimization model considering attribute relaxation also includes: The research octane number blending effect model is expressed as: in, Represents the octane value of component oil i after compensation for blending effect, RON i Indicates the octane value of component oil i, pb i Indicates the blending effect compensation value of component oil i, RON b represents the predicted value of the octane number attribute of the blended product, r i represents the mass formula of component oil i, ρ i Represents the density of component oil i.
10. The gasoline blending and rescheduling method according to claim 9, characterized in that: The second constraint condition includes: Attribute constraints: in, represents the property j of component oil i, and the relaxation parameter δ j >0 means that the attribute j of the refined oil does not satisfy the corresponding lower limit constraint, and the relaxation parameter δ j <0 means that the attribute j of the refined oil does not satisfy the corresponding upper limit constraint, and the relaxation parameter δ j = 0 means that the attribute j of the refined oil does not violate the attribute constraint; and / or Recipe constraints: r i min ≤r i ≤r i max i=1,2,…,n Among them, r i min represents the lower limit of the formulation of component oil i, r i max represents the upper limit of the formulation of component oil i; and / or Recipe change step constraints: d≥r i_last -r i d≥r i -r i_last Where d represents the maximum step size of a single change, r i_last Indicates the blending formula of component oil i calculated in the last optimization; and / or Component oil flow constraints: r c ·total_fiow<Power c Among them, total_fiow represents the total blending flow, Power c Represents the flow rate parameter of the oil pump in the movement path of component oil c.
11. The gasoline blending and rescheduling method according to claim 10, characterized in that: The step of adopting the attribute relaxation feedback method to correct the objective function and the first constraint condition of the multi-objective model to construct a gasoline blending rescheduling model based on real-time optimization relaxation feedback includes: In response to the relaxation parameter δ i >0, increase the lower limit of the finished oil property constraint: In response to the relaxation parameter δ i <0, lower the upper limit of the refined oil property constraint: as well as In response to the relaxation parameter δ corresponding to the octane value attribute ron > 0, add the relaxation parameter δ to the second objective function of minimizing the excess octane number ron :
12. A gasoline blending rescheduling system based on real-time optimization relaxation feedback, characterized in that: include: a memory having computer instructions stored thereon; as well as A processor is connected to the memory and configured to execute computer instructions stored in the memory to implement the gasoline blending rescheduling method based on real-time optimized relaxation feedback according to any one of claims 1 to 11.
13. A computer-readable storage medium having computer instructions stored thereon, characterized in that: When the computer instructions are executed by a processor, the gasoline blending rescheduling method based on real-time optimized relaxation feedback according to any one of claims 1 to 11 is implemented.
