A deterministic scenario-guided multi-stage limited adaptability robust optimization method and system

Abstract

The present invention relates to the field of production scheduling and uncertain optimization in process industries, and specifically discloses a multi-stage finite adaptability robust optimization method guided by a deterministic scenario, comprising the following steps: Step 1, clarifying the slowly time-varying uncertainty problem in the production scheduling process, establishing a nominal formula for the uncertainty parameters and a long-period mixed integer optimal control problem under a deterministic scenario; Step 2, solving the deterministic mixed integer optimal control problem using a control parameter vectorization method to obtain an initial decision plan; Step 3, using the initial decision plan under the deterministic scenario as a benchmark, using a candidate path generation algorithm to obtain a series of candidate solutions, constructing a candidate solution set, etc. The present invention establishes a robust optimization model for an uncertain production scheduling process with multi-stage and decision-dependent characteristics in process industries and proposes an efficient solution method.

Description

Technical Field

[0001] The present invention relates to the technical field of production scheduling and uncertainty optimization, and in particular to a deterministic scenario-guided multi-stage finite adaptability robust optimization method, and also to a deterministic scenario-guided multi-stage finite adaptability robust optimization system. Background Art

[0002] In process industries, production and operations often face various uncertainties, such as fluctuations in raw material prices, demand, and sudden equipment failures. These uncertainties directly impact the safety, stability, and efficiency of the production process. In actual production, production scheduling is a critical process that coordinates various steps in the process, ensures the smooth implementation of production plans, and maximizes production efficiency and quality. However, due to the long-term, dynamic nature of industrial production processes and the multiple sources of uncertainty, the effectiveness of scheduling decisions is often affected by these uncertainties.

[0003] The uncertainty of industrial processes is a widely concerned issue in the fields of production scheduling and uncertainty optimization. Robust optimization, as a common method for dealing with uncertainty problems, has been widely used in production scheduling processes such as crude oil scheduling optimization.

[0004] Currently, the most commonly used robust optimization methods are static robust optimization, where all decision variables are determined before the scheduling is executed and are not adjusted during the scheduling process. However, static robust optimization methods have the following disadvantages:

[0005] First, since all decisions need to be determined in advance based on the worst-case scenario, there may be a problem of being too conservative;

[0006] Secondly, it lacks consideration of the time-varying nature of uncertainty sets and decision-making dependencies, making it unsuitable for optimization scenarios involving binary variables, such as equipment maintenance cycle scheduling problems.

[0007] Finally, its applicability to production scheduling problems with nonlinear constraints is low, and developing robust optimization methods with multi-stage decision-making dependencies under uncertainty is one of the challenging problems facing process industries.

[0008] Multi-stage decision-dependent robust optimization methods have three distinct properties: Property 1: The decision-making process consists of multiple stages, and different stages correspond to different uncertainty sets, which are time-varying. Property 2: The decision and the range of uncertainty parameters influence each other, and the construction of the uncertainty set is decision-dependent. Property 3: Considering the nonlinearity of the production scheduling model. Based on Property 1, the multi-stage adjustable robust method allows variables to be adaptively adjusted after the uncertainty is observed, to some extent alleviating the problem of overconservatism in conventional robust optimization. Based on Property 2, the candidate solution set production algorithm and the finite adaptability method transform the infinite-dimensional robust optimization problem into a finite-dimensional problem by constructing multiple candidate paths and partitioning the uncertainty set in advance. Based on Property 3, the worst-case analysis method avoids the dual reconstruction of the robust optimization model, allowing the direct use of heuristic algorithms for solution. Summary of the Invention

[0009] In view of this, the present invention provides a deterministic scenario-guided multi-stage finite adaptability robust optimization method for modeling and solving uncertain production scheduling processes with multi-stage and decision-dependent characteristics in process industries.

[0010] To achieve the above objectives, the basic solution of the present invention provides a deterministic scenario-guided multi-stage limited adaptive robust optimization method, comprising the following steps:

[0011] Step 1: Identify the slow time-varying uncertainty problem in the production scheduling process, establish the nominal formula of the uncertainty parameters and the long-period mixed integer optimal control problem in the deterministic scenario;

[0012] Step 2: Solve the deterministic mixed integer optimal control problem using the control parameter vectorization method to obtain the initial decision solution;

[0013] Step 3: Based on the initial decision plan in the deterministic scenario, a candidate path generation algorithm is used to obtain a series of candidate solutions and construct a candidate solution set;

[0014] Step 4: Establish a robust optimization model for multi-stage decision-dependent production scheduling process, construct the uncertainty set of slowly time-varying parameters, and at the same time, divide the uncertainty set of decision-dependent parameters into finite partitions based on the finite adaptability algorithm to generate finite evolution paths of uncertainty parameters.

[0015] Step 5: For different evolution paths and time-varying uncertainty sets, the multi-stage robust optimization problem is solved based on the worst-case analysis method.

[0016] In one possible design, step 1 includes,

[0017] Step 1.1: Identify the key slow-varying parameters that have a significant impact on system performance and are uncertain during production scheduling. Add this to the performance degradation of the device over time to obtain the nominal formula:

[0018]

[0019] Where α(t)>0,λ(t)>0 are parameters that affect the degree and rate of parameter change, θ(t) is a slow time-varying parameter, and θ0 is the parameter value at the initial moment;

[0020] In step 1.2, consider the impact of slowly time-varying parameters on system performance and establish a long-period mixed integer optimal control problem in a deterministic scenario:

[0021]

[0022] Where, the objective function J(·) consists of two parts: t∈[t0,t f ] and the terminal time t f The final value term φ0[x(t f )], x is the state variable, u is the binary decision variable, q is the continuous decision variable, θ is the slowly time-varying parameter, g(·) is the inequality constraint; F(·) is the state equation of the system, and U is the value range of the decision variable u.

[0023] In a possible design, step 2 includes transforming the dynamic optimization problem in a deterministic scenario into a finite-dimensional nonlinear programming problem using the control parameter vectorization method, and converting the optimization period t∈[t0,t f ] is discretized into several intervals t0<t1<t2<…<t L-1 <t L , an approximation strategy based on piecewise constant function, t l The distance between t and t0 is described as:

[0024] t l =t0+l×(t f -t0) / L,l=0,1,...,L

[0025] At the same time, the control vector and the objective function are approximately expressed as:

[0026]

[0027] Where u l (t) is the lth interval t∈[t l-1 ,t l ] is the approximate value of the control trajectory, X l (t) is a parameterized function whose value is:

[0028]

[0029] In a possible design, in step 3, guided by the decision-making of the deterministic scenario, a candidate path generation algorithm is defined to obtain a series of candidate solutions, construct a candidate solution set, and make full use of empirical knowledge to guide uncertainty optimization.

[0030] In a possible design, step 3 is specifically as follows: after obtaining the optimal solution of the mixed integer optimal control problem in a deterministic scenario, the candidate decision solution set in the first stage is obtained by using the candidate path generation algorithm based on the optimal solution.

[0031] In a possible design, in step 4, based on the time-varying characteristics of the uncertainty parameters and the decision dependency, a multi-stage decision-dependent robust optimization model and a time-varying uncertainty set are established. Based on the candidate solution set and the finite adaptive algorithm in step 3, the decision-dependent time-varying uncertainty set is divided into finite-dimensional uncertain parameter evolution paths, providing a means to decompose the decision-dependent robust optimization problem.

[0032] In one possible design, step 4 includes,

[0033] Step 4.1: Considering the interdependence between decision-making and uncertainty sets, a robust optimization model of multi-stage decision-making dependence and time-varying uncertainty sets is established for the uncertainty in process industry production scheduling. The robust optimization model is

[0034]

[0035] Where u 1 ,...,u L and q 1 ,...,q L is the decision from stage 1 to stage l, and u l ∈{0,1} N is a binary decision variable connecting two consecutive stages l and l+1. It is the first stage (u l ,q l ) depends on the feasible region of u l-1 and ξ l The cost function of each stage is defined as ξ l ∈Ξ l is the uncertainty vector of the first stage, and the uncertainty set in the first stage is represented by l Indicates that this depends on the decision variable u in the previous stage l-1 Without loss of generality, the uncertainty set is described by a polyhedral uncertainty set:

[0036]

[0037] Where, parameter W∈R N×K and is the nominal value of the uncertainty vector at stage l, is the value of the uncertain parameter observed in stage l-1. It represents the element-wise multiplication, i.e. the Schur product;

[0038] Step 4.2: Based on the candidate solution set and finite adaptive algorithm guided by deterministic scenarios, the decision-dependent time-varying uncertainty set is divided into finite-dimensional uncertain parameter evolution paths:

[0039]

[0040] Where, is the uncertainty set of the p-th path in stage l; for the entire decision stage, the uncertainty set in the p-th path is defined as:

[0041]

[0042] In a possible design, in step 5, the multi-stage decision-dependent robust optimization model is converted into an optimization problem for solving the worst path, and the complex robust optimization is converted into an optimization problem that can be directly solved based on the worst-case analysis method.

[0043] In one possible design, step 5 is specifically as follows:

[0044] Step 5.1: Using a deterministic scenario-guided limited adaptability method, the robust optimization model with multi-stage decision-making dependencies is transformed into an optimization problem for solving the worst path:

[0045]

[0046] Where P is the set of all candidate paths;

[0047] Step 5.2, based on the worst-case analysis, convert the robust optimization into multiple deterministic optimization problems that can be solved directly:

[0048]

[0049] η f -ΔJ r0 ≤0

[0050] η g ≤0

[0051]

[0052] Where ηf is the target robustness, which is used to measure the sensitivity of the objective function to uncertainty; η g is the constraint robustness, which is used to reflect the sensitivity of uncertainty constraints; f c Used to describe the value of the objective function.

[0053] The present invention provides a system comprising a memory, a control processor, and a computer program stored in the memory and executable on the control processor, wherein the control processor executes the program to implement the deterministic scenario-guided multi-stage finite adaptive robust optimization method as described above.

[0054] Compared with the prior art, the present invention has the following beneficial effects:

[0055] The present invention establishes a robust optimization model and proposes an efficient solution method for the uncertain production scheduling process with multi-stage and decision-dependent characteristics in the process industry. The present invention takes into account the uncertainty problems with long periodicity, dynamics, decision-dependence and other characteristics in the process industry, and establishes a multi-stage decision-dependent robust optimization model for the production scheduling process. At the same time, by integrating the mixed integer optimal control process, a candidate path generation method for slow time-varying industrial processes is proposed. Guided by deterministic scenarios, multiple feasible paths are constructed using a finite adaptability method, and the complex multi-stage robust optimization problem is converted into a finite-dimensional optimization process. Furthermore, based on the worst-case analysis, the above-mentioned robust optimization problem is converted into a form that can be directly solved. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0057] Figure 1 shows a schematic diagram of an industrial crude oil heat exchanger network;

[0058] Figure 2 A schematic diagram of a process flow of a deterministic scenario-guided multi-stage limited adaptability robust optimization method proposed in an embodiment of the present application is shown;

[0059] Figure 3 Schematic diagrams of the scaling values ​​of each heat exchanger under deterministic and uncertain scenarios are shown, where Figures (a)-(e) are scaling trend diagrams for heat exchangers 1-5;

[0060] Figure 4Schematic diagrams of utility energy consumption under deterministic and uncertain scenarios are shown, where (a)-(b) are schematic diagrams of energy consumption of cold fluids C1 and C2, and (c)-(d) are schematic diagrams of energy consumption of hot fluids H1 and H2;

[0061] Figure 5 Schematic diagram of an immunoassay showing uncertain parameters. DETAILED DESCRIPTION

[0062] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention. They are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. By referring to these contents, ordinary technicians in this field should be able to understand other possible implementation methods and advantages of the present invention. The components in the figures are not drawn to scale, and similar component symbols are generally used to represent similar components.

[0063] A deterministic scenario-guided multi-stage finite adaptability robust optimization method is used to model and solve the uncertain production scheduling process with multi-stage and decision-dependent characteristics in process industry. The energy management and cleaning scheduling problem of the heat exchanger network commonly used in the process industry heat transfer process is taken as an example. The process equipment schematic diagram is shown as follows Figure 1 As shown in Figure 1, this heat exchange network consists of two hot streams and two cold streams, with heat exchanged through the five heat exchangers listed. When a piece of equipment is cleaned, the energy management of the entire heat exchange network is restructured. In industrial applications, the hot streams are typically processed finished oil products, represented by H1 and H2 in the figure. Meanwhile, the cold streams are crude oil, represented by C1 and C2, which need to be heated to a set temperature to meet subsequent processing requirements. Furthermore, C1 and C2 are coolers that further cool the hot streams to the target outlet temperature. H1 and H2 are heaters responsible for raising the crude oil temperature to the desired value after heat exchange. Initially, the heat exchange network structure is determined during the design phase, with each heat transfer unit assigned its own heat load. However, as the equipment operates, the accumulation of fouling inevitably leads to a continuous decrease in the heat transfer efficiency of the heat exchangers. Consequently, energy consumption for the coolers and heaters increases. To achieve heat transfer targets, it is necessary to determine a cleaning schedule for each heat exchanger and the additional heat required for C1, C2, and H1, H2 during each scheduling cycle.

[0064] The slowly time-varying uncertainty parameter in this heat transfer process is the accumulated fouling of the heat exchanger. This parameter is difficult to measure online in industry, but it significantly impacts energy management and cleaning scheduling for the heat transfer process. This example involves a long-term production scheduling problem involving multiple decision-making stages. Cleaning and maintenance decisions directly impact the distribution of the uncertainty parameter, creating decision-dependency. The heat transfer process exhibits typical nonlinearity.

[0065] For the production scheduling problem under uncertain conditions, this embodiment adopts the following steps to model and solve it:

[0066] Step 1: Establish the nominal formula of the uncertainty parameters and the long-period mixed integer optimal control problem in the deterministic scenario; specifically,

[0067] In step 1.1, the heat transfer coefficient with scaling accumulation is selected as the key slow time-varying parameter, and the nominal formula for the heat transfer coefficient is obtained as follows:

[0068]

[0069] Where U is the heat transfer coefficient and U0 is its initial value. Function Shows the accumulation of scale over time.

[0070] In step 1.2, considering the impact of scaling accumulation on heat transfer efficiency, a long-period mixed integer optimal control problem is established under a deterministic scenario:

[0071]

[0072] The objective function J(·) consists of two parts: t∈[t0,t f ] and the terminal time t f The final value term φ0[x(t f )], x is the state variable, u is the binary decision variable, q is the continuous decision variable, θ is the slowly varying parameter, g(·) is the inequality constraint; F(·) is the state equation of the system. U is the value range of the decision variable u.

[0073] Step 2: Use the control parameter vectorization method to transform the dynamic optimization problem in the deterministic scenario into a finite-dimensional nonlinear programming problem, and set the optimization period t∈[t0,t f ] is discretized into several intervals t0<t1<t2<…<t L-1 <t L . Based on the approximation strategy of piecewise constant function, t l The distance between t and t0 is described as:

[0074] t l =t0+l×(t f -t0) / L,l=0,1,...,L

[0075] At the same time, the control vector and the objective function are approximately expressed as:

[0076]

[0077]

[0078] Where ul (t) is the lth interval t∈[t l-1 ,t l ] is an approximation of the control trajectory in X. l (t) is a parameterized function whose value is:

[0079]

[0080] Step 3: After obtaining the optimal solution of the mixed integer optimal control problem in a deterministic scenario, the candidate decision solution set in the first stage is obtained by using the candidate path generation algorithm based on the optimal solution.

[0081]

[0082] Step 4: Establish a robust optimization model of the multi-stage decision-dependent heat transfer process and construct the uncertainty set of scaling accumulation. At the same time, the decision-dependent uncertainty set is divided into finite partitions based on the finite adaptability algorithm to generate finite evolution paths of uncertainty parameters. Specifically,

[0083] Step 4.1: Consider the interdependence between energy management and cleaning decisions and the fouling uncertainty set, and establish a time-varying uncertainty set:

[0084]

[0085] Where, and They represent the upper and lower bounds of the fouling value of the nth heat exchanger, respectively, and are defined as:

[0086]

[0087] Step 4.2: Based on the candidate solution set and finite adaptive algorithm guided by deterministic scenarios, the decision-dependent time-varying uncertainty set is divided into finite-dimensional uncertain parameter evolution paths:

[0088]

[0089] Where, is the uncertainty set of the p-th path in stage l. For the entire decision stage, the uncertainty set in the p-th path is defined as:

[0090]

[0091] Step 5: For different evolution paths of scaling parameters, the multi-stage robust optimization problem is solved based on the worst-case analysis method. Specifically,

[0092] Step 5.1: Using a deterministic scenario-guided limited adaptability method, the robust optimization model with multi-stage decision-making dependencies is transformed into an optimization problem for solving the worst path:

[0093]

[0094] Where P is the set of all candidate paths.

[0095] Step 5.2, based on the worst-case analysis, convert the robust optimization into multiple deterministic optimization problems that can be solved directly:

[0096]

[0097] η f -ΔJ r0 ≤0

[0098] η g ≤0

[0099]

[0100] Where η f is the target robustness, which is used to measure the sensitivity of the objective function to uncertainty; η g is the constraint robustness, which is used to reflect the sensitivity of uncertainty constraints; f c Used to describe the value of the objective function.

[0101] Based on the same inventive concept, the present invention provides a multi-stage finite adaptability robust optimization method guided by a deterministic scenario. By obtaining a set of candidate solutions according to a candidate path generation algorithm in a deterministic scenario, a solution set is constructed, and then a robust optimization model in multi-stage decision-making is constructed and solved using a finite adaptability algorithm. Then, a time-varying uncertainty set is constructed, and a worst-case analysis strategy is used to transform the robust optimization problem. The method can model and solve the robust optimization problem that depends on multi-stage decision-making in production scheduling under uncertainty conditions in process industries.

[0102] In this embodiment, Figure 3 The variation of the fouling value of each heat exchanger under deterministic and uncertain scenarios is shown. Figure 3 (a) to Figure 3 (e) shows the variation patterns of fouling values ​​of the five heat exchangers under deterministic and uncertain scenarios, respectively. These patterns are related to the optimized cleaning decisions. For example, Figure 3The worst-case decision in (a) is made at the 3rd, 5th, and 12th month, so the fouling curve has a clear downward trend (because the fouling is cleaned). The deterministic case is cleaned at the 3rd, 6th, and 12th month, respectively. The description of the remaining heat exchangers is similar. Significant differences can be observed between the results obtained in the worst-case and deterministic cases, highlighting the significant impact of uncertainty on fouling behavior. In addition, the variation trend of fouling is intricately intertwined with the scheduling decision, showing a clear dependency between fouling variation and decision making. The significant variation in fouling resistance values ​​between the worst-case and deterministic cases further emphasizes the necessity of a robust optimization method that can adapt to the fluctuations of uncertain fouling to ensure optimal performance of the heat exchanger in practical applications.

[0103] In this embodiment, Figure 4 The energy consumption of the utility is shown for the worst-case and deterministic scenarios, where Figure 4 (a) to Figure 4 (d) shows the energy consumption of cold fluid C1, cold fluid C2, hot fluid H1, and hot fluid H2 in the deterministic and uncertain scenarios, respectively. These energy consumption changes are consistent with the optimized cleaning decision results, that is, Figure 3 The decision on when to perform cleaning is consistent across the entire system. The comparison shows that the worst-case energy scheduling is remarkably robust, consistent with the principles of resilient design. Robust optimization methods can help mitigate the potential impact of uncertainty on the system, ensuring a more resilient and adaptive energy management strategy.

[0104] In this embodiment, Figure 5 Results from the uncertainty parameter immunity test were presented, further validating the effectiveness of the present invention in mitigating uncertainty risks. In the uncertainty parameter immunity test, the present invention demonstrated an energy margin at the end of the scheduling cycle. This margin emphasizes the adaptability of the scheduling method in decision-making, especially when considering worst-case scenarios. Insufficient energy management reserves due to uncertainty can threaten the stability of the entire plant operation, thereby affecting production efficiency. In severe cases, the disruption of the thermal and electrical balance can lead to unforeseen production consequences.

[0105] The present invention also provides a deterministic scenario-guided multi-stage finite adaptive robust optimization system, comprising a memory, a control processor, and a computer program stored on the memory and executable on the control processor. The control processor executes the program to implement the aforementioned deterministic scenario-guided multi-stage finite adaptive robust optimization method.

[0106] The present invention also provides a computer-readable storage medium, which stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to implement the aforementioned deterministic scenario-guided multi-stage finite adaptability robust optimization method.

[0107] 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 different orders and / or concurrently with other actions from the diagrams and descriptions herein or not illustrated and described herein but understood by those skilled in the art. Those skilled in the art will further appreciate that the various illustrative logic blocks, modules, circuits, and algorithm steps described in conjunction with the embodiments disclosed herein can be implemented as electronic hardware, computer software, or a combination of the two. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits, and steps are generally described above in terms of their functionality. Whether such functionality is implemented as hardware or software depends on the specific application and the design constraints imposed on the overall system. Technicians can implement the described functionality in different ways for each specific application, but such implementation decisions should not be interpreted as resulting in a departure from the scope of the present invention. The various illustrative logic blocks, modules, and circuits described in conjunction with the embodiments disclosed herein may be implemented or executed using a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A general-purpose processor may be a microprocessor, but in the alternative, the processor may be any conventional processor, a battery compartment control board, a micro-battery compartment control board, or a state machine. A processor may also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors in conjunction with a DSP core, or any other such configuration. The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. The software module may reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor so that the processor can read and write information from / to the storage medium. In an alternative, the storage medium may be integrated into the processor. The processor and storage medium may reside in an ASIC. The ASIC may reside in a user terminal. In an alternative, the processor and storage medium may reside in the user terminal as discrete components. In one or more exemplary embodiments, the functions described may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software as a computer program product, the functions may be stored or transmitted as one or more instructions or codes on a computer-readable medium.Computer-readable media include both computer storage media and communication media, including any media that facilitates the transfer of a computer program from one place to another. Storage media can be any available media that can be accessed by a computer. As an example and not limitation, such computer-readable media may include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, disk storage or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer. Any connection is also properly referred to as a computer-readable medium. For example, if the software is transmitted from a website, a central control computer, or other remote source using a coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwaves, then the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwaves are included in the definition of medium. As used herein, disk and disc include compact disc (CD), laser disc, optical disc, digital versatile disc (DVD), floppy disk and Blu-ray disc, where disks typically reproduce data magnetically, while discs reproduce data optically with lasers. Combinations of the above should also be included within the scope of computer-readable media.

[0108] Although the above methods are illustrated and described as a series of acts for simplicity of explanation, it is to be understood and appreciated that these methods are not limited by the order of the acts, as some acts may occur in a different order and / or concurrently with other acts from those illustrated and described herein or not illustrated and described herein but understandable to those skilled in the art according to one or more embodiments.

[0109] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as above in terms of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can, without departing from the scope of the technical solution of the present invention, make some changes or modifications to equivalent embodiments using the technical contents disclosed above. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A deterministic scenario-guided multi-stage limited adaptability robust optimization method, characterized by: The following steps are involved: Step 1: Identify the slow time-varying uncertainty problem in the production scheduling process, establish the nominal formula of the uncertainty parameters and the long-period mixed integer optimal control problem in the deterministic scenario; Step 2: Solve the deterministic mixed integer optimal control problem using the control parameter vectorization method to obtain the initial decision solution; Step 3: Based on the initial decision plan in the deterministic scenario, a candidate path generation algorithm is used to obtain a series of candidate solutions and construct a candidate solution set; Step 4: Establish a robust optimization model for multi-stage decision-dependent production scheduling process, construct the uncertainty set of slowly time-varying parameters, and at the same time, divide the uncertainty set of decision-dependent parameters into finite partitions based on the finite adaptability algorithm to generate finite evolution paths of uncertainty parameters. Step 5: For different evolution paths and time-varying uncertainty sets, the multi-stage robust optimization problem is solved based on the worst-case analysis method.

2. The deterministic scenario-guided multi-stage limited adaptive robust optimization method according to claim 1, characterized in that: Step 1 includes, Step 1.1: Identify the key slow-varying parameters that have a significant impact on system performance and are uncertain during production scheduling. Add this to the performance degradation of the device over time to obtain the nominal formula: Where α(t) affects the degree of parameter change, λ(t) affects the rate of parameter change; θ(t) is a slowly varying parameter, and θ0 is the parameter value at the initial moment; Step 1.2, considering the impact of slow time-varying parameters on system performance, establish the long-period mixed integer optimal control problem in a deterministic scenario, as shown in the following formula: g(x(t),u(t),q(t),θ(t),t)≤0,t0≤t≤t f u(t)∈U,U={0,1} Where, the objective function J(·) consists of two parts: t∈[t0,t f ] and the terminal time t f The final value term φ0[x(t f )], x is the state variable, u is the binary decision variable, q is the continuous decision variable, θ is the slowly time-varying parameter, g(·) is the inequality constraint; F(·) is the state equation of the system, and U is the value range of the decision variable u.

3. The deterministic scenario-guided multi-stage limited adaptive robust optimization method according to claim 2, characterized in that: Step 2 includes using the control parameter vectorization method to transform the dynamic optimization problem in the deterministic scenario into a finite-dimensional nonlinear programming problem, and to reduce the optimization period t∈[t0,t f ] is discretized into several intervals t0<t1<t2<…<t L-1 <t L , an approximation strategy based on piecewise constant function, t l The distance between t and t0 is described as: t l =t0+l×(t f -t0) / L,l=0,1,...,L At the same time, the control vector and the objective function are approximately expressed as, Where u l (t) is the lth interval t∈[t l-1 ,t l ] is the approximate value of the control trajectory, X l (t) is a parameterized function whose value is, 4. A deterministic scenario-guided multi-stage limited adaptability robust optimization method according to any one of claims 1 to 3, characterized in that: In step 3, guided by the decision-making of the deterministic scenario, a candidate path generation algorithm is defined to obtain a series of candidate solutions, construct a candidate solution set, and make full use of empirical knowledge to guide uncertainty optimization.

5. The deterministic scenario-guided multi-stage limited adaptive robust optimization method according to claim 4, characterized in that: Specifically, step 3 is to obtain the optimal solution of the mixed integer optimal control problem in a deterministic scenario, and then use the optimal solution as a benchmark to obtain the candidate decision solution set in the first stage through the candidate path generation algorithm. Where p is the subscript of the path, which indicates the number of the evolution path.

6. The deterministic scenario-guided multi-stage limited adaptive robust optimization method according to claim 5, characterized in that: In step 4, based on the time-varying characteristics of the uncertainty parameters and the decision dependency, a multi-stage decision-dependent robust optimization model and a time-varying uncertainty set are established. Based on the candidate solution set and the finite adaptive algorithm in step 3, the decision-dependent time-varying uncertainty set is divided into finite-dimensional uncertain parameter evolution paths, providing a means for decomposing the decision-dependent robust optimization problem.

7. A deterministic scenario-guided multi-stage limited adaptive robust optimization method according to claim 5 or 6, characterized in that: Step 4 includes, Step 4.1: Considering the interdependence between decision-making and uncertainty sets in the process of process industry production scheduling, a robust optimization model of multi-stage decision-making dependence and time-varying uncertainty sets is established, where the robust optimization model is Where u 1 ,...,u L and q 1 ,...,q L is the decision from stage 1 to stage l, and u l ∈{0,1} N is a binary decision variable connecting two consecutive stages l and l+1, is the feasible region of stage l (ul,ql), depending on u l-1 and ξl, the cost function of each stage is defined as ξ l ∈Ξ l is the uncertainty vector of the first stage, and the uncertainty set in the first stage is represented by l Indicates that this depends on the decision variable u in the previous stage l-1 ; Without loss of generality, the uncertainty set is described by a polyhedral uncertainty set: Where, parameter W∈R N×K and is the nominal value of the uncertainty vector at stage l, is the value of the uncertain parameter observed in the l-1th stage, and the symbol It represents the element-wise multiplication, i.e. the Schur product; Step 4.2: Based on the candidate solution set and finite adaptive algorithm guided by deterministic scenarios, the decision-dependent time-varying uncertainty set is divided into finite-dimensional uncertain parameter evolution paths. Where, is the uncertainty set of the p-th path in stage l. For the entire decision stage, the uncertainty set in the p-th path is defined as:

8. The deterministic scenario-guided multi-stage limited adaptability robust optimization method according to claim 7, characterized in that: In step 5, the multi-stage decision-dependent robust optimization model is converted into an optimization problem for solving the worst path, and the complex robust optimization is converted into an optimization problem that can be directly solved based on the worst-case analysis method.

9. The deterministic scenario-guided multi-stage limited adaptive robust optimization method according to claim 8, characterized in that: Step 5 is as follows: Step 5.1: Using a deterministic scenario-guided limited adaptability method, the robust optimization model with multi-stage decision-making dependencies is transformed into an optimization problem for solving the worst path: Where P is the set of all candidate paths; Step 5.2, based on the worst-case analysis, convert the complex robust optimization into an optimization problem that can be solved directly: or f -ΔJ r0 ≤0 or g ≤0 Where η f is the target robustness, which is used to measure the sensitivity of the objective function to uncertainty; η g is the constraint robustness, which is used to reflect the sensitivity of uncertainty constraints; f c Used to describe the value of the objective function.

10. A system, characterized in that: The method comprises a memory, a control processor, and a computer program stored in the memory and executable on the control processor, wherein the control processor executes the program to implement the deterministic scenario-guided multi-stage finite adaptive robust optimization method according to any one of claims 1 to 9.

Images