Collaborative optimization method and system based on adaptive relaxation function
By introducing adaptive relaxation factors into the collaborative optimization method, the convergence difficulties and local optimization problems of the collaborative optimization method when meeting the consistency constraints are solved, and the robustness and efficiency of optimization are improved.
Patent Information
- Application Number
- CN202411173545.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-26
- Publication Date
- 2024-11-26
AI Technical Summary
The collaborative optimization method is difficult to converge when meeting the consistency constraints and is prone to falling into local optimization, resulting in system-level optimization difficulties and reduced robustness.
The adaptive relaxation factor is introduced, and the initial consistency requirements are relaxed by adding adaptive relaxation functions to the system-level optimization objective function, and the constraints are adjusted through iteratively, thereby improving convergence speed and global search capabilities.
The robustness and convergence speed of the optimization results are improved, the impact of the starting point on the results is reduced, the optimization results are avoided from falling into local optimization, and the computing efficiency is enhanced.
Smart Images

Figure CN119026365A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of computer technology, and in particular to a collaborative optimization method and system based on an adaptive relaxation function. Background Art
[0002] With the rapid development of science and technology, complex engineering systems and the complexity problems they generate have become increasingly prominent. When designing complex systems, the multidisciplinary design optimization method (MDO) must fully consider the mutual coupling relationship between various disciplines, use appropriate methods to decompose the system into discipline-based models, and then coordinate and control these subsystems through a specific framework based on the relationship between disciplines, so as to finally obtain the global optimal solution of the system.
[0003] Among them, the collaborative optimization method (CO) has a high degree of disciplinary autonomy. It makes disciplinary decisions through disciplinary-level optimizers and can process in parallel. It is a MDO method that is currently studied more and widely used. This method is particularly suitable for distributed design environments such as aircraft, ships and marine structures. Each sub-discipline of the CO method has an optimizer, and algorithms suitable for each sub-discipline can be selected for optimization. The analysis modules of each sub-discipline can be combined with any optimization algorithm in a portable manner. In the sub-disciplines of the CO method, sub-discipline optimization information that is only related to a certain discipline, such as local design variables and local constraints, is only analyzed in the sub-discipline-level optimization module, thereby greatly reducing the complex requirements for coupling information exchange between different disciplines.
[0004] However, the CO method still has many imperfections that need to be improved. For example, the consistency equality constraint in the collaborative optimization system-level problem belongs to the constraint under ideal conditions. However, in most cases, this condition is difficult to meet, so the system-level optimization will have difficulty converging due to the non-existence of the design feasible domain; for example, the Lagrange multiplier may not exist during system-level optimization, or the constraint gradient function at the optimal design point is discontinuous, causing the system-level optimization problem to fail to meet the Kuhn-Tucker condition. The consistency constraint condition is not completely determined by the system-level design variables, resulting in a non-smooth constraint function, and sometimes there will be discontinuities, which will cause CO to fall into a local optimum and reduce the robustness of the CO method. Summary of the invention
[0005] The present invention aims at the problems that the collaborative optimization algorithm itself is difficult to meet the consistency constraint conditions, resulting in convergence difficulties, and the optimization results of the standard collaborative optimization algorithm are prone to fall into local optimality. This method introduces an adaptive relaxation factor, overcomes the inherent convergence difficulties of the standard CO algorithm, and has excellent global search capabilities, solves the problem of starting point selection, and avoids the optimization results from falling into local optimality.
[0006] In a first aspect, the present invention provides a collaborative optimization method based on an adaptive relaxation function, comprising:
[0007] Step 1. Establish a mathematical model for the optimization problem;
[0008] Step 2. Write the mathematical model of the optimization problem in step 1 into a collaborative optimization form, decompose the system optimization problem into n sub-disciplines, and establish the system-level optimization objective function and the discipline-level optimization objective function respectively;
[0009] Step 3. Introducing an adaptive relaxation function into the system-level optimization objective function;
[0010] Step 4. Jointly solve the system-level optimization objective function and the discipline-level optimization objective function.
[0011] Furthermore, the mathematical model of the optimization problem in step 1 is:
[0012] min f=f(x)
[0013] stg i (x)≤0
[0014] x lb ≤x≤x ub
[0015] Where x is the optimization variable, f is the objective function, f(x) means the objective function f is a function of the optimization variable x, g i (x) is the constraint condition, x lb , x ub are the upper and lower bounds of the optimization variable, and min represents the minimum value.
[0016] Furthermore, the system-level optimization objective function in step 2 is:
[0017] min f(z)
[0018] wrtz
[0019]
[0020] Where wrt indicates that the optimization variable of the system-level optimization objective function is z, f(z) is the system-level optimization objective function, z indicates the set of design variables of the system-level optimization objective function, st indicates the constraint condition, and J i represents the optimization objective function of sub-discipline i, z j is the j-th design variable assigned to sub-discipline i at the system level, represents the optimization variable of the jth design variable of subject i, s j represents the number of design variables of the subject-level optimization objective function, ε iis the relaxation factor of subject i.
[0021] Furthermore, the subject-level optimization objective function in step 2 is:
[0022]
[0023] in, It is a discipline-level constraint.
[0024] Furthermore, the adaptive relaxation function is defined as,
[0025]
[0026] Among them, α is the adaptive factor, ε0 is the initial relaxation factor, α1, α2 are constants between (0,1), and α1<α2, r is the radius factor, and tan -1 is the inverse tangent function.
[0027] Furthermore, the adaptive factor is expressed as:
[0028]
[0029] where d max =||x ub -x lb ||2 represents the maximum range of the design domain, Represents the distance between the i-th sub-discipline variable and the system-level variable, α∈[0,1].
[0030] Furthermore, in step 4, initial values of optimization variables are assigned and an optimization algorithm is used to jointly solve the system-level optimization objective function and the discipline-level optimization objective function.
[0031] On the other hand, the present invention provides a collaborative optimization system based on an adaptive relaxation function, comprising:
[0032] Module 1. It is used to establish a mathematical model for optimization problems;
[0033] Module 2. It is used to write the mathematical model of the optimization problem in module 1 into a collaborative optimization form, decompose the system optimization problem into n sub-disciplines, and establish the system-level optimization objective function and the discipline-level optimization objective function respectively;
[0034] Module 3. It is used to introduce an adaptive relaxation function into the system-level optimization objective function;
[0035] Module 4. It is used to jointly solve the system-level optimization objective function and the discipline-level optimization objective function.
[0036] Compared with the prior art, the present invention has the following beneficial effects:
[0037] 1. Compared with the traditional collaborative optimization algorithm, the present invention adopts an adaptive relaxation function to relax the consistency requirements of system-level optimization variables and subject-level optimization variables in the initial stage of optimization, so the optimal point can be found in a larger range. Therefore, the optimization result is less affected by the starting point value.
[0038] 2. The present invention uses an adaptive relaxation function to actively tighten the consistency constraints between the system-level optimization variables and the subject-level optimization variables as the optimization iteration proceeds, thereby increasing the convergence speed and improving the computational efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0040] Figure 1 is a flow chart of the method of the present invention;
[0041] Figure 2 is the definition of the relaxation function of the present invention;
[0042] Figure 3 is the objective function iteration curve of the present invention;
[0043] Figure 4 It is the tolerance iteration curve of the present invention. DETAILED DESCRIPTION
[0044] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0045] It should be noted that, in the description of the embodiments of the present invention, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "include one..." do not exclude the existence of other identical elements in the process, method, article or device including the elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to the specific circumstances.
[0046] The terms "first", "second", etc. in this application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are generally of one type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.
[0047] Example 1
[0048] like Figure 1 As shown, this embodiment provides a collaborative optimization method based on an adaptive relaxation function, taking the gear reducer optimization problem as an example:
[0049] Step 1. Establish a mathematical model for the optimization problem;
[0050] The reducer design problem is one of the 10 standard examples used by NASA to evaluate the performance of the MDO method. The goal of reducer design is to minimize the reducer volume while satisfying various constraints.
[0051] Step 1: Establish a mathematical model for the optimization problem.
[0052]
[0053] wrtx1,x2,x3,x4,x5,x6,x7
[0054]
[0055] g7=x2x3 / 40≤1.0
[0056] g8=x1 / (12x2)≤1.0
[0057] g9=5x2 / x1≤1.0
[0058] g 10 =(1.5x6+1.9) / x4≤1.0
[0059] g 11 =(1.1x7+1.9) / x5≤1.0
[0060] 2.6≤x1≤3.6,0.7≤x2≤0.8,17≤x3≤28,7.3≤x4≤8.3,
[0061] 7.3≤x5≤8.3,2.9≤x6≤3.9,5.0≤x7≤5.5
[0062] Among them, f(x) is the objective function, g1-g 11 is the constraint condition, including 7 design variables, tooth surface width x1, gear module x2, pinion teeth number x3, bearing spacing x4 and x5, pinion shaft diameter x6 and gear shaft diameter x7. g1-g 11 They represent the maximum bending stress of the gear teeth, the maximum contact stress of the gear teeth, the maximum lateral deflection of the shaft, the maximum stress of the shaft, and the constraints such as size and space limitations.
[0063] Step 2. Write the mathematical model of the optimization problem in step 1 into a collaborative optimization form, decompose the system optimization problem into n sub-disciplines, and establish the system-level optimization objective function and the discipline-level optimization objective function respectively;
[0064] The optimization problem in this embodiment includes a system-level optimization problem and three sub-discipline-level optimization problems, which are written as the system-level optimization objective function of the standard collaborative optimization algorithm:
[0065] minf(z)
[0066] wrtz1,z2,z3,z4,z5,z6,z7
[0067]
[0068] Among them, f(z) is the system-level optimization objective function, wrt represents the optimization variable of the system-level optimization objective function, z1, z2, z3, z4, z5, z6, z7 are the optimization variables of this problem, st represents the constraint condition, J i represents the optimization objective function of sub-discipline i, z j is the j-th design variable assigned to sub-discipline i at the system level, represents the optimization variable of the jth design variable of subject i, s j represents the number of design variables of the subject-level optimization objective function, and ε1, ε2, and ε3 are the relaxation factors of subjects 1, 2, and 3, respectively.
[0069] The subject-level optimization objective functions include:
[0070] The first sub-discipline optimization problem is:
[0071]
[0072] stc j -1.0≤0,j=1,2,7,8,9
[0073] The second sub-discipline optimization problem is:
[0074]
[0075] stc j -1.0≤0,j=1,2,4,6,7,8,9,11
[0076] The third sub-discipline optimization problem is:
[0077]
[0078] stc j -1.0≤0,j=1,2,3,5,7,8,9,10
[0079] Step 3. Introducing an adaptive relaxation function into the system-level optimization objective function;
[0080]
[0081] Among them, α is the adaptive factor, ε0 is the initial relaxation factor, α1, α2 are constants between (0,1), and α1<α2, and r is the radius factor. Further, the adaptive factor α is defined as,
[0082]
[0083] Among them, d max =||x ub -x lb ||2, represents the maximum range of the design domain. represents the distance between the ith sub-discipline variable and the system-level variable. Obviously, d i <d max , so α i ∈[0,1]. ε0 is the initial relaxation factor, which is taken as 0.1 in this embodiment.
[0084] like Figure 2As shown, the function is divided into three parts, and the three parts are segmented by α1, α2, for example, α1 = 0.25, α2 = 0.75. Each part has a different mathematical expression. The first part is represented by a linear function, the second part is represented by an inverse tangent function, and the third part is represented by another linear function. If α is small, ε is large.
[0085] Step 4: Perform system-level and discipline-level optimizations.
[0086] Assign the initial values of the optimization variables (2.8, 0.71, 25, 7.9, 7.599, 3, 5.099), use the optimization algorithm to start the system-level and subject-level optimization problems, and solve the optimization problems. The objective function iteration curve is as follows: Figure 3 As shown in Figure 1. ACO is the method and CO is the traditional collaborative optimization method. It can be seen that the method can quickly converge to the optimal solution with a small number of iterations. The tolerance iteration curves of the three sub-disciplines are shown in Figure 1. Figure 4 As shown, it can be seen that in the initial stage of optimization, the tolerance is large, which relaxes the difference between each sub-discipline variable and the system-level variable. As the iteration proceeds, the tolerance is adaptively adjusted to smaller and smaller levels to ensure the accuracy of the calculation results.
[0087] Example 2
[0088] This embodiment provides a collaborative optimization system based on an adaptive relaxation function, including:
[0089] Module 1. It is used to establish a mathematical model for optimization problems;
[0090] Module 2. It is used to write the mathematical model of the optimization problem in module 1 into a collaborative optimization form, decompose the system optimization problem into n sub-disciplines, and establish the system-level optimization objective function and the discipline-level optimization objective function respectively;
[0091] Module 3. It is used to introduce an adaptive relaxation function into the system-level optimization objective function;
[0092] Module 4. It is used to jointly solve the system-level optimization objective function and the discipline-level optimization objective function.
[0093] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.
[0094] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A collaborative optimization method based on an adaptive relaxation function, characterized in that: include: Step 1. Establish a mathematical model for the optimization problem; Step 2. Write the mathematical model of the optimization problem in step 1 into a collaborative optimization form, decompose the system optimization problem into n sub-disciplines, and establish the system-level optimization objective function and the discipline-level optimization objective function respectively; Step 3. Introducing an adaptive relaxation function into the system-level optimization objective function; Step 4. Jointly solve the system-level optimization objective function and the discipline-level optimization objective function.
2. The collaborative optimization method based on adaptive relaxation function according to claim 1, characterized in that: The mathematical model of the optimization problem in step 1 is: minf=f(x) s.t.g i (x)≤0 x lb ≤x≤x ub Where x is the set of optimization variables, f is the objective function, f(x) means that the objective function f is a function of the optimization variable x, and g i (x) is the constraint condition, x lb , x ub are the upper and lower bounds of the optimization variable, and min represents the minimum value.
3. The collaborative optimization method based on adaptive relaxation function according to claim 2, characterized in that: The system-level optimization objective function in step 2 is: min f(z) wrtz Where wrt indicates that the optimization variable of the system-level optimization objective function is z, f(z) is the system-level optimization objective function, z indicates the set of design variables of the system-level optimization objective function, st indicates the constraint condition, and J i represents the optimization objective function of sub-discipline i, z j is the j-th design variable assigned to sub-discipline i at the system level, represents the optimization variable of the jth design variable of subject i, s j represents the number of design variables of the subject-level optimization objective function, ε i is the relaxation factor of subject i.
4. The collaborative optimization method based on adaptive relaxation function according to claim 3, characterized in that: The subject-level optimization objective function in step 2 is: in, It is a discipline-level constraint.
5. The collaborative optimization method based on adaptive relaxation function according to claim 4, characterized in that: The adaptive relaxation function is defined as: Among them, α is the adaptive factor, ε0 is the initial relaxation factor, α1, α2 are constants between (0,1), and α1<α2, r is the radius factor, and tan -1 is the inverse tangent function.
6. The collaborative optimization method based on adaptive relaxation function according to claim 5, characterized in that: The adaptive factor is expressed as: where d max =||x ub -x lb ||2 represents the maximum range of the design domain, Represents the distance between the i-th sub-discipline variable and the system-level variable, α∈[0,1].
7. The collaborative optimization method based on adaptive relaxation function according to claim 5, characterized in that: In step 4, initial values of optimization variables are assigned and an optimization algorithm is used to jointly solve the system-level optimization objective function and the subject-level optimization objective function.
8. A collaborative optimization system based on an adaptive relaxation function, characterized in that: include: Module 1. It is used to establish a mathematical model for optimization problems; Module 2. It is used to write the mathematical model of the optimization problem in module 1 into a collaborative optimization form, decompose the system optimization problem into n sub-disciplines, and establish the system-level optimization objective function and the discipline-level optimization objective function respectively; Module 3. It is used to introduce an adaptive relaxation function into the system-level optimization objective function; Module 4. It is used to jointly solve the system-level optimization objective function and the discipline-level optimization objective function; The collaborative optimization system based on the adaptive relaxation function is used to execute the steps in the collaborative optimization method based on the adaptive relaxation function according to any one of claims 1 to 7.
