Adaptive feedback method and system based on two-stage consensus optimization model

Abstract

The application provides a kind of adaptive feedback method, system, storage medium and electronic equipment based on two-stage consensus optimization model, and relates to information processing field.The application constructs corresponding expert decision matrix according to known satellite emergency scheme attribute;SOM algorithm clustering is used to obtain several subgroups;If the group consensus level is less than the threshold value, then on the basis of the preset adaptive feedback mechanism, a two-stage consensus optimization model is constructed to realize the consensus reaching process, and the final group decision matrix is obtained;According to the final group decision matrix, the MULTIMOORA method is used to obtain the satellite emergency scheme ranking result for multi-attribute large group emergency decision.The two-stage consensus optimization model uses an improved automatic strategy, which ensures the CRP automatic function of satellite scheme evaluation while considering the willingness of experts;Combined with the minimum adjustment amount optimization model and the minimum iteration number optimization model, the CRP continuously runs under the constraint condition, and the adaptive optimization of group consensus level and weight is realized.

Description

Technical Field

[0001] This invention relates to the field of information processing technology, and specifically to an adaptive feedback method, system, storage medium, and electronic device based on a two-stage consensus optimization model. Background Technology

[0002] Multi-attribute group decision-making (MAGDM) in the satellite field involves multiple stakeholders making decisions based on various emergency attributes, such as mission completion benefits, program performance, and satellite resource utilization. It is widely used in management practices, including emergency management and medical waste management. In recent years, frequent emergencies have occurred worldwide, often resulting in catastrophic consequences. Typical characteristics of emergency decision-making include time constraints, partial or incomplete information, and the pressure of potentially severe consequences. This necessitates collaborative decision-making by numerous experts from multiple fields, making large-group emergency decision-making (LGEDM) an important method for emergency management. Generally, when the number of participants in the emergency decision-making group exceeds 20, the MAGDM problem is termed the multi-attribute large-group emergency decision-making (MALGEDM) problem.

[0003] When all experts on a given issue reach a consensus, the decision of the potentially influencing group is more likely to be accepted. This typically means incorporating a consensus-building process (CRP) before selecting the best alternative satellite contingency plan for a GDM problem. Scholars have developed various CRP methods and structures in different contexts, such as CRP in consensus and minimum cost, CRP in social networks, CRP in data-driven / web environments, and adaptive CRP. Among these models, adaptive CRP can adaptively feed back opinions to dynamic parameters such as consensus level and expert weights, thereby achieving consensus.

[0004] While previous research has made some progress in adaptive consensus mechanisms (CRP), some shortcomings remain. For example, there is always a tendency to minimize the total adjustment of preference information to save adjustment time and preserve the initial preferences of experts. Therefore, it is necessary to soften expert preference supervision and reduce the time consumption of CRP while taking into account expert intentions to guide the adaptive consensus process. Summary of the Invention

[0005] (a) Technical problems to be solved

[0006] To address the shortcomings of existing technologies, this invention provides an adaptive feedback method, system, storage medium, and electronic device based on a two-stage consensus optimization model, which solves the technical problem that the adaptive consensus process cannot simultaneously save time and retain the initial preferences of experts.

[0007] (II) Technical Solution

[0008] To achieve the above objectives, the present invention utilizes the following technical satellite emergency response solution:

[0009] An adaptive feedback method based on a two-stage consensus optimization model includes:

[0010] S1. Based on the known attributes of the satellite emergency response plan, obtain the decision opinions of various experts and construct the corresponding expert decision matrix;

[0011] S2. Based on the expert decision matrix, use the SOM algorithm to cluster and obtain several subgroups;

[0012] S3. Based on the expert decision matrix and subgroups, determine whether the group consensus level meets the threshold; if the group consensus level is less than the threshold, then on the basis of the preset adaptive feedback mechanism, construct a two-stage consensus optimization model to realize the consensus achievement process and obtain the final group decision matrix.

[0013] S4. Based on the final group decision matrix, the MULTIMOORA method is used to obtain the ranking results of satellite emergency solutions for multi-attribute large-group emergency decision-making.

[0014] Preferably, in S1, the expert decision matrix is ​​represented using the hesitant fuzzy binary semantic set HF2TLSs, including:

[0015]

[0016] Among them, the discrete finite set X = {x1,…,x} of the satellite emergency response plan m}, where m is the number of satellite emergency response plans and m≥2;

[0017] Expert set e = {e 1 ,…,e N}, where N is the number of experts and N≥20;

[0018] Attribute set C = {c1,…,c n}, where n is the number of attributes and n≥2;

[0019] Different positions For expert e l Selected satellite emergency plan x i For attribute c j Decision information;

[0020] At the granularity of HF2TLSs, It consists of several HF2TLSs, the number of which is determined by expert opinion;

[0021] For language tags, the values ​​are S = {s0, s1, ..., s} g}, where g is the potential of S; This represents the sign conversion value and its range is [-0.5, 0.5].

[0022] Preferably, the process of determining the group consensus level in S3 includes:

[0023] Calculate the subgroup weights.

[0024] All experts were divided into K subgroups using SOM clustering, with each subgroup having a C0 ... k Contains #C k An expert calculated the cohesion of a subgroup. Get ρ is the weight parameter, e l Representing the lth expert, Representing the expert decision matrix subgroup C k Opinion Matrix G k The subgroup weights are calculated based on the Hausdorff distance between the two subgroups:

[0025]

[0026] Computational group consensus level,

[0027]

[0028]

[0029] Specifically, the subgroup decision matrix is ​​obtained based on the decision matrices of all experts in the subgroup. For the subgroup decision matrix, x i Satellite emergency response plan in c j Evaluation values ​​of attributes; determination of the group decision matrix using a weighted average operator. Group decision matrix based on x i Satellite emergency response plan in c j Evaluation value of the attribute;

[0030] SIM(C k ) represents the similarity between subgroups; G k and G f D represents the decision matrix for different subgroups; D represents the Hausdorff distance between different groups; GCL(C k () represents the level of consensus among the groups.

[0031] Preferably, in S3, if GCL(C) k ) less than the threshold ψ and less than the first parameter δ1: GCL(C kIf δ1 < ψ, a low-level feedback strategy (FS-LL) is adopted to achieve the consensus process, which specifically includes:

[0032] The identification rules for FS-LL are as follows:

[0033] First, determine the satellite emergency plan. i The consensus level of the set ALT1 is lower than the average consensus level of all satellite contingency plans to be corrected.

[0034]

[0035]

[0036] in, The consensus level among experts for each satellite contingency plan; d represents the Hausdorff distance between different hesitant, fuzzy binary semantic sets;

[0037] Second, determine the location that needs to be modified (x) i ,c j The set POS1 contains;

[0038]

[0039]

[0040] Where POS1 represents all the locations that need to be modified (x i ,c j A set of ) The level of consensus among experts on each satellite contingency plan for each attribute;

[0041] The directional rule DR.1 of FS-LL is as follows:

[0042] DR.1.1: If Then expert e l Add to attribute c j Assessment;

[0043]

[0044] DR.1.2: If Then expert e l Reduce the impact on attribute c j Assessment;

[0045]

[0046] in, For subgroup C k The expert in the middle e l In the t-th iteration, for the satellite emergency plan x i In attribute cj The evaluation values ​​are: λ1∈[0.5,1],λ2∈[0,0.5], where λ1 and λ2 are the opinion adjustment parameters.

[0047] Preferably, in S3, if GCL(C) k If the threshold ψ is less than or equal to the first parameter δ1 and less than the second parameter δ1 (δ1≤GCL<δ2<ψ), the consensus-reaching process is achieved using the medium consensus level feedback strategy FS-ML, which specifically includes:

[0048] The FS-ML identification rules are as follows:

[0049] (1) When ICL(C k When ) < δ2,

[0050] First, identify subgroups whose consistency level is lower than the consistency parameter as...

[0051] SUB 21 ={C k |ICL(C k )<δ2}

[0052] ICL(C k )=1-D(G k ,R c )

[0053] Among them, ICL(C k ) is a subgroup C k Group decision matrix The level of consensus among them;

[0054] Second, for the identified subgroups, experts whose consistency level is lower than that of the subgroup are identified as...

[0055] EXP 21 ={e l |ECL(e l ) <SCL(C k )∧e l ∈C k}

[0056] ECL(H l )=1-D(H l ,R c )

[0057]

[0058] Among them, SCL(C k ) represents subgroup consistency; ECL(H) l (for expert e) l The level of consensus;

[0059] Third, satellite emergency response plans with a consistency level below the threshold will be identified as...

[0060]

[0061] Fourth, the location that needs to be modified (x) i ,c j ) is identified as

[0062]

[0063] (2) When ICL(C k When )≥δ2,

[0064] First, identify the subgroup with high consensus but low consistency.

[0065] SUB 22 ={C k |ICL(C k )≥δ2∪SCL(C k )<ψ}

[0066] Second, for the identified subgroups, experts whose consensus level is lower than that of the subgroup are identified as...

[0067] EXP 22 ={e l |ECL(e l ) <SCL(C k )∧e l ∈C k}

[0068] Third, satellite emergency response plans with a consensus level lower than the average consensus level of all satellite emergency response plans are identified as...

[0069]

[0070] Fourth, the location that needs to be modified (x) i ,c j ) is identified as

[0071]

[0072] The FS-ML direction rule DR.2 is as follows:

[0073] DR.2.1: If So, expert e l Increase their understanding of satellite contingency plans. i In attribute c j Assessment;

[0074]

[0075] DR.2.2: If Then expert e l Reduce the impact on attribute c j Assessment;

[0076]

[0077] Where E∈[0,g] refers to the expected value of the evaluation; λ1∈[0.5,1],λ2∈[0,0.5]; Indicates the evaluation value HF2TLSs and The probability between them; Δ- 1 This is a conversion function.

[0078] Preferably, in S3, if GCL(C) k If the threshold ψ is less than or equal to the second parameter δ1, and δ2≤GCL<ψ, the consensus-reaching process is achieved using the medium-to-high consensus level feedback strategy FS-MHL, which specifically includes:

[0079] The identification rules for FS-MHL are as follows:

[0080] First, identify experts (EXP3) whose consensus level is lower than the subgroup consensus.

[0081] EXP3 = {e l |ECL(e l ) <SCL(C k )}

[0082]

[0083] Among them, SCL(C k Subgroup consistency;

[0084] Second, for any expert e l ∈EXP3, Identify satellite emergency response plan x i The rules are as follows;

[0085]

[0086] Third, for any satellite emergency plan x i ∈ALT3, identify the attribute c that needs to be modified. j for;

[0087]

[0088] The directional rule DR.3 of FS-MHL is as follows:

[0089] DR.3.1: If So, expert e lIncrease their understanding of satellite contingency plans. i In attribute c j Assessment;

[0090]

[0091] DR.3.2: If Then expert e l Reduce the impact on attribute c j Assessment;

[0092]

[0093] Where λ1∈[0.5,1], λ2∈[0,0.5]; Indicates the evaluation value HF2TLSs and The probability between them.

[0094] Preferably, the two-stage consensus optimization model in S3 includes,

[0095] The objective function of model M1 corresponding to the first stage is:

[0096]

[0097] Where t = {1,…T} represents the t-th iteration, and k = {1,…K} represents the k-th subgroup C. k ; Subgroup C k The expert in the middle e l In the t-th iteration, for the satellite emergency plan x i In attribute c j The evaluation value on;

[0098] And the objective function of the second-stage corresponding model M2:

[0099] minT

[0100] Where t is the t-th iteration; the total number of iterations is T.

[0101] Preferably, the two-phase consensus optimization model in S3 further includes:

[0102] The constraints for the first stage corresponding to model M1 are as follows:

[0103] (1-1) Deviation in Expert Decision-Making Opinions

[0104]

[0105] (1-2) Direction of adjustment of expert decision-making opinions

[0106]

[0107]

[0108] in, For the group decision matrix R c For x in the (t-1)th iteration i Satellite emergency response plan in c j Evaluation value of the attribute For subgroup C k Decision matrix G k For satellite emergency response plan x in the (t-1)th iteration i In attribute c j The evaluation value is given by P, which is the formula for the probability of design.

[0109] (1-3) Adjustment range of parameters λ1 and λ2

[0110] 0.5≤λ1≤1

[0111] 0≤λ²≤0.5

[0112] (1-4) Upper limit of the number of iteration rounds

[0113] T≤T1

[0114] Where T1 is the upper limit of the preset number of iteration rounds in the first stage;

[0115] (1-5) The adjusted group consensus level reaches the consensus threshold.

[0116] GCL(C 1(T) C 2(T) ,…,C K(T) )≥ψ;

[0117] And the constraints of the second stage corresponding model M2:

[0118] (2-1) Obtain the minimum adjustment amount from model M1

[0119]

[0120] in, To minimize adjustment deviation;

[0121] (2-2) Direction of Adjustment of Expert Decision-Making Opinions

[0122]

[0123] Among them, EWR represents the expert intention rule for design; DR.3.1 and DR.3.2, as well as DR.4.1 and DR.4.2 are all based on DR.1 and DR.2, with the addition of the expert satellite emergency plan intention rule EWR to form the direction rule;

[0124] (2-3) Same as (1-3) above;

[0125] (2-4) Upper limit of the number of iteration rounds

[0126] T≤T2

[0127] Where T2 is the preset upper limit of the number of iteration rounds in the second stage;

[0128] (2-5) Same as (1-5) above.

[0129] Preferably, the directional rules DR.3 and DR.4 are as follows:

[0130] Directional rule DR.3 includes:

[0131] (1) DR.3.1:

[0132] correspond

[0133] if Then expert e l No modifications are needed to the satellite emergency plan. i In attribute c j The decision opinion is as follows; otherwise, the expert e shall follow the direction rule D1.2. l The satellite emergency plan needs to be added. i In attribute c j Decision-making opinions;

[0134] (2)DR.3.2

[0135] correspond According to the aforementioned direction rule DR.1, expert e l The satellite emergency response plan needs to be adjusted. i In attribute c j The above decision-making opinions;

[0136] Directional rule DR.4 includes:

[0137] (1) DR.4.1:

[0138] correspond

[0139] if Then expert e l No modifications are needed to the satellite emergency plan. i In attribute c jThe decision opinion is as follows; otherwise, the expert's opinion is as follows, according to the aforementioned directional rule DR.2.2. l The satellite emergency plan needs to be added. i In attribute c j The above decision-making opinions;

[0140] (2) DR.4.2:

[0141] correspond Then, according to the aforementioned direction rule DR.2, expert e l The satellite emergency response plan needs to be adjusted. i In attribute c j The decision-making opinions.

[0142] An adaptive feedback system based on a two-stage consensus optimization model includes:

[0143] The module is used to obtain the decision opinions of various experts based on the known attributes of satellite emergency plans and to construct the corresponding expert decision matrix.

[0144] The clustering module is used to obtain several subgroups by using the SOM algorithm based on the expert decision matrix;

[0145] The feedback module is used to determine whether the group consensus level meets the threshold based on the expert decision matrix and the subgroup; if the group consensus level is less than the threshold, a two-stage consensus optimization model is constructed on the basis of the preset adaptive feedback mechanism to realize the consensus reaching process and obtain the final group decision matrix.

[0146] The ranking module is used to obtain the ranking results of satellite emergency plans for multi-attribute large-group emergency decision-making based on the final group decision matrix using the MULTIMOORA method.

[0147] A storage medium storing a computer program for adaptive feedback based on a two-phase consensus optimization model, wherein the computer program causes a computer to execute the adaptive feedback method based on the two-phase consensus optimization model as described above.

[0148] An electronic device, comprising:

[0149] One or more processors;

[0150] Memory; and

[0151] One or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs including an adaptive feedback method for performing the two-phase consensus optimization model as described above.

[0152] (III) Beneficial Effects

[0153] This invention provides an adaptive feedback method, system, storage medium, and electronic device based on a two-stage consensus optimization model. Compared with existing technologies, it has the following advantages:

[0154] This invention includes: obtaining expert opinions based on known satellite emergency response plan attributes and constructing a corresponding expert decision matrix; using the SOM algorithm to cluster several subgroups based on the expert decision matrix; determining whether the group consensus level meets a threshold based on the expert decision matrix and the subgroups; if the group consensus level is less than the threshold, constructing a two-stage consensus optimization model based on a preset adaptive feedback mechanism to achieve the consensus process and obtain the final group decision matrix; and using the MULTIMOORA method to obtain the satellite emergency response plan ranking results for multi-attribute large-group emergency decision-making based on the final group decision matrix. This two-stage consensus optimization model employs an improved automatic strategy, ensuring the automatic CRP function of satellite plan evaluation while considering the experts' opinions; it combines a minimum adjustment optimization model and a minimum iteration optimization model, enabling CRP to run continuously under constraints and achieving adaptive optimization of the group consensus level and weights. Attached Figure Description

[0155] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0156] Figure 1 A flowchart illustrating an adaptive feedback method based on a two-stage consensus optimization model provided in an embodiment of the present invention;

[0157] Figure 2 This is a structural block diagram of an adaptive feedback system based on a two-stage consensus optimization model, provided for an embodiment of the present invention. Detailed Implementation

[0158] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention are described clearly and completely. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0159] This application provides an adaptive feedback method, system, storage medium, and electronic device based on a two-phase consensus optimization model, which solves the technical problem of difficulty in reaching consensus.

[0160] The technical solution in this application embodiment is to solve the above-mentioned technical problems, and the overall idea is as follows:

[0161] Due to the complexity and uncertainty of emergency decision-making problems involving large groups with multiple attributes, efforts are needed to address consensus challenges. In this embodiment of the invention, a two-stage consensus optimization model is proposed to guide the MALGEDM problem in a hesitant, fuzzy, binary language environment, reducing adjustment bias and the number of consensus rounds.

[0162] The main steps are as follows:

[0163] First, the Self-Organizing Map (SOM) algorithm is used to perform cluster analysis on the experts. For the MALGEDM problem, based on the known attributes of satellite emergency response plans, each expert provides their individual decision opinion, i.e., a decision matrix composed of HF2TLSs. Based on the decision matrix, clustering is performed using the SOM algorithm, grouping all experts into several subgroups.

[0164] Then, the consensus-reaching process is implemented. A method for determining subgroup weights based on cohesion is proposed, and the group decision matrix is ​​determined based on the subgroup decision matrix. Next, an adaptive feedback mechanism is constructed; on this basis, a two-stage consensus optimization model is established, incorporating expert opinions into the model. Using a two-stage objective function, the minimum adjustment amount of CRP is first obtained, and the minimum number of iterations is obtained as a constraint.

[0165] Finally, the satellite emergency response plan selection process is conducted. Based on the consensus reached in the previous step, the group decision preference matrix and the MULTIMOORA method are used to select the final satellite emergency response plan, improving the robustness and accuracy of the results, and outputting the final selected satellite emergency response plan.

[0166] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0167] Example:

[0168] like Figure 1 As shown, this embodiment of the invention provides an adaptive feedback method based on a two-stage consensus optimization model, including:

[0169] S1. Based on the known attributes of the satellite emergency response plan, obtain the decision opinions of various experts and construct the corresponding expert decision matrix;

[0170] S2. Based on the expert decision matrix, use the SOM algorithm to cluster and obtain several subgroups;

[0171] S3. Based on the expert decision matrix and subgroups, determine whether the group consensus level meets the threshold; if the group consensus level is less than the threshold, then on the basis of the preset adaptive feedback mechanism, construct a two-stage consensus optimization model to realize the consensus achievement process and obtain the final group decision matrix.

[0172] S4. Based on the final group decision matrix, the MULTIMOORA method is used to obtain the ranking results of satellite emergency solutions for multi-attribute large-group emergency decision-making.

[0173] The two-stage consensus optimization model introduced in this embodiment of the invention adopts an improved automatic strategy, which ensures the automatic function of CRP in satellite scheme evaluation while taking into account the wishes of experts; it combines the minimum adjustment optimization model and the minimum iteration number optimization model, so that CRP can run continuously under constraints and achieve adaptive optimization of the group consensus level and weight.

[0174] The following sections will detail each step of the aforementioned emergency satellite response plan:

[0175] First, it should be noted that, in the implementation of this invention, the MALGEDM problem is defined as: a large number of experts must make timely and high-quality decisions to address multiple attributes in response to an emergency by selecting from a set of feasible alternative satellite contingency plans.

[0176] Formally, the elements of the MALGEDM problem consist of the following:

[0177] 1) The discrete finite set X = {x1, ..., x} for emergency satellite emergency response plans m}(m≥2) represents a possible satellite contingency plan for solving the MALGEDM problem.

[0178] 2) A group of experts e = {e 1 ,…,e N (N≥20) experts were invited to evaluate the satellite contingency plan. Regarding the number of experts required for large group decision-making problems, there are two widely used standards: one defines it as having more than 11 experts, and the other as having more than 20 experts. In this study, the latter standard was used.

[0179] 3) A set of attributes C = {c1,…,c...} n (n≥2).

[0180] 4) Expert Hesitation-Fuzzy Two-Language (HF2TLS) Evaluation Score

[0181] Let m be the decision matrix for any expert, with n columns. Each row represents one of the m satellite emergency response plans, and each column represents an attribute of one of the n plans. Each position represents the expert's decision information regarding the satellite emergency response plan for each attribute. For HF2TLSs granularity; It consists of several HF2TLSs, the number of which is determined by expert opinion; It is a language tag, with values ​​S = {s0, s1, ..., s} g}(e.g., S = {s0 = extremely poor, s1 = very poor, s2 = poor, s3 = average, s4 = good, s5 = very good, s6 = exceptionally good}); Let g represent the sign transformation value and its range be [-0.5, 0.5), where g is the potential of S.

[0182] As expert e 1 The given decision matrix is

[0183]

[0184] In step S1, based on the known attributes of the satellite emergency response plan, the decision opinions of each expert are obtained, and the corresponding expert decision matrix is ​​constructed.

[0185] As mentioned above, in this step, the expert decision matrix is ​​represented using the hesitant fuzzy binary semantic set HF2TLSs, denoted as:

[0186] In step S2, based on the expert decision matrix, the SOM algorithm is used to cluster and obtain several subgroups; including:

[0187] S21. Initialization: Initialize the weights of each node to the standardized expert decision matrix, taking any number between 0 and 1; for each expert decision matrix, calculate the expectation of the hesitant fuzzy binary semantic set at each position according to the expectation formula, and obtain the input decision matrix.

[0188]

[0189] S22. Sampling: Randomly select an input decision matrix;

[0190] S23. Matching: Calculate the node BMU that is most similar to the input decision matrix among all nodes based on the Euler distance formula;

[0191]

[0192] When taking node BMU, Eulerian distance The minimum value is taken; i′ is the i′-th input decision matrix, with a total of m′; j′ is the j′-th neuron node in the winning neighborhood, with a total of n′; Indicates the weight of the corresponding node;

[0193] S24. Update: Determine the nodes adjacent to BMU and update the weights w of BMU and its adjacent nodes; among which, a Gaussian function is used to determine the modification of the weights of the neighborhood neurons.

[0194] W v (s+1)=W v (s)+θ(u,v,s)·α(s)

[0195] Among them, W v (s), W v (s+1) represent the current weight and the updated weight, respectively; θ(u,v,s) is the constraint on the update; α(s) is the learning rate.

[0196] S25, Continuous: Complete one iteration, return to S22, until the set number of iterations is met and the clustering result is output.

[0197] In step S3, based on the expert decision matrix and subgroups, it is determined whether the group consensus level meets the threshold; if the group consensus level is less than the threshold, a two-stage consensus optimization model is constructed on the basis of the preset adaptive feedback mechanism to realize the consensus achievement process and obtain the final group decision matrix.

[0198] After SOM clustering in step S2, all experts are divided into K subgroups, and each subgroup contains #C k One expert; based on the decision matrices of all experts in the subgroup, obtain the subgroup decision matrix. The subgroup weights are obtained based on the number of experts in each subgroup, and the group decision matrix is ​​determined using a weighted average operator.

[0199] The process of determining the group consensus level in this step includes:

[0200] Calculate the subgroup weights.

[0201] All experts were divided into K subgroups using SOM clustering, with each subgroup having a C0 ... k Contains #C k An expert calculated the cohesion of a subgroup. Get ρ is the weight parameter, e l Representing the lth expert, Representing the expert decision matrix subgroup C k Opinion Matrix G k The subgroup weights are calculated based on the Hausdorff distance between the two subgroups:

[0202]

[0203] Computational group consensus level,

[0204]

[0205]

[0206] Specifically, the subgroup decision matrix is ​​obtained based on the decision matrices of all experts in the subgroup. For the subgroup decision matrix, x i Satellite emergency response plan in c j Evaluation values ​​of attributes; determination of the group decision matrix using a weighted average operator. Group decision matrix based on x i Satellite emergency response plan in c j Evaluation value of the attribute;

[0207] SIM(C k ) represents the similarity between subgroups; G k and G f D represents the decision matrix for different subgroups; D represents the Hausdorff distance between different groups; GCL(C k () represents the level of consensus among the groups.

[0208] Before detailing the adaptive feedback mechanism in this step, we need to define consensus measures such as subgroup similarity and internal consensus level, as follows:

[0209] [1] Calculate the distance between decision matrices to lay the foundation for weight calculation and consensus measurement.

[0210]

[0211] in, express and The Hausdorff distance between hesitant, ambiguous binary semantic sets; and Let be the decision matrix for any two experts.

[0212] [2] Calculate the consensus level.

[0213] ·Define Expert e l Consensus level ECL(H) l )for:

[0214] ECL(H l )=1-D(H l ,R c ), where D is the Hausdorff distance metric in the formula.

[0215] Subgroup Consistency SCL(C) k )for:

[0216] · Among them, #C k For subgroup C k The number of experts in the field.

[0217] Subgroup C k Group decision matrix The consensus level between them is defined as:

[0218] ICL(C k )=1-D(G k ,R c ), D(G k ,R c This also reflects the distance between subgroups and groups.

[0219] ●The similarity between subgroups can be defined as:

[0220]

[0221] ●Based on similarity, the level of group consensus is obtained:

[0222]

[0223] ● The level of consensus among experts on each satellite contingency plan for each attribute:

[0224]

[0225] • Level of consensus among experts on each satellite contingency plan:

[0226] All of these consensus measures are within the interval [0,1].

[0227] In CRP, a predefined threshold ψ should be set in advance. If GCL(C k If ψ ≥ 0.5, the entire group reaches a consensus and then applies the selection process; otherwise, another iteration should be performed. The value of ψ depends on the specific problem to be solved: when the problem is very important, a higher ψ is needed, such as 0.9 or 0.85; when the problem is not very important or when a quick decision must be made for an urgent problem, a lower value is needed.

[0228] The adaptive feedback mechanism proposed in this embodiment of the invention requires activation when the group consensus level does not meet a predefined threshold. First, the group consensus level is categorized into four types: 1) low level, 2) medium level, 3) medium-high level, and 4) high level. Before using the adaptive feedback strategy, three parameters δ1, δ2, and a predefined group consensus threshold ψ should be determined at the start of the CRP. Simultaneously, the feedback process consists of identification rules and direction rules (DR). The overall strategy steps are shown in Table 1.

[0229] Table 1

[0230]

[0231] To elaborate:

[0232] First scenario: If GCL(C k ) less than the threshold ψ and less than the first parameter δ1: GCL(C k When δ1 < ψ, a low-level feedback strategy (FS-LL) is employed to achieve consensus. Typically, in the early stages of CRP, experts' opinions differ significantly, resulting in low consensus. In this situation, to reduce discrepancies, all subgroups need to address the lower consensus level (x) at the points of lower consensus. i ,c j Modifications were made; specifically, these included:

[0233] The FS-LL identification rules are as follows (to determine the object of changing preferences):

[0234] First, determine the satellite emergency plan. i The consensus level of the set ALT1 is lower than the average consensus level of all satellite contingency plans to be corrected.

[0235]

[0236]

[0237] in, To establish the level of consensus among experts on each satellite contingency plan;

[0238] Second, determine the location that needs to be modified (x) i ,c j The set POS1 contains;

[0239]

[0240]

[0241] Where POS1 represents all the locations that need to be modified (x i ,c j A set of ) The level of consensus among experts on each satellite contingency plan for each attribute;

[0242] The FS-LL directional rule DR.1 is as follows (determining the direction of preference changes to guide experts in CRP to adjust details):

[0243] DR.1.1: If Then expert e l Add to attribute c j Assessment;

[0244]

[0245] DR.1.2: If Then expert e l Reduce the impact on attribute c j Assessment;

[0246]

[0247] in, For subgroup C k The expert in the middle e l In the t-th iteration, for the satellite emergency plan x i In attribute c j The evaluation values ​​are: λ1∈[0.5,1],λ2∈[0,0.5], where λ1 and λ2 are the opinion adjustment parameters.

[0248] The second scenario: If GCL(C k If the threshold ψ is less than or equal to the first parameter δ1 and less than the second parameter δ1: δ1≤GCL<δ2<ψ, the consensus-reaching process is achieved by using the medium consensus level feedback strategy FS-ML.

[0249] In this situation, it is logical to reduce the number of experts who need to make modifications. By considering consistency, SCL(C k The process involves identifying the experts who need modification. If the consensus level of a subgroup is less than parameter δ2, then all experts in that subgroup should make minor modifications to the satellite contingency plan with low consensus. Conversely, if the subgroup consensus is greater than parameter δ2, but there is insufficient subgroup consensus, then only the experts affecting subgroup consensus are considered. Therefore, the feedback strategy will be divided into two aspects: FS-ML. 1 FS-ML 2 Next, we will introduce these two strategies in detail:

[0250] (1) When ICL(C k When ) < δ2,

[0251] FS-ML 1 The recognition rules are as follows:

[0252] First, identify subgroups whose consistency level is lower than the consistency parameter as...

[0253] SUB 21 ={C k |ICL(C k )<δ2}

[0254] ICL(C k )=1-D(G k ,R c )

[0255] Among them, ICL(C k ) is a subgroup C k Group decision matrix The level of consensus among them;

[0256] Second, for the identified subgroups, experts whose consistency level is lower than that of the subgroup are identified as...

[0257] EXP 21 ={e l |ECL(e l ) <SCL(C k )∧e l ∈C k}

[0258] ECL(H l )=1-D(H l ,R c )

[0259]

[0260] Among them, SCL(C k ) represents subgroup consistency; ECL(H) l (for expert e) l The level of consensus;

[0261] Third, satellite emergency response plans with a consistency level below the threshold will be identified as...

[0262]

[0263] Fourth, the location that needs to be modified (x) i ,c j ) is identified as

[0264]

[0265]

[0266] in, The level of consensus among experts on each satellite contingency plan for each attribute;

[0267] (2) When ICL(C k When )≥δ2,

[0268] The FS-ML 2 The recognition rules are as follows:

[0269] First, identify the subgroup with high consensus but low consistency.

[0270] SUB 22 ={C k |ICL(C k )≥δ2∪SCL(C k )<ψ}

[0271] Second, for the identified subgroups, experts whose consensus level is lower than that of the subgroup are identified as...

[0272] EXP 22 ={e l |ECL(e l ) <SCL(C k )∧e l ∈C k}

[0273] Third, satellite emergency response plans with a consensus level lower than the average consensus level of all satellite emergency response plans are identified as...

[0274]

[0275] Fourth, the location that needs to be modified (x) i ,c j ) is identified as

[0276]

[0277] The FS-ML (FS-ML) 1 and FS-ML 2 The direction rule (DR.2) is as follows:

[0278] DR.2.1: If So, expert e l Increase their understanding of satellite contingency plans. i In attribute c j Assessment;

[0279]

[0280] DR.2.2: If Then expert e l Reduce the impact on attribute c jAssessment;

[0281]

[0282] Where E∈[0,g] refers to the expected value of the evaluation; λ1∈[0.5,1],λ2∈[0,0.5]; Indicates the evaluation value HF2TLSs and The probability between them; Δ -1 This is a conversion function.

[0283] The third scenario: If GCL(C k If the threshold ψ is less than or equal to the second parameter δ1, and δ2 ≤ GCL < ψ, the FS-MHL (Medium-High Consensus Level Feedback) strategy is used to achieve consensus. After adjustments using the first two strategies, the group consensus level is very close to the pre-set threshold ψ. In this case, it means that the similarity of the subgroups is already high, and only a few experts affecting the consistency within the subgroups need to be modified; specifically including:

[0284] The identification rules for FS-MHL are as follows:

[0285] First, identify experts (EXP3) whose consensus level is lower than the subgroup consensus.

[0286] EXP3 = {e l |ECL(e l ) <SCL(C k )}

[0287]

[0288] Among them, SCL(C k Subgroup consistency;

[0289] Second, for any expert e l ∈EXP3, Identify satellite emergency response plan x i The rules are as follows;

[0290]

[0291] Third, for any satellite emergency plan x i ∈ALT3, identify the attribute c that needs to be modified. j for;

[0292]

[0293]

[0294] in, The level of consensus among experts on each satellite contingency plan for each attribute;

[0295] The directional rules (DR.3) of the FS-MHL are exactly the same as those of DR.2, and will not be repeated here.

[0296] Fourth case: If GCL(C k If the threshold ψ is greater than the threshold ψ, i.e., GCL≥ψ, no processing is required to achieve the consensus process.

[0297] In particular, this invention proposes a new formula for comparing the probability of HF2TLS (which is involved in the second and third cases mentioned above), to make a reasonable and accurate comparison of the expert's evaluation value, thereby providing a basis for the identification-direction rules in the adaptive feedback process.

[0298] See below for a specific formula for the probability degree of pairwise comparisons of HF2TLSs based on the uniform distribution probability criterion and the binary semantic distance measure, and further the properties of the probability degree ranking formula are given.

[0299] make and There are two in S = {s0, s1, ..., s} g HF2TLSs on}

[0300] definition and The probability between for:

[0301]

[0302] in,

[0303]

[0304]

[0305]

[0306]

[0307]

[0308] in, #T i (i = 1, 2, 3) represents the number of elements in the set T1, T2, T3.

[0309] Property: Makes If these are three hesitant, ambiguous, binary semantic sets on a set of linguistic terms, then...

[0310] 1) Normative:

[0311] 2) Intuitiveness: If but if but

[0312] 3) Complementarity:

[0313] 4) Reflexivity: If but

[0314] 5) Transitivity: If but

[0315] Proof: Since the proofs of properties 1)-4) are quite intuitive, we will only prove property 5) here.

[0316]

[0317] At the same time

[0318]

[0319] so

[0320]

[0321] Furthermore, step S3 introduces a two-stage consensus optimization model: an appropriate consensus-reaching mechanism can improve consensus efficiency. Undoubtedly, minimizing the adjustment of expert opinions and retaining as many original expert opinions as possible during the opinion adjustment process is a crucial factor affecting consensus efficiency. Moreover, respecting the experts' own preferences regarding satellite contingency plans is a noteworthy issue. In some practical situations, experts are only willing to add opinions to their preferred alternative satellite contingency plans; otherwise, they will refuse to modify them. This affects both the number of iterations and the amount of adjustment.

[0322] To address this, this invention proposes a two-stage consensus optimization model. The first-stage model (minimum adjustment model M1) aims to obtain the minimum adjustment amount under normal circumstances. Based on this, the second-stage model (minimum iteration count model M2) aims to reach a consensus as quickly as possible, taking into account the experts' willingness to implement emergency satellite solutions.

[0323] Phase 1: The first objective of optimizing the model is to minimize the overall adjustment opinions of the experts, i.e., to reduce adjustment bias. This helps to retain the initial opinions of the experts to the greatest extent possible and reduce the amount of adjustment. The corresponding objective function for model M1 is:

[0324]

[0325] Where t = {1,…T} represents the t-th iteration, and k = {1,…K} represents the k-th subgroup C. k ;

[0326] Subgroup C k The expert in the middle e l In the t-th iteration, for the satellite emergency plan x i In attribute c j The evaluation value.

[0327] and constraints:

[0328] (1-1) Deviation in Expert Decision-Making Opinions

[0329]

[0330] (1-2) Direction of adjustment of expert decision-making opinions

[0331]

[0332]

[0333] in, For the group decision matrix R c For x in the (t-1)th iteration i Satellite emergency response plan in c j Evaluation value of the attribute For subgroup C k Decision matrix G k For satellite emergency response plan x in the (t-1)th iteration i In attribute c j The evaluation value is given by P, which is the formula for the probability of design.

[0334] (1-3) Adjustment range of parameters λ1 and λ2

[0335] 0.5≤λ1≤1

[0336] 0≤λ²≤0.5

[0337] (1-4) Upper limit of the number of iteration rounds

[0338] T≤T1

[0339] Where T1 is the upper limit of the preset number of iteration rounds in the first stage; for example, in this embodiment of the invention, T1 = 15;

[0340] (1-5) The adjusted group consensus level reaches the consensus threshold.

[0341] GCL(C 1(T) C 2(T) ,…,C K(T))≥ψ;

[0342] It's worth noting that model M1 only adjusts the preference information of experts who haven't reached the consensus threshold, and its optimization objective is to minimize the change in expert opinions. By analyzing the model, we can determine the minimum adjustment bias. It should be noted that the above model does not take into account the experts' willingness to provide satellite emergency response plans. To minimize bias, the number of iterations should be further considered to ensure timely emergency response.

[0343] Second stage: Obtaining the minimum adjustment amount in the first stage Building upon the previous stage, the second phase incorporates the Expert Decision-Making (EWR) rule for satellite contingency plans into the adaptive feedback adjustment process. This aims to minimize the number of iterations through model optimization while respecting decision-makers' preferences. The first phase's directional rules do not consider the preferences of various experts regarding the satellite contingency plan itself. However, this is unavoidable in practical applications, for example, e 1 The preference is to choose satellite emergency plan x1, while e 2 A preference is placed on satellite contingency plan x2, which should also be considered. Therefore, to make the results more accurate, this stage adds expert satellite contingency plan intention rules (EWR) formation direction rules DR.3 and DR.4 to DR.1 and DR.2. Note that the specific design of the EWR rules is determined by the decision problem, as follows:

[0344] Directional rule DR.3 includes:

[0345] (1) DR.3.1:

[0346] correspond

[0347] if Then expert e l No modifications are needed to the satellite emergency plan. i In attribute c j The decision opinion is as follows; otherwise, the expert e shall follow the direction rule D1.2. l The satellite emergency plan needs to be added. i In attribute c j The above decision-making opinions;

[0348] (2)DR.3.2

[0349] correspond According to the aforementioned direction rule DR.1, expert e l The satellite emergency response plan needs to be adjusted. i In attribute e j The above decision-making opinions;

[0350] Directional rule DR.4 includes:

[0351] (1) DR.4.1:

[0352] correspond

[0353] if Then expert e l No modifications are needed to the satellite emergency plan. i In attribute c j The decision opinion is as follows; otherwise, the expert's opinion is as follows, according to the aforementioned directional rule DR.2.2. l The satellite emergency plan needs to be added. i In attribute c j The above decision-making opinions;

[0354] (2) DR.4.2:

[0355] correspond Then, according to the aforementioned direction rule DR.2, expert e l The satellite emergency response plan needs to be adjusted. i In attribute c j The decision-making opinions.

[0356] Guided by the rules in each of the above directions, the minimum adjustment objective is achieved through model M2 as follows:

[0357] Objective function of model M2:

[0358] minT

[0359] And the constraints of the second stage corresponding model M2:

[0360] (2-1) Obtain the minimum adjustment amount from model M1

[0361]

[0362] (2-2) Direction of Adjustment of Expert Decision-Making Opinions

[0363]

[0364] Among them, EWR represents the expert intention rule for design; DR.3.1 and DR.3.2, as well as DR.4.1 and DR.4.2 are all based on DR.1 and DR.2, with the addition of the expert satellite emergency plan intention rule EWR to form the direction rule;

[0365] (2-3) Same as (1-3) above;

[0366] (2-4) Upper limit of the number of iteration rounds

[0367] T≤T2

[0368] Where T2 is the preset upper limit of the number of iteration rounds in the second stage;

[0369] (2-5) Same as (1-5) above.

[0370] In step S4, based on the final group decision matrix, the MULTIMOORA method is used to obtain the ranking results of satellite emergency solutions for multi-attribute large-group emergency decision-making.

[0371] After consensus is reached through the adaptive feedback mechanism in step S3, the final group decision matrix R is obtained. c The normalization matrix R c =(x ij ) m×n First, three sorting algorithms, RS, RP, and FMF, are mapped using Algorithm 2 as shown in Table 2, and then integrated into a final sorting result to improve the accuracy and robustness of the result.

[0372] Table 2

[0373]

[0374] The MOORA (Multiobjective Optimization By Ratio Analysis) method was originally introduced by Brauers and Zavadskas and further enhanced by adding the Full Multiplication Method (FMF) to generate the Multiplicative MOORA (MULTIMOORA) method.

[0375] As an emerging multi-attribute decision-making method, MULTIMOORA is simple, effective, and easy to rank and optimize satellite emergency response plans compared to other methods. MULTIMOORA is based on three subordinate methods: Ratio System (RS), Reference Point (RP), and FMF; and uses dominance theory to calculate the final ranking. Therefore, the MULTIMOORA method can improve the accuracy of results in the final selection process of MALGEDM.

[0376] like Figure 2 As shown, this embodiment of the invention provides an adaptive feedback system based on a two-stage consensus optimization model, comprising:

[0377] The module is used to obtain the decision opinions of various experts based on the known attributes of satellite emergency plans and to construct the corresponding expert decision matrix.

[0378] The clustering module is used to obtain several subgroups by using the SOM algorithm based on the expert decision matrix;

[0379] The feedback module is used to determine whether the group consensus level meets the threshold based on the expert decision matrix and the subgroup; if the group consensus level is less than the threshold, a two-stage consensus optimization model is constructed on the basis of the preset adaptive feedback mechanism to realize the consensus reaching process and obtain the final group decision matrix.

[0380] The ranking module is used to obtain the ranking results of satellite emergency plans for multi-attribute large-group emergency decision-making based on the final group decision matrix using the MULTIMOORA method.

[0381] This invention provides a storage medium storing a computer program for adaptive feedback based on a two-phase consensus optimization model, wherein the computer program causes a computer to execute the adaptive feedback method based on the two-phase consensus optimization model as described above.

[0382] An electronic device, comprising:

[0383] One or more processors;

[0384] Memory; and

[0385] One or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs including an adaptive feedback method for performing the two-phase consensus optimization model as described above.

[0386] It is understood that the adaptive feedback system, storage medium, and electronic device based on the two-stage consensus optimization model provided in the embodiments of the present invention correspond to the adaptive feedback method based on the two-stage consensus optimization model provided in the embodiments of the present invention. The explanations, examples, and beneficial effects of the relevant contents can be referred to the corresponding parts of the adaptive feedback method based on the two-stage consensus optimization model, and will not be repeated here.

[0387] In summary, compared with existing technologies, it has the following beneficial effects:

[0388] 1. Based on refined expert identification rules and direction rules and specific modification details, the embodiments of the present invention introduce an adaptive feedback mechanism to realize CRP, and iterate multiple times to make the consensus level reach a given consensus threshold.

[0389] 2. By designing a probability formula applicable to hesitant fuzzy binary semantic sets, we can make reasonable and accurate comparisons of expert evaluation values, thereby providing a basis for the identification-direction rules in the adaptive feedback process.

[0390] 3. Design a novel adaptive feedback process. This process, belonging to an improved automatic strategy, can achieve automatic feedback while reducing time consumption and the involvement of coordinators.

[0391] 4. The method for determining the weights of the designed subgroups can simultaneously consider the subgroup size and cohesion.

[0392] 5. A two-stage consensus optimization model was constructed. This model adopts an improved automatic strategy, which ensures the automatic function of CRP in satellite scheme evaluation while taking into account the wishes of experts. The proposed two-stage adaptive consensus model combines the minimum adjustment optimization model and the minimum iteration number optimization model, which enables CRP to run continuously under constraints and achieves adaptive optimization of the group consensus level and weight.

[0393] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0394] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An adaptive feedback method based on a two-stage consensus optimization model, characterized in that, include: S1. Based on the known attributes of the satellite emergency response plan, obtain the decision opinions of various experts and construct the corresponding expert decision matrix; S2. Based on the expert decision matrix, use the SOM algorithm to cluster and obtain several subgroups; S3. Based on the expert decision matrix and subgroups, determine whether the group consensus level meets the threshold; if the group consensus level is less than the threshold, then on the basis of the preset adaptive feedback mechanism, construct a two-stage consensus optimization model to realize the consensus achievement process and obtain the final group decision matrix. S4. Based on the final group decision matrix, the MULTIMOORA method is used to obtain the ranking results of satellite emergency schemes for multi-attribute large-group emergency decision-making. The two-stage consensus optimization model in S3 includes: The objective function of model M1 corresponding to the first stage is: in, Indicates the first iteration Indicates the first Subgroup ; , Subgroups Chinese experts In the , The next iteration addresses satellite emergency plans. In attributes The evaluation value on; For the number of experts, The number of subgroups, For the number of satellite emergency response plans, The number of attributes for satellite emergency response plans; Hausdorff distance represents different hesitant and ambiguous binary semantic sets; And the objective function of the second-stage corresponding model M2: in, For the first Round of iterations; total number of iterations is ; The two-phase consensus optimization model in S3 also includes: The constraints for the first stage corresponding to model M1 are as follows: (1-1) Deviation in adjusting expert decision-making opinions (1-2) Direction of adjustment of expert decision-making opinions in, Group decision matrix In the In the next iteration, targeting Satellite emergency response plan Evaluation value of the attribute For subgroups Decision matrix In the The next iteration addresses satellite emergency plans. In attributes The evaluation value on the above Formula for the probability of design; For conversion functions; , and Adjust parameters based on feedback; For language tags, the value is... , for The momentum; Represents the symbolic conversion value and its range is ; (1-3) parameters and Adjustment range (1-4) Upper limit of the number of iteration rounds in, This is the upper limit for the number of iteration rounds preset in the first phase; (1-5) The adjusted group consensus level reaches the consensus threshold. ； And the constraints of the second stage corresponding model M2: (2-1) Obtain the minimum adjustment amount from model M1 in, Indicates the minimum adjustment deviation; (2-2) Direction of adjustment of expert decision-making opinions in, The expert consent rules represent the design; DR.3.1 and DR.3.2, as well as DR.4.1 and DR.4.2, are based on DR.1 and DR.2, with the addition of expert satellite emergency plan consent rules. Forming directional rules; (2-3) Same as (1-3) above; (2-4) Upper limit of the number of iteration rounds in, This is the upper limit of the preset number of iteration rounds in the second stage; (2-5) Same as (1-5) above.

2. The adaptive feedback method based on a two-stage consensus optimization model as described in claim 1, characterized in that, The expert decision matrix in S1 is represented by the hesitant fuzzy binary semantic set HF2TLSs, including: Among them, the discrete finite set of satellite emergency response plans , For the number of satellite emergency plans and ; Experts , For the number of experts and ; Attribute Collection , The number of attributes and ; Different positions For experts Selected satellite emergency plan For attributes Decision information; At the granularity of HF2TLSs, It consists of several HF2TLSs, the number of which is determined by expert opinion; For language tags, the value is... , for The momentum; Represents the symbolic conversion value and its range is .

3. The adaptive feedback method based on a two-stage consensus optimization model as described in claim 2, characterized in that, The process of determining the group consensus level in S3 includes: Calculate the subgroup weights. All experts were divided into groups using SOM clustering. Each subgroup Inside An expert calculated the cohesion of a subgroup. , obtain , For weight parameters, Representing the One expert, Representing the expert decision matrix and subgroup Subgroup decision matrix The subgroup weights are calculated based on the Hausdorff distance between the two subgroups: Computational group consensus level, Specifically, the subgroup decision matrix is ​​obtained based on the decision matrices of all experts in the subgroup. , For subgroup decision matrix Satellite emergency response plan Evaluation values ​​of attributes; determination of the group decision matrix using a weighted average operator. , Group decision matrix based on Satellite emergency response plan Evaluation value of the attribute; Similarity between subgroups; and For different subgroup decision matrices; Hausdorff distance between different groups; The level of consensus among the group.

4. The adaptive feedback method based on a two-stage consensus optimization model as described in claim 3, characterized in that, If in S3 Less than the threshold And less than the first parameter : The consensus-reaching process is achieved using a low-level feedback strategy (FS-LL), specifically including: The identification rules for FS-LL are as follows: First, determine the satellite emergency plan. Set Its consensus level is lower than the average consensus level of all satellite contingency plans to be revised; in, For experts Consensus level for each satellite contingency plan; Hausdorff distance represents different hesitant and ambiguous binary semantic sets; Second, determine the location that needs to be modified. Set ; in, For all locations that need modification The set; expert Consensus level for each satellite contingency plan across each attribute; The FS-LL direction rule as follows: :if Then experts Add attributes Assessment; :if Then experts Reduce attributes Assessment; in, For subgroups Chinese experts In the The next iteration addresses satellite emergency plans. In attributes The evaluation value on; , and Adjust parameters based on feedback; And / or if in S3 Less than the threshold And greater than or equal to the first parameter Less than the second parameter : The consensus-reaching process is achieved using a medium-level consensus feedback strategy, FS-ML, which specifically includes: The FS-ML identification rules are as follows: (1) When hour, First, identify subgroups whose consistency level is lower than the consistency parameter as... in, For subgroups Group decision matrix The level of consensus among them; Second, for the identified subgroups, experts whose consistency level is lower than that of the subgroup are identified as... in, For subgroup consistency; For experts The level of consensus; Third, satellite emergency response plans with a consistency level below the threshold will be identified as... Fourth, the areas that need to be modified Identified as (2) When hour, First, identify the subgroup with high consensus but low consistency. Second, for the identified subgroups, experts whose consensus level is lower than the subgroup's consistency are identified as... Third, satellite emergency response plans with a consensus level lower than the average consensus level of all satellite emergency response plans are identified as... Fourth, the areas that need to be modified Identified as ； The FS-ML direction rules as follows: :if So, experts Increase their understanding of satellite contingency plans In attributes Assessment; :if Then experts Reduce attributes Assessment; in, The expected value of the assessment; ; Indicates the evaluation value HF2TLSs and The degree of possibility between them; For conversion functions; And / or if in S3 Less than the threshold And greater than or equal to the second parameter : The consensus-reaching process is achieved using the medium-to-high consensus level feedback strategy FS-MHL, which specifically includes: The identification rules for FS-MHL are as follows: First, identify experts whose consensus level is lower than that of the subgroup consensus. ; in, For subgroup consistency; Second, for any expert Identify satellite emergency response plans The rules are as follows; Third, for any satellite emergency plan Identify the attributes that need to be modified. for; The FS-MHL direction rules as follows: :if So, experts Increase their understanding of satellite contingency plans In attributes Assessment; :if Then experts Reduce attributes Assessment; in, ; Indicates the evaluation value HF2TLSs and The probability between them.

5. The adaptive feedback method based on a two-stage consensus optimization model as described in claim 1, characterized in that, Directional rules DR.3 and DR.4 are as follows: Directional rule DR.3 includes: (1) DR.3.1: correspond ; if Then experts No modifications are needed to the satellite emergency plan. In attributes The expert's decision-making opinion; otherwise, in accordance with the aforementioned directional rule D1.2, the expert's opinion. There is a need to add contingency plans for satellites. In attributes Decision-making opinions; (2) DR.3.2 correspond According to the aforementioned direction rule DR.1, experts The emergency response plan for satellites needs to be adjusted. In attributes Decision-making opinions; Directional rule DR.4 includes: (1) DR.4.1: correspond ; if Then experts No modifications are needed to the satellite emergency plan. In attributes The expert's decision-making opinion; otherwise, in accordance with the aforementioned directional rule DR.2.2, the expert's opinion. There is a need to add contingency plans for satellites. In attributes Decision-making opinions; (2) DR.4.2: correspond Then, according to the aforementioned direction rule DR.2, the expert The emergency response plan for satellites needs to be adjusted. In attributes The decision-making opinions.

6. An adaptive feedback system based on a two-stage consensus optimization model, characterized in that, An adaptive feedback method for implementing the two-phase consensus optimization model as described in any one of claims 1 to 5, comprising: The module is used to obtain the decision opinions of various experts based on the known attributes of satellite emergency plans and to construct the corresponding expert decision matrix. The clustering module is used to obtain several subgroups by using the SOM algorithm based on the expert decision matrix; The feedback module is used to determine whether the group consensus level meets the threshold based on the expert decision matrix and the subgroup; if the group consensus level is less than the threshold, a two-stage consensus optimization model is constructed on the basis of the preset adaptive feedback mechanism to realize the consensus reaching process and obtain the final group decision matrix. The ranking module is used to obtain the ranking results of satellite emergency response schemes for multi-attribute large-group emergency decision-making based on the final group decision matrix using the MULTIMOORA method.

7. A storage medium, characterized in that, It stores a computer program for adaptive feedback based on a two-phase consensus optimization model, wherein the computer program causes a computer to execute the adaptive feedback method based on a two-phase consensus optimization model as described in any one of claims 1 to 5.

8. An electronic device, characterized in that, include: One or more processors; Memory; as well as One or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs including an adaptive feedback method based on a two-phase consensus optimization model as described in any one of claims 1 to 5.

Images