Industrial personal computer product selecting and sorting method based on multi-scale language evaluation scale

Abstract

The invention discloses an industrial personal computer product selecting and sorting method based on multi-scale language evaluation scales. The method comprises the steps that t evaluation matrixes of industrial personal computer products are aggregated into an initial probability language decision matrix, and the expected value of each probability language term set is calculated; calculating a weighted average value of expected values of all probability language term sets by utilizing a probability language ratio system, and taking the weighted average value as a sorting value of the probability language ratio system; a probability language reference point system is utilized, a reference point of the criterion is obtained through a unary quadratic score function, the distance between a probability language term set of each criterion and the reference point is calculated through distance measurement, and then a sorting value of the probability language reference point system is obtained; calculating a weighted geometric mean value expected by all criteria as a sorting value of each probabilistic language full multiplication model by utilizing the probabilistic language full multiplication model; and respectively calculating the comprehensive sorting score of each industrial personal computer product and sorting. According to the invention, a plurality of evaluation systems are integrated to obtain a more reasonable final sequence.

Description

Technical Field

[0001] The present invention relates to the field of information science, and in particular to an industrial computer product selection and sorting method based on a multi-scale language evaluation scale. Background Art

[0002] Industrial control computers, commonly known as industrial computers, are the core components of industrial control systems. Compared with ordinary computers, industrial computers have higher stability, reliability, security and anti-interference capabilities, and are suitable for long-term operation and stable control of industrial production processes in harsh environments. Therefore, the selection of industrial computers is a typical decision-making problem, which is not only important but also requires high accuracy. Currently, the commonly used multi-criteria decision-making methods include: ELECTRE method, MSM method, TOPSIS method, VIKOR method, etc. Multi-criteria decision-making methods usually only consider the opinions of a single decision maker, while multi-criteria group decision-making methods integrate the opinions of multiple decision makers, so the results are more objective.

[0003] Because fuzzy theory can well express the uncertainty of decision makers' evaluation, multi-criteria group decision-making methods characterized by fuzzy information can be used. Some methods propose intuitionistic fuzzy support functions, risk functions and credibility functions based on intuitionistic fuzzy sets, and construct an extended intuitionistic fuzzy sorting method, combined with group consensus adaptive search and adjustment methods to solve risk assessment problems. Some methods propose intuitionistic interval information aggregation operators based on intuitionistic interval numbers and use the framework of the TOPSIS method to deal with multi-criteria group decision-making problems. Some methods use the intuitionistic fuzzy multiplicative best-worst method with intuitionistic fuzzy multiplicative preference relations to solve the preference relations in the criteria.

[0004] Probabilistic language term sets have a good ability to express uncertain information in the process of dealing with multi-criteria group decision-making problems, and can represent decision information from both qualitative and quantitative aspects. Although multi-criteria group decision-making based on probabilistic language term sets has been widely studied and applied, there are still the following shortcomings: (1) The distance measurement method of probabilistic language term sets is often used in the decision-making problem solving process, but the current distance measurement method still has defects; (2) When representing decision information, the situation that different criteria use different language evaluation scales is ignored, and the current theories and methods in this regard are relatively lacking; (3) There is a lack of research on the framework of probabilistic language multi-criteria group decision-making methods based on multi-scale language evaluation scales. Summary of the invention

[0005] The purpose of the present invention is to provide an industrial computer product selection and ranking method based on a multi-scale language evaluation scale. The probabilistic language term set does not need to be standardized, the information of the original data will not be lost, and a more reasonable final ranking can be obtained by integrating multiple evaluation systems.

[0006] The technical solution adopted by the present invention is:

[0007] A method for selecting and ranking industrial computer products based on a multi-scale language evaluation scale comprises the following steps:

[0008] Step 1: Aggregate the t evaluation matrices of m industrial computer products into an initial probabilistic language decision matrix, and calculate the expected value of each probabilistic language term set in the probabilistic language decision matrix;

[0009] Step 2: Based on the expected value of each probabilistic language term set, the probabilistic language ratio system is used to calculate the weighted average of the expected values ​​of all probabilistic language term sets as the ranking value of the probabilistic language ratio system. The probability language ratio system ranking value calculation formula is as follows:

[0010]

[0011] Among them, w j represents the weight of the jth criterion; E(L ij (p)) is the probabilistic language term set L ij (p) expected value;

[0012] Specifically, we first calculate the expected value of each probabilistic language term set, and then find the weighted average of the expected values ​​of each criterion of the industrial computer product. The larger the weighted average value is, the better the industrial computer product is. Sort the industrial computer products in descending order of value.

[0013] Step 3: Using the probabilistic language reference point system, obtain the reference point of the criterion through a univariate quadratic score function and calculate the distance between the probabilistic language term set of each criterion and the reference point through a distance measure to obtain the ranking value of the probabilistic language reference point system Ranking values ​​of probabilistic language reference point systems;

[0014]

[0015] Among them, w j represents the weight of the jth criterion, represents the optimal probabilistic language term set for the jth criterion, represents the worst probability language term set for the jth criterion, Represents the distance between the best probability language term set and the worst probability language term set; Represents the distance between each data under the criterion and the optimal probability language term set;

[0016] Step 4: Use the probability language full multiplication model to calculate the weighted geometric mean of all criterion expectations as the ranking value of each probability language full multiplication model The formula is as follows:

[0017]

[0018] Among them, w j represents the weight of the jth criterion; E(L ij (p)) is the probabilistic language term set L ij (p) expected value;

[0019] Specifically, the expected value of the probability language term set is calculated, and then the weighted geometric mean of the expected values ​​of each criterion of the alternative solution is calculated. The larger the average value, the better the solution, so according to Sort the industrial computer products in descending order of value.

[0020] Step 5: Normalize the ranking values ​​of each industrial computer product in different systems to obtain a comprehensive ranking score and rank each industrial computer product according to the comprehensive ranking score. The comprehensive ranking score y i The calculation formula is as follows,

[0021] Furthermore, step 1 specifically includes the following steps:

[0022] Step 1-1, aggregate t evaluation matrices. The specific expression is as follows:

[0023]

[0024] Among them, i represents the number of alternative options, j represents the number of evaluation criteria, and L ij (p) represents the evaluation aggregation result of each expert on the jth criterion of the i-th alternative, represents the evaluation result of the first expert on the jth criterion of the i-th alternative, represents the evaluation result of the tth expert, ω g represents the weight of the g-th expert; Linguistic terms that convey qualitative information; Probability that represents quantitative information; The linguistic term representing the qualitative information given by the g-th expert; represents the probability of the quantitative information given by the g-th expert; represents the evaluation result given by the g-th expert;

[0025] Step 1-2, obtaining the initial probability language decision matrix L based on the aggregation results;

[0026]

[0027] L mn (p) represents the aggregated evaluation result of each expert on the nth criterion of the mth product;

[0028] Steps 1-3, and calculate the expected value of each probabilistic linguistic term set in the probabilistic linguistic decision matrix;

[0029]

[0030] Among them, #L(p) represents the number of probabilistic language terms; E(L ij (p)) is the probabilistic language term set L ij The expected value of (p).

[0031] Specifically, the specific steps of step 3 are as follows:

[0032] Step 3-1, first use the univariate quadratic score function to find the reference point of the criterion, as follows:

[0033]

[0034] in, represents the optimal probabilistic language term set for the jth criterion, represents the worst probability language term set of the jth criterion; sf(L 1j (p)) indicates L 1j (p) unary quadratic score, L 1j (p) represents the evaluation aggregation result of each expert on the jth criterion of the first alternative, and A linguistic term that represents the qualitative information of the jth criterion of the first alternative; The probability of the quantitative information of the jth criterion of the first alternative; #L 1j (p) indicates L 1j (p) the number of language terms; and so on, sf(L mj (p)) indicates L mj (p) unary quadratic score, L mj (p) represents the aggregated evaluation result of each expert on the jth criterion of the mth alternative, and A linguistic term that represents the qualitative information of the jth criterion of the mth alternative; The probability of the quantitative information of the jth criterion of the mth alternative; #L mj (p) indicates L mj (p) the number of language terms;

[0035] Step 3-2, calculate the distance between the best probability language term set and the worst probability language term set

[0036]

[0037] in, represents the optimal probabilistic language term set for the jth criterion, represents the worst probability language term set for the jth criterion, express The unary quadratic score value of express The unary quadratic score value of ;

[0038] Step 3-3, calculate the distance between each data under the criterion and the optimal probability language term set

[0039]

[0040] Among them, L lj represents the probabilistic language term set of the jth criterion; sf(L lj ) indicates L lj The unary quadratic score value of ;

[0041] Step 3-4, calculate and obtain the ranking value of the probabilistic language reference point system;

[0042]

[0043] In the above formula, w j represents the weight of the jth criterion, Represents the distance between the best probability language term set and the worst probability language term set; It represents the distance between each data under the criterion and the optimal probability language term set. The smaller the distance, the better the solution. Therefore, the probability language reference point system is based on Sort the industrial computer products in ascending order of value.

[0044] Furthermore, in step 5, according to the sorting score y i The industrial computer products are finally sorted from large to small, and the best product is selected based on the sorting.

[0045] The present invention adopts the above technical scheme, and has the following technical advantages over the prior art: 1. The probabilistic language term sets used in the prior art calculations are all based on the same language evaluation scale. When probabilistic language term sets based on different language evaluation scales appear in a decision problem, unified operations cannot be performed. The probabilistic language term sets involved in the calculation of the algorithm proposed by the present invention do not need to be standardized, and the information of the original data will not be lost. 2. The improved distance measurement method proposed by the present invention uses the difference of the univariate quadratic score function to calculate the distance between different probabilistic language term sets, because the one-to-one comparison between language terms is avoided. This method can not only calculate the distance between probabilistic language term sets under the same language evaluation scale, but also calculate the distance between probabilistic language term sets under different language evaluation scales. At the same time, because the univariate quadratic score function value can compare the advantages and disadvantages of different probabilistic language term sets, other comparison links can be omitted in the decision method, thereby simplifying the calculation process and improving the decision efficiency. 3. The traditional MULTIMOORA method uses the dominance theory to obtain the comprehensive ranking of the three models. Not only is the process cumbersome, but there is also the problem of circular reasoning. If the ranking results of the three models are simply integrated, there will also be a situation where the schemes are parallel. The present invention normalizes the probability language ratio system ranking value, the probability language reference point system ranking value and the probability language full multiplication model ranking value respectively. This process can not only reflect the ranking order of alternative options, but also show the relative differences between different ranking results, so as to obtain a more reasonable final ranking. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments;

[0047] Figure 1 The present invention is a flowchart of a method for selecting and ranking industrial computer products based on a multi-scale language evaluation scale. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0049] Current probabilistic linguistic multi-criteria group decision-making methods focus on the theories of distance measurement, correlation measurement, similarity, etc. of probabilistic linguistic term sets, but rarely consider probabilistic linguistic term sets based on multi-scale language evaluation scales, thus ignoring the research on the decision-making method framework required for such problems.

[0050] like Figure 1As shown, the present invention discloses a method for selecting and ranking industrial computer products based on a multi-scale language evaluation scale. In view of the situation where different criteria in group decision-making problems use different language evaluation scales, semantic scaling functions in the form of multiple language terms are introduced to represent the probabilistic language term set under the multi-scale language evaluation scale. With respect to the probabilistic language term set under the multi-scale language evaluation scale, the existing distance measure is analyzed. In order to avoid the distortion of decision maker evaluation information caused by the standardized operation, a univariate quadratic score function and an improved distance measure are proposed, which are respectively used to compare the pros and cons of two different probabilistic language term sets and calculate the distance between them. The specific principles of the present invention are described in detail below:

[0051] The definition of the probabilistic language terminology set is as follows: Let S1 = {s α |α=0,1,…,τ} is a given reference language evaluation scale, then the probabilistic language term set defined on S1 is:

[0052] L(p)={l (k) (p (k) )|l (k) ∈S1,p (k) ≥0,k=1,2,…,#L(p)}

[0053] Among them, l (k) (p (k) ) is a probabilistic language term. Each probabilistic language term consists of two parts: a language term representing qualitative information l (k) and the probability p representing quantitative information (k) , #L(p) represents the number of probabilistic language terms.

[0054] In order to solve the problem of operating on probabilistic language term sets with different language evaluation scales, we first introduce the following semantic scaling function: f:s α →δ α ,δ α ∈[0,1], because f is a strictly monotonically increasing function, so f -1 :δ α →s α ,δ α ∈[0,1]. Therefore, the semantic scaling function of the above form is given below.

[0055] Definition 1 (1) When the language evaluation scale is an unbalanced language evaluation scale, the semantic scaling function is:

[0056] f(s α )=α / 2τ,α∈[0,2τ]

[0057] (2) When the language evaluation scale is a balanced language evaluation scale, the semantic scaling function is:

[0058] f(s α )=(α+τ) / 2τ,α∈[-τ,τ]

[0059] (3) When the language evaluation scale is an unbalanced form of a non-uniform language evaluation scale with a gradually increasing deviation from the middle to both sides, the semantic scaling function is:

[0060]

[0061] Among them, the μ value is used to control the degree of gradual increase of the deviation. The larger the μ is, the greater the degree of increase of the deviation is.

[0062] (4) When the language evaluation scale is a balanced form of a non-uniform language evaluation scale with a gradually increasing deviation from the middle to both sides, the semantic scaling function is:

[0063]

[0064] Here, the μ value has the same meaning as above.

[0065] (5) When the language evaluation scale is an unbalanced form of a non-uniform language evaluation scale with a gradually decreasing deviation from the middle to both sides, the semantic scaling function is:

[0066]

[0067] Among them, the γ value is used to control the degree to which the deviation gradually decreases. The larger the γ, the greater the degree of deviation reduction.

[0068] (6) When the language evaluation scale is a balanced form of a non-uniform language evaluation scale with a gradually decreasing deviation from the middle to both sides, the semantic scaling function is:

[0069]

[0070] Since the operations of the current probabilistic language term set require normalization of the probability, such as normalizing L(p) = {s3(0.7)} to It is obvious that the normalization changes the evaluation information of the decision maker and loses the authenticity of the decision information to a certain extent. In order to solve the above problem, the present invention proposes a new expected value and deviation function, so that the decision process no longer performs probability normalization processing on the probability language term set, but directly calculates the original probability language term set.

[0071] Definition 2 Assume that L(p) = {l (k) (p (k) )|l (k) ∈S,p (k) ≥0,k=1,2,…,#L(p)} is a probabilistic language term set, where S can be any form of language evaluation scale, and its expected value is defined as follows:

[0072]

[0073] Definition 3 Assume that \(L(p)=\{l (k) (p (k) )|l (k) \in S, p (k) \geq0, k = 1, 2, \ldots, \#L(p)\}\) is a probabilistic linguistic term set, where \(S\) can be a linguistic evaluation scale in any form, and its deviation value is defined as follows:

[0074]

[0075] Use the expected value and deviation value to compare the sizes of probabilistic linguistic term sets. For any two probabilistic linguistic term sets \(L_1(p)\) and \(L_2(p)\):

[0076] If \(E(L_1(p)) > E(L_2(p))\), then \(L_1(p) > L_2(p)\);

[0077] If \(E(L_1(p)) < E(L_2(p))\), then \(L_1(p) < L_2(p)\);

[0078] If \(E(L_1(p)) = E(L_2(p))\), then if \(\sigma(L_1(p)) > \sigma(L_2(p))\) then \(L_1(p) > L_2(p)\), if \(\sigma(L_1(p)) < \sigma(L_2(p))\) then \(L_1(p) < L_2(p)\), if \(\sigma(L_1(p)) = \sigma(L_2(p))\) then \(L_1(p) = L_2(p)\).

[0079] To address the problem that the existing distance measures for probabilistic linguistic term sets cannot handle unbalanced linguistic evaluation scales, the present invention proposes a quadratic scoring function, which is defined as follows:

[0080] Definition 4 Assume that \(L(p)=\{l (k) (p (k) )|l (k) \in S, p (k) \geq0, k = 1, 2, \ldots, \#L(p)\}\) is a probabilistic linguistic term set, where \(S\) can be a linguistic evaluation scale in any form. Its quadratic scoring function formula is as follows:

[0081]

[0082] Integrate the concepts of expectation and variance into one formula. By calculating the unary quadratic score value, the superiority and inferiority of two probabilistic linguistic term sets can be directly judged, which is more concise than the method of comparing the score function and the deviation value. The judgment rule is as follows: For any two probabilistic linguistic term sets, if sf(L1(p)) > sf(L2(p)), then L1(p) > L2(p); if sf(L1(p)) < sf(L2(p)), then L1(p) < L2(p); if sf(L1(p)) = sf(L2(p)), then L1(p) = L2(p).

[0083] Based on the above score function, the following improved distance measure method is proposed and defined as follows:

[0084] Definition 5 Suppose there are two probabilistic linguistic term sets L1(p) = {l1 (k) (p1 (k) ) | l1 (k) ∈S 1 , p (k) ≥0, k = 1, 2, …, #L1(p)} and L2(p) = {l2 (k) (p2 (k) ) | l2 (k) ∈S 2 , p2 (k) ≥0, k = 1, 2, …, #L2(p)}, where SS 1 and S 2 are reference linguistic evaluation scales in any form. The improved distance measure formula is as follows:

[0085] d(L1(p), L2(p)) = |sf(L1(p)) - sf(L2(p))|

[0086] where

[0087]

[0088] Since Definition 5 uses the unary quadratic function f(x) = 2x - x 2 , which is strictly monotonically increasing in (0, 1) and has f(0) = 0, f(1) = 1. Therefore, the improved distance measure satisfies the following properties:

[0089] (1) Boundedness, 0 ≤ d(L1(p), L2(p)) ≤ 1;

[0090] (2) Reflexivity, d(L1(p), L1(p)) = 0;

[0091] (3) Symmetry, d(L1(p), L2(p)) = d(L2(p), L1(p)).

[0092] The present invention includes three ranking systems: a probability language ratio model, a probability language reference point system and a probability language full multiplication model. The three ranking systems can obtain three rankings, and the final ranking of the scheme is obtained by integrating the three rankings. Compared with the existing probability language MULTIMOORA method, the method used in the present invention expands the calculation range of the probability language term set and improves the probability language reference point system, so it can solve more complex decision problems. In order to more intuitively express the decision steps of the probability language MULTIMOORA method based on the multi-scale language evaluation scale, a method flow chart is given, as shown in FIG. Figure 1 shown.

[0093] The evaluation criteria in the decision-making process include benefit-based and cost-based. In the traditional MULTIMOORA method, the two types of criterion data need to be calculated separately, but in the method used in the present invention, the calculation of the two types of criteria can be unified by setting different language evaluation scales. For benefit-based criteria, a positive language evaluation scale is set, that is, the larger the subscript of the language term, the higher the evaluation, which means "high return"; for cost-based criteria, a negative language evaluation scale is set, that is, the larger the subscript of the language term, the lower the evaluation, which means "low cost". In summary, the larger the subscript of the language term, the better the result.

[0094] The above method is applied to the selection problem of industrial computer products. Assume that a company wants to purchase a batch of industrial computers, and there are m products (A1, A2, ..., A m ) are available, and the enterprise invites t experts in related fields (D1, D2, …, D t ) from n aspects (C1, C2, ..., C n ) (such as price, reliability, safety and other criteria) to conduct a comprehensive evaluation of industrial computer products. The expert group negotiated and decided that the weight of each criterion is w = (w1, w2, ..., w n ) T Each expert gives the probabilistic language evaluation data of each criterion according to the given language evaluation scale information, and obtains t probabilistic language evaluation matrices in

[0095] Step 1: Aggregate t evaluation matrices into an initial probabilistic language decision matrix. The aggregation method is as follows:

[0096]

[0097] Among them, i represents the number of alternative options, j represents the number of evaluation criteria, and L ij (p) represents the evaluation aggregation result of each expert on the jth criterion of the i-th alternative, represents the evaluation result of the first expert on the jth criterion of the i-th alternative, represents the evaluation result of the tth expert, ω g Represents the weight of the g-th expert, the default is ω g =1 / t, and different weights can also be assigned according to the identity of the expert; Linguistic terms that convey qualitative information; Probability that represents quantitative information; The linguistic term representing the qualitative information given by the g-th expert; represents the probability of the quantitative information given by the g-th expert; represents the evaluation result given by the g-th expert;

[0098] Then the initial probability language decision matrix L is obtained.

[0099]

[0100] L mn (p) represents the aggregated evaluation result of each expert on the nth criterion of the mth product.

[0101] Step 2: Use the probability language ratio system to calculate the comprehensive ratio value of each industrial computer product. The calculation formula is as follows:

[0102]

[0103] Among them, w j represents the weight of the jth criterion. First, use the formula in Definition 2 to calculate the expected value of each probabilistic language term set, and then find the weighted average of the expected values ​​of each criterion of the industrial computer product. The larger the weighted average value is, the better the industrial computer product is. Therefore, according to Sort the industrial computer products in descending order of value.

[0104] Step 3: Use the probabilistic language reference point system to calculate the ranking value of each industrial computer product. First, use the univariate quadratic score function to find the reference point of the criterion, as shown below.

[0105]

[0106] in, represents the optimal probabilistic language term set for the jth criterion, represents the worst probabilistic language term set of the jth criterion. Then the distance between the probabilistic language term sets is calculated by the formula in Definition 5 to obtain the ranking value.

[0107]

[0108] In the above formula, wj represents the weight of the jth criterion, Represents the distance between the best probability language term set and the worst probability language term set. It represents the distance between each data under the criterion and the optimal probability language term set. The smaller the distance, the better the solution. Therefore, the probability language reference point system is based on Sort the industrial computer products in ascending order of value.

[0109] Step 4: Use the probability language full multiplication model to calculate the comprehensive ratio value of each industrial computer product. The formula is as follows:

[0110]

[0111] Among them, w j represents the weight of the jth criterion. This method uses the formula in Definition 2 to calculate the expected value of the probabilistic language term set, and then calculates the weighted geometric mean of the expected values ​​of each criterion of the alternative solution. The larger the average value, the better the solution, so according to Sort the industrial computer products in descending order of value.

[0112] Step 5: Use the following formula to sort the industrial computer products.

[0113]

[0114] In the above formula, They are the ranking values ​​of the probability language ratio system, the probability language reference point system and the probability language full multiplication model respectively. is the maximum value of the probability language ratio system ranking value, is the minimum value of the probability language ratio system ranking value, and the other two are the same. Finally, according to y i The industrial computer products are finally sorted from large to small, and the best product is selected based on the sorting.

[0115] The present invention adopts the above technical scheme, and has the following technical advantages over the prior art: 1. The probabilistic language term sets used in the prior art calculations are all based on the same language evaluation scale. When probabilistic language term sets based on different language evaluation scales appear in a decision problem, unified operations cannot be performed. The probabilistic language term sets involved in the calculation of the algorithm proposed by the present invention do not need to be standardized, and the information of the original data will not be lost. 2. The improved distance measurement method proposed by the present invention uses the difference of the univariate quadratic score function to calculate the distance between different probabilistic language term sets, because the one-to-one comparison between language terms is avoided. This method can not only calculate the distance between probabilistic language term sets under the same language evaluation scale, but also calculate the distance between probabilistic language term sets under different language evaluation scales. At the same time, because the univariate quadratic score function value can compare the advantages and disadvantages of different probabilistic language term sets, other comparison links can be omitted in the decision method, thereby simplifying the calculation process and improving the decision efficiency. 3. The traditional MULTIMOORA method uses the dominance theory to obtain the comprehensive ranking of the three models. Not only is the process cumbersome, but there is also the problem of circular reasoning. If the ranking results of the three models are simply integrated, there will also be a situation where the schemes are parallel. The present invention normalizes the probability language ratio system ranking value, the probability language reference point system ranking value and the probability language full multiplication model ranking value respectively. This process can not only reflect the ranking order of alternative options, but also show the relative differences between different ranking results, so as to obtain a more reasonable final ranking.

[0116] Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. In the absence of conflict, the embodiments in the present application and the features in the embodiments can be combined with each other. The components of the embodiments of the present application generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the detailed description of the embodiments of the present application is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians of the art without making creative work are within the scope of protection of the present application.

Claims

1. A method for selecting and ranking industrial computer products based on a multi-scale language evaluation scale, characterized by: The following steps are involved: Step 1: Aggregate the t evaluation matrices of m industrial computer products into an initial probabilistic language decision matrix, and calculate the expected value of each probabilistic language term set in the probabilistic language decision matrix; Step 2: Using the probability language ratio system, calculate the weighted average of the expected values ​​of all probability language term sets as the ranking value of the probability language ratio system The probability language ratio system ranking value calculation formula is as follows: Among them, w j represents the weight of the jth criterion; E(L ij (p)) is the probabilistic language term set L ij (p) expected value; Step 3: Using the probabilistic language reference point system, obtain the reference point of the criterion through a univariate quadratic score function, and calculate the distance between the probabilistic language term set of each criterion and the reference point through a distance measure, and then obtain the ranking value of the probabilistic language reference point system The formula for calculating the ranking value of the probabilistic language reference point system is as follows: Among them, w j represents the weight of the jth criterion, represents the optimal probabilistic language term set for the jth criterion, represents the worst probability language term set for the jth criterion, Represents the distance between the best probability language term set and the worst probability language term set; Represents the distance between each data under the criterion and the optimal probability language term set; Step 4: Using the probability language full multiplication model, calculate the weighted geometric mean of all criterion expectations as the ranking value of each probability language full multiplication model The formula is as follows: Among them, w j represents the weight of the jth criterion; E(L ij (p)) is the probabilistic language term set L ij (p) expected value; Step 5: Normalize the ranking values ​​of each industrial computer product in different systems to obtain a comprehensive ranking score and rank each industrial computer product according to the comprehensive ranking score. The comprehensive ranking score y i The calculation formula is as follows, in, They are the ranking values ​​of the probability language ratio system, the probability language reference point system and the probability language full multiplication model respectively; is the maximum value of the probability language ratio system ranking value, The minimum value of the probability language ratio system ranking value; is the maximum value of the probability language reference point system ranking value, The minimum value of the probability language reference point system ranking value; is the maximum value of the probability language full multiplication model ranking value, Minimum value for the probabilistic linguistic fully multiplicative model ranking value.

2. The method for selecting and ranking industrial computer products based on a multi-scale language evaluation scale according to claim 1, characterized in that: Step 1 specifically includes the following steps: Step 1-1, aggregate t evaluation matrices. The specific expression is as follows: Among them, i represents the number of alternative options, j represents the number of evaluation criteria, and L ij (p) represents the evaluation aggregation result of each expert on the jth criterion of the i-th alternative, represents the evaluation result of the first expert on the jth criterion of the i-th alternative, represents the evaluation result of the tth expert, ω g represents the weight of the g-th expert; Linguistic terms that convey qualitative information; Probability that represents quantitative information; The linguistic term representing the qualitative information given by the g-th expert; represents the probability of the quantitative information given by the g-th expert; represents the evaluation result given by the g-th expert; Step 1-2, obtaining the initial probability language decision matrix L based on the aggregation results; L mn (p) represents the aggregated evaluation result of each expert on the nth criterion of the mth product; Steps 1-3, and calculate the expected value of each probabilistic linguistic term set in the probabilistic linguistic decision matrix; Among them, #L(p) represents the number of probabilistic language terms; E(L ij (p)) is the probabilistic language term set L ij The expected value of (p).

3. The method for selecting and ranking industrial computer products based on a multi-scale language evaluation scale according to claim 2 is characterized in that: The weight ω of the g-th expert g =1 / t; or, assign different weights according to the identity of the expert.

4. The method for selecting and ranking industrial computer products based on a multi-scale language evaluation scale according to claim 1, characterized in that: The specific steps of step 3 are as follows: Step 3-1, first use the univariate quadratic score function to find the reference point of the criterion, as follows: in, represents the optimal probabilistic language term set for the jth criterion, represents the worst probability language term set of the jth criterion; sf(L 1j (p)) indicates L 1j (p) is a quadratic score, L 1j (p) represents the evaluation aggregation result of each expert on the jth criterion of the first alternative, and A linguistic term that represents the qualitative information of the jth criterion of the first alternative; The probability of the quantitative information of the jth criterion of the first alternative; #L 1j (p) indicates L 1j (p) the number of language terms; and so on, sf(L mj (p)) indicates L mj (p) is a quadratic score, L mj (p) represents the aggregated evaluation result of each expert on the jth criterion of the mth alternative, and A linguistic term that represents the qualitative information of the jth criterion of the mth alternative; The probability of the quantitative information of the jth criterion of the mth alternative; #L mj (p) indicates L mj (p) the number of language terms; Step 3-2, calculate the distance between the best probability language term set and the worst probability language term set in, represents the optimal probabilistic language term set for the jth criterion, represents the worst probability language term set for the jth criterion, express The unary quadratic score value of express The unary quadratic score value of ; Step 3-3, calculate the distance between each data under the criterion and the optimal probability language term set Among them, L lj represents the probabilistic language term set of the jth criterion; sf(L lj ) indicates L lj The unary quadratic score value of ; Step 3-4, calculate and obtain the ranking value of the probabilistic language reference point system; Among them, w j represents the weight of the jth criterion, Represents the distance between the best probability language term set and the worst probability language term set; Represents the distance between each data under the criterion and the optimal probability language term set.

5. The method for selecting and ranking industrial computer products based on a multi-scale language evaluation scale according to claim 1, characterized in that: In step 5, according to the sorting score y i The industrial computer products are finally sorted from large to small, and the best product is selected based on the sorting.

Images