A multi-scenario selection multi-factor structural weight analysis method

Abstract

The present application relates to multi-factor selection of multi-scheme, and provides a multi-factor structural weight analysis method for multi-scheme selection, which is used for determining the influence degree of multi-factor on the selection process of multi-scheme. Since the scheme has not been implemented, the scheme implementation results under the influence of multi-factor are lacking. The method represents the integrity of factor influence through the relationship between the factor value domain and the whole domain of the factor value in each scheme. The distinctness of factor influence is represented through the value domain relationship of the same factor in different schemes. The integrity and distinctness are described by interval numbers, and the factor structural weight is formed by defining the integrity degree and distinctness degree. The method includes the construction of an analysis system, a decision matrix, a normalization matrix, the determination of the factor integrity degree and distinctness degree, the factor influence degree and weight. The method is applied to the determination of the structural weight of each factor in the selection of mining methods. The method can be applied to the determination of the influence of factors on the selection of schemes under the condition of lacking scheme results.

Description

Technical Field

[0001] This invention relates to the selection of multiple options based on multiple factors, and in particular to the determination of the structural weights of multiple factors in the selection of multiple options, clarifying the role of each factor in the process of determining the options. Background Technology

[0002] Multiple-option comparison is a common problem faced by various industries in production and daily life, and it is also a complex decision-making process. This process requires considering multiple influencing factors and the implementation results of each option under these influencing factors. However, the problem is obvious: it's impossible to select a suitable option based solely on the implementation results of each option under the influence of multiple factors. Especially since the selection process doesn't yield a final result, only a certain expectation. Therefore, we need to consider the relationship between each option and the changing influencing factors, including the value range relationship of each factor in different options, and the relationship between the value range of each factor in each option and the entire value range of that factor, to determine the role of each factor in the selection of each option. That is, the degree of influence of multiple factors on the multiple-option selection process is called factor structural weight. This differs from traditional weight determination methods. Traditional methods generally require the system's input values ​​(i.e., influencing factors) and the system's output values ​​(i.e., the option results) as a correspondence to determine the relationship between factor changes and result changes. The weight of the factor's influence on the result is then determined based on the degree of change. Obviously, this method cannot be used without option results. Therefore, this approach proposes determining the importance of a factor in option selection solely through the relationship between the factor and the option. This is only related to the system structure composed of multiple factors and multiple options, and is independent of the option results. This is the basic idea behind the structural weighting of factors.

[0003] Currently, there are numerous studies on the weights of factors, conditions, and attributes under the influence of various systems and factors, providing effective analytical methods for the relationship between system state changes and factor changes in related fields. The results of these methods constitute the factor influence weights. However, these studies primarily employ methods such as causal relationship analysis and input-output relationship analysis. These methods are not applicable to determining the structural weights of the aforementioned factors and cannot be used because they lack implementation results and system outputs.

[0004] To address the aforementioned issues, this invention characterizes the overall influence of a factor by relating its value range to the entire value range of that factor in each scheme; and characterizes the distinguishability of the factor's influence by relating the value ranges of the same factor in different schemes. Interval numbers are used to define the overall influence and the distinguishability, ultimately enabling the calculation of the structural weights of the factors. Summary of the Invention

[0005] The purpose of this invention is to provide a multi-factor structural weight analysis method for multiple alternative selection, so as to solve the technical problem in the prior art that there is no way to realize multi-factor structural weight analysis calculation when selecting multiple alternatives.

[0006] Factor structural weights;

[0007] Weight generally refers to the degree of importance of a factor or indicator relative to a certain thing, reflecting the percentage that a factor or indicator accounts for. It emphasizes the relative importance of the factor or indicator, tending to focus on contribution or importance. Weight represents the correspondence between a change in a factor and a change in the system state, or the absolute change in the system state caused by a one-unit change in a factor. When all other conditions are equal, if a change in one factor leads to a greater change in the system state than the influence of another factor, then the weight of the first factor is greater than the weight of the second factor. There are many methods for determining weights, including analytical methods, analytic hierarchy process (AHP), logical reasoning, and expert experience.

[0008] Analytical methods are the most accurate way to determine factor weights or importance. In this case, the value of the influencing factor is taken as the independent variable, the system state result as the dependent variable, and the weight is the rate of change of the dependent variable with respect to the independent variable. Essentially, this involves solving for the rate of change of the factor's effect. When a precise analytical expression exists, this rate of change can be obtained by taking partial derivatives. For example, the factor importance mentioned in spatial fault tree theory. A single factor is used as the independent variable, and the system failure probability is used as the dependent variable to form a characteristic function. Considering the characteristic functions of all factors, the failure probability distribution of a certain component is constructed. Through the structural form of the system composed of components, the failure probability distributions of all components are used to construct the system failure probability distribution P(x1,x2,…,x…). N ), where x n For factor f n The value of is given by n = 1, ..., N. Therefore, P(x1, x2, ..., x...) N ) is a function determined by all factors. Therefore, factor f n The weight or importance can be expressed as This solution method is the most explicit and simple, but it is generally difficult to obtain an analytical expression to implement the above process.

[0009] The Analytic Hierarchy Process (AHP) is a semi-empirical, semi-computational method. Its foundational data comes from expert experience, and calculations are performed based on that experience. Expert data forms a judgment matrix, which is then used for calculations based on Saaty's nine importance levels and assigned values. The final result is the weight of each factor.

[0010] When there are only logical relationships but no explicit and sufficient data, the weights of each factor can be obtained through these logical relationships. The state of the result can be obtained by changing the state of the factors rather than through data. If, through deduction, it is found that the state of the system changes after a factor changes from one state to its opposite state, this indicates that the factor played a significant role. For example, structural importance in fault tree analysis.

[0011] When the logical relationships are unclear, the only recourse is generally through intelligent algorithms or expert systems. Based on expert experience and the actual situation on-site, experts will rank the importance of each factor, similar to the ranking of weights. However, obtaining precise quantitative weight values ​​is difficult. Alternatively, multiple factor values ​​can be used as inputs, and the corresponding system states as outputs, forming a finite set of correspondences. The scaling factors of these correspondences are the weights. This is the inherent mechanism of artificial neural network models.

[0012] The above discussion actually considers a system with both input and output. Whether based on human experience or data, weights are a mapping relationship between input and output, i.e., deformation and scaling coefficients. However, in practice, especially in prediction problems, there are often no explicit system output values ​​as the basis for constructing relationships.

[0013] This invention provides a multi-factor structural weight analysis method for multiple alternative selection. To determine the degree of influence of multiple factors on the multiple alternative selection process, this method is proposed. Since there is a lack of comparison of alternative implementation results under the influence of multiple factors before the alternatives are implemented, the method characterizes the overall influence of factors by the relationship between the factor's value range and the entire value range of that factor in each alternative; it characterizes the distinguishability of factor influence by the relationship between the value ranges of the same factors in different alternatives; it describes the overall influence and distinguishability using interval numbers, and defines the overall degree and distinguishability, forming the structural weights of factors; it includes constructing an analysis system, a decision matrix, a normalization matrix, determining the overall degree and distinguishability of factors, the influence degree of factors, and their weights; it is applied to the determination of the structural weights of each factor when selecting mining methods; and it is applied to the determination of the influence of factors on alternative selection under the condition of lacking alternative results.

[0014] For multiple alternatives, the weights of factors can only be determined through comparative analysis of factor values ​​on the input side, and are defined as factor structural weights. Structural weights do not depend on the system results; they are calculated solely based on the degree of influence of the factor itself on the alternatives. Factor structural weights examine both the relationship between the value range of a factor across all alternatives and the global value range of that factor across all alternatives, and the relationship between the value ranges of the same factor across different alternatives.

[0015] A factor's value range covering a wider range of factors indicates a broader influence on the selection of alternatives. A larger range signifies a greater influence, or a higher weight. The relationship between the value ranges of the same factor across different alternatives represents the factor's distinctiveness. If all alternatives have the same value range for a factor, it means the factor has the same effect on all alternatives, making it impossible to distinguish the factor's influence across different alternatives. If the factor value ranges are different and do not overlap, it indicates that the factor's influence varies across different alternatives, representing different, distinguishable factor weights.

[0016] In summary, the relationship between the factor's value range and the entire range of that factor's value in each scheme is defined as the holistic nature of the factor weight; the relationship between the same factor's value range in different schemes is defined as the distinguishability of the factor weight. Thus, even if changes in factors lead to unknown outcomes in the system state, the selection effect of each factor on each scheme can be preliminarily determined through the relationship between the scheme and the factors, i.e., the structural weight of the factors.

[0017] Interval representation of wholeness and distinctiveness;

[0018] The value range of the factors mentioned above is clearly not a single numerical value, but rather an interval of values ​​that varies within a certain range. Interval numbers can effectively represent range values. For example, a = [a...]. - ,a + ] represents the interval number a, and its maximum value is a. + The minimum value is a - Interval numbers can represent the range of factor values. The normalized interval number a1 = [a1...] - a1 + ] and a2 = [a2 - a2 + Draw on a line segment of unit length.

[0019] There are six scenarios where the two interval numbers are in different relative positions; these six scenarios are: the global range of a1 is less than the global range of a2; the global range of a2 is less than the global range of a1; the total value of the global range of a1 is less than the total value of the global range of a2 and they overlap; the total value of the global range of a2 is less than the total value of the global range of a1 and they overlap; the global range of a2 contains the total value of the global range of a1; and the global range of a1 contains the total value of a2. The relationship between the factor value range and the global value range of the factor can be determined by the ratio of the interval number length to the unit length line segment, i.e., the globality, and its value is |a + -a - The length relationship between the union and intersection of the value ranges of the same factors in different schemes can represent the distinguishability.

[0020] When the ranges represented by two interval numbers overlap, it means that the two interval numbers have the same value, that is, the same factor has the same range of values ​​in different schemes. This represents the correlation between different schemes and the same factor, where ||a1∩a2|| is the proportion α of the common part. 12 || represents the length of the interval. ||a1∪a2|| represents the proportion β of the entire related range. 12 Furthermore, if α 12 =0 proves that the two intervals do not intersect, meaning there is no correlation. In this case, the correlation between the two intervals is minimal, and the discrimination is maximum, which is beneficial for differentiating the effects of the factor on different options. ||a1∩a2||=||a1∪a2|| indicates that the two intervals overlap, with a maximum correlation of 1. In this case, the discrimination is minimal, which is detrimental to the effect of the factor on different options. Therefore, the discrimination is defined as 1-α.12 / β 12 .

[0021] The combination of wholeness and discrimination can represent the degree of influence of a factor on a certain plan, thus defining the degree of influence and laying the foundation for determining the structural weights of each factor.

[0022] Regarding algorithm construction:

[0023] The impact of factors on a solution depends on their wholeness and discriminativeness. Therefore, the factor structural weighting algorithm constructed here is based on the corresponding wholeness and discriminativeness. The algorithm construction process is given below.

[0024] 1) Construct the analysis system. Let the set of possible solutions for the system be A = {A1, ..., A...} M}, m=1,…,M,A m Let A ∈ M, and M be the total number of alternatives. The set of factors influencing the selection of a suitable alternative in the system is F = {f1, ..., f2}. N}, n=1,…,N,f n ∈F, where N is the number of influencing factors, i.e., the evaluation index for selecting a suitable scheme.

[0025] 2) Construct the decision matrix Γ0. Factor f n For option A m The range of factors affecting the scope of influence is: This is clearly an interval number. The resulting decision matrix Γ0 is shown in equation (1).

[0026]

[0027] 3) Construct the normalization matrix Γ1. Normalize all interval numbers in Γ0. The normalization standard is the factor f. n For all schemes A 1～M The maximum value of the influence range corresponds to 1, and the minimum value corresponds to 0. At this point... The normalization algorithm is shown in equation (2). The normalization matrix is...

[0028]

[0029] 4) Determine f n The overall coherence and distinctiveness. Analytical factors f n For all schemes A 1～M The impact can be analyzed from the perspectives of overallity and differentiation.

[0030] Wholeness represents factor f n For a certain plan A m The scope of its influence. Due to f n For A 1～MThe scope of influence has been normalized, therefore this is equivalent to f n and A m The length of the corresponding interval number in Γ1.

[0031] Two normalized interval numbers, and They represent f respectively n For A i and A j The range of influence, i,j∈{1,...,M}. From the diagram, it can be seen that due to... This shows that f n For A j The scope of effect is greater than f n For A i The former has a wider scope of influence, therefore it is more holistic. Therefore, in terms of overall impact... Superior Numerically available f n For A m The overall impact.

[0032] and They are f n For option A m and A j The normalized influence interval number. Where α mjn for and The intersection of interval numbers is an interval number with length . β mjn for and The union of interval numbers is an interval number with length .

[0033] 5) Determine f n Influence and weight ω n Factor f n For option A m The overall degree and the distinguishability are combined to define the influence of factors on the scheme, as shown in Equation (3).

[0034]

[0035] This leads to the influence matrix Γ2 = (δ mn ) MN In equation (3), ||α mjn || and ||β mjn The calculation of || is cumbersome, requiring consideration of six cases. Summarizing these six cases, ||α is... mjn || and ||β mjnThe calculation method and conditions for || are shown in equation (4).

[0036]

[0037] Factor f n For all schemes A 1～M The impact is All factors f 1～N For all schemes A 1～M The impact is Therefore, factor f n Weight ω in the structure n As shown in equation (5).

[0038]

[0039] Unless otherwise specified, all letters in this application are intermediate variables.

[0040] The beneficial technical effects of this invention are as follows: The above-described method for analyzing the influence of factors on scheme selection ultimately quantifies and forms the structural weights of these factors. Its key feature is that it does not require the results of each scheme's execution as basic data for analysis; it only requires identifying a limited number of factors and schemes, as well as the range of values ​​within which each factor influences each scheme. This provides an effective method for analyzing the importance of factors in scheme selection and prediction when no data is available. It is also important to note that the basic assumption of the above calculation method is that changes in factors and changes in results are benefit-oriented. That is, an increase in the factor value leads to a positive outcome. Besides benefit-oriented changes, cost-oriented and neutral changes also exist. These two types can be easily transformed into benefit-oriented changes. Attached Figure Description

[0041] Figure 1 A schematic diagram showing six relative positions of two interval numbers is provided;

[0042] Figure 2 This demonstrates the significance of wholeness;

[0043] Figure 3 This demonstrates the significance of the discrimination factor. Detailed Implementation

[0044] The embodiments of this application will be further described below with reference to the accompanying drawings:

[0045] Regarding the structural weights of factors:

[0046] Weight generally refers to the degree of importance of a factor or indicator relative to a certain thing, reflecting the percentage that a factor or indicator accounts for. It emphasizes the relative importance of the factor or indicator, tending to focus on contribution or importance. Weight represents the correspondence between a change in a factor and a change in the system state, or the absolute change in the system state caused by a one-unit change in a factor. When all other conditions are equal, if a change in one factor leads to a greater change in the system state than the influence of another factor, then the weight of the first factor is greater than the weight of the second factor. There are many methods for determining weights, including analytical methods, analytic hierarchy process (AHP), logical reasoning, and expert experience.

[0047] Analytical methods are the most accurate way to determine factor weights or importance. In this case, the value of the influencing factor is taken as the independent variable, the system state result as the dependent variable, and the weight is the rate of change of the dependent variable with respect to the independent variable. Essentially, this involves solving for the rate of change of the factor's effect. When a precise analytical expression exists, this rate of change can be obtained by taking partial derivatives. For example, the factor importance mentioned in spatial fault tree theory. A single factor is used as the independent variable, and the system failure probability is used as the dependent variable to form a characteristic function. Considering the characteristic functions of all factors, the failure probability distribution of a certain component is constructed. Through the structural form of the system composed of components, the failure probability distributions of all components are used to construct the system failure probability distribution P(x1,x2,…,x…). N ), where x n For factor f n The value of is given by n = 1, ..., N. Therefore, P(x1, x2, ..., x...) N ) is a function determined by all factors. Therefore, factor f n The weight or importance can be expressed as This solution method is the most explicit and simple, but it is generally difficult to obtain an analytical expression to implement the above process.

[0048] The Analytic Hierarchy Process (AHP) is a semi-empirical, semi-computational method. Its foundational data comes from expert experience, and calculations are performed based on that experience. Expert data forms a judgment matrix, which is then used for calculations based on Saaty's nine importance levels and assigned values. The final result is the weight of each factor.

[0049] When there are only logical relationships but no explicit and sufficient data, the weights of each factor can be obtained through these logical relationships. The state of the result can be obtained by changing the state of the factors rather than through data. If, through deduction, it is found that the state of the system changes after a factor changes from one state to its opposite state, this indicates that the factor played a significant role. For example, structural importance in fault tree analysis.

[0050] When the logical relationships are unclear, the only recourse is generally through intelligent algorithms or expert systems. Based on expert experience and the actual situation on-site, experts will rank the importance of each factor, similar to the ranking of weights. However, obtaining precise quantitative weight values ​​is difficult. Alternatively, multiple factor values ​​can be used as inputs, and the corresponding system states as outputs, forming a finite set of correspondences. The scaling factors of these correspondences are the weights. This is the inherent mechanism of artificial neural network models.

[0051] The above discussion actually considers a system with both input and output. Whether based on human experience or data, weights are a mapping relationship between input and output, i.e., deformation and scaling coefficients. However, in practice, especially in prediction problems, there are often no explicit system output values ​​as the basis for constructing relationships. For multiple alternatives, the weights of factors can only be determined by comparing and analyzing factor values ​​on the input side, defined as factor structural weights. Structural weights do not depend on the system results, but only on the degree of influence of the factor itself on the alternatives. Factor structural weights examine, on the one hand, the relationship between the value range of a factor in all alternatives and the entire value range of that factor in each alternative, and on the other hand, the relationship between the value ranges of the same factor in different alternatives.

[0052] A factor's value range covering a wider range of factors indicates a broader influence on the selection of alternatives. A larger range signifies a greater influence, or a higher weight. The relationship between the value ranges of the same factor across different alternatives represents the factor's distinctiveness. If all alternatives have the same value range for a factor, it means the factor has the same effect on all alternatives, making it impossible to distinguish the factor's influence across different alternatives. If the factor value ranges are different and do not overlap, it indicates that the factor's influence varies across different alternatives, representing different, distinguishable factor weights.

[0053] In summary, the relationship between the factor's value range and the entire range of that factor's value in each scheme is defined as the holistic nature of the factor weight; the relationship between the same factor's value range in different schemes is defined as the distinguishability of the factor weight. Thus, even if changes in factors lead to unknown outcomes in the system state, the selection effect of each factor on each scheme can be preliminarily determined through the relationship between the scheme and the factors, i.e., the structural weight of the factors.

[0054] Interval representation of wholeness and distinctiveness:

[0055] The value range of the factors mentioned above is clearly not a single numerical value, but rather an interval of values ​​that varies within a certain range. Interval numbers can effectively represent range values. For example, a = [a...]. - ,a + ] represents the interval number a, and its maximum value is a. + The minimum value is a - Interval numbers can represent the range of factor values. The normalized interval number a1 = [a1...]- a1 + ] and a2 = [a2 - a2 + Draw on a line segment of unit length, such as Figure 1 As shown.

[0056] Figure 1 There are six cases that result from the two interval numbers being in different relative positions. These six cases are: the global a1 is less than the global a2; the global a2 is less than the global a1; the total global a1 is less than the total global a2 and they overlap; the total global a2 is less than the total global a1 and they overlap; the global a2 contains the total global a1; and the global a1 contains the total global a2.

[0057] The relationship between the factor value range and the global factor value range can be determined using the ratio of the interval length to the unit length line segment, i.e., the globality, and its value is |a|. + -a - The length relationship between the union and intersection of the value ranges of the same factors in different schemes can represent the distinguishability.

[0058] like Figure 1 As shown, when the ranges represented by two interval numbers overlap, it indicates that the two interval numbers have the same value, meaning that the same factor has the same range of values ​​in different schemes. This represents the correlation between different schemes and the same factor, where ||a1∩a2|| is the proportion α of the common part. 12 || represents the length of the interval. ||a1∪a2|| represents the proportion β of the entire related range. 12 Furthermore, if α 12 =0 proves that the two intervals do not intersect, meaning there is no correlation. In this case, the correlation between the two intervals is minimal, and the discrimination is maximum, which is beneficial for differentiating the effects of the factor on different options. ||a1∩a2||=||a1∪a2|| indicates that the two intervals overlap, with a maximum correlation of 1. In this case, the discrimination is minimal, which is detrimental to the effect of the factor on different options. Therefore, the discrimination is defined as 1-α. 12 / β 12 .

[0059] The combination of wholeness and discrimination can represent the degree of influence of a factor on a certain plan, thus defining the degree of influence and laying the foundation for determining the structural weights of each factor.

[0060] Regarding algorithm construction:

[0061] The impact of factors on a solution depends on their wholeness and discriminativeness. Therefore, the factor structural weighting algorithm constructed here is based on the corresponding wholeness and discriminativeness. The algorithm construction process is given below.

[0062] 1) Construct the analysis system. Let the set of possible solutions for the system be A = {A1, ..., A...} M}, m=1,…,M,A m Let A ∈ M, and M be the total number of alternatives. The set of factors influencing the selection of a suitable alternative in the system is F = {f1, ..., f2}. N}, n=1,…,N,f n ∈F, where N is the number of influencing factors, i.e., the evaluation index for selecting a suitable scheme.

[0063] 2) Construct the decision matrix Γ0. Factor f n For option A m The range of factors affecting the scope of influence is: This is clearly an interval number. The resulting decision matrix Γ0 is shown in equation (1).

[0064]

[0065] 3) Construct the normalization matrix Γ1. Normalize all interval numbers in Γ0. The normalization standard is the factor f. n For all schemes A 1～M The maximum value of the influence range corresponds to 1, and the minimum value corresponds to 0. At this point... The normalization algorithm is shown in equation (2). The normalization matrix is...

[0066]

[0067] 4) Determine f n The overall coherence and distinctiveness. Analytical factors f n For all schemes A 1～M The impact can be analyzed from the perspectives of overallity and differentiation.

[0068] Wholeness represents factor f n For a certain plan A m The scope of its influence. Due to f n For A 1～M The scope of influence has been normalized, therefore this is equivalent to f n and A m The length of the corresponding interval number in Γ1.

[0069] Figure 2 There are two normalized interval numbers. and They represent f respectively n For A i and A j The range of influence, i,j∈{1,...,M}. From the diagram, it can be seen that due to... This shows that f n For A jThe scope of effect is greater than f n For A i The former has a wider scope of influence, therefore it is more holistic. Therefore, in terms of overall impact... Superior Numerically available f n For A m The overall impact.

[0070] Figure 3 This represents the significance of differentiation. (See the image.) and They are f n For option A m and A j The normalized influence interval number. Where α mjn for and The intersection of interval numbers is an interval number with length . β mjn for and The union of interval numbers is an interval number with length .

[0071] 5) Determine f n Influence and weight ω n Factor f n For option A m The overall degree and the distinguishability are combined to define the influence of factors on the scheme, as shown in Equation (3).

[0072]

[0073] This leads to the influence matrix Γ2 = (δ mn ) MN In equation (3), ||α mjn || and ||β mjn The calculation of || is cumbersome and requires consideration of, for example Figure 1 There are six possible scenarios. Summarize the six scenarios where ||α mjn || and ||β mjn The calculation method and conditions for || are shown in equation (4).

[0074]

[0075] Factor f n For all schemes A 1～M The impact is All factors f 1～N For all schemes A 1～M The impact is Therefore, factor f n Weight ω in the structuren As shown in equation (5).

[0076]

[0077] The structural weights of various factors in mining scheme selection were analyzed using the methods described above. An analytical system was constructed, considering influencing factors including loss rate f1, dilution rate f2, production capacity f3, mining-to-cut ratio f4, mining efficiency f5, mining cost f6, safety f7, ventilation conditions f8, technological complexity f9, and adaptability f1. 10 Explosive consumption f 11 Construction difficulty 12 Mechanization 13 Labor intensity f 14 Among them, f1, f2, f4, f6 and f 11 The first indicator is cost-based, while the others are benefit-based. Alternative solutions include full-scale stope filling method A1, shallow-hole room-pillar filling method A2, longwall stope filling method A3, multi-level strip mining filling method A4, hydraulic support cut-and-fill filling method A5, shortwall mining filling method A6, and segmented stope filling method A7. A decision matrix Γ0 is constructed based on the above methods, as shown in Table 1.

[0078] Table 1. Decision matrix composed of basic data

[0079]

[0080] Semantic quantization: Let the range of values ​​corresponding to the semantics in Table 1 be [0.6, 0.7], [0.7, 0.8], [0.8, 0.9], and [0.9, 1], respectively. Construct the normalization matrix Γ1 according to equation (2).

[0081] Determine f n The overallity and discriminative power are calculated according to the definition, and the specific values ​​are shown in Table 2. The specific values ​​obtained according to the definition of discriminative power are shown in Table 3.

[0082] Table 2. Overall Degree of Factors

[0083]

[0084] Table 3. Discrimination of Factors

[0085]

[0086] Factor f n The degree of influence is obtained by multiplying the corresponding elements of the overall degree in Table 2 and the distinguishability in Table 3.

[0087] Factor f n For all schemes A 1～M The impact is as follows:

[0088] δ 1～14 ={1.4172,2.7535,1.2497,2.0643,2.0160,2.5714,1.7500,1.7500,2.3333,2.3333,4.4762,2.3333,2.3333,2.3333}. All factors f 1～N For all schemes A 1～M The total impact is 31.7149. Therefore, factor f... n Weights on the structure composed of all schemes:

[0089] ω 1～14 ={0.0447,0.0868,0.0394,0.0651,0.0636,0.0811,0.0552,0.0552,0.0736,0.0736,0.1411,0.0736,0.0736,0.0736}. The structural importance of each factor is ranked as follows: explosive consumption f 11 >Dilution rate f2>Mining cost f6>Process complexity f9=Adaptability f 10 =Construction difficulty f 12 =Mechanization f 13 =Labor intensity f 14 > Mining ratio f4 > Mining efficiency f5 > Safety f7 = Ventilation conditions f8 > Loss rate f1 > Production capacity f3. This illustrates the factors f 11 f3 played the greatest role in selecting from the above 7 options, while f3 played the least role.

[0090] In summary, the multi-factor structural weighting analysis method for multiple alternative selection determines factor weights from the structural perspective of the system composed of factors and alternatives by considering the proportion of the factor's value range to the full range of the factor's value in each alternative, as well as the relative position of the value range of the same factor in different alternatives.

Claims

1. A multi-scenario selection multi-factor structural weight analysis method, characterized in that, In order to determine the influence degree of multiple factors on the process of multi-scheme selection, the method is proposed. Since the scheme has not been implemented, the comparison of the scheme implementation results under the influence of multiple factors is lacking. The method represents the integrity of factor influence through the relationship between the value domain of the factor and the value domain of the factor in each scheme. The distinctiveness of factor influence is represented by the relationship between the value domain of the same factor in different schemes. The integrity and distinctiveness are described by interval numbers, and the overall degree and distinct degree are defined to form the structural weight of the factor; It includes constructing analysis system, decision matrix, normalized matrix, determining factor overall degree and distinct degree, factor influence degree and weight; applied to the determination of structural weight of each factor in mining method selection; applied to the determination of the influence of factors on scheme selection under the condition of lacking scheme results; For multiple alternative schemes, the weight of the factor is determined by comparing and analyzing the factor value on the input side, which is defined as the structural weight of the factor. The structural weight does not depend on the system results, but only depends on the influence degree of the factor itself on the scheme. The structural weight of the factor examines the relationship between the value domain of a certain factor in all schemes and the value domain of the factor in each scheme, and the relationship between the value domain of the same factor in different schemes; Interval number representation of integrity and distinctiveness; the value domain of the factor is not a single numerical value, but an interval value that changes within a certain range; Interval numbers effectively represent range values; let represent an interval number with maximum value and minimum value ; interval number represents a value range of a factor; the normalized interval number and is plotted on a line segment of unit length; The six cases are caused by the different relative positions of two interval numbers, and the six cases are respectively: Total domain is less than Total domain; Total domain is less than Total domain; Total domain is less than Total domain and has overlap; Total domain is less than Total domain and has overlap; Total domain contains Total domain; Total domain contains Total domain; the relationship between the factor value domain and the factor value total domain is determined by the ratio of the length of the interval number and the unit length line segment, that is, the integrity, and the value is ; while the length relationship of the union and intersection of the same factor value domain of different schemes represents the distinction. When two interval numbers have intersection, it means that two interval numbers have the same value, i.e. the same factor in different schemes has the same range, which represents the relevance of the same factor in different schemes, is the proportion of the same part , represents the length of the interval number; represents the proportion of the whole range ; further, if proves that two interval numbers are disjoint, i.e. there is no relevance; at this time, the relevance of two interval numbers is the smallest, and the discrimination degree is the largest, which is conducive to distinguishing the effect of the factor on different schemes; , it means that two interval numbers coincide, and the relevance is the largest, i.e. 1; at this time, the discrimination degree is the smallest, which is not conducive to distinguishing the effect of the factor on different schemes; therefore, the discrimination degree is defined as ; The comprehensive representation of overall degree and distinct degree represents the influence degree of a certain factor on a certain scheme, thereby defining the influence degree, which lays the foundation for determining the structural weight of each factor; The role of the factor on the scheme depends on the integrity and distinctiveness, and the method construction process; 1) Constructing analysis system; let the set of schemes selected by the system be , , , be the total number of schemes; let the set of influence factors affecting the selection of suitable schemes by the system be , , , be the number of influence factors, i.e. evaluation indexes for selecting suitable schemes; 2) Constructing the decision matrix ; Factors To the scheme The range of the impact of the generated factors is This is obviously an interval number; the decision matrix formed at this time As shown in equation (1); （1） 3) Construct the normalized matrix ;right Normalize all interval numbers in the data, and the normalization standard is the factor. For all schemes The maximum value of the influence range corresponds to 1, and the minimum value corresponds to 0; at this time... The normalization algorithm is shown in equation (2); the normalization matrix is... ; （2） 4) determine overall and discriminant; analyze factors impact on all scenarios from overall and discriminant; analyze; The overall degree represents the factor The impact of a certain solution on the range of influence; since The range of influence on is normalized, this is equivalent to and The corresponding interval length in ; Two normalized interval numbers, and , respectively represent for and Scope of influence ;because This shows right The scope of effect is greater than right The former has a wider scope of influence, therefore it is more holistic; therefore, in terms of overall impact... Superior Numerically using express right The overall impact; and are respectively the normalized influence interval number of the scheme and ; wherein is and the intersection of the interval numbers, is an interval number, whose length is ; is and the union of the interval numbers, is an interval number, whose length is ; 5) determining the degree of influence and weight of each factor on the overall degree and differentiation of the solution as shown in equation (3). （3） Further, the influence degree matrix is obtained ; in formula (3) and The calculation is complicated, and six cases need to be considered; the calculation method and conditions of and are summarized as shown in formula (4); （4） factor The impact of all factors on all scenarios is The impact of all factors on all scenarios is The structural weight of factors is shown in equation (5): （5）； When the structural weight analysis method of multiple factors for multi-scheme selection is used to analyze the structural weight of each factor in mining scheme selection, it specifically includes: A construction analysis system is built, and the influencing factors considered include loss rate , dilution rate , production capacity , cutting-mining ratio , mining efficiency , mining cost , safety , ventilation condition , process complexity , adaptability , unit consumption of explosive , construction difficulty , mechanization , labor intensity ; wherein and are cost-type indexes, and the rest are benefit-type indexes Alternative options include full-mine backfilling , shallow hole room and pillar backfilling , longwall mining backfilling , multi-layer strip mining backfilling , hydraulic support wall cutting backfilling , shortwall mining backfilling , sublevel room and pillar backfilling .

2. The multi-scenario selected multi-factor structural weight analysis method according to claim 1, characterized in that, Constructing the decision matrix As shown in Table 1 : ； Semantic quantification, set the value domain corresponding to the semantics in Table 1 as [0.6, 0.7], [0.7, 0.8], [0.8, 0.9], and [0.9, 1] for better, good, very good, respectively; The normalization matrix is constructed according to formula (2) ; determine the overall degree and the discrimination degree; the overall degree is calculated according to definition, and the specific value is shown in Table 2; the specific value obtained according to the definition of the discrimination degree is shown in Table 3; ； ； Influence of factors The degree of influence of the factors is obtained by multiplying the overall degree of Table 2 and the discrimination degree of Table 3 for the corresponding elements; Factors The impact for all scenarios is: ； All factors The sum of the effects on all scenarios is 31.7149; then the weight of the factor on the structure of all scenarios ； The structural importance of each factor is sorted as: explosive unit consumption > depletion rate > mining cost > process complexity = adaptability = ease of construction = mechanization = labor intensity > cut-to-fill ratio > mining efficiency > safety = ventilation conditions > loss rate > production capacity ; explanatory factor Among the above-mentioned 7 options, the one that plays the greatest role in the selection is The one that plays the least role is

Images