Response potential model evaluation method based on construction of load cluster

Abstract

The invention discloses a response potential model evaluation method based on construction of a load cluster, and relates to the technical field of power system management. The method comprises the steps of S1, obtaining load measurement data, S2, constructing a load decomposition model, S3, determining load curve platform power, S4, calculating demand response potential, S5, constructing user portrait feature indexes, S6, determining the demand response potential and S7, evaluating and sorting. According to the method, the response potentials of different loads can be effectively evaluated and sorted through S1-S7, and a scientific basis is provided for management, regulation and control of a power system; through the SWARA method and the CRITIC method, the subjective weight and the objective weight are obtained respectively, then the comprehensive weight is obtained, it is guaranteed that the obtained comprehensive weight is close to the subjective weight and the objective weight, the error rate of the comprehensive weight is increased, and therefore the response potential sorting and evaluation result of various loads is more accurate.

Description

Technical Field

[0001] The present invention relates to the technical field of power system management, and in particular to a response potential model evaluation method based on building a load cluster. Background Art

[0002] The total load of the power system is the sum of the total power consumed by all electrical equipment in the system; the power consumed by industry, agriculture, post and telecommunications, transportation, municipal administration, commerce, and urban and rural residents is added together to get the comprehensive power load of the power system; the power consumed by the comprehensive power load plus the power loss of the network is the power that each power plant in the system should supply, which is called the power supply load of the power system;

[0003] Under the market competition mechanism, the evaluation of demand response potential can help users, load aggregators and power grid dispatching centers to clarify the source, expected scale and response effect of demand response resources, thereby helping to formulate the goals and strategies for demand response development to meet the needs of maximizing their respective interests;

[0004] However, existing technologies still have some shortcomings when evaluating the response potential model of power grid load clusters:

[0005] 1. Most existing technologies are not convenient for effectively evaluating and ranking the response potential of different loads during the evaluation process, thus failing to provide a scientific basis for the management of the power system;

[0006] 2. At the same time, in the evaluation process, the existing technology uses a single and incomplete calculation method when calculating the weights of various indicators, which makes it difficult for the calculated weights to fit the actual situation, resulting in a high error rate in the results of the load response potential ranking and evaluation;

[0007] In view of the above problems, the inventors propose a response potential model evaluation method based on building a load cluster to solve the above problems. Summary of the invention

[0008] In order to solve the above problems, the present invention aims to provide a response potential model evaluation method based on building load clusters.

[0009] In order to solve the above technical problems, the present invention adopts the following technical solution: a response potential model evaluation method based on building a load cluster, comprising the following steps:

[0010] S1. Obtain load measurement data

[0011] First, collect and analyze the load measurement data of power users, including historical load data and power consumption behavior;

[0012] S2. Build load decomposition model

[0013] The load decomposition model of the dedicated transformer user is constructed through the STL algorithm, and the load is decomposed into periodic components to better understand the periodic changes of the load;

[0014] S3. Determine the load curve platform power

[0015] The load periodic component is processed by the SG filtering algorithm to determine the power that can represent the load curve platform. Each local minimum point can represent the power of the load curve platform.

[0016] S4. Calculation of demand response potential

[0017] According to the specified demand response start time, the real-time load power of the load periodic component at the response start time is calculated using the dedicated transformer user load decomposition model and the load curve platform power determination model. The demand response potential power is determined by comparing the real-time load power with the platform power.

[0018] S5. Constructing user portrait feature indicators

[0019] Based on the theoretical response potential and historical data of power user load, the user portrait characteristic index is constructed to characterize the power consumption characteristics of power users;

[0020] S6. Determine demand response potential

[0021] Based on the user portrait characteristic indicators, determine the demand response potential of power users, and determine the regional adjustable power according to the demand response potential;

[0022] S7. Evaluation and ranking

[0023] The SWARA method and CRITIC method are used to calculate the comprehensive weight of each indicator, and then the response potential of each type of load is evaluated and ranked.

[0024] Preferably, in S2, the specific steps of constructing the load decomposition model of the dedicated transformer user by the STL algorithm are as follows:

[0025] A1): First, collect the load data of the dedicated transformer user, wherein the load data includes the load value of the time series;

[0026] A2): Set the parameters of the STL algorithm, and then gradually decompose the load data into trend components, seasonal components and residual components through multiple iterations, and then output the decomposed trend components, seasonal components and residual components for subsequent analysis;

[0027] A3): Construct a load decomposition model for dedicated transformer users, wherein the load decomposition model includes a trend prediction model, a seasonal prediction model and residual processing.

[0028] Preferably, in S3, the specific steps of the SG filtering algorithm for processing the periodic component of the load are as follows:

[0029] B1): Determine the filter parameters of the SG filter algorithm, the number of neighboring points used to fit the polynomial at each location, so as to produce a smoother output;

[0030] B2): For each point in the signal, construct a data window with the specified window length;

[0031] B3): In each data window, a polynomial of a specified order is fitted using the least squares method;

[0032] B4): Use the obtained polynomial coefficients to calculate the filter value of the center point of the window, and then repeat B2) and B4) until all points in the signal are processed;

[0033] B5): Obtain smoothed load data, in which the load periodicity component is retained and the noise is reduced.

[0034] Preferably, in S7, the specific steps of the SWARA method and the CRITIC method are as follows:

[0035] C1): First, define several load response potential influencing factor indicators, and the indicator set is expressed as:

[0036] C j =(C1,C2…C n ),J=1,2…,n

[0037] If the number of electricity users to be evaluated is m, the set of alternative solutions can be expressed as:

[0038] A i =(A1,A2,…,A m ),i=1,2,…,m;

[0039] The above alternatives are evaluated by the decision set, which is expressed as:

[0040] D k =(D1,D2,…,D k ),k=1,2,…,K

[0041] C2): Determine the subjective weight and objective weight of the influencing factor indicators in the load response potential assessment in a fuzzy environment, and then determine the comprehensive weight.

[0042] Preferably, the specific steps of obtaining the subjective weight are as follows:

[0043] The importance of each indicator is obtained by evaluating the decision set according to the semantic variables. The semantic variables include equally important, relatively less important, less important, unimportant, and extremely unimportant. Equally important is represented by EI, relatively less important is represented by MI, less important is represented by LI, unimportant is represented by VLI, and extremely unimportant is represented by MLI.

[0044] The fuzzy scale of EI is (1, 1, 1), the fuzzy scale of MI is (2 / 3, 4 / 5, 1), the fuzzy scale of LI is (2 / 5, 1 / 2, 2 / 3), the fuzzy scale of VLI is (2 / 7, 1 / 3, 2 / 5), and the fuzzy scale of MLI is (2 / 9, 1 / 4, 2 / 7). The specific steps for obtaining the subjective weights are as follows:

[0045] C21): The importance of indicators is evaluated through semantic variables. The semantic variables evaluated by decision set k on indicator j are expressed as The corresponding triangular fuzzy scale is

[0046] C22): Get the evaluation result matrix of K decision makers for indicator weights

[0047]

[0048] Calculate the average triangular fuzzy number for each index j:

[0049]

[0050] C23): Determined by formula (3) The clarity value And according to Sort the indicators in descending order;

[0051]

[0052] C24): Starting from the second ranked index, calculate the difference C between index j and the previous index j-1 j , to determine the relative importance of each indicator score;

[0053]

[0054] C25): Relative coefficient r of the evaluation index j :

[0055]

[0056] C26): Calculate the modified weight value π of the indicator j :

[0057]

[0058] C27): Calculate the subjective weight α of each influencing factor.

[0059]

[0060] Preferably, the specific steps of obtaining the objective weight are as follows:

[0061] C28): Form the same indicator weight evaluation result matrix as formula (1)

[0062] C29): Setting up a standardized evaluation matrix The indicators include: benefit-type indicator B and cost-type indicator C, which represent that the evaluation data of the indicator is positively correlated and negatively correlated with the load response potential, respectively;

[0063] when When it is benefit indicator B:

[0064]

[0065] when When it is a cost indicator C:

[0066]

[0067] Where, k = 1, 2, ..., K, j = 1, 2, ..., n, is the value of the evaluation matrix after normalization in formula (1)

[0068] C30): Defuzzify the triangular fuzziness of each element in the normalized matrix and calculate its triangular fuzzy clarity value, as shown in equations (10)-(12):

[0069]

[0070] Where: and are the standardized left and right endpoint values, is the total clarity value of each element in the standardized matrix;

[0071] C31): The standard deviation is used to represent the contrast between indicators, and the correlation coefficient between indicators is used to quantify the conflict. The higher the conflict, the lower the correlation between indicators. The standard deviation σ between each indicator j And the correlation coefficient ρ ij Calculate through equations (13)-(14);

[0072]

[0073] Where: jis the standard deviation of the jth indicator, represents the mean value of the jth indicator in the standardized evaluation matrix, ρ ij j is the correlation coefficient between the i-th indicator and the j-th indicator, P i D and are the i-th and j-th columns of the standardized evaluation matrix, respectively;

[0074] C32): Calculate the objective weight β i , the information content E contained in each indicator is obtained by the correlation coefficient and standard deviation j :

[0075]

[0076] When the amount of information contained in an indicator E j The larger the value is, the greater its weight is. The objective weight of the indicator β i It is expressed as the proportion of each indicator's information to the total information.

[0077]

[0078] Preferably, the specific steps of determining the comprehensive weight are as follows:

[0079] C33): After obtaining the subjective weight and objective weight of each indicator, this paper uses a linear weighting method to combine the two weights to determine the comprehensive weight ω j :

[0080] ω j =ξα j +(1-ξ)β j Formula (17)

[0081] Among them: 0<ζ<1, α j represents the subjective weight of indicator j obtained by the SWARA method, β j Represents the objective weight of indicator j obtained by the CRITIC method

[0082] Compared with the prior art, the present invention has the following beneficial effects:

[0083] 1. In the present invention, through S1-S7, the response potential of different loads can be effectively evaluated and ranked, providing a scientific basis for the management and regulation of the power system;

[0084] 2. In the present invention, the subjective weight and the objective weight are obtained respectively through the SWARA method and the CRITIC method, and then the comprehensive weight is obtained, so as to ensure that the obtained comprehensive weight is close to the subjective weight and the objective weight, and improve the error rate of the comprehensive weight, so that the response potential ranking and evaluation results of various loads are more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0086] Figure 1 The present invention is a flow chart of a response potential model evaluation method based on building a load cluster. DETAILED DESCRIPTION

[0087] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0088] Example: Figure 1 As shown, the present invention provides a response potential model evaluation method based on building a load cluster, comprising the following steps:

[0089] S1. Obtain load measurement data

[0090] First, collect and analyze the load measurement data of power users, including historical load data and power consumption behavior;

[0091] S2. Build load decomposition model

[0092] The load decomposition model of the dedicated transformer user is constructed by the STL algorithm, and the load is decomposed into periodic components to better understand the periodic changes of the load. The specific steps of the STL algorithm to construct the load decomposition model of the dedicated transformer user are as follows:

[0093] A1): First, collect the load data of the dedicated transformer user, wherein the load data includes the load value of the time series;

[0094] A2): Set the parameters of the STL algorithm, and then gradually decompose the load data into trend components, seasonal components and residual components through multiple iterations, and then output the decomposed trend components, seasonal components and residual components for subsequent analysis;

[0095] A3): Construct a load decomposition model for dedicated transformer users, wherein the load decomposition model includes a trend prediction model, a seasonal prediction model and residual processing.

[0096] S3. Determine the load curve platform power

[0097] The load periodicity component is processed by the SG filtering algorithm to determine the power that can represent the load curve platform. Each local minimum point can represent the power of the load curve platform. The specific steps of the SG filtering algorithm to process the load periodicity component are as follows:

[0098] B1): Determine the filter parameters of the SG filter algorithm, the number of neighboring points used to fit the polynomial at each location, so as to produce a smoother output;

[0099] B2): For each point in the signal, construct a data window with the specified window length;

[0100] B3): In each data window, a polynomial of a specified order is fitted using the least squares method;

[0101] B4): Use the obtained polynomial coefficients to calculate the filter value of the center point of the window, and then repeat B2) and B4) until all points in the signal are processed;

[0102] B5): Obtain smoothed load data, in which the load periodicity component is retained and the noise is reduced.

[0103] S4. Calculation of demand response potential

[0104] According to the specified demand response start time, the real-time load power of the load periodic component at the response start time is calculated using the dedicated transformer user load decomposition model and the load curve platform power determination model. The demand response potential power is determined by comparing the real-time load power with the platform power.

[0105] S5. Constructing user portrait feature indicators

[0106] Based on the theoretical response potential and historical data of power user load, the user portrait characteristic index is constructed to characterize the power consumption characteristics of power users;

[0107] S6. Determine demand response potential

[0108] Based on the user portrait characteristic indicators, determine the demand response potential of power users, and determine the regional adjustable power according to the demand response potential;

[0109] S7. Evaluation and ranking

[0110] The SWARA method and CRITIC method are used to calculate the comprehensive weight of each indicator, and then the response potential of each type of load is evaluated and ranked;

[0111] The specific steps of the SWARA method and the CRITIC method are as follows:

[0112] C1): First, define several load response potential influencing factor indicators, and the indicator set is expressed as:

[0113] C j =(C1,C2…C n ),J=1,2…,n

[0114] If the number of electricity users to be evaluated is m, the set of alternative solutions can be expressed as:

[0115] A i =(A1,A2,…,A m ),i=1,2,…,m;

[0116] The above alternatives are evaluated by the decision set, which is expressed as:

[0117] D k =(D1,D2,…,D k ),k=1,2,…,K

[0118] C2): Determine the subjective weight and objective weight of the influencing factor indicators in the load response potential assessment in a fuzzy environment, and then determine the comprehensive weight.

[0119] The specific steps of obtaining the subjective weight are as follows:

[0120] The importance of each indicator is obtained by evaluating the decision set according to the semantic variables. The semantic variables include equally important, relatively less important, less important, unimportant, and extremely unimportant. Equally important is represented by EI, relatively less important is represented by MI, less important is represented by LI, unimportant is represented by VLI, and extremely unimportant is represented by MLI.

[0121] The fuzzy scale of EI is (1, 1, 1), the fuzzy scale of MI is (2 / 3, 4 / 5, 1), the fuzzy scale of LI is (2 / 5, 1 / 2, 2 / 3), the fuzzy scale of VLI is (2 / 7, 1 / 3, 2 / 5), and the fuzzy scale of MLI is (2 / 9, 1 / 4, 2 / 7). The specific steps for obtaining the subjective weights are as follows:

[0122] C21): The importance of indicators is evaluated through semantic variables. The semantic variables evaluated by decision set k on indicator j are expressed as The corresponding triangular fuzzy scale is

[0123] C22): Get the evaluation result matrix of K decision makers for indicator weights

[0124]

[0125] Calculate the average triangular fuzzy number for each index j:

[0126]

[0127] C23): Determined by formula (3) The clarity value And according to Sort the indicators in descending order;

[0128]

[0129] C24): Starting from the second ranked index, calculate the difference C between index j and the previous index j-1 j , to determine the relative importance of each indicator score;

[0130]

[0131] C25): Relative coefficient r of the evaluation index j :

[0132]

[0133] C26): Calculate the modified weight value π of the indicator j :

[0134]

[0135] C27): Calculate the subjective weight α of each influencing factor.

[0136]

[0137] The specific steps of obtaining the objective weight are as follows:

[0138] C28): Form the same indicator weight evaluation result matrix as formula (1)

[0139] C29): Setting up a standardized evaluation matrix The indicators include: benefit-type indicator B and cost-type indicator C, which represent that the evaluation data of the indicator is positively correlated and negatively correlated with the load response potential, respectively;

[0140] when When it is benefit indicator B:

[0141]

[0142] when When it is a cost indicator C:

[0143]

[0144] Where, k = 1, 2, ..., K, j = 1, 2, ..., n, is the value of the evaluation matrix after normalization in formula (1)

[0145] C30): Defuzzify the triangular fuzziness of each element in the normalized matrix and calculate its triangular fuzzy clarity value, as shown in equations (10)-(12):

[0146]

[0147] Where: and are the standardized left and right endpoint values, is the total clarity value of each element in the standardized matrix;

[0148] C31): The standard deviation is used to represent the contrast between indicators, and the correlation coefficient between indicators is used to quantify the conflict. The higher the conflict, the lower the correlation between indicators. The standard deviation σ between each indicator j And the correlation coefficient ρ ij Calculate through equations (13)-(14);

[0149]

[0150] Where: j is the standard deviation of the jth indicator, represents the mean value of the jth indicator in the standardized evaluation matrix, ρ ij j is the correlation coefficient between the i-th indicator and the j-th indicator, P i D and are the i-th and j-th columns of the standardized evaluation matrix, respectively;

[0151] C32): Calculate the objective weight β i , the information content E contained in each indicator is obtained by the correlation coefficient and standard deviation j :

[0152]

[0153] When the amount of information contained in an indicator E j The larger the value is, the greater its weight is. The objective weight of the indicator β i It is expressed as the proportion of each indicator's information to the total information.

[0154]

[0155] The specific steps of determining the comprehensive weight are as follows:

[0156] C33): After obtaining the subjective weight and objective weight of each indicator, this paper uses a linear weighting method to combine the two weights to determine the comprehensive weight ω j :

[0157] ω j =ξα j +(1-ξ)β j Formula (17)

[0158] Among them: 0<ζ<1, α j represents the subjective weight of indicator j obtained by the SWARA method, β j Represents the objective weight of indicator j obtained by the CRITIC method.

[0159] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A response potential model evaluation method based on building load clusters, characterized in that: The following steps are involved: S1. Obtain load measurement data First, collect and analyze the load measurement data of power users, including historical load data and power consumption behavior; S2. Build load decomposition model The load decomposition model of the dedicated transformer user is constructed through the STL algorithm, and the load is decomposed into periodic components to better understand the periodic changes of the load; S3. Determine the load curve platform power The load periodic component is processed by the SG filtering algorithm to determine the power that can represent the load curve platform. Each local minimum point can represent the power of the load curve platform. S4. Calculate demand response potential According to the specified demand response start time, the real-time load power of the load periodic component at the response start time is calculated using the dedicated transformer user load decomposition model and the load curve platform power determination model. The demand response potential power is determined by comparing the real-time load power with the platform power. S5. Constructing user portrait feature indicators Based on the theoretical response potential and historical data of power user load, the user portrait characteristic index is constructed to characterize the power consumption characteristics of power users; S6. Determine demand response potential Based on the user portrait characteristic indicators, determine the demand response potential of power users, and determine the regional adjustable power according to the demand response potential; S7. Evaluation and ranking The SWARA method and CRITIC method are used to calculate the comprehensive weight of each indicator, and then the response potential of each type of load is evaluated and ranked.

2. A response potential model evaluation method based on building a load cluster as claimed in claim 1, characterized in that: In S2, the specific steps of the STL algorithm to construct the load decomposition model of the dedicated transformer user are as follows: A1): First, collect the load data of the dedicated transformer user, wherein the load data includes the load value of the time series; A2): Set the parameters of the STL algorithm, and then gradually decompose the load data into trend components, seasonal components and residual components through multiple iterations, and then output the decomposed trend components, seasonal components and residual components for subsequent analysis; A3): Construct a load decomposition model for dedicated transformer users, wherein the load decomposition model includes a trend prediction model, a seasonal prediction model and residual processing.

3. A response potential model evaluation method based on building a load cluster as claimed in claim 1, characterized in that: In S3, the specific steps of the SG filtering algorithm for processing the periodic component of the load are as follows: B1): Determine the filter parameters of the SG filter algorithm, the number of neighboring points used to fit the polynomial at each location, so as to produce a smoother output; B2): For each point in the signal, construct a data window with the specified window length; B3): In each data window, a polynomial of a specified order is fitted using the least squares method; B4): Use the obtained polynomial coefficients to calculate the filter value of the center point of the window, and then repeat B2) and B4) until all points in the signal are processed; B5): Obtain smoothed load data, in which the load periodicity component is retained and the noise is reduced.

4. A response potential model evaluation method based on building a load cluster as claimed in claim 1, characterized in that: In S7, the specific steps of the SWARA method and the CRITIC method are as follows: C1): First, define several load response potential influencing factor indicators, and the indicator set is expressed as: C j =(C1,C2…C n ),J=1,2…,n If the number of electricity users to be evaluated is m, the set of alternative solutions can be expressed as: A i =(A1,A2,…,A m ),i=1,2,…,m; The above alternatives are evaluated by the decision set, which is expressed as: D k =(D1,D2,…,D k ),k=1,2,…,K C2): Determine the subjective weight and objective weight of the influencing factor indicators in the load response potential assessment in a fuzzy environment, and then determine the comprehensive weight.

5. A response potential model evaluation method based on building a load cluster as claimed in claim 4, characterized in that: The specific steps of obtaining the subjective weight are as follows: The importance of each indicator is obtained by evaluating the decision set according to the semantic variables. The semantic variables include equally important, relatively less important, less important, unimportant, and extremely unimportant. Equally important is represented by EI, relatively less important is represented by MI, less important is represented by LI, unimportant is represented by VLI, and extremely unimportant is represented by MLI. The fuzzy scale of EI is (1, 1, 1), the fuzzy scale of MI is (23, 45, 1), the fuzzy scale of LI is (25, 12, 23), the fuzzy scale of VLI is (27, 13, 25), and the fuzzy scale of MLI is (29, 14, 27). The specific steps to obtain the subjective weight are as follows: C21): The importance of indicators is evaluated through semantic variables. The semantic variables evaluated by decision set k on indicator j are expressed as The corresponding triangular fuzzy scale is C22): Get the evaluation result matrix of K decision makers for indicator weights Calculate the average triangular fuzzy number for each index j: C23): Determined by formula (3) The clarity value And according to Sort the indicators in descending order; C24): Calculate the difference C between index j and the previous index j-1 starting from the second ranked index j , to determine the relative importance of each indicator score; C25): Relative coefficient r of the evaluation index j : C26): Calculate the modified weight value π of the indicator j : C27): Calculate the subjective weight α of each influencing factor.

6. A response potential model evaluation method based on building a load cluster as claimed in claim 5, characterized in that: The specific steps of obtaining the objective weight are as follows: C28): Form the same indicator weight evaluation result matrix as formula (1) C29): Setting up a standardized evaluation matrix The indicators include: benefit-type indicator B and cost-type indicator C, which represent that the evaluation data of the indicator is positively correlated and negatively correlated with the load response potential, respectively; when When it is benefit indicator B: when When it is a cost indicator C: Where, k = 1, 2, ..., K, j = 1, 2, ..., n, is the value of the evaluation matrix after normalization in formula (1) C30): Defuzzify the triangular fuzziness of each element in the normalized matrix and calculate its triangular fuzzy clarity value, as shown in equations (10)-(12): Where: and are the standardized left and right endpoint values, is the total clarity value of each element in the standardized matrix; C31): The standard deviation is used to represent the contrast between indicators, and the correlation coefficient between indicators is used to quantify the conflict. The higher the conflict, the lower the correlation between indicators. The standard deviation σ between each indicator j And the correlation coefficient ρ ij Calculate through equations (13)-(14); Where: j is the standard deviation of the jth indicator, represents the mean value of the jth indicator in the standardized evaluation matrix, ρ ij j is the correlation coefficient between the i-th indicator and the j-th indicator, P i D and P j D are the i-th and j-th columns of the standardized evaluation matrix, respectively; C32): Calculate the objective weight β i , the information content E contained in each indicator is obtained by the correlation coefficient and standard deviation j : When the amount of information contained in an indicator E j The larger the value is, the greater its weight is. The objective weight of the indicator β i It is expressed as the proportion of each indicator's information to the total information.

7. A response potential model evaluation method based on building a load cluster as claimed in claim 6, characterized in that: The specific steps of determining the comprehensive weight are as follows: C33): After obtaining the subjective weight and objective weight of each indicator, this paper uses a linear weighting method to combine the two weights to determine the comprehensive weight ω j : oh j =now j +(1-ξ)β j formula(17) Among them: 0<ζ<1, α j represents the subjective weight of indicator j obtained by the SWARA method, β j Represents the objective weight of indicator j obtained by the CRITIC method.

Images