CN118157104A - Comprehensive performance evaluation method for power grid rolling scheduling strategy - Google Patents

Abstract

The invention relates to the technical field of generalization evaluation of power grid active cruising scheduling, in particular to a comprehensive performance evaluation method for a power grid rolling scheduling strategy. The method carries out active cruising scheduling on a scheduling strategy under different source load scenes by a parallel computing and evaluating method, and computes each sub-index of a scheduling result; and finally, calculating the weight of each index by using an AHP-inverse entropy weight model, and comprehensively grading the scheduling results in different scenes by using a fuzzy evaluation method.

Description

Comprehensive performance evaluation method for power grid rolling scheduling strategy

Technical Field

The invention relates to the technical field of generalization evaluation of active cruising scheduling of a power grid, in particular to a comprehensive performance evaluation method for a rolling scheduling strategy of the power grid, which is used for evaluating decision performance of the scheduling strategy in various complex source load scenes and partial fault scenes.

Background

Along with the gradual promotion of the construction of novel electric power systems in China, the number of main bodies involved in the optimized dispatching of the electric power systems is increased, the fluctuation of power supplies is enhanced, and the difficulty of the prospective dispatching of the power grid is greatly improved. To address this challenge, various data-driven based or partially data-driven intelligent algorithms are applied to grid look-ahead scheduling aid decisions to promote the automation and intelligence level of grid look-ahead scheduling in multiple operating scenarios. Under the background, aiming at the characteristic of strong adaptability of the prospective scheduling intelligent agent to multiple scenes, the decision effect of the intelligent agent is comprehensively evaluated, so that the construction of the prospective scheduling intelligent algorithm is guided, and the method has practical significance.

In the field of power grid economic dispatch evaluation, the existing research mainly uses the economic dispatch agent design as a link of the economic dispatch agent design, and is used for comparing with the traditional dispatch method to verify the effectiveness of the agent design, the whole flow is simpler, and the research specially aiming at the economic dispatch evaluation is less. The system is developed for daily power generation dispatching planning, a power grid economic dispatching pre-evaluation system is provided, a basic data acquisition subsystem, a standardized reference building subsystem and a deviation rate calculation subsystem are constructed, and the automation level of power grid economic dispatching pre-evaluation is improved. The method is researched and oriented to various data driving methods, and the intelligent algorithm for power grid optimization scheduling is evaluated from the dimensions of decision optimality, computational complexity and the like, so that the computational efficiency advantage of the data driving method is fully embodied. The influence of bad data in the power grid look-ahead scheduling is researched, and a sensitivity matrix of the look-ahead scheduling on the bad data is provided so as to rapidly evaluate the influence of data loss or errors on the power grid look-ahead scheduling decision when a line is blocked. However, the research is not performed on the quality of the scheduling policy decision under a large number of different source load and fault scenes, so that the scheduling policy is evaluated according to the performance of a plurality of source load scenes with different characteristics.

Disclosure of Invention

Therefore, it is necessary to provide a parallel computing and evaluating method for performing active cruising scheduling on a scheduling strategy under different source load scenes, and computing each sub-index of a scheduling result; and finally, calculating the weight of each index by using an AHP-inverse entropy weight model, and comprehensively grading the scheduling results in different scenes by using a fuzzy evaluation method.

In order to achieve the above purpose, the present invention provides the following technical solutions:

a comprehensive performance evaluation method for a power grid rolling scheduling strategy comprises the following steps:

S1, selecting a plurality of new energy day-ahead output curves and load prediction curves, multiplying the curves by Gaussian distribution sampling curves with different mean values and variances, and generating a plurality of groups of source load curve scenes with different deviation degrees; the scheduling strategy is subjected to parallel decision calculation under the scenes, and a daily rolling scheduling result obtained by scheduling calculation is obtained;

S2, constructing generalized sub-indexes comprising safety, economy and cleanliness, calculating sub-indexes under different scenes, and finally carrying out normalization treatment and extremely-large treatment on the indexes;

s3, establishing a fuzzy judgment matrix F by adopting an improved fuzzy AHP analytic hierarchy process, and calculating the weight of each index by using the AHP method to obtain subjective weight

S4, calculating the weight of each index by adopting an inverse entropy weight model to obtain objective weight

S5, subjective weight calculated according to the steps S3 and S4And objective weight/>Calculating a variable weight coefficient omega _i of the equalization function, and forming an index weight matrix omega;

S6, calculating membership functions of the generalization indexes by using the normalized result in the step S2;

and S7, carrying out quantitative evaluation on the scheduling results of different source load scenes according to membership functions of the generalization indexes, and obtaining scheduling performance results under different source load scenes.

According to further optimization of the technical scheme, the step S1 specifically comprises the following steps:

S1.1, taking a group of new energy output curves PG _renewable, wherein PG _i is the output value of the new energy at the ith moment, and taking one-day output curves at 15 minutes intervals, so i takes 96 as shown in the following formula:

PG_renewable＝[pg₁,pg₂,...,pg_i],i＝96

Taking a set of load prediction curves P _L, wherein pl _i is a predicted value of the load at the ith moment, and the formula is as follows:

P_L＝[pl₁,pl₂,...,pl_i],i＝96

And generating N sets of bias curves a= { a ₁,a₂,...,a_i,...a_N}^T by using a gaussian distribution model, wherein a _i is a set of matrices which contain 96 bias values and conform to gaussian distribution, N _j is a bias value, and the formula is as follows:

a_i＝[n₁,n₂,...n_j],n_j～Normal(μ_j,σ_j ²),j∈[1,96]

Finally, generating N groups of source load scene curves PG _renewable,P_L with different deviation degrees, wherein:

P_L′＝A_N×96·P_L96×1;PG_renewable′＝A_N×96·PG_{renewable,96×1}

S1.2, respectively loading the source load curves into different threads, and then loading scheduling strategies to be evaluated into the threads in parallel; and finally, carrying out 96-point parallel calculation to obtain an active cruising path result.

The technical scheme is further optimized, and the step S2 specifically refers to: the calculation method for giving the index by comprehensively considering various factors such as system cost, section power flow out-of-limit condition, load balance, new energy consumption rate, balance machine out-of-limit power and the like comprises the following steps:

S2.1, system cost:

wherein P _i is the decision output of the corresponding unit i at a certain moment, C _i is the unit running cost of the corresponding unit, N units are in total in the system, C _max is a normalization constant, the lower the system cost is, the higher the score is, and the upper limit is 100 score;

S2.2, section tide out-of-limit:

Wherein S is the total section number, r ^s is the rewarding value of the S-th section, and the calculation method of r ^s is as follows:

Wherein P _s is the tide of section s, P _s ^max is the upper limit of section s, and P _s ^min is the lower limit of section s;

s2.3, load balancing:

Wherein L represents the total system real load at the moment t, P _i is the output of the unit i, N is the total number of the units, and the reward is 0 when the load is balanced;

s2.4, new energy consumption rate:

Wherein, P _i represents the output of the current new energy, P _i ^max represents the predicted maximum output of the current new energy, and M new energy units are all arranged.

The technical scheme is further optimized, and the step S3 specifically comprises the following steps:

S3.1, constructing a fuzzy complementary judgment matrix F, namely constructing F _m×n by a 1-9 scale method, wherein the definition of the 1-9 scale method is shown in the following table:

s3.2, calculating a weight formula by using the fuzzy complementary matrix, wherein the weight formula is as follows:

Wherein F _ij represents an element in the fuzzy complementary determining matrix F _m×n, and W _i represents a weight of the i-th index;

s3.3, carrying out consistency test through fuzzy complementary judgment matrix F:

the matrix F is first summed row by row as follows:

Then transformed by the result of the above equation: Finally obtaining a fuzzy consistency matrix R

Wherein, the relation of each element in R needs to satisfy:

0≤r_ij≤1,r_ii＝0.5,r_ij+r_ji＝1

And is also provided with There is r _ij＝r_ik-r_jk +0.5.

The technical scheme is further optimized, and the step S4 specifically comprises the following steps:

S4.1, forward processing of evaluation indexes: converting the minimum index, the intermediate index and the interval index into the maximum index; the formula is as follows:

Very small- > very large: x _i＝max{x₁,x₂,...,x_i}-x_i, if all elements are positive, the reciprocal can be directly taken

Intermediate- > very large: m=max { |x _i-x_best | },

M＝max{a-min{x_i},max{x_i}-b}

Regional- > very large:

S4.2, a normalization process, in which n evaluation objects, m indexes, i.e., x=x _n×m, are normalized as follows:

s4.3, calculating the proportion of each element, wherein the sum of the proportion is 1, namely normalizing the standardized matrix, and calculating a probability matrix p, wherein the formula is as follows:

S4.4, calculating the information entropy e of each index

S4.5, calculating the weight, wherein 1-e is the useful value of the information, normalizing to obtain the weight,

The technical scheme is further optimized, and the step S5 specifically comprises the following steps:

S5.1, calculating the fixed weight

S5.2, calculating an evaluation value v _i:

Wherein, The constant weight coefficient of the jth key index in the ith item is given, and x _ij is the evaluation value of the jth key index in the ith item;

S5.3, variable weight coefficient

S5.4 equalization function

S5.5, the variable weight coefficient of the equalization function is as follows:

Wherein a is an equalization factor.

The technical scheme is further optimized, and the step S6 specifically comprises the following steps:

Calculating membership functions of all indexes by using the result x _i obtained by normalizing and summing the step S2 to be the maximum, wherein the fuzzy relation of all indexes and state evaluation indexes is reflected by adopting a normal membership function A (x _i) according to the characteristic decision of the active cruise generalization evaluation indexes of the scheduling strategy

Where a represents the boundary of the membership function and σ represents the standard deviation of the normal curve.

The technical scheme is further optimized, and the step S7 specifically comprises the following steps:

S7.1, calculating a single factor evaluation matrix R through the membership function in the step S6:

Wherein m is the number of generalization indexes;

S7.2, establishing a factor set of comprehensive evaluation: u= (U ₁,u₂,...,u_m) where element U ₁ represents the ith factor affecting the evaluation object, an evaluation set of comprehensive evaluation is established: v= (V ₁,v₂,...,v_m) where element V ₁ represents the j-th evaluation result;

s7.3, building a comprehensive evaluation model:

B＝ΩR＝(ω₁,ω₂,…ω_m)R＝(b₁,b₂,b₃,b₄)

Wherein the method comprises the steps of

And S7.4, according to the maximum membership degree principle and the evaluation set V, obtaining the evaluation results of the scheduling strategy under different source load scenes through the evaluation model B.

Compared with the prior art, the technical scheme has the beneficial effects that: firstly, constructing a multi-operation source load scene set based on actual data, and carrying out multi-scene active scheduling on a scheduling strategy in parallel; secondly, the invention adopts a subjective and objective comprehensive weighting method, which is more scientific compared with other weighting design methods; third, the improved fuzzy comprehensive evaluation method of the invention uses a normal membership function to reflect the fuzzy relation between each sub-index and the state evaluation.

Drawings

FIG. 1 is a flow chart of parallel decision making of a scheduling strategy in multiple scenarios;

FIG. 2 is a graph of a scheduling policy generalization evaluation index system;

FIG. 3 is a flow chart of a scheduling policy state evaluation method.

Detailed Description

In order to describe the technical content, constructional features, achieved objects and effects of the technical solution in detail, the following description is made in connection with the specific embodiments in conjunction with the accompanying drawings.

As shown in fig. 1, a comprehensive performance evaluation method for a rolling scheduling strategy of a power grid includes the following sequential steps:

S1, selecting a plurality of new energy day-ahead output curves and load prediction curves, multiplying the curves by Gaussian distribution sampling curves with different mean values and variances, and generating a plurality of groups of source load curve scenes with different deviation degrees; and the scheduling strategies to be evaluated are calculated in parallel under the scenes, and a calculated daily rolling scheduling result is obtained.

The step S1 specifically comprises the following steps:

S1.1, a group of new energy output curves PG _renewable are taken, wherein PG _i is the output value of the new energy at the ith moment, and the output curves of one day are taken at intervals of 15 minutes. So i takes 96, as shown in the following formula:

PG_renewable＝[pg₁,pg₂,…,pg_i]，i＝96

Taking a set of load prediction curves P _L, wherein pl _i is a predicted value of the load at the ith moment, and the formula is as follows:

P_L＝[pl₁,pl₂,...,pl_i]，i＝96

And generating N sets of bias curves a= { a ₁,a₂,...,a_i,...a_N}^T by using a gaussian distribution model, wherein a _i is a set of matrices which contain 96 bias values and conform to gaussian distribution, and N _j is a bias value. The formula is as follows:

a_i＝[n₁,n₂,…n_j],n_j～Normal(μ_j,σ_j ²),j∈[1,96]

And finally generating N groups of source load scene curves PG _renewable,P_L with different deviation degrees. Wherein:

S1.2, respectively loading the source load curves into different threads, and then loading scheduling strategies to be evaluated into the threads in parallel; and finally, carrying out 96-point parallel calculation to obtain an active cruising path result.

S2, constructing generalized sub-indexes including safety, economy and cleanliness, and calculating sub-indexes under different scenes. And finally, carrying out normalization treatment and huge treatment on the indexes.

The step S2 specifically refers to: the invention comprehensively considers the system cost, the cross-section power flow out-of-limit condition, the load balance, the new energy consumption rate, the out-of-limit power of the balancing machine and other factors to give out the calculation method of the index.

S2.1, system Cost _th (positive rewards):

where P _i is the decision output of the corresponding unit i at a certain moment, C _i is the unit running cost of the corresponding unit, and the system has N units in total. C _max is the normalization constant. The lower the system cost, the higher the score, with an upper limit of 100 points.

S2.2, section power flow out-of-limit Cost _pf (positive rewards):

wherein S is the total section number, and r ^s is the prize value of the S-th section. The calculation method of r ^s is as follows:

Wherein P _s is the power flow of section s, Is the upper limit of section s,/>Is the lower limit of the section s.

S2.3, load balancing Cost _bal (positive rewards):

Wherein L represents the total system real load at time t, P _i is the output of the unit i, and N is the total number of units. The prize is 0 when the load is balanced.

S2.4, new energy consumption rate (positive rewards):

Wherein, P _i represents the output of the current new energy, P _i ^max represents the predicted maximum output of the current new energy, and M new energy units are all arranged.

S3, establishing a fuzzy judgment matrix F by adopting an improved fuzzy AHP analytic hierarchy process, and calculating the weight of each index by using the AHP method to obtain subjective weight

The step S3 specifically comprises the following steps:

S3.1, constructing a fuzzy complementary judgment matrix F, namely constructing F _m×n by a 1-9 scale method, wherein the definition of the 1-9 scale method is shown in the following table:

s3.2, calculating a weight formula by using the fuzzy complementary matrix, wherein the weight formula is as follows:

where F _ij represents an element in the fuzzy complementary determining matrix F _m×n, and W _i represents a weight of the i-th index.

S3.3, carrying out consistency test through fuzzy complementary judgment matrix F:

the matrix F is first summed row by row as follows:

Then transformed by the result of the above equation: Finally obtaining a fuzzy consistency matrix R

Wherein, the relation of each element in R needs to satisfy:

0≤r_ij≤1,r_ii＝0.5,r_ij+r_ji＝1

And is also provided with With r _ij＝r_ik-r_jk +0.5

S4, calculating the weight of each index by adopting an inverse entropy weight model to obtain objective weight

The step S4 specifically includes the following steps:

S4.1, forward processing of evaluation indexes: converting the minimum index, the intermediate index and the interval index into the maximum index; the formula is as follows:

very small- > very large: x _i＝max{x₁,x₂,...,x_i}-x_i. If all elements are positive numbers, the reciprocal can be directly taken

Intermediate- > very large: m=max { |x _i-x_best | },

M＝max{a-min{x_i},max{x_i}-b}

Regional- > very large:

S4.2, normalization (balancing because of differences between indices or differences in dimensions), where n evaluation objects, m indices, i.e. x=x _n×m. The normalization formula is as follows:

S4.3, calculating the proportion of each element (the sum of the proportion is 1, namely, the normalized matrix is normalized), and calculating the probability matrix p, wherein the formula is as follows:

s4.4, calculating the information entropy e (uncertainty) of each index

S4.5, calculating the weight (1-e is the useful value of the information, and normalizing to obtain the weight)

S5 subjective weight calculated according to the steps S3 and S4And objective weight/>And calculating a variable weight coefficient omega _i of the equalization function, and forming an index weight matrix omega.

The step S5 specifically includes the following steps:

S5.1, calculating the fixed weight

S5.2, calculating an evaluation value v _i:

Wherein, And x _ij is the evaluation value of the jth key index in the ith item.

S5.3, variable weight coefficient

S5.4 equalization function

S5.5, the variable weight coefficient of the equalization function is as follows:

Wherein a is an equalization factor.

S6, calculating membership functions of the generalization indexes by using the normalized result in the step S2.

The step S6 specifically includes the following steps:

S6.1, calculating membership functions of all indexes by utilizing the result x _i obtained by normalizing and summing to be maximum in the step S2, wherein the fuzzy relation of all indexes and state evaluation indexes is reflected by adopting a normal membership function A (x _i) according to the characteristic decision of the scheduling strategy active cruise generalization evaluation index

Where a represents the boundary of the membership function and σ represents the standard deviation of the normal curve.

S7, carrying out quantitative evaluation on the scheduling results of different source load scenes according to membership functions of various generalization indexes, and obtaining scheduling performance results under different source load scenes;

The step S7 specifically includes the following steps:

S7.1, calculating a single factor evaluation matrix R through the membership function in the step S6:

Wherein m is the number of generalization indexes.

S7.2, establishing a factor set of comprehensive evaluation: u= (U ₁,u₂,...,u_m), where element U ₁ represents the ith factor affecting the evaluation object. Establishing an evaluation set of comprehensive evaluation: v= (V ₁,v₂,...,v_m) where element V ₁ represents the j-th evaluation result.

S7.3, building a comprehensive evaluation model:

B＝Ω·R＝(ω₁,ω₂,…ω_m)·R＝(b₁,b₂,b₃,b₄)

Wherein the method comprises the steps of

And S7.4, according to the maximum membership degree principle and the evaluation set V, obtaining the evaluation results of the scheduling strategy under different source load scenes through the evaluation model B.

According to the invention, a plurality of operation scenes of new energy and load prediction fluctuation are constructed according to an actual data curve, and a scheduling strategy is operated in a plurality of scenes in parallel by using a computer multithreading technology. Active scheduling results under a plurality of scenes are obtained; meanwhile, the weight of each sub-target of the scheduling strategy in the active scheduling cruising of the power grid is calculated by combining the AHP-inverse entropy weight model and an improved fuzzy analytic hierarchy process, and compared with the traditional analytic hierarchy process, the method is more accurate and scientific.

It is noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or terminal. Without further limitation, an element defined by the statement "comprising … …" or "comprising … …" does not exclude the presence of additional elements in a process, method, article, or terminal device that includes the element. Further, herein, "greater than," "less than," "exceeding," and the like are understood to not include the present number; "above", "below", "within" and the like are understood to include this number.

While the embodiments have been described above, other variations and modifications will occur to those skilled in the art once the basic inventive concepts are known, and it is therefore intended that the foregoing description and drawings illustrate only embodiments of the invention and not limit the scope of the invention, and it is therefore intended that the invention not be limited to the specific embodiments described, but that the invention may be practiced with their equivalent structures or with their equivalent processes or with their use directly or indirectly in other related fields.

Claims (8)

1. The comprehensive performance evaluation method for the power grid rolling scheduling strategy is characterized by comprising the following steps of:

S1, selecting a plurality of new energy day-ahead output curves and load prediction curves, multiplying the curves by Gaussian distribution sampling curves with different mean values and variances, and generating a plurality of groups of source load curve scenes with different deviation degrees; the scheduling strategy is subjected to parallel decision calculation under the scenes, and a daily rolling scheduling result obtained by scheduling calculation is obtained;

S2, constructing generalized sub-indexes comprising safety, economy and cleanliness, calculating sub-indexes under different scenes, and finally carrying out normalization treatment and extremely-large treatment on the indexes;

s3, establishing a fuzzy judgment matrix F by adopting an improved fuzzy AHP analytic hierarchy process, and calculating the weight of each index by using the AHP method to obtain subjective weight

S4, calculating the weight of each index by adopting an inverse entropy weight model to obtain objective weight

S5, subjective weight calculated according to the steps S3 and S4And objective weight/>Calculating a variable weight coefficient omega _i of the equalization function, and forming an index weight matrix omega;

S6, calculating membership functions of the generalization indexes by using the normalized result in the step S2;

and S7, carrying out quantitative evaluation on the scheduling results of different source load scenes according to membership functions of the generalization indexes, and obtaining scheduling performance results under different source load scenes.

2. The comprehensive performance evaluation method for the power grid rolling scheduling strategy according to claim 1, wherein the step S1 specifically includes the following steps:

S1.1, taking a group of new energy output curves PG _renewable, wherein PG _i is the output value of the new energy at the ith moment, and taking one-day output curves at 15 minutes intervals, so i takes 96 as shown in the following formula:

PG_renewable＝[pg₁,pg₂,...,pg_i],i＝96

Taking a set of load prediction curves P _L, wherein pl _i is a predicted value of the load at the ith moment, and the formula is as follows:

P_L＝[pl₁,pl₂,...,pl_i],i＝96

And generating N sets of bias curves a= { a ₁,a₂,...,a_i,...a_N}^T by using a gaussian distribution model, wherein a _i is a set of matrices which contain 96 bias values and conform to gaussian distribution, N _j is a bias value, and the formula is as follows:

a_i＝[n₁,n₂,...n_j],n_j～Normal(μ_j,σ_j ²),j∈[1,96]

Finally, generating N groups of source load scene curves PG _renewable,P_L with different deviation degrees, wherein:

P_L′＝A_N×96·P_L96×1;PG_renewable′＝A_N×96·PG_{renewable,96×1}

S1.2, respectively loading the source load curves into different threads, and then loading scheduling strategies to be evaluated into the threads in parallel; and finally, carrying out 96-point parallel calculation to obtain an active cruising path result.

3. The comprehensive performance evaluation method for the power grid rolling scheduling strategy according to claim 1, wherein the step S2 specifically refers to: the calculation method for giving the index by comprehensively considering various factors such as system cost, section power flow out-of-limit condition, load balance, new energy consumption rate, balance machine out-of-limit power and the like comprises the following steps:

S2.1, system cost:

wherein P _i is the decision output of the corresponding unit i at a certain moment, C _i is the unit running cost of the corresponding unit, N units are in total in the system, C _max is a normalization constant, the lower the system cost is, the higher the score is, and the upper limit is 100 score;

S2.2, section tide out-of-limit:

Wherein S is the total section number, r ^s is the rewarding value of the S-th section, and the calculation method of r ^s is as follows:

Wherein P _s is the power flow of section s, Is the upper limit of section s,/>Is the lower limit of the section s;

s2.3, load balancing:

Wherein L represents the total system real load at the moment t, P _i is the output of the unit i, N is the total number of the units, and the reward is 0 when the load is balanced;

s2.4, new energy consumption rate:

Wherein, P _i represents the output of the current new energy, P _i ^max represents the predicted maximum output of the current new energy, and M new energy units are all arranged.

4. The comprehensive performance evaluation method for the power grid rolling scheduling strategy according to claim 1, wherein the step S3 specifically includes the following steps:

S3.1, constructing a fuzzy complementary judgment matrix F, namely constructing F _m×n by a 1-9 scale method, wherein the definition of the 1-9 scale method is shown in the following table:

s3.2, calculating a weight formula by using the fuzzy complementary matrix, wherein the weight formula is as follows:

Wherein F _ij represents an element in the fuzzy complementary determining matrix F _m×n, and W _i represents a weight of the i-th index;

s3.3, carrying out consistency test through fuzzy complementary judgment matrix F:

the matrix F is first summed row by row as follows:

Then transformed by the result of the above equation: Finally obtaining a fuzzy consistency matrix R

Wherein, the relation of each element in R needs to satisfy:

0≤r_ij≤1,r_ii＝0.5,r_ij+r_ji＝1

And is also provided with There is r _ij＝r_ik-r_jk +0.5.

5. The comprehensive performance evaluation method for the power grid rolling scheduling strategy according to claim 1, wherein the step S4 specifically comprises the following steps:

S4.1, forward processing of evaluation indexes: converting the minimum index, the intermediate index and the interval index into the maximum index; the formula is as follows:

Very small- > very large: x _i＝max{x₁,x₂,...,x_i}-x_i, if all elements are positive, the reciprocal can be directly taken

Intermediate- > very large:

M＝max{a-min{x_i},max{x_i}-b}

regional- > very large:

S4.2, a normalization process, in which n evaluation objects, m indexes, i.e., x=x _n×m, are normalized as follows:

s4.3, calculating the proportion of each element, wherein the sum of the proportion is 1, namely normalizing the standardized matrix, and calculating a probability matrix p, wherein the formula is as follows:

S4.4, calculating the information entropy e of each index

S4.5, calculating the weight, wherein 1-e is the useful value of the information, normalizing to obtain the weight,

6. The comprehensive performance evaluation method for the power grid rolling scheduling strategy according to claim 1, wherein the step S5 specifically comprises the following steps:

S5.1, calculating the fixed weight

S5.2, calculating an evaluation value v _i:

Wherein, The constant weight coefficient of the jth key index in the ith item is given, and x _ij is the evaluation value of the jth key index in the ith item;

S5.3, variable weight coefficient

S5.4 equalization function

S5.5, the variable weight coefficient of the equalization function is as follows:

Wherein a is an equalization factor.

7. The comprehensive performance evaluation method for the power grid rolling scheduling strategy according to claim 1, wherein the step S6 specifically includes the following steps:

Calculating membership functions of all indexes by using the result x _i obtained by normalizing and summing the step S2 to be the maximum, wherein the fuzzy relation of all indexes and state evaluation indexes is reflected by adopting a normal membership function A (x _i) according to the characteristic decision of the active cruise generalization evaluation indexes of the scheduling strategy

Where a represents the boundary of the membership function and σ represents the standard deviation of the normal curve.

8. The method for evaluating comprehensive performance of a rolling scheduling strategy for a power grid according to claim 7, wherein the step S7 specifically includes the following steps:

S7.1, calculating a single factor evaluation matrix R through the membership function in the step S6:

Wherein m is the number of generalization indexes;

S7.2, establishing a factor set of comprehensive evaluation: u= (U ₁,u₂,...,u_m) where element U ₁ represents the ith factor affecting the evaluation object, an evaluation set of comprehensive evaluation is established: v= (V ₁,v₂,...,v_m) where element V ₁ represents the j-th evaluation result;

s7.3, building a comprehensive evaluation model:

B＝ΩR＝(ω₁,ω₂,…ω_m)R＝(b₁,b₂,b₃,b₄)

Wherein the method comprises the steps of

And S7.4, according to the maximum membership degree principle and the evaluation set V, obtaining the evaluation results of the scheduling strategy under different source load scenes through the evaluation model B.