Layered optimization scheduling method and system for wind power cluster and medium

Abstract

The invention provides a layered optimization scheduling method and system for a wind power cluster and a medium, and relates to the technical field of power system optimization scheduling, and the method comprises the steps: determining evaluation indexes, obtaining the variable weight of each evaluation index, employing a membership function and a fuzzy comprehensive evaluation method, and obtaining the variable weight of each evaluation index; obtaining a membership score of each wind power plant of the wind power cluster under each evaluation index, and obtaining a comprehensive degradation score of each wind power plant; taking the ultra-short-term power predicted value of the wind power plant as the basis for judging the power change trend of the wind power plant, determining the climbing classification of the wind power plant, and determining the scheduling sequence of the wind power plant; on the premise that the power balance constraint between the wind power plants is met, the total output power of all the wind power plants is adjusted and controlled in real time according to the dispatching requirement for power increasing or power reducing of the wind power cluster and based on the dispatching sequence of the wind power plants of all the wind power plant groups. According to the method, the time-varying property and uncertainty of the wind power plant are fully considered, and the fatigue and fluctuation conditions of the whole wind power cluster are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of power system optimization and dispatching, and in particular to a wind power cluster hierarchical optimization and dispatching method, system and medium. Background Art

[0002] Wind power generation is developing rapidly as an important part of renewable energy, but the grid's ability to accept and regulate wind power faces challenges. Especially under the new trend of large-scale and clustered wind power, the effective management and optimized dispatch of wind farms are becoming increasingly prominent. How to achieve efficient and stable dispatch of wind power clusters and ensure the stability of the power system has become a key issue in the field of wind power generation.

[0003] In the optimal scheduling of wind power clusters, traditional control methods may not be able to fully adapt to the time-varying and uncertain nature of wind farms, as well as the differences between wind farms, resulting in low efficiency and poor reliability in the power output management of wind power clusters. Summary of the invention

[0004] To solve the above problems, the present invention provides a hierarchical optimization scheduling method for wind power clusters. The method is based on hierarchical optimization scheduling of large wind power clusters with entropy variable weight scoring, fully considering the time-varying and uncertainty of wind farms, and reducing the fatigue and fluctuation of the entire wind power cluster through scientific analysis and reasonable resource allocation.

[0005] To achieve the above object, the present invention provides a wind power cluster hierarchical optimization scheduling method, comprising the following steps:

[0006] The entropy method is used to determine the weights of evaluation indicators related to fatigue damage differences and output power fluctuations between wind farms in the wind power cluster; the degradation type of each evaluation indicator is determined, and based on the weight of the evaluation indicator, the variable weight theory is used to balance the horizontal correlation between the evaluation indicators to obtain the variable weight;

[0007] Adopting the membership function and using the fuzzy comprehensive evaluation method, the membership scores of each wind farm in the wind power cluster under each evaluation index are obtained;

[0008] The membership score is weighted according to the variable weight of each evaluation index to obtain the comprehensive degradation score of each wind farm;

[0009] According to the ultra-short-term power forecast value of the wind farm, the wind farm group is divided into climbing group, transition group and down-climbing group respectively; within the same wind farm group, the dispatching order of the wind farm is determined according to the comprehensive degradation score of each wind farm, and on the premise of meeting the power balance constraint between the wind farms, the total output power of each wind farm is adjusted and controlled in real time according to the dispatching requirements of the wind power cluster to increase or decrease the power.

[0010] Preferably, the comprehensive fatigue coefficient is used to quantitatively evaluate the fatigue damage caused by active power output and turbulence effect during the operation of the wind farm, as shown in the following formula:

[0011]

[0012] Where: is the cumulative fatigue coefficient of wind farm i at time t0, the second term is the working fatigue caused by the operation of the wind farm, and the third term is the wind farm fatigue caused by turbulence; P i (t) is the current output power of wind farm i; is the rated power of wind farm i; is the design life of unit i; D is the maintenance compensation coefficient of wind farm i; dis is the turbulence fatigue equivalent coefficient of wind farm i; is the turbulence intensity of wind farm i at time t.

[0013] Preferably, the power fluctuation coefficient is an indicator for measuring the power stability of a wind farm in each operation cycle, as shown in the following formula:

[0014]

[0015] In the formula, g i (t) is the power standard deviation of unit i in the past N0 cycles, which is used to measure the power fluctuation of the unit; P i a It is the average power value of the unit in the past N0 cycles. This indicator takes into account the fluctuations within N0 cycles and quantifies it as the volatility indicator.

[0016] Preferably, the wind abandonment rate is the difference between the ratio of the actual power generation of the wind farm to its theoretical maximum power generation capacity and 1, as shown in the following formula:

[0017]

[0018] In the formula, δ i is the wind curtailment rate of unit i; P i (t) is the current power of unit i; is the current theoretical maximum power obtained by unit i according to the MPPT strategy, which is determined by interpolating the unit power curve according to the current wind speed.

[0019] Preferably, the power regulation margin includes a power up regulation margin And the power reduction margin, as shown in the following formula:

[0020]

[0021] In the formula, is the power prediction value of unit i in the next cycle, that is, the maximum power that unit i can achieve in the next cycle; It is the lower limit of the unit power.

[0022] Preferably, the acquisition of the variable weights of the evaluation indicators includes the following steps:

[0023] Using the concept of information entropy, we can assume that the multi-attribute decision matrix M of the sample data is calculated as follows:

[0024]

[0025] In the formula, {A1, A2, ..., A m} represents m wind farms; x mn represents the value of the nth indicator of the mth wind farm; m and n represent the number of wind farms and indicators respectively;

[0026] Normalize the indicator values:

[0027] x′ ij =x ij / max(x ij );

[0028] In the formula, x′ ij Relative value of the representative indicator;

[0029] Calculate the contribution of the i-th wind farm under the j-th indicator:

[0030]

[0031] Where P ij It represents the proportion of the i-th wind farm in the j-th index, and is regarded as the probability used in the relative entropy calculation, that is, the proportion of continuity;

[0032] Calculate the entropy value of the jth indicator:

[0033]

[0034] In the formula, K = 1 / ln(m), ensuring 0 ≤ E j ≤1, that is, the maximum is 1; for information entropy, e j The larger it is, the greater the information entropy is, indicating that the jth indicator obtains less information and the greater the degree of dispersion;

[0035] The entropy weight of the jth indicator is:

[0036]

[0037] Determine the degradation type of each evaluation index:

[0038] For the smaller the better indicator:

[0039]

[0040] For intermediate indicators:

[0041]

[0042] For the larger the better indicator:

[0043]

[0044] In the formula, g(x) is the degradation degree of the indicator; x is the actual value of the indicator; [α, β] is the operating limit of the indicator, [x a ,x b ] is the optimal value range of the indicator;

[0045] The variable weight formula is:

[0046]

[0047] In the formula, w i ' j is the variable weight of indicator j; w j is the constant weight of indicator j; g ij is the indicator degradation degree; T is the variable weight coefficient, which is 0.5; n is the number of indicators;

[0048] Determine the membership function:

[0049]

[0050] The membership function combines the triangular and semi-trapezoidal functions and is divided into four fuzzy sets {excellent, good, fair, poor};

[0051] By formulating the membership function, the membership score of wind farm i under index j is obtained as follows:

[0052] h ij =0.4v 1ij +0.3v 2ij +0.2v 3ij +0.1v 4ij ;

[0053] In the formula, g is the degradation degree, and {v1, v2, v3, v4} are the membership degrees of the four sets, with weights set to 40%, 30%, 20%, and 10% respectively;

[0054] Then the index score set of wind farm i is:

[0055] H i =[h i1 ,h i2 ,…,hin ];

[0056] According to the weights of each index of each wind farm after entropy value weighting and the index scores of each wind farm, the comprehensive score of the degradation degree of wind farm i is obtained as follows:

[0057]

[0058] Normalize the overall score to a percentage:

[0059]

[0060] Preferably, the method of using the ultra-short-term power forecast value of the wind farm as a basis for judging the power change trend of the wind farm and determining the ramp classification of the wind farm comprises the following steps:

[0061] The ultra-short-term power forecast value of the wind farm from time t to the next four cycles is used as the basis for judging the power change trend of the wind farm;

[0062] The wind farm ramp classification index is defined as follows:

[0063]

[0064] Where, sign() is the sign function; Predict the power value of the wind farm for the nth cycle in the future; is the power output value of the wind farm at time t;

[0065] According to K i The wind farm groups are divided into uphill groups, transition groups and downhill groups.

[0066] Preferably, the step of determining the dispatching order of wind farms in the same wind farm group according to the comprehensive degradation score of each wind farm comprises the following steps:

[0067] The degradation degree matrix of each evaluation index under different states is calculated, and two types of sample data, the upward adjustment matrix and the downward adjustment matrix, are established. The degradation degree of different sample matrices is calculated under different cluster scheduling states.

[0068] When the wind power cluster needs to increase power, the increase matrix is ​​calculated to obtain the increase comprehensive degradation score for sorting. Since the smaller the wind farm degradation comprehensive score is, the stronger its increase ability is, the wind farms are sorted and dispatched from low to high scores;

[0069] When the power is reduced, the reduction matrix is ​​calculated to obtain the comprehensive score of the degradation degree of the reduced wind farm, and the scores are sorted from low to high and dispatched;

[0070] In both states, wind farms with lower comprehensive scores in their respective states are given priority.

[0071] The present invention also provides a wind power cluster hierarchical optimization scheduling system, the system comprising:

[0072] processor;

[0073] a memory having stored thereon a computer program executable on the processor;

[0074] Wherein, when the computer program is executed by the processor, the steps of the wind power cluster hierarchical optimization scheduling method are implemented.

[0075] The present invention also provides a computer-readable storage medium, on which a data processing program is stored, and when the data processing program is executed by a processor, the steps of the wind power cluster hierarchical optimization scheduling method are implemented.

[0076] Beneficial effects of the present invention:

[0077] The present invention proposes a hierarchical optimization scheduling method for wind power clusters. The method adopts entropy value variable weight technology and can dynamically adjust weights according to real-time data of wind farms, so that the power output management of wind power clusters is more accurate and adaptable; by comprehensively considering evaluation indicators such as comprehensive fatigue coefficient, power fluctuation coefficient, wind abandonment rate and power regulation margin of wind farms, a comprehensive evaluation of wind farm performance is achieved; the method introduces a hierarchical optimization scheduling strategy, which can dynamically group and sort wind farms according to power change trends, and give priority to scheduling wind farms with better performance and higher scheduling flexibility; the method effectively reduces the power fluctuation of wind power clusters and improves the stability of wind farm operation through a scientific hierarchical scheduling method; the scheduling strategy reduces the overall fatigue load of the wind farm by reducing unnecessary scheduling actions, and enhances the long-term operation reliability of the wind farm; the scheduling method of the method is helpful to balance the load of the power grid, reduce the impact of wind farms on the stability of the power grid, and improve the friendly operation capability of the power grid to wind power. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1 This is a diagram of the architecture of a hierarchical optimization scheduling method for a large wind power cluster according to an embodiment of the present invention;

[0079] Figure 2 This is a flow chart of dynamic hierarchical sorting and scheduling of wind power clusters according to an embodiment of the present invention;

[0080] Figure 3 is a layout diagram of wind farm sites in a base according to an embodiment of the present invention;

[0081] Figure 4 This is a comparison diagram of the output effects of each wind farm before and after power reduction under the cluster dispatching mode of an embodiment of the present invention;

[0082] Figure 5This is a comparison diagram of the output effects of each wind farm before and after power reduction under the proportional dispatch mode of an embodiment of the present invention;

[0083] Figure 6 The output results of the large-scale wind power cluster hierarchical optimization scheduling method based on entropy value variable weight scoring in the embodiment of the present invention are compared.

[0084] Figure 7 This is a diagram comparing the output of the power instruction of an embodiment of the present invention with three scheduling methods. DETAILED DESCRIPTION

[0085] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0086] Example 1

[0087] The present invention provides a wind power cluster hierarchical optimization scheduling method, and the specific scheduling method architecture diagram is as follows: Figure 1 As shown, the following steps are included:

[0088] S1: Determine the evaluation indicators related to fatigue damage differences and output power fluctuations among wind farms in the wind power cluster, use the entropy method to determine the weight of each evaluation indicator, and perform normalization.

[0089] Among them, the evaluation indicators include comprehensive fatigue coefficient, power fluctuation coefficient, wind abandonment rate and power regulation margin.

[0090] S2: Determine the degradation type of each evaluation indicator, use variable weight theory to balance the horizontal correlation between the evaluation indicators, and obtain the variable weight of each evaluation indicator.

[0091] S3: Adopt the membership function and use the fuzzy comprehensive evaluation method to obtain the membership score of each wind farm in the wind power cluster under each evaluation index; according to the variable weight and membership score of each evaluation index, obtain the comprehensive score of the degradation degree of each wind farm.

[0092] S4: The ultra-short-term power forecast value of the wind farm is used as the basis for judging the power change trend of the wind farm, the climbing classification of the wind farm is determined, and the wind farm group is divided into climbing group, transition group and down-climbing group respectively; within the same wind farm group, the dispatch order of the wind farm is determined according to the comprehensive degradation score of each wind farm.

[0093] S5: Under the premise of satisfying the power balance constraint between wind farms, according to the scheduling requirements of wind farm clusters to increase or decrease power, and based on the scheduling order of wind farms in each wind farm group, the total output power of each wind farm is adjusted and controlled in real time.

[0094] Specifically, the present invention aims to study the performance evaluation of wind farms, achieve accurate tracking of wind power cluster power instructions, and optimize wind farm operations. The specific goals include reducing the fatigue damage differences between wind farms and reducing the volatility of output power. To this end, four evaluation indicators are selected: the comprehensive fatigue coefficient of the wind farm, the power fluctuation coefficient, the wind abandonment rate, and the flexibility of power adjustment.

[0095] (1) Comprehensive fatigue coefficient

[0096] In order to optimize the fatigue balance in wind farm operation, a comprehensive fatigue coefficient of wind farm is proposed to quantitatively evaluate the fatigue damage caused by active power output and turbulence effect during the operation of wind farm. The method for calculating this coefficient is as follows:

[0097]

[0098] Where: is the cumulative fatigue coefficient of wind farm i at time t0, the second term is the working fatigue caused by the operation of the wind farm, and the third term is the wind farm fatigue caused by turbulence; P i (t) is the current output power of wind farm i; is the rated power of wind farm i; is the design life of unit i; D is the maintenance compensation coefficient of wind farm i; dis is the turbulence fatigue equivalent coefficient of wind farm i; is the turbulence intensity of wind farm i at time t.

[0099] (2) Power fluctuation coefficient

[0100] The power fluctuation coefficient is an indicator to measure the power stability of a wind farm in each operation cycle. It aims to improve the stability of wind farm operation and reduce the fatigue damage caused by it by reducing power fluctuations, as shown in the following formula:

[0101]

[0102] In the formula, g i (t) is the power standard deviation of unit i in the past N0 cycles, which is used to measure the power fluctuation of the unit; P i a It is the average power value of the unit in the past N0 cycles. This indicator takes into account the fluctuations within N0 cycles and quantifies it as the volatility indicator.

[0103] (3) Wind curtailment rate

[0104] The wind curtailment rate is the difference between the ratio of the actual power generation of a wind farm to its theoretical maximum power generation capacity and 1, as shown in the following formula:

[0105]

[0106] In the formula, δ i is the wind curtailment rate of unit i; P i (t) is the current power of unit i; is the current theoretical maximum power obtained by unit i according to the MPPT strategy, which is determined by interpolating the unit power curve according to the current wind speed.

[0107] (4) Power regulation margin

[0108] The power regulation margin of a wind farm directly reflects the potential of the wind farm to adjust power. The power regulation margin includes the power increase margin. And the power reduction margin, as shown in the following formula:

[0109]

[0110] In the formula, is the power prediction value of unit i in the next cycle, that is, the maximum power that unit i can achieve in the next cycle; It is the lower limit of the unit power.

[0111] Furthermore, in S2, the acquisition of the variable weights includes the following steps:

[0112] S2.1: Using the concept of information entropy, we can assume that the multi-attribute decision matrix M of the sample data is calculated as follows:

[0113]

[0114] In the formula, {A1, A2, ..., A m} represents m wind farms; x mn represents the value of the nth indicator of the mth wind farm; m and n represent the number of wind farms and indicators respectively.

[0115] S2.2: Normalize the indicator values:

[0116] x′ ij =x ij / max(x ij );

[0117] In the formula, x′ ij Relative value of the index. Considering that the units of various indicators are not uniform, in order to calculate the comprehensive index, it is necessary to first perform standardization and convert the absolute value of the index into a relative value to solve the problem of homogenization of different quality index values.

[0118] S2.3: Calculate the contribution of the i-th wind farm under the j-th indicator:

[0119]

[0120] Where P ij It represents the proportion of the i-th wind farm in the j-th index, and is regarded as the probability used in the relative entropy calculation, that is, the proportion of continuity.

[0121] S2.4: Calculate the entropy value of the jth indicator:

[0122]

[0123] In the formula, K = 1 / ln(m), ensuring 0 ≤ E j ≤1, that is, the maximum is 1; for information entropy, e j The larger it is, the greater the information entropy is, indicating that the j-th indicator obtains less information and the greater the degree of dispersion.

[0124] S2.5: The entropy weight of the jth indicator is:

[0125]

[0126] According to the definition, information entropy is a real number between [0,1]. Since information entropy and weight are negatively correlated, the information utility value is used for calculation, and finally the entropy weight of each indicator is obtained by normalization, that is, the weight of each indicator.

[0127] The entropy method determines the index weights according to the variation of the index values. This is an objective weighting method that avoids the deviation caused by human factors. However, since the importance of the index itself is ignored and the horizontal influence between the indexes cannot be considered, the index weights determined sometimes differ greatly from the expected results. In this regard, the present invention introduces the variable weight theory to make up for the shortcoming that the entropy method cannot consider the correlation between the indexes.

[0128] S2.6: The degradation value describes the degree of degradation of the actual power generation performance of the wind turbine compared with the fault state. Therefore, the better the index value, the smaller the degradation degree, and the degradation degree range is [0,1].

[0129] Determine the degradation type of each evaluation index:

[0130] For the smaller the better indicator:

[0131]

[0132] For intermediate indicators:

[0133]

[0134] For the larger the better indicator:

[0135]

[0136] In the formula, g(x) is the degradation degree of the indicator; x is the actual value of the indicator; [α, β] is the operating limit of the indicator, [x a ,x b ] is the optimal value range of the indicator.

[0137] S2.7: The variable weight formula is:

[0138]

[0139] In the formula, w i ' j is the variable weight of indicator j; w j is the constant weight of indicator j; g ij is the indicator degradation degree; T is the variable weight coefficient, which is 0.5; n is the number of indicators.

[0140] The variable weight theory reflects the trade-off between each indicator in the overall decision-making process, which is achieved by correcting the weights of various undesirable indicators. In the comprehensive evaluation of wind farm power generation performance, if a certain indicator value is high but the weight is low, its degradation performance will be ignored. In the comprehensive evaluation, it has a great impact on the accuracy of wind farm performance evaluation. Therefore, the variable weight theory is introduced to balance the horizontal correlation between indicators to accurately evaluate the performance of wind farms.

[0141] In S3, obtaining the comprehensive degradation score includes the following steps:

[0142] S3.1: Determine the membership function:

[0143]

[0144]

[0145] The membership function is a mathematical tool used to describe fuzzy sets, which is used to indicate the degree to which an element belongs to a fuzzy set. The membership function combines the triangular and semi-trapezoidal functions to divide the fuzzy sets into four categories: {excellent, good, fair, poor}.

[0146] S3.2: Formulate the membership function and obtain the membership score of wind farm i under index j as follows:

[0147] h ij =0.4v 1ij +0.3v 2ij +0.2v 3ij +0.1v 4ij ;

[0148] Where g is the degradation degree, and {v1, v2, v3, v4} are the membership degrees of the four sets, with weights set to 40%, 30%, 20%, and 10% respectively.

[0149] Then the index score set of wind farm i is:

[0150] H i =[h i1 ,h i2 ,…,h in ];

[0151] S3.3: According to the weights of each index of each wind farm after entropy value weighting and the index scores of each wind farm, the comprehensive score of the degradation degree of wind farm i is obtained as follows:

[0152]

[0153] Normalize the overall score to a percentage:

[0154]

[0155] Furthermore, in S4, in the dynamic grouping process, the ultra-short-term power forecast value of the wind farm from time t to the next four cycles is used as the basis for judging the power change trend of the wind farm. This means that the forecast value will be continuously updated over time, so it has certain dynamic performance. Grouping is performed based on the change trend of the forecast value, that is, the ramp rate. However, it should be emphasized that these groups are not fixed and will be re-grouped according to the indicators in each cycle.

[0156] For the same wind farm group, the dispatching order of wind farms is determined according to the comprehensive scores of wind farms. However, for different indicators, when the power is adjusted up or down, the size of the indicator has different effects on the operation status of the wind farm in the next cycle. Therefore, the impact of indicators on cluster output under different states should be classified and discussed, as shown in Table 1.

[0157] Table 1 Classification of indicator degradation types.

[0158]

[0159] The degradation degree matrix of the indicators under different states is calculated, and two types of sample data, the upward adjustment matrix and the downward adjustment matrix, are established. The degradation degree of different sample matrices is calculated under different cluster dispatching states. That is, when the wind power cluster needs to increase power, the upward adjustment matrix is ​​calculated, and the upward adjustment comprehensive degradation degree score is obtained for sorting. Since the smaller the comprehensive score of the wind farm degradation degree, the stronger its upward adjustment ability, the wind farms are sorted and dispatched from low to high scores; similarly, when the power is reduced, the downward adjustment matrix is ​​calculated, and the comprehensive score of the downward adjustment wind farm degradation degree is obtained, and the scores are sorted and dispatched from low to high. That is, in both states, the wind farms with lower comprehensive scores in their respective states are preferentially dispatched.

[0160] Furthermore, in S5, the specific constraints are:

[0161] The hierarchical sorting of wind power clusters in the present invention has two main objectives: one is to adjust and control the total output power of each wind farm in real time according to the requirements of the wind power cluster system scheduling to keep it consistent with the system requirements and realize cluster grid-connected power command tracking; the other is to reduce the scheduling and output fluctuations of a single wind farm and improve the reliability of command tracking by sorting the wind farm fluctuation indicators.

[0162] In order to make the regulation results more realistic, the total output power of the wind farm should be kept within a certain range to ensure the power supply stability of the entire wind power cluster. This can be achieved by setting upper and lower limits on the total power, that is, the power balance constraint between wind farms.

[0163] Specifically, the dynamic hierarchical scheduling process is as follows:

[0164] The base wind power cluster dynamic hierarchical sorting algorithm proposed by the present invention dynamically groups and sorts the wind farms in the cluster according to the grouping index and the comprehensive score of the wind farm over time. The cluster needs to increase or decrease power according to the difference between the current output and the instruction, and the predicted value of the wind farm, and then dispatches according to the grouping sorting.

[0165] In this embodiment, an example of a hierarchical optimization scheduling method for a large wind power cluster based on entropy variable weight scoring is provided, and key parameters such as the wake impact and power output of each wind farm in the wind power base are calculated and evaluated in detail. This embodiment is based on the data of each wind farm in a large northwest base. The base layout is as follows: Figure 3 As shown in the figure, there are 10 wind farms in total. The specific information of each wind farm is shown in Table 2. Since the wind measurement cycle of the base is 10 minutes, there are 144 cycles of data in one day. The main wind direction during the sampling period is westerly, and the wind speed is concentrated around 10m / s.

[0166] Table 2 Installed capacity information of each wind farm in this large base.

[0167] Wind Farm Total installation Installation information #1 600MW 65 units of 4.8MW, 64 units of 4.5MW #2 828.9MW 48 units of 4.8MW, 133 units of 4.5MW #3 401.2MW 99 units 3.4MW, 19 units 3.4MW #4 401.2MW 107 units 3.4MW, 11 units 3.4MW #5 701.7MW 29 units of 4.8MW, 125 units of 4.5MW #6 234MW 52 units 4.5MW #7 752MW 188 4MW #8 500MW 125 4MW #9 298.2MW 71 units of 4.2MW, 20 units of 5MW #10 596.4MW 142 units 4.2MW

[0168] The wind power clusters are divided into different layers and sorted according to the power variation capability and other key performance indicators of the wind farms within a certain period of time:

[0169] In the dynamic grouping process, the ultra-short-term power forecast value of the wind farm from time t to the next four cycles is used as the basis for judging the power change trend of the wind farm.

[0170] The wind farm ramp classification index is defined as follows:

[0171]

[0172] Where, sign() is the sign function; Predict the power value of the wind farm for the nth cycle in the future; is the power output value of the wind farm at time t.

[0173] The above formula shows how to calculate the ramp classification index of a wind farm, which can evaluate the performance and change trend of the wind farm. Since the predicted value is constantly updated and the clusters are regrouped every cycle, this process has a certain dynamic performance, which can more accurately predict the power changes of the wind farm and group them accordingly.

[0174] It can be seen from the definition. According to K i The wind farm groups are divided into uphill group, transition group and downhill group. The grouping indicators and their respective characteristics are shown in Table 3:

[0175] Table 3 Indicators and characteristics of climbing clustering.

[0176]

[0177] Therefore, according to the ramp rate indicator, the dispatching order of the field group when the power command changes is: when the power needs to be increased, the upper climbing group transitions to the upper climbing group; when the power needs to be reduced, the lower climbing group transitions to the upper climbing group. It should be noted that when the cluster power is increased, the lower climbing group is in a power reduction trend and cannot be dispatched to increase power.

[0178] Calculate the degradation matrix of the indicators under different states and establish the upward adjustment matrix M up and the down-regulation matrix M dw Two types of sample data are used to calculate the degradation degree of different sample matrices under different cluster scheduling states. That is, when the wind power cluster needs to increase power, the increase matrix M up Calculate and get the upward adjustment comprehensive degradation score for sorting. Since the smaller the wind farm degradation comprehensive score is, the stronger its upward adjustment ability is, the wind farms are sorted and dispatched from low to high scores; similarly, when the power is reduced, the downward adjustment matrix M dw Calculation is performed to obtain the comprehensive score of the deterioration degree of the downgraded wind farm, and the wind farms are sorted from low to high and dispatched. That is, in both states, the wind farms with lower comprehensive scores in each state are dispatched first.

[0179] Further, such as Figure 2 As shown in the figure, under the premise of meeting the constraints, the total output power of each wind farm is adjusted and controlled in real time according to the requirements of the wind power cluster system dispatch:

[0180] In order to make the control results more realistic, the total output power of the wind farm should be kept within a certain range to ensure the power supply stability of the entire wind power cluster. This can be achieved by setting the upper and lower limits of the total power, that is, the power balance constraint between wind farms:

[0181]

[0182] Where Pmin and P max They are the upper and lower limits of the total output power of the wind power cluster respectively.

[0183] The dispatching power of each wind farm should be less than its installed capacity to ensure that the wind farm operates within a reasonable load range. That is, the wind farm capacity constraint is:

[0184] P i ref ≤P i rate

[0185] Where P i rate is the installed capacity of wind farm i, i.e. the rated output power.

[0186] Considering the physical constraints in the wind power cluster, the wind farm output forecast constraint is:

[0187] P i min ≤P i ref ≤P i for

[0188] Where P i min ,P i ref ,P i for are the minimum dispatch power, dispatch active power command and predicted power of wind farm i respectively, and P i min =0.3P i for .

[0189] The speed regulation capability of a wind farm is limited, so the output power change rate needs to be limited. This can be achieved by setting the speed slope, i.e. the wind farm speed regulation capability constraint:

[0190] P i real -P i ref ≤k·d i

[0191] Where P i real is the actual output of wind farm i, d i is the time interval, and the subsequent examples use a 10-min interval, i.e., 600 s; k is the velocity slope, which is generally 0.2-0.5, and 0.3 is selected here.

[0192] The output fluctuation of the wind farm should be controlled within a certain range to reduce the impact on the system. This can be achieved by setting an upper limit for the output fluctuation. Here, the output fluctuation limit coefficient is 3, which is the wind farm operation stability constraint:

[0193] |P i real -P i ref |≤P i rate / 3

[0194] Where P i real is the actual output of wind farm i.

[0195] In addition, in order to reduce the cluster power tracking error, the output error is: the actual cluster output and P r The error does not exceed 1%P r .

[0196] Further: Dynamically group and sort the wind farms in the cluster according to the grouping index and the comprehensive score of the wind farm over time:

[0197] The wind farms in the cluster are dynamically grouped and sorted according to the changes in the grouping index and the comprehensive score of the wind farm over time. The cluster needs to increase or decrease power according to the difference between the current output and the command, as well as the predicted value of the wind farm, and then dispatched according to the grouping sorting. The dispatching process is as follows: Figure 1 shown.

[0198] Take the moment when the ramp rates of all wind farms in the base are at (-4,4), that is, when all wind farms in the cluster belong to the transition group, the power dispatch does not consider the grouping situation at this time, and only sorts and dispatches according to the downward adjustment score of each wind farm. At this time, the cluster power is in a downward adjustment state. The output comparison of each wind farm before and after power reduction under the two methods is shown in Figure 4 , 5 Since there is no downward adjustment score for proportional allocation, only the cluster scheduling method proposed in this embodiment is compared and analyzed. The comprehensive evaluation scores of each wind farm when reducing power are shown in Table 4. The scores are sorted from low to high in the table, and the first line is the corresponding wind farm number.

[0199] Table 4: Downward ratings of wind farms before power reduction.

[0200]

[0201] Through Figure 4The comparative analysis of wind farm output before and after power reduction shows that only five wind farms numbered 7, 2, 3, 10, and 1 implemented power reduction actions, and these actions were performed in order from low to high scores. The order of this scheduling result is completely consistent with the downward adjustment score ranking listed in Table 4, which fully reflects the rationality and expected effect of the cluster allocation scheduling strategy proposed in this embodiment. Figure 5 Compared with the proportional allocation method shown in the figure, this method only selects some wind farms to perform power reduction actions, rather than all wind farms participating in the scheduling. This differentiated scheduling method can effectively reduce the number of wind farms participating in the scheduling, thereby reducing the overall fatigue load of the entire wind power cluster, enhancing the operating stability of the system and the long-term operating reliability of the wind farm.

[0202] Determination of evaluation indicators and comparison methods:

[0203] In order to compare the superiority of the wind power cluster optimization scheduling strategy considering the comprehensive score of wind farms proposed in this embodiment, the cluster instruction tracking effects of the three methods of proportional allocation, sequential scheduling and cluster scheduling proposed in this embodiment are compared. The three scheduling methods are introduced as follows:

[0204] Method 1: Proportional allocation. Different from the traditional proportional allocation method, this strategy allocates power based on the ratio of the predicted power of each unit in the next time period to the total predicted power of the wind farm, aiming to more accurately match the power regulation capacity of each unit:

[0205]

[0206] Method 2: Sequential scheduling. The main difference between this method and the cluster method proposed in this embodiment is that this method does not involve the process of grouping wind farms and ranking by comprehensive scores, and when executing power adjustment commands, the power output is increased or decreased only according to the order of wind farm numbers.

[0207] Method 3: Cluster scheduling: Based on the clustering and sorting scheduling strategy proposed in Chapter 4, clustering is based on the climbing rate and sorting is based on the entropy value variable weighted comprehensive score.

[0208] Cluster instructions are set for the above three methods, and the scheduling performance under different methods is compared and analyzed, including instruction tracking, volatility, fatigue, error, etc. For this purpose, the base power instructions are set as shown in Table 5.

[0209] Table 5 Power command settings.

[0210] Number of cycles 0-50 50-75 75-100 100-144 Command power(MW) 3000 3500 4500 2000

[0211] Figure 6 To compare the output results of the hierarchical optimization scheduling method for large wind power clusters using entropy-based variable weight scoring.

[0212] In wind farm management, power tracking accuracy is one of the important indicators for evaluating the stability and efficiency of the power system. Figure 7 As shown, the present invention compares and analyzes the power tracking percentage errors of three different power scheduling schemes in each cycle, aiming to reveal the impact of different scheduling strategies on the output power tracking performance of wind farms.

[0213] The analysis results show that the power tracking deviations of various schemes are mainly concentrated in two key moments: one is the moment when the cluster power command changes, and the other is the moment when the actual power output of the wind farm fluctuates. In both cases, the size of the power tracking error directly reflects the adaptability and accuracy of different scheduling schemes.

[0214] Optimization effect analysis:

[0215] After careful comparative analysis, the results show that the three scheduling strategies can effectively track cluster power instructions. Among them, the hierarchical optimization scheduling strategy proposed in this embodiment performs best in terms of MRE (maximum relative error) and RMSE (root mean square error), reaching 2.9266% and 3.7022×10 5 MW is significantly better than the other two methods.

[0216] The above is a wind power cluster hierarchical optimization scheduling method provided by one embodiment of this embodiment. Based on the same idea, this embodiment also provides a corresponding wind power cluster hierarchical optimization scheduling system. For the specific definition of the wind power cluster hierarchical optimization scheduling system, please refer to the definition of the wind power cluster hierarchical optimization scheduling method in the above text, which will not be repeated here. Each module in the above-mentioned wind power cluster hierarchical optimization scheduling system can be implemented in whole or in part through software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above-mentioned modules.

[0217] This embodiment also provides a computer-readable storage medium, which stores a computer program, which can be used to execute the above Figure 1 A hierarchical optimization scheduling method for wind power clusters is provided.

[0218] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0219] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A wind power cluster hierarchical optimization scheduling method, characterized in that: The following steps are involved: The entropy method is used to determine the weights of evaluation indicators related to fatigue damage differences and output power fluctuations among wind farms in the wind power cluster; Determine the degradation type of each evaluation index, and based on the weight of the evaluation index, use the variable weight theory to balance the horizontal correlation between the evaluation indicators to obtain the variable weight; Adopting the membership function and using the fuzzy comprehensive evaluation method, the membership scores of each wind farm in the wind power cluster under each evaluation index are obtained; The membership score is weighted according to the variable weight of each evaluation index to obtain the comprehensive degradation score of each wind farm; According to the ultra-short-term power forecast value of the wind farm, the wind farm group is divided into climbing group, transition group and down-climbing group respectively; within the same wind farm group, the dispatching order of the wind farm is determined according to the comprehensive degradation score of each wind farm, and on the premise of meeting the power balance constraint between the wind farms, the total output power of each wind farm is adjusted and controlled in real time according to the dispatching requirements of the wind power cluster to increase or decrease the power.

2. The wind power cluster hierarchical optimization scheduling method according to claim 1 is characterized in that: The evaluation indexes include comprehensive fatigue coefficient, power fluctuation coefficient, wind abandonment rate and power regulation margin; the comprehensive fatigue coefficient is used to quantitatively evaluate the fatigue damage caused by active power output and turbulence effect during the operation of the wind farm, as shown in the following formula: Where: is the cumulative fatigue coefficient of wind farm i at time t0, the second term is the working fatigue caused by the operation of the wind farm, and the third term is the wind farm fatigue caused by turbulence; P i (t) is the current output power of wind farm i; is the rated power of wind farm i; is the design life of unit i; D is the maintenance compensation coefficient of wind farm i; dis is the turbulence fatigue equivalent coefficient of wind farm i; is the turbulence intensity of wind farm i at time t.

3. The wind power cluster hierarchical optimization scheduling method according to claim 2 is characterized in that: The power fluctuation coefficient is an indicator for measuring the power stability of a wind farm in each operation cycle, as shown in the following formula: In the formula, g i (t) is the power standard deviation of unit i in the past N0 cycles, which is used to measure the power fluctuation of the unit; P i a It is the average power value of the unit in the past N0 cycles. This indicator takes into account the fluctuations within N0 cycles and quantifies it as the volatility indicator.

4. The hierarchical optimization scheduling method for wind power clusters according to claim 3 is characterized in that: The wind abandonment rate is the difference between the ratio of the actual power generation of the wind farm and its theoretical maximum power generation capacity and 1, as shown in the following formula: In the formula, δ i is the wind curtailment rate of unit i; P i (t) is the current power of unit i; is the current theoretical maximum power obtained by unit i according to the MPPT strategy, which is determined by interpolating the unit power curve according to the current wind speed.

5. The wind power cluster hierarchical optimization scheduling method according to claim 4 is characterized in that: The power regulation margin includes a power increase margin And the power reduction margin, as shown in the following formula: In the formula, is the power prediction value of unit i in the next cycle, that is, the maximum power that unit i can achieve in the next cycle; It is the lower limit of the unit power.

6. The wind power cluster hierarchical optimization scheduling method according to claim 1 is characterized in that: The acquisition of the variable weights of the evaluation indicators includes the following steps: Using the concept of information entropy, we can assume that the multi-attribute decision matrix M of the sample data is calculated as follows: In the formula, {A1, A2, ..., A m } represents m wind farms; x mn represents the value of the nth indicator of the mth wind farm; m and n represent the number of wind farms and indicators respectively; Normalize the indicator values: x i ′ j =x ij / max(x ij ); In the formula, x i ' j Relative value of the representative indicator; Calculate the contribution of the i-th wind farm under the j-th indicator: Where P ij It represents the proportion of the i-th wind farm in the j-th index, and is regarded as the probability used in the relative entropy calculation, that is, the proportion of continuity; Calculate the entropy value of the jth indicator: In the formula, K = 1 / ln(m), ensuring 0 ≤ E j ≤1, that is, the maximum is 1; for information entropy, e j The larger it is, the greater the information entropy is, indicating that the jth indicator obtains less information and the greater the degree of dispersion; The entropy weight of the jth indicator is: Determine the degradation type of each evaluation index: For the smaller the better indicator: For intermediate indicators: For the larger the better indicator: In the formula, g(x) is the degradation degree of the indicator; x is the actual value of the indicator; [α, β] is the operating limit of the indicator, [x a ,x b ] is the optimal value range of the indicator; The variable weight formula is: In the formula, w i ' j is the variable weight of indicator j; w j is the constant weight of indicator j; g ij is the indicator degradation degree; T is the variable weight coefficient, which is 0.5; n is the number of indicators; Determine the membership function: The membership function combines the triangular and semi-trapezoidal functions and is divided into four fuzzy sets {excellent, good, fair, poor}; By formulating the membership function, the membership score of wind farm i under index j is obtained as follows: h ij =0.4v 1ij +0.3v 2ij +0.2v 3ij +0.1v 4ij ; In the formula, g is the degradation degree, and {v1, v2, v3, v4} are the membership degrees of the four sets, with weights set to 40%, 30%, 20%, and 10% respectively; Then the index score set of wind farm i is: H i =[h i1 ,h i2 ,…,h in ]; According to the weights of each index of each wind farm after entropy value weighting and the index scores of each wind farm, the comprehensive score of the degradation degree of wind farm i is obtained as follows: Normalize the overall score to a percentage:

7. The wind power cluster hierarchical optimization scheduling method according to claim 1 is characterized in that: The grouping of the wind farm groups comprises the following steps: The ultra-short-term power forecast value of the wind farm from time t to the next four cycles is used as the basis for judging the power change trend of the wind farm; The wind farm ramp classification index is defined as follows: Where, sign() is the sign function; Predict the power value of the wind farm for the nth cycle in the future; is the power output value of the wind farm at time t; According to K i The wind farm groups are divided into uphill groups, transition groups and downhill groups.

8. The wind power cluster hierarchical optimization scheduling method according to claim 7 is characterized in that: The dispatching order of wind farms is determined according to the comprehensive degradation scores of the wind farms in the same wind farm group, and includes the following steps: The degradation degree matrix of each evaluation index under different states is calculated, and two types of sample data, the upward adjustment matrix and the downward adjustment matrix, are established. The degradation degree of different sample matrices is calculated under different cluster scheduling states. When the wind power cluster needs to increase power, the increase matrix is ​​calculated to obtain the increase comprehensive degradation score for sorting. Since the smaller the wind farm degradation comprehensive score is, the stronger its increase ability is, the wind farms are sorted and dispatched from low to high scores; When the power is reduced, the reduction matrix is ​​calculated to obtain the comprehensive score of the degradation degree of the reduced wind farm, and the scores are sorted from low to high and dispatched; In both states, wind farms with lower comprehensive scores in their respective states are given priority.

9. A wind power cluster hierarchical optimization dispatching system, characterized in that: The system comprises: processor; a memory having stored thereon a computer program executable on the processor; Wherein, when the computer program is executed by the processor, the steps of the wind power cluster hierarchical optimization scheduling method according to any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a data processing program, and when the data processing program is executed by a processor, the steps of the hierarchical optimization scheduling method for wind power clusters according to any one of claims 1 to 8 are implemented.

Images