Active and reactive multi-level integrated coordination control method for wind-solar power generation cluster

Abstract

The invention relates to an active and reactive multi-level integrated coordination control method for a wind-solar power generation cluster. The method comprises the following steps: (1) establishing a machine-field-cluster multi-level scheduling architecture; (2) establishing a machine-field-group optimization target and constraint conditions; (3) establishing a multi-time-scale joint optimization scheduling model composed of a day-ahead scheduling plan, an intra-day rolling plan and real-time scheduling; and (4) improving a raccoon algorithm in combination with the migration factor and an elite selection strategy, and solving the machine-field-group multi-time scale joint optimization scheduling model by using the improved raccoon algorithm to obtain a final optimization scheduling result. The method is high in reliability and good in practicability, compared with the prior art, active and reactive integrated coordination control can be achieved at the same time, and a new method is provided for large-scale wind and light cluster optimization scheduling.

Description

Technical Field

[0001] The present invention relates to the field of power grid dispatching, and in particular to a multi-level integrated coordinated control method for active and reactive power of a wind and solar power generation cluster. Background Art

[0002] With the large-scale access of renewable energy such as wind and solar energy to the power grid, the power system is shifting from a system dominated by traditional rotating motors to a complex system dominated by new power electronic equipment. However, due to the randomness and volatility of natural resources such as wind energy and light, the controllability and dispatchability of intermittent power output are poor. In addition, the inherent particularity of sources, loads, and networks in the northwest region has caused profound changes in the operating characteristics of the power grid, bringing huge challenges to the consumption of renewable energy. Therefore, it is urgent to propose a coordinated control method for the access of large-scale renewable energy such as wind and light to the power grid.

[0003] At present, there are many studies on technical issues such as the control of wind turbines / photovoltaic modules, and the active and reactive optimization control of wind farms / photovoltaic power stations. The literature "Equivalent modeling method for regional decentralized photovoltaic clusters based on cluster analysis" [Wu HB, Liu ZQ, Chen Y, et al. CPSS Transactions on Power Electronics and Applications, 2018, 3 (2): 146-153.] proposes that cluster division indicators are usually based on electrical distance, focusing on the coupling of clusters and the utilization of power sources within clusters. The literature "Renewable energy generation cluster division method for improving the absorption capacity" [Bi Rui, Liu Xianfang, Ding Ming, et al. Proceedings of the CSEE, 2019, 39 (22): 6583-6591.] proposes a cluster division method to improve the local absorption capacity of renewable energy and the source-load matching level. However, the active power optimization control of the "machine-farm-group" of large-scale wind and photovoltaic bases aims to maximize the wind and photovoltaic output, and the reactive power optimization control aims to minimize network losses. Simply weighted combination of the two optimization objectives has no clear physical meaning and is not optimizable.

[0004] In summary, most existing technologies ignore the temporal and spatial correlations between various stations and clusters, and do not completely solve the coordination and optimization control problems brought about by the large-scale access of renewable energy such as wind and solar energy to the power grid. Therefore, it is necessary to establish an active-reactive integrated optimization model to achieve integrated optimization control of active power and reactive voltage. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a multi-level integrated coordinated control method for active and reactive power of a wind and solar power generation cluster with high reliability and good practicality.

[0006] In order to solve the above problems, a multi-level integrated coordinated control method for active and reactive power of a wind-solar power generation cluster described in the present invention comprises the following steps:

[0007] ⑴ Establish a multi-level dispatching framework of airport, airport and group;

[0008] (2) Establish the optimization objectives and constraints of the airport-field-group;

[0009] ⑶ Establish a multi-time scale joint optimization scheduling model consisting of day-ahead scheduling, intraday rolling planning and real-time scheduling;

[0010] (4) The raccoon algorithm is improved by combining the migration factor and the elite selection strategy. The improved raccoon algorithm is used to solve the airport-field-group multi-time scale joint optimization scheduling model to obtain the final optimization scheduling result.

[0011] In step (1), the multi-level dispatching architecture of machine-farm-cluster consists of a unit layer, a station layer, a cluster layer and an upper-level dispatching layer; the cluster layer is composed of all stations gathered together; the dispatching process of this architecture is as follows: the wind / light power prediction data and safety constraints of the unit layer are uploaded to the station layer for optimization, the operation information of each station is uploaded to the cluster layer, the upper-level dispatching center of the cluster layer receives the operation information of the cluster, and performs optimization calculations in combination with the load prediction data of the power grid and the safety constraints of the cluster; after the optimization is completed, the upper-level dispatching center issues dispatching instructions to the cluster, station and machine level by level.

[0012] The establishment of the optimization objectives and constraints of the airport-airport-group in step (2) consists of a day-ahead operation phase, an intra-day operation phase, and a real-time scheduling phase;

[0013] ① In the day-ahead operation stage, the optimization goal is to minimize the system operation cost and network loss, and its objective function is:

[0014] minC1=C wt +C pv +C loss

[0015]

[0016] Where: C wt is the operating cost of the wind farm, in RMB; C pv is the operating cost of the photovoltaic power station, in yuan; C loss is the network loss cost, in RMB; the specified time length is divided into N using Δt as the scale T time period; is the cost coefficient of the i-th wind farm; is the cost coefficient of the i-th photovoltaic power station; is the power generation of the i-th wind farm at time t, in MW; is the power generation of the i-th PV power station at time t, in MW;

[0017] The constraints are composed of power balance constraints, wind-solar operation constraints, wind-solar ramp constraints and station output regulation rate constraints; among which:

[0018] iPower balance constraints:

[0019]

[0020] Where: P L,t is the electrical load, in MW;

[0021] ii. Wind and solar operation constraints:

[0022]

[0023] Where: P wt is the output of the wind farm, in MW; P pv is the output of the photovoltaic power station, in MW; is the maximum and minimum output of the wind farm, in MW; The maximum and minimum output of the photovoltaic power station, in MW;

[0024] iii. Wind and solar climbing constraints:

[0025]

[0026] Where: w,t is the wind power ramp rate; min,t , max,t is the minimum and maximum ramp rate of wind power; Δt is the time interval between two wind power values ​​at time t, in hours; is the lower limit of photovoltaic power, P pv,t is the photovoltaic power generation at time t, P pv,t-1 is the photovoltaic power generation power at time t-1, in MW; P is the limit value of photovoltaic down and up climbing rate; w,t is the wind power generation power at time t, P w,t+1 is the wind power generation power at time t+1, in MW;

[0027] ⅳ Station output regulation rate constraints:

[0028]

[0029] Where: α w is the power change rate coefficient of the wind farm; α pv is the power change rate coefficient of the photovoltaic power station; and are the power generation of the i-th wind farm and the j-th photovoltaic power station at time t, in MW; and are the changing power of the wind farm and the changing power of the photovoltaic power station during the period t-Δt, in MW; P w,i and P pv,j are the power generation of the i-th wind farm and the j-th photovoltaic power station, in MW;

[0030] ② Intraday operation phase, to minimize the wind curtailment cost C wt,loss and the abandoned light cost C pv,loss To optimize the target, the objective function is:

[0031]

[0032]

[0033] Where: K wt,loss , K pv,loss are the penalty coefficients for wind and solar power abandonment respectively; They are the abandoned wind and solar power, in MW;

[0034] The constraints for curtailing wind and solar power are:

[0035]

[0036] Where: P t w,roll , P t pv,roll They are wind power forecast output and photovoltaic forecast output, in MW; They are wind power abandonment and solar power abandonment, in MW; They are wind power actual power and photovoltaic actual power, in MW;

[0037] ③ In the real-time scheduling stage, the total deviation between the output of each station of the airport-field-group and the scheduled instructions is minimized as the optimization goal, and the influence of the prediction error is taken into account at the same time; its objective function is:

[0038]

[0039] Where: M is the number of time periods in the real-time scheduling phase; are the power forecast values ​​of wind power and photovoltaic power, in MW; It is the planned power generation output of the joint operation of the machine, farm and cluster, in MW;

[0040] The constraints for airport-field-group joint operation are:

[0041]

[0042] Where: P dev is the offset power during joint operation of airport, airport and cluster, in MW; P wt,roll , P pv,roll They are the output plans of wind power and photovoltaic power, in MW; are the wind power and photovoltaic power generation at time t, in MW; is the combined output plan at time t, in MW.

[0043] The scheduling process of the multi-time scale joint optimization scheduling model in step (3) is as follows:

[0044] a. In the day-ahead operation stage, based on the wind-solar forecast data and power station parameters, the day-ahead optimization scheduling objective function based on the minimum system operation cost and network loss is constructed to generate the planned output of the wind-solar combined power generation system;

[0045] b. Update the intraday wind and solar forecast data, construct the intraday optimization scheduling objective function based on minimizing the wind abandonment cost and solar abandonment cost, and output the planned output of the wind and solar combined power generation system within the day;

[0046] c. Update the real-time wind and solar forecast data, minimize the total deviation between the output of each station in the machine-field-group and the predetermined instruction as the optimization goal, build a real-time correction model, and output the output plan for the min period;

[0047] If the optimization period within the day is not completed, return to step c;

[0048] If the full-day scheduling is not completed, return to step b.

[0049] Compared with the prior art, the present invention has the following advantages:

[0050] 1. The present invention establishes a multi-level dispatching architecture of machine-farm-cluster, and solves the spatiotemporal correlation problem between each station and cluster through the distributed dispatching mode of machine-farm-cluster; then, considering the coupling relationship between different time scales, a multi-time scale optimization method combined with the improved raccoon algorithm is proposed. Through day-ahead dispatching, daily rolling dispatching and real-time dispatching, the influence of uncertain factors on the operation of the power system is eliminated to the greatest extent, thereby solving the problems of inconsistent levels and serious power abandonment in the dispatching mode of renewable energy.

[0051] 2. The present invention fully considers the coupling relationship between different time scales, and realizes the step-by-step refinement and global and local coordination of power grid energy management.

[0052] 3. The present invention has high reliability and good practicality. Compared with the prior art, the present invention can simultaneously realize integrated coordinated control of active and reactive power, and provides a new method for optimizing the scheduling of large-scale wind and solar clusters. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] The specific implementation modes of the present invention are further described in detail below in conjunction with the accompanying drawings.

[0054] Figure 1 It is a control framework diagram of the present invention.

[0055] Figure 2 This is a flowchart of the multi-time scale joint optimization scheduling of the present invention.

[0056] Figure 3 It is a schematic diagram of the output of each station of the present invention.

[0057] Figure 4 It is a schematic diagram of cluster scheduling instructions and cluster total output according to the present invention.

[0058] Figure 5 This is a curve diagram of active power loss of the system of the present invention.

[0059] Figure 6 This is a real-time optimization result graph of the present invention. DETAILED DESCRIPTION

[0060] To improve the level of wind and solar power consumption and smooth out the randomness and volatility of the output of a single unit. The present invention formulates a machine-farm-group active output plan based on ultra-short-term load forecast data and power station power forecast data within a scheduling cycle and sends it to wind farms and photovoltaic power stations. When the lower-level stations receive commands from the upper layer, information is exchanged between the individual stations, and the active output of each station is reasonably formulated to optimize reactive power loss and achieve distributed on-site distribution. Finally, the power station side uploads the real-time information and operating status of each individual station to the upper-level scheduling end by adjusting the start and stop of the unit and the station output adjustment rate. The upper and lower layers cooperate with each other to form a step-by-step refined output coordination to maximize the utilization rate of wind power and photovoltaic energy and the level of refined operation control of wind and photovoltaic power station groups.

[0061] A multi-level integrated coordinated control method for active and reactive power of a wind and solar power generation cluster comprises the following steps:

[0062] ⑴ In order to solve the problems of coordinated control and optimized scheduling of large-scale high-density distributed power sources connected to the power grid, a multi-level scheduling architecture of machine-field-group is established.

[0063] In actual operation, the dispatching and management of renewable energy power stations have problems such as complex dispatching control levels, rough dispatching control strategies, and insufficient dispatching automation. Centralized dispatching relies on the data processing capabilities of the control center, and has problems such as poor risk resistance and system flexibility. To solve the above problems, the present invention proposes a distributed dispatching mode for renewable energy machines, farms, and clusters. The distributed interactive relationships at each level of the machine, farm, and cluster are as follows: Figure 1 shown.

[0064] The machine-farm-cluster multi-level dispatching architecture consists of a unit layer, a station layer, a cluster layer and an upper-level dispatching layer; the cluster layer is composed of all stations together; the dispatching process of this architecture is as follows: the wind / solar power forecast data and safety constraints of the unit layer are uploaded to the station layer for optimization, and the operation information of each station is uploaded to the cluster layer. The upper-level dispatching center of the cluster layer receives the operation information of the cluster and performs optimization calculations in combination with the load forecast data of the power grid and the safety constraints of the cluster; after the optimization is completed, the upper-level dispatching center issues dispatching instructions to the cluster, station and machine level by level.

[0065] ⑵ Establish optimization objectives and constraints for machine-field-group, and minimize overall operating costs by optimizing the output of each unit.

[0066] Based on the time characteristics of load and power forecasts, the accuracy gradually increases as the time scale decreases. A multi-time scale optimization operation model of machine-field-group is established to refine the output plan step by step, so as to use the forecast data with lower error to deal with the uncertainty and volatility of renewable energy and load. Based on the multi-time scale response strategy, it is divided into three stages: day-ahead, intraday and real-time. The power forecast and load forecast data are used as the basis for the formulation of the day-ahead scheduling plan. Intraday scheduling is to carry out rolling optimization and correction of the day-ahead plan. Combine ultra-short-term load and power forecasts and real-time backup data to formulate a real-time scheduling plan.

[0067] Therefore, the establishment of the optimization objectives and constraints of the airport-airport-group consists of the day-ahead operation phase, the intraday operation phase, and the real-time scheduling phase;

[0068] ① In the day-ahead operation stage, the optimization goal is to minimize the system operation cost and network loss, and its objective function is:

[0069] minC1=C wt +C pv +C loss

[0070]

[0071] Where: C wt is the operating cost of the wind farm, in RMB; C pv is the operating cost of the photovoltaic power station, in yuan; Closs is the network loss cost, in RMB; the specified time length is divided into N using Δt as the scale T time period; is the cost coefficient of the i-th wind farm; is the cost coefficient of the i-th photovoltaic power station; is the power generation of the i-th wind farm at time t, in MW; is the power generation of the i-th PV power station at time t, in MW;

[0072] The constraints are composed of power balance constraints, wind-solar operation constraints, wind-solar ramp constraints and station output regulation rate constraints; among which:

[0073] iPower balance constraints:

[0074]

[0075] Where: P L,t is the electrical load, in MW;

[0076] ii. Wind and solar operation constraints:

[0077]

[0078] Where: P wt is the output of the wind farm, in MW; P pv is the output of the photovoltaic power station, in MW; is the maximum and minimum output of the wind farm, in MW; The maximum and minimum output of the photovoltaic power station, in MW;

[0079] iii. Wind and solar climbing constraints:

[0080]

[0081] Where: w,t is the wind power ramp rate; min,t , max,t is the minimum and maximum ramp rate of wind power; Δt is the time interval between two wind power values ​​at time t, in hours; is the lower limit of photovoltaic power, P pv,t is the photovoltaic power generation at time t, P pv,t-1 is the photovoltaic power generation power at time t-1, in MW; P is the limit value of photovoltaic down and up climbing rate; w,t is the wind power generation power at time t, P w,t+1 is the wind power generation power at time t+1, in MW;

[0082] ⅳ Station output regulation rate constraints:

[0083]

[0084] Where: α w is the power change rate coefficient of the wind farm; α pv is the power change rate coefficient of the photovoltaic power station; and are the power generation capacity of the i-th wind farm and the j-th photovoltaic power station at time t.

[0085] rate, in MW; and are the changing power of the wind farm and the changing power of the photovoltaic power station during the period t-Δt, in MW; P w,i and P pv,j are the power generation of the i-th wind farm and the j-th photovoltaic power station, in MW;

[0086] ② Intraday operation phase, to minimize the wind curtailment cost C wt,loss and the abandoned light cost C pv,loss To optimize the target, the objective function is:

[0087]

[0088]

[0089] Where: K wt,loss , K pv,loss are the penalty coefficients for wind and solar power abandonment respectively; They are the abandoned wind and solar power, in MW;

[0090] The constraints for curtailing wind and solar power are:

[0091]

[0092] Where: P t w,roll , P t pv,roll They are wind power forecast output and photovoltaic forecast output, in MW; They are wind power abandonment and solar power abandonment, in MW; They are wind power actual power and photovoltaic actual power, in MW;

[0093] ③ In the real-time scheduling stage, the total deviation between the output of each station of the airport-field-group and the scheduled instructions is minimized as the optimization goal, and the influence of the prediction error is taken into account at the same time; its objective function is:

[0094]

[0095] Where: M is the number of time periods in the real-time scheduling phase; are the power forecast values ​​of wind power and photovoltaic power, in MW; It is the planned power generation output of the joint operation of the machine, farm and cluster, in MW;

[0096] The constraints for airport-field-group joint operation are:

[0097]

[0098] Where: P dev is the offset power during joint operation of airport, airport and cluster, in MW; P wt,roll , P pv,roll They are the output plans of wind power and photovoltaic power, in MW; are the wind power and photovoltaic power generation at time t, in MW; is the combined output plan at time t, in MW.

[0099] ⑶ Considering that as the prediction time scale is shortened, the power prediction error will also decrease. Therefore, the present invention is oriented to the machine-field-group scenario, based on engineering practicality and accuracy, and formulates a multi-level coordinated and step-by-step refined scheduling cycle system, and establishes a multi-time scale joint optimization scheduling model consisting of day-ahead scheduling plan, intra-day rolling plan and real-time scheduling, so as to improve the scheduling accuracy and grid operation stability, and reduce the impact of renewable energy and load uncertainty on the stable and economic operation of the grid.

[0100] like Figure 2 As shown in the figure, the scheduling process of the multi-time scale joint optimization scheduling model is as follows:

[0101] a. In the day-ahead operation stage, based on the wind-solar forecast data and power station parameters, the day-ahead optimization scheduling objective function based on the minimum system operation cost and network loss is constructed to generate the planned output of the wind-solar combined power generation system;

[0102] b. Update the intraday wind and solar forecast data, construct the intraday optimization scheduling objective function based on minimizing the wind abandonment cost and solar abandonment cost, and output the planned output of the wind and solar combined power generation system within the day;

[0103] c. Update the real-time wind and solar forecast data, minimize the total deviation between the output of each station in the machine-field-group and the predetermined instruction as the optimization goal, build a real-time correction model, and output the output plan for the min period;

[0104] If the optimization period within the day is not completed, return to step c;

[0105] If the full-day scheduling is not completed, return to step b.

[0106] (4) The raccoon algorithm is improved by combining the migration factor and the elite selection strategy. The improved raccoon algorithm is used to solve the airport-field-group multi-time scale joint optimization scheduling model to obtain the final optimization scheduling result.

[0107] Among them: The improved raccoon algorithm is performed as follows:

[0108] AInitialization

[0109] The Coati Optimization Algorithm (COA) is a population-based metaheuristic method in which coatis are considered as population members of the algorithm. The position of each raccoon in the search space determines the value of the decision variable. Therefore, in COA, the position of the coatis represents a candidate solution to the problem. At the beginning of the COA implementation, the population is randomly initialized in the optimization space:

[0110] x i,j =lb j +r(ub j -lb j )

[0111] Where: x i,j For individuals; lb j To find the optimal lower boundary; ub j is the upper boundary of optimization; r is a random number between [0,1].

[0112] B Hunting and attacking phase

[0113] The first stage of modeling the update of the raccoon population in the search space is to simulate the strategy of raccoons when attacking iguanas. A group of raccoons climbed up a tree to reach an iguana and scare it, while several other raccoons waited under the tree until the iguana fell to the ground. After the iguana fell to the ground, the raccoons attacked and hunted it.

[0114] In the COA algorithm, the position of the best member of the population is assumed to be the position of the iguana. Assume that half of the raccoons climb the tree and the other half wait for the iguana to fall to the ground. Therefore, the mathematical simulation of the position is:

[0115]

[0116] Where: is the new position of the i-th raccoon in the j-th dimension; r is a random number between [0,1]; Iguana j is the position of the iguana in the jth dimension, which actually refers to the position of the best member; I is a number randomly selected from the set {1,2}; N is the number of raccoons; [N / 2] is the maximum integer not exceeding [N / 2]; m is the number of decision variables.

[0117] After the iguana falls to the ground, it is placed at a random position in the search space. Based on this random position, the raccoon on the ground moves in the search space. This step is simulated by two formulas.

[0118]

[0119] Where: is the position of the iguana on the ground in the jth dimension, where j = 1, 2, ..., m.

[0120]

[0121] Where: is the objective function value after the iguana falls to the ground in the jth dimension; is the objective function value of the i-th raccoon in the j-th dimension.

[0122] If the raccoon remains in the same position during multiple iterations, the algorithm is likely to fall into a local optimum. To address this, the concept of migration factor is introduced. When the raccoon remains in the same position during M iterations of the migration factor, the raccoon is forced to migrate to a nearby random position to increase its search range and to help it escape the local optimum as much as possible.

[0123] In order to record the number of iterations in which the raccoon's current position remains unchanged as the number of iterations increases, the present invention introduces the concept of retention frequency f. If the raccoon's position is still the same as before the iteration after the current iteration, the retention frequency f is increased by 1, otherwise f is set to 0. If the raccoon's position remains unchanged in M ​​consecutive iterations, the raccoon needs to be migrated to a nearby random position outside the current activity range. The specific forced migration conditions are as follows:

[0124]

[0125] Where: Indicates the current location of the raccoon; Represents the random position of the raccoon.

[0126] C. Escape from predator stage

[0127] The second stage of the process of updating the raccoon's position in the search space is mathematically modeled based on the raccoon's natural behavior when encountering and fleeing from a predator. When a predator attacks a raccoon, the raccoon flees its location. A random position is generated near each raccoon's location using the following formula.

[0128]

[0129] Where: is the local lower bound of the j-th decision variable; is the local upper bound of the jth decision variable; t is the number of iterations; T is the maximum number of iterations.

[0130]

[0131] i=1,2,...,N,j=1,2,...,m

[0132] Where: is the new position of the i-th raccoon in the j-th dimension.

[0133] D Optimized selection operation

[0134] Since the optimization goal of the multi-mode dynamic constrained resource allocation scheduling model established by the present invention is to minimize the total cost, a solution with higher fitness means a lower total scheduling cost, and the corresponding solution quality is higher, and the fitness is higher, which is specifically manifested as follows:

[0135]

[0136] Wherein: C(i) represents the size of the objective function value corresponding to the i-th candidate solution among all candidate solutions, that is, the total cost of the scheduling strategy corresponding to the i-th candidate solution of the mathematical model established by the present invention, and f(i) represents the fitness of the candidate solution.

[0137] In the real-time scheduling stage, the present invention combines elite retention and roulette ideas to optimize the selection operation. This not only retains excellent individuals but also greatly enriches the diversity of population genes, so that the algorithm converges in the expected direction as soon as possible. The elite selection strategy is adopted to retain N high-quality candidate solutions. single The specific process is: sort the fitness of each candidate solution in the current activity range, and select the top N with higher fitness. single Individuals are directly copied into the offspring without being inherited, while N single The specific calculation method is:

[0138] N single =(1-G)·N sum

[0139] Where: G represents the generation gap value in the interval (0,1), N sum It reflects the size of the population.

[0140] In order to retain the better individuals in the population, N single Individuals with higher quality are retained using the elite selection strategy, and the remaining G·N sum Each individual uses a roulette wheel to make a decision. First, calculate the fitness of each individual; second, calculate the fitness of the i-th individual Nsingle The retention probability Then, the cumulative probability of each individual is calculated Finally, randomly generate G·N sum A random number in the interval (0,1] is selected based on the number of intervals it falls into. Individuals with a smaller probability of being retained also have a probability of being selected, which greatly enriches the diversity of candidate solutions and also increases the ability of the algorithm to search for better solutions.

[0141]

[0142]

[0143] Where: For individuals The corresponding fitness value; It reflects the individual The probability of being selected and retained; represents the cumulative probability of the i-th individual.

[0144] Example

[0145] Using data from a certain place in the northwest region, taking a machine-farm-group consisting of 7 stations as an example, the total installed capacity of the machine-farm-group is 480MW, of which there are 5 60MW wind farms and 2 90MW photovoltaic power stations, and the network loss cost is 0.80 yuan / (kW·h).

[0146] In the day-ahead dispatch stage, the system operating cost and network loss rate under two different operating conditions are shown in Table 1.

[0147] Table 1 Day-ahead optimization results for different operating conditions

[0148] Scenario Operating cost (yuan) Network loss rate Each station operates independently 436637.99 2.41％ Airport-Airport-Group Joint Operation 437435.08 0.35％

[0149] As can be seen from Table 1, compared with the optimization results of the independent operation system of each station without considering the machine-farm-cluster division, the joint operation cluster optimization can effectively solve the uncertainty and volatility problems of station processing. Cluster joint operation conducts power feedback through mutual cooperation between stations and clusters, and information exchange is carried out within each independent power station, which greatly improves the absorption rate of renewable energy and significantly reduces the operating cost of the system. At the same time, through the joint optimization of machine-farm-cluster, the power flow on the branches of each independent power station is reduced, and the network loss of the system is reduced.

[0150] The improved raccoon algorithm is used to solve the airport-field-group optimization model. The results are as follows: Figure 3 and Figure 4 shown. Figure 3It is the iterative convergence process of the active power output of each power station, where P1-P5 are the outputs of five wind farms, and P6 and P7 are the outputs of two photovoltaic power stations. Figure 4 The process of tracking instructions for the total output of the cluster.

[0151] Depend on Figure 5 It can be seen that during the period when the active output of distributed photovoltaics is less than the active power demand of the load, the system network loss is reduced by reducing the power purchase from the power grid and reducing the active power transmitted to this cluster from other clusters.

[0152] In order to accurately reflect the fluctuations of renewable energy and load and reduce the deviation of day-ahead scheduling, the intraday rolling optimization scheduling strategy is adopted in the intraday optimization scheduling. The cost results of the intraday rolling optimization scheduling are shown in Table 2.

[0153] Table 2 Analysis of intraday optimization results

[0154] time Amount of wind and solar power abandoned / (MWh) Wind and solar power abandonment rate / % Airport-Airport-Group Joint Operation 91 0.8 Each station operates independently 653 1.9

[0155] It can be seen from Table 2 that when the machine-farm-cluster joint operation mode is considered, the system's wind and solar power abandonment rate is significantly reduced. It can be seen that the machine-farm-cluster joint operation effectively reduces the wind and solar power abandonment losses caused by the uncertainty of wind and solar power output during the daily scheduling period, and improves the utilization rate of renewable energy.

[0156] Depend on Figure 6 It can be seen that the introduction of real-time scheduling effectively reduces the deviation of wind-solar combined output caused by the uncertainty of wind-solar forecasting during the day, and the real-time output of the combined power generation system throughout the day deviates from the daily forecast output.

[0157] Based on the above optimization results and comparative analysis, it can be seen that the cluster operation mode is conducive to the local consumption of wind and solar power generation and ensures the economic operation of the system. The dispatch center can constrain the network loss rate and the wind and solar abandonment rates according to demand, and obtain the distributed photovoltaic output and generator output strategy that meets the lowest economic cost.

Claims

1. A multi-level integrated coordinated control method for active and reactive power of a wind and solar power generation cluster, comprising the following steps: ⑴ Establish a multi-level dispatching framework of airport, airport and group; (2) Establish the optimization objectives and constraints of the airport-field-group; ⑶ Establish a multi-time scale joint optimization scheduling model consisting of day-ahead scheduling, intraday rolling planning and real-time scheduling; (4) The raccoon algorithm is improved by combining the migration factor and the elite selection strategy. The improved raccoon algorithm is used to solve the airport-field-group multi-time scale joint optimization scheduling model to obtain the final optimization scheduling result.

2. A multi-level integrated coordinated control method for active and reactive power of a wind and solar power generation cluster as claimed in claim 1, characterized in that: In step (1), the multi-level dispatching architecture of machine-farm-cluster is composed of a unit layer, a station layer, a cluster layer and an upper dispatching layer; the cluster layer is composed of all stations; the dispatching process of the architecture is as follows: the wind / light power prediction data and safety constraints of the unit layer are uploaded to the station layer for optimization, the operation information of each station is uploaded to the cluster layer, the upper dispatching center of the cluster layer receives the operation information of the cluster, and performs optimization calculation in combination with the load prediction data of the power grid and the safety constraints of the cluster; After the optimization is completed, the superior dispatching center will issue dispatching instructions to the cluster, station, and unit step by step.

3. A multi-level integrated coordinated control method for active and reactive power of a wind and solar power generation cluster as claimed in claim 1, characterized in that: The establishment of the optimization objectives and constraints of the airport-airport-group in step (2) consists of a day-ahead operation phase, an intra-day operation phase, and a real-time scheduling phase; ① In the day-ahead operation stage, the optimization goal is to minimize the system operation cost and network loss, and its objective function is: min C1=C wt +C pv +C loss Where: C wt is the operating cost of the wind farm, in RMB; C pv is the operating cost of the photovoltaic power station, in yuan; C loss is the network loss cost, in RMB; the specified time length is divided into N using Δt as the scale T time period; is the cost coefficient of the i-th wind farm; is the cost coefficient of the i-th photovoltaic power station; is the power generation of the i-th wind farm at time t, in MW; is the power generation of the i-th PV power station at time t, in MW; The constraints are composed of power balance constraints, wind-solar operation constraints, wind-solar ramp constraints and station output regulation rate constraints; among which: iPower balance constraints: Where: P L,t is the electrical load, in MW; ii. Wind and solar operation constraints: Where: P wt is the output of the wind farm, in MW; P pv is the output of the photovoltaic power station, in MW; is the maximum and minimum output of the wind farm, in MW; The maximum and minimum output of the photovoltaic power station, in MW; iii. Wind and solar climbing constraints: Where: w,t is the wind power ramp rate; min,t , max,t is the minimum and maximum ramp rate of wind power; Δt is the time interval between two wind power values ​​at time t, in hours; is the upper limit of photovoltaic power, P pv,t is the photovoltaic power generation power at time t, P pv,t-1 is the photovoltaic power generation power at time t-1, in MW; P is the limit value of photovoltaic down and up climbing rate; w,t is the wind power generation power at time t, P w,t+1 is the wind power generation power at time t+1, in MW; ⅳ Station output regulation rate constraints: Where: α w is the power change rate coefficient of the wind farm; α pv is the power change rate coefficient of the photovoltaic power station; and are the power generation of the i-th wind farm and the j-th photovoltaic power station at time t, in MW; and are the changing power of the wind farm and the changing power of the photovoltaic power station during the period t-Δt, in MW; P w,i and P pv,j are the power generation of the i-th wind farm and the j-th photovoltaic power station, in MW; ② Intraday operation phase, to minimize the wind curtailment cost C wt,loss and the abandoned light cost C pv,loss To optimize the target, the objective function is: Where: K wt,loss , K pv,loss are the penalty coefficients for wind and solar power abandonment respectively; They are the abandoned wind and solar power, in MW; The constraints for curtailing wind and solar power are: Where: They are wind power forecast output and photovoltaic forecast output, in MW; They are wind power abandonment and solar power abandonment, in MW; They are wind power actual power and photovoltaic actual power, in MW; ③ In the real-time scheduling stage, the total deviation between the output of each station of the airport-field-group and the scheduled instructions is minimized as the optimization goal, and the influence of the prediction error is taken into account at the same time; its objective function is: Where: M is the number of time periods in the real-time scheduling phase; are the power forecast values ​​of wind power and photovoltaic power, in MW; It is the planned power generation output of the joint operation of the machine, farm and cluster, in MW; The constraints for airport-airport-group joint operation are: Where: P dev is the offset power during joint operation of airport, airport and cluster, in MW; P wt,roll , P pv,roll They are the output plans of wind power and photovoltaic power, in MW; are the wind power and photovoltaic power generation at time t, in MW; is the combined output plan at time t, in MW.

4. A multi-level integrated coordinated control method for active and reactive power of a wind and solar power generation cluster as claimed in claim 1, characterized in that: The scheduling process of the multi-time scale joint optimization scheduling model in step (3) is as follows: a. In the day-ahead operation stage, based on the wind-solar forecast data and power station parameters, the day-ahead optimization scheduling objective function based on the minimum system operation cost and network loss is constructed to generate the planned output of the wind-solar combined power generation system; b. Update the intraday wind and solar forecast data, construct the intraday optimization scheduling objective function based on minimizing the wind abandonment cost and solar abandonment cost, and output the planned output of the wind and solar combined power generation system within the day; c. Update the real-time wind and solar forecast data, minimize the total deviation between the output of each station in the machine-field-group and the predetermined instruction as the optimization goal, build a real-time correction model, and output the output plan for the min period; If the optimization period within the day is not completed, return to step c; If the full-day scheduling is not completed, return to step b.

Images