Photovoltaic power station active power optimal distribution method for ultra-short-term photovoltaic power prediction

Abstract

The application relates to the field of new energy power generation and energy storage, and discloses a photovoltaic power station active power optimization distribution method for ultra-short-term photovoltaic power prediction. The application solves the problem that, in the existing photovoltaic power station power generation control method, when the energy storage system cannot consume the remaining power after the power frequency is adjusted, the photovoltaic power station itself is completely used for power adjustment, energy is discarded, and economic loss is serious. The application uses a PSO-LSTM neural network to perform ultra-short-term photovoltaic power prediction, uses a prediction result, combines energy storage charging and discharging capacity, and calculates the power P r consumed by the photovoltaic power station and the power P ur uploaded to the power grid, so as to optimize the distribution of the predicted active power output by the photovoltaic power station with the energy storage system. The photovoltaic power station and the power grid are combined to participate in power adjustment, the photovoltaic power station is classified, and a deviation penalty is introduced to reduce economic loss of the photovoltaic power station. The application is mainly used for controlling photovoltaic power station power generation.

Description

Technical Field

[0001] This invention relates to the field of new energy power generation and energy storage, and specifically to a novel AGC strategy for photovoltaic power plants with energy storage configuration. Background Technology

[0002] Solar energy is a clean energy source with randomness and volatility, so the output power of photovoltaic (PV) power generation will vary due to factors such as weather conditions, seasonal changes, and cloud cover. This randomness and volatility can lead to insufficient ability of PV power plants to respond to grid frequency regulation commands, thus affecting grid stability and frequency quality. To ensure system active power balance and frequency quality, Automatic Generation Control (AGC) has become one of the important means to ensure stable grid operation. With the development of my country's new energy industry, multi-energy systems such as photovoltaic-storage combined systems have also been introduced into the power system. When allocating active power using existing PV power plant generation control methods, if the predicted PV power exceeds the frequency regulation power, and the energy storage system cannot absorb the remaining capacity power after meeting the frequency regulation power, there are two methods for power absorption. First, from the perspective of the PV power plant, the PV power plant itself participates in power regulation, actively reducing its output power for power absorption. This method is the mainstream power regulation method, but it suffers from energy waste, serious economic losses, and poor power allocation effect. Secondly, from the perspective of the entire power grid, the energy generated by the remaining capacity after meeting the frequency regulation power can be uploaded to the power grid. However, uploading the remaining energy to the power grid will cause instability in the power grid frequency. The power grid needs to coordinate with other frequency regulation equipment to stabilize the frequency. The more electrical energy uploaded to the power grid, the higher the cost of stabilizing the power grid frequency. Therefore, this frequency regulation method is not used in the current technology. The mainstream method is the first one for power absorption. Summary of the Invention

[0003] The purpose of this invention is to address the problem in existing photovoltaic (PV) power generation control methods where, when the energy storage system cannot absorb the surplus power exceeding the frequency regulation power, the PV power plant itself is solely responsible for power regulation, resulting in significant energy waste and economic losses. This invention provides a method for optimizing the allocation of active power in PV power plants based on ultra-short-term PV power prediction.

[0004] An optimal allocation method for active power in photovoltaic power plants based on ultra-short-term photovoltaic power prediction is used to optimize the allocation of the predicted active power output of photovoltaic power plants with energy storage systems. The optimization allocation method is implemented based on the AGC controller of the photovoltaic power plant and includes the following steps:

[0005] S1. The AGC controller of the photovoltaic power station is used to receive the active power output of the photovoltaic power station at the predicted time, which is obtained by the trained photovoltaic power prediction model. and the given frequency regulation power issued by the power dispatch center The photovoltaic power prediction model is implemented using a PSO-LSTM neural network.

[0006] Determine if it satisfies If the result is yes, proceed to step S2; if the result is no, proceed to step S4.

[0007] S2. Determine the energy storage and charging capacity ΔP of the energy storage system. ch And calculate excess power Determine the energy storage and charging capacity ΔP of the energy storage system ch Does it meet the requirements? The result is that the energy storage system absorbs the excess power. At this point, there is no need to adjust the photovoltaic power station to output photovoltaic power. The result is negative, so proceed to step S3.

[0008] S3, Calculate the power difference Excess power First, the energy storage system absorbs ΔP. ch Then, by adjusting the output of the photovoltaic power station and the power grid to jointly absorb ΔP1, the optimal allocation of the active power output of the photovoltaic power station is completed.

[0009] When adjusting the output of the photovoltaic power plant and the power transmitted to the grid to be absorbed by the grid, the power P absorbed by the photovoltaic power plant is calculated with the goal of minimizing the penalty cost F. r and the power P uploaded to the grid ur Specifically:

[0010]

[0011] ρ1 and ρ2 are the electricity sales price and the curtailment penalty factor for solar power, respectively. r =P Ⅰ +P Ⅱ +P Ⅲ ;P Ⅰ P Ⅱ and P Ⅲ These represent the power absorbed by the first to third types of photovoltaic arrays in a photovoltaic power plant, and the three types of photovoltaic arrays have different temperatures;

[0012] S4. Determine the discharge capacity ΔP of the energy storage system. dis And calculate the power deficit. Determine the discharge capacity ΔP of the energy storage system dis Does ΔP satisfy? dis ≥ΔP′ PV The result is that the energy storage system supplies power to the photovoltaic power plant to compensate for the power deficit. The result is negative; the power deficit ΔP′ is supplied to the photovoltaic power station through a combination of energy storage system and power plant. PVFurthermore, the energy storage system supplies ΔP to the photovoltaic power plant. dis This allows for the optimized allocation of active power output from the photovoltaic power station.

[0013] Preferably, the process of obtaining the trained photovoltaic power prediction model includes:

[0014] S11. Construct a training set based on historical data of photovoltaic power plants. The input data of each training sample in the training set includes the active power output of the photovoltaic power plant at each sampling point within any consecutive N hours, as well as the average temperature and wind speed within that consecutive N hours. The output data includes the active power output of the photovoltaic power plant at time M after N consecutive hours.

[0015] S12. Train the photovoltaic power prediction model using the training set to obtain the trained photovoltaic power prediction model; wherein, the input data and output data of the training samples are used as the input and output of the photovoltaic power prediction model, respectively.

[0016] Preferably, in step S11, a sampling point is formed at 15-minute intervals, and a total of 4N sampling points are formed over N consecutive hours.

[0017] Preferably, the active power output of the photovoltaic power station at the time to be predicted is predicted using the trained photovoltaic power prediction model. The implementation method is as follows:

[0018] The output power of the photovoltaic power station at each sampling point within a consecutive N hours of historical data corresponding to the time to be predicted, along with the average temperature and wind speed within that consecutive N hours, are input into the trained photovoltaic power prediction model to predict the active power output of the photovoltaic power station at the time to be predicted.

[0019] Preferably, in step S4, the power deficit ΔP′ is supplied to the photovoltaic power station jointly by the energy storage system and the power plant. PV The implementation method is as follows:

[0020] Energy storage system supplies ΔP to photovoltaic power plant dis The power plant supplies ΔP′ to the photovoltaic power station PV -ΔP dis .

[0021] Preferably, in photovoltaic power plants, photovoltaic arrays with 10°≤T<25° are defined as Class I photovoltaic arrays;

[0022] Photovoltaic arrays with 25°≤T<40° are defined as Class II photovoltaic arrays;

[0023] Photovoltaic arrays with 40°≤T are defined as Class III photovoltaic arrays.

[0024] Preferably, PⅠ =λ1P r P Ⅱ =λ2P r P Ⅲ =λ3P r , and λ1+λ2+λ3=1;

[0025] Wherein, λ1, λ2, and λ3 are the absorption coefficients of the first to third types of photovoltaic arrays, respectively.

[0026] Preferably, the charging and discharging power constraints of the energy storage system are as follows:

[0027] When the energy storage system is charging, it meets the requirements.

[0028]

[0029] When the energy storage system discharges, it meets the following requirements:

[0030]

[0031] In the formula: and These are the maximum charging and discharging power of the energy storage system, respectively; P ch (t) and P dis (t) represents the charging and discharging power of the energy storage system at time t; σ is the self-discharge rate of the energy storage system; η c η d These represent the charging and discharging efficiencies of the energy storage system, respectively; E max E represents the maximum capacity of the energy storage system. min E represents the minimum capacity of the energy storage system. ess (t-1) represents the remaining power of the energy storage system at time t-1, and Δt is the time difference between time t and time t-1.

[0032] Preferably, the state-of-charge constraints of the energy storage system are as follows:

[0033] SOC min ≤SOC(t)≤SOC max ;

[0034]

[0035] Among them, SOC min SOC max These are the lower and upper limits of the state of charge (SOC) of the energy storage system, respectively. SOC(t) is the SOC of the energy storage system at time t, and σ is the self-discharge rate of the energy storage system. c η d These represent the charge and discharge efficiencies of the energy storage system; w c (t), w d (t) represents the on / off and relational numbers of the energy storage system at time t, respectively.bess (t) represents the power of the energy storage system in its charging or discharging state at time t, and P bess (t)>0 indicates the power of the energy storage system in the charging state at time t, P bess (t) < 0 indicates the power of the energy storage system in the discharge state at time t.

[0036] Preferably, the expression for the State of Health (SOH) of the battery in the energy storage system is:

[0037]

[0038] Among them, Q max C represents the current maximum discharge capacity of the battery. rated This refers to the battery's capacity at the time of manufacture.

[0039] Advantages of this invention:

[0040] The proposed method for optimizing the active power allocation of photovoltaic power plants based on ultra-short-term photovoltaic power prediction utilizes a particle swarm optimization long short-term memory (PSO-LSTM) neural network for ultra-short-term photovoltaic power prediction. The prediction results, combined with energy storage charging and discharging capabilities, and with the goal of minimizing the penalty cost F, are used to calculate the power P that the photovoltaic power plant can absorb. r and the power P uploaded to the grid ur It is used to optimize the allocation of the predicted active power output of a photovoltaic power station with an energy storage system.

[0041] This invention employs a combined approach of photovoltaic (PV) power plant and grid connection for power regulation, introducing PV array classification and deviation penalties to reduce economic losses for the PV power plant. Furthermore, given a fixed PV power plant output, it rationally allocates the output of different types of PV arrays, further minimizing economic losses. This solves the problem of energy waste and severe economic losses caused by PV power plant-only power regulation in existing technologies.

[0042] Further research will be conducted on the energy storage constraints and lifespan of energy storage systems in photovoltaic power plants with energy storage, thereby further improving the economic efficiency of photovoltaic power plant operation. Attached Figure Description

[0043] Figure 1 This is a flowchart of the photovoltaic power plant active power optimization allocation method for ultra-short-term photovoltaic power prediction as described in this invention;

[0044] Figure 2 This is a schematic diagram of the AGC controller network structure;

[0045] Figure 3 This is a graph of the output power of a photovoltaic power plant predicted using a PSO-LSTM neural network;

[0046] Figure 4 It is a power output diagram of the photovoltaic power station and energy storage system after receiving a given frequency modulation power;

[0047] Figure 5 This is a diagram showing the output power of various photovoltaic arrays in a photovoltaic power station;

[0048] Figure 6 This is a cost comparison chart between the method of the present invention and the existing technology that relies solely on photovoltaic power plants for power generation regulation. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0051] The purpose of this invention is to propose a photovoltaic power station with energy storage configuration that responds to a given frequency regulation power from the grid. A method for optimizing the allocation of active power in photovoltaic (PV) power plants is proposed. First, a particle swarm optimization long short-term memory (PSO-LSTM) neural network is used for ultra-short-term PV power prediction. Then, the prediction results, combined with energy storage charging and discharging capabilities, and with the goal of minimizing the penalty cost F, are used to calculate the power P that the PV power plant can absorb. r and the power P uploaded to the grid ur This method is used to optimize the allocation of the predicted active power output of a photovoltaic power plant with an energy storage system. Furthermore, an energy storage constraint model and a lifespan model for the energy storage system in a photovoltaic power plant with energy storage are established. Finally, a numerical example analysis demonstrates the effectiveness and feasibility of the strategy. The implementation of this strategy can effectively improve the economic efficiency of photovoltaic power plant operation.

[0052] See Figure 1 and Figure 2 This embodiment describes a method for optimizing the allocation of active power in a photovoltaic power plant based on ultra-short-term photovoltaic power prediction. This method is used to optimize the allocation of the predicted active power output of a photovoltaic power plant containing an energy storage system. The optimization allocation method is implemented based on the AGC controller of the photovoltaic power plant and includes the following steps:

[0053] S1. The AGC controller of the photovoltaic power station is used to receive the active power output of the photovoltaic power station at the predicted time, which is obtained by the trained photovoltaic power prediction model. and the given frequency regulation power issued by the power dispatch center The photovoltaic power prediction model is implemented using a PSO-LSTM neural network.

[0054] Determine if it satisfies If the result is yes, proceed to step S2; if the result is no, proceed to step S4.

[0055] S2. Determine the energy storage and charging capacity ΔP of the energy storage system. ch And calculate excess power Determine the energy storage and charging capacity ΔP of the energy storage system ch Does it meet the requirements? The result is that the energy storage system absorbs the excess power. At this point, there is no need to adjust the photovoltaic power station to output photovoltaic power. The result is negative, so proceed to step S3.

[0056] S3, Calculate the power difference Excess power First, the energy storage system absorbs ΔP. ch Then, by adjusting the output of the photovoltaic power station and the power grid to jointly absorb ΔP1, the optimal allocation of the active power output of the photovoltaic power station is completed.

[0057] When adjusting the output of the photovoltaic power plant and the power transmitted to the grid to be absorbed by the grid, the power P absorbed by the photovoltaic power plant is calculated with the goal of minimizing the penalty cost F. r and the power P uploaded to the grid ur Specifically:

[0058]

[0059] ρ1 and ρ2 are the electricity sales price and the curtailment penalty factor for solar power, respectively. r =P Ⅰ +P Ⅱ +P Ⅲ ;P Ⅰ P Ⅱ and P Ⅲ These represent the power absorbed by the first to third types of photovoltaic arrays in a photovoltaic power plant, and the three types of photovoltaic arrays have different temperatures;

[0060] S4. Determine the discharge capacity ΔP of the energy storage system. dis And calculate the power deficit. Determine the discharge capacity ΔP of the energy storage system dis Does ΔP satisfy? dis ≥ΔP′ PV The result is that the energy storage system supplies power to the photovoltaic power plant to compensate for the power deficit. The result is negative; the power deficit ΔP′ is supplied to the photovoltaic power station through a combination of energy storage system and power plant. PV Furthermore, the energy storage system supplies ΔP to the photovoltaic power plant.dis This allows for the optimized allocation of active power output from the photovoltaic power station.

[0061] Furthermore, in step S4, the power deficit ΔP′ is supplied to the photovoltaic power station through the energy storage system and the power plant together. PV The implementation method is as follows: the energy storage system supplies ΔP to the photovoltaic power station. dis The power plant supplies ΔP′ to the photovoltaic power station PV -ΔP dis .

[0062] In practical applications, the root mean squared error (RMSE) and mean absolute error (MAE), commonly used objective evaluation metrics for photovoltaic power prediction models, are employed to fully assess the algorithm's performance.

[0063]

[0064]

[0065] In the formula: y i This represents the true power value of the i-th sample. denoted as the predicted power of the i-th sample; N is the number of samples. The smaller the MAE and RMSE indices, the better the prediction effect of the algorithm.

[0066] For photovoltaic power plants, when the light intensity is constant, the relationship between the output power of photovoltaic panels and their output voltage exhibits a certain nonlinear characteristic, and there is a unique maximum power point. As the temperature increases, the maximum output power tends to decrease. The output power of photovoltaic power can be controlled by adjusting the output voltage of the photovoltaic array.

[0067] (one)

[0068] When the total predicted power of the photovoltaic power station Greater than or equal to the given frequency regulation power issued by the power dispatch center At this point, the output power of the photovoltaic power station needs to be reduced to meet the frequency regulation power requirement; the excess power to be reduced is...

[0069]

[0070]

[0071] The predicted power of each photovoltaic array;

[0072] For photovoltaic power plants with energy storage, if the energy storage charging capacity... Energy storage can then absorb the remaining capacity after frequency regulation, without needing to adjust photovoltaic output. Otherwise, a power difference ΔP1 will occur.

[0073]

[0074] If ΔP1 is entirely regulated by the photovoltaic power station, it will result in significant curtailment and increase the cost of curtailment penalties. If ΔP1 is not regulated, it will result in deviation penalties. Therefore, a penalty cost objective function is introduced to control the output power of the photovoltaic power station, balance curtailment and deviation penalties, and achieve optimal economic benefits.

[0075]

[0076] ΔP1=P r +P ur (7);

[0077] In the formula: F is the penalty cost; ρ1 and ρ2 are the electricity sales price and the curtailment penalty factor, respectively; P r The power absorbed by the photovoltaic power station, also known as the regulation capacity of the photovoltaic power station; P ur This refers to the power uploaded to the grid, which is also known as the unregulated power of the photovoltaic power plant.

[0078] When adjusting the output power of photovoltaic arrays, in order to improve the power generation efficiency of the photovoltaic arrays and prevent the temperature of high-temperature photovoltaic arrays from increasing further, the adjustment power of each photovoltaic array should be reasonably allocated. Photovoltaic arrays with a temperature T of 10° ≤ T < 25° are classified as Class I, those with a temperature T of 25° ≤ T < 40° are classified as Class II, and those with a temperature T of 40° ≤ T are classified as Class III.

[0079] P r =P Ⅰ +P Ⅱ +P Ⅲ (8);

[0080]

[0081] In the formula: P Ⅰ P Ⅱ P Ⅲ These represent the power absorbed by photovoltaic arrays of types I, II, and III, respectively; in specific applications, λ1 = 0.7, λ2 = 0.2, and λ3 = 0.1. λ1, λ2, and λ3 are the absorption coefficients for photovoltaic arrays of types I to III, respectively.

[0082] (two)

[0083] When the frequency modulation power Greater than the total predicted power of the photovoltaic power plant At this point, it is necessary to utilize the discharge capacity of energy storage and coordinate with other power plants to participate in frequency regulation to fill the power deficit, which is ΔP′. PV .

[0084]

[0085] If the energy storage and discharge capacity ΔP dis ≥ΔP′ PV If so, energy storage can be used to fill the power deficit. Otherwise, other power plants need to be coordinated to participate in frequency regulation to fill the remaining power deficit, which is ΔP2:

[0086] ΔP2=ΔP′ PV -ΔP dis (12);

[0087] The process of obtaining the trained photovoltaic power prediction model for application includes:

[0088] S11. Construct a training set based on historical data of photovoltaic power plants. The input data of each training sample in the training set includes the active power output of the photovoltaic power plant at each sampling point within any consecutive N hours, as well as the average temperature and wind speed within that consecutive N hours. The output data includes the active power output of the photovoltaic power plant at time M after N consecutive hours.

[0089] Specifically, a sampling point is formed at 15-minute intervals, and a total of 4N sampling points are formed over N consecutive hours. The maximum time interval between time M and N hours is 4 hours. Industry standards define predicting the next 24 hours and sampling at 1-hour intervals as short-term; predicting the next 4 hours and sampling at 15-minute intervals is called ultra-short-term.

[0090] S12. Train the photovoltaic power prediction model using the training set to obtain the trained photovoltaic power prediction model; wherein, the input data and output data of the training samples are used as the input and output of the photovoltaic power prediction model, respectively.

[0091] In application, the trained photovoltaic power prediction model is used to predict the active power output of the photovoltaic power station at the time to be predicted. The implementation method is as follows:

[0092] The output power of the photovoltaic power station at each sampling point within a consecutive N hours of historical data corresponding to the time to be predicted, along with the average temperature and wind speed within that consecutive N hours, are input into the trained photovoltaic power prediction model to predict the active power output of the photovoltaic power station at the time to be predicted.

[0093] Constraints of energy storage systems:

[0094] In the process of absorbing and supplementing photovoltaic power, the basic characteristics of the energy storage system itself lead to certain limitations in its state of charge, charging and discharging capacity, and charging and discharging power. In order to enable the energy storage system to better absorb and supplement photovoltaic power and extend the service life of the energy storage system, this invention adds some constraints to the energy storage system.

[0095] (A) The charging and discharging power constraint of the energy storage system mainly depends on the maximum charging and discharging power of the energy storage system. At time t, the maximum charging and discharging power of the energy storage system is determined by the remaining energy and characteristics of the energy storage system. Therefore, the charging and discharging power constraint condition of the energy storage system is:

[0096] When the energy storage system is charging, it meets the requirements.

[0097]

[0098] When the energy storage system discharges, it meets the following requirements:

[0099]

[0100] In the formula: and These are the maximum charging and discharging power of the energy storage system, respectively; P ch (t) and P dis (t) represents the charging and discharging power of the energy storage system at time t; σ is the self-discharge rate of the energy storage system; η c η d These represent the charging and discharging efficiencies of the energy storage system, respectively; E max E represents the maximum capacity of the energy storage system. min E represents the minimum capacity of the energy storage system. ess (t-1) represents the remaining power of the energy storage system at time t-1, and Δt is the time difference between time t and time t-1.

[0101] (B) To optimize the charging and discharging process of the energy storage system and avoid the impact of excessively high or low state of charge (SOC), constraints are required. Therefore, the SOC of the energy storage system should meet the following constraints:

[0102] SOC min ≤SOC(t)≤SOC max (14);

[0103]

[0104] Among them, SOC min SOC max These are the lower and upper limits of the state of charge (SOC) of the energy storage system, respectively. SOC(t) is the SOC of the energy storage system at time t, and σ is the self-discharge rate of the energy storage system. c η dThese represent the charge and discharge efficiencies of the energy storage system; w c (t), w d (t) represents the on / off and relational numbers of the energy storage system at time t, respectively. bess (t) represents the power of the energy storage system in its charging or discharging state at time t, and P bess (t)>0 indicates the power of the energy storage system in the charging state at time t, P bess (t) < 0 indicates the power of the energy storage system in the discharge state at time t.

[0105] Energy storage system life monitoring:

[0106] Because energy storage systems are constantly charging and discharging, battery health status testing is necessary. There are many definitions of battery health status (SOH), but the most common definition is based on capacity.

[0107]

[0108] In the formula: C aged C represents the current capacity of the battery. rated This indicates the battery's capacity at the time of manufacture.

[0109] In practice, battery capacity is difficult to measure directly, especially for lithium-ion batteries. Therefore, this invention redefines the state of harmonics (SOH) of a battery by using its maximum discharge capacity instead of its capacity.

[0110]

[0111] This definition method is more in line with the usage of energy storage devices, where Q max C represents the current maximum discharge capacity of the battery. rated This refers to the battery's capacity at the time of manufacture.

[0112] Verification experiment:

[0113] To verify the proposed photovoltaic power prediction model implemented using a PSO-LSTM neural network, AGC (Automatic Generation Control) was performed using actual data from the Delingha photovoltaic power station. The photovoltaic power station has a capacity of 40MW and an energy storage capacity of 2.5MW / 5MWh. The photovoltaic power station participates in system regulation at t=0s. First, the photovoltaic power prediction model implemented using the PSO-LSTM neural network is used to predict the photovoltaic power generation of the photovoltaic power station in the next 15 minutes. Then, AGC is performed in conjunction with the frequency regulation command (i.e., the given frequency regulation power) received by the photovoltaic power station within 300s.

[0114] Figure 3 The output power of the photovoltaic power plant is predicted using the PSO-LSTM neural network. Figure 4 Is using Figure 3The output diagram of the photovoltaic power station and energy storage after receiving a frequency regulation command (i.e., given frequency regulation power) within a certain period of time, as predicted by the PSO-LSTM neural network.

[0115] Figure 4 The three curves shown are the photovoltaic power forecast curve, photovoltaic output curve, and energy storage output curve for the next 300 seconds. Figure 4 The curve for the frequency modulation command (i.e., the given frequency modulation power) is not shown in the diagram. Figure 4 At 90s, 120s, and 150s, there is a deviation between the photovoltaic predicted power and the frequency regulation command (i.e., the given frequency regulation power). At this time, it is necessary to change the output voltage of the photovoltaic power station to adjust the output power of the photovoltaic power station (i.e., the output of the photovoltaic array, also called photovoltaic output) to meet the frequency regulation command.

[0116] Figure 5 The output power diagrams of various photovoltaic arrays in a photovoltaic power plant under power generation regulation using the method of this invention are presented. Figure 5 The dashed line represents the adjustment amount for three types of photovoltaic arrays. Figure 5 The solid line represents the predicted power of three types of photovoltaic arrays; Figure 6 Given Figure 5 The diagram shows a cost comparison between the method of the present invention and the existing technology that relies solely on photovoltaic power plants for power generation regulation.

[0117] Using various photovoltaic arrays of the present invention in Figure 5 Under the output conditions shown, the economic optimum can be achieved under frequency modulation commands (i.e., given frequency modulation power). Figure 6 The left-hand bar chart shows the cost of regulating power generation using traditional methods, without classifying photovoltaic arrays, and only adjusting the output of the photovoltaic arrays. Figure 6 The left-hand bar chart shows the cost of regulation using three types of photovoltaic arrays under the proposed method of joint regulation of photovoltaic arrays and the upstream grid. Figure 6 A comparison of the two bar charts shows that the proposed method of joint regulation of photovoltaic arrays and grid power supply can reduce costs by 11.1% through regulation using three types of photovoltaic arrays.

[0118] While the invention has been described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the invention. Therefore, it should be understood that many modifications can be made to the exemplary embodiments, and other arrangements can be designed without departing from the spirit and scope of the invention as defined by the appended claims. It should be understood that different dependent claims and features described herein can be combined in ways different from those described in the original claims. It is also understood that features described in conjunction with individual embodiments can be used in other described embodiments.

Claims

1. A method for optimizing the allocation of active power in photovoltaic power plants based on ultra-short-term photovoltaic power prediction, used to optimize the allocation of the predicted active power output of photovoltaic power plants including energy storage systems, characterized in that... The optimized allocation method is based on the AGC controller of the photovoltaic power plant and includes the following steps: S1. The AGC controller of the photovoltaic power station is used to receive the active power output of the photovoltaic power station at the predicted time, which is obtained by the trained photovoltaic power prediction model. and the given frequency regulation power issued by the power dispatch center The photovoltaic power prediction model is implemented using a PSO-LSTM neural network. Determine if it satisfies If the result is yes, proceed to step S2; if the result is no, proceed to step S4. S2. Determine the energy storage and charging capacity ΔP of the energy storage system. ch And calculate excess power Determine the energy storage and charging capacity ΔP of the energy storage system ch Does it meet the requirements? The result is that the energy storage system absorbs the excess power. At this point, there is no need to adjust the photovoltaic power station to output photovoltaic power. The result is negative, so proceed to step S3. S3, Calculate the power difference Excess power First, the energy storage system absorbs ΔP. ch Then, by adjusting the output of the photovoltaic power station and the power grid to jointly absorb ΔP1, the optimal allocation of the active power output of the photovoltaic power station is completed. When adjusting the output of the photovoltaic power plant and the power transmitted to the grid to be absorbed by the grid, the power P absorbed by the photovoltaic power plant is calculated with the goal of minimizing the penalty cost F. r and the power P uploaded to the grid ur Specifically: ρ1 and ρ2 are the electricity sales price and the curtailment penalty factor for solar power, respectively. r =P Ⅰ +P Ⅱ +P Ⅲ ;P Ⅰ P Ⅱ and P Ⅲ These represent the power absorbed by the first to third types of photovoltaic arrays in a photovoltaic power plant, and the three types of photovoltaic arrays have different temperatures; S4. Determine the discharge capacity ΔP of the energy storage system. dis And calculate the power deficit. Determine the discharge capacity ΔP of the energy storage system dis Does ΔP satisfy? dis ≥ΔP′ PV The result is that the energy storage system supplies power to the photovoltaic power plant to compensate for the power deficit. The result is negative; the power deficit ΔP′ is supplied to the photovoltaic power station through a combination of energy storage system and power plant. PV Furthermore, the energy storage system supplies ΔP to the photovoltaic power station. dis This allows for the optimized allocation of active power output from the photovoltaic power station.

2. The method for optimizing the allocation of active power in a photovoltaic power plant based on ultra-short-term photovoltaic power prediction according to claim 1, characterized in that, The process of obtaining the trained photovoltaic power prediction model includes: S11. Construct a training set based on historical data of photovoltaic power plants. The input data of each training sample in the training set includes the active power output of the photovoltaic power plant at each sampling point within any consecutive N hours, as well as the average temperature and wind speed within that consecutive N hours. The output data includes the active power output of the photovoltaic power plant at time M after N consecutive hours. S12. Train the photovoltaic power prediction model using the training set to obtain the trained photovoltaic power prediction model; wherein, the input data and output data of the training samples are used as the input and output of the photovoltaic power prediction model, respectively.

3. The method for optimizing the allocation of active power in a photovoltaic power plant based on ultra-short-term photovoltaic power prediction according to claim 2, characterized in that, In step S11, a sampling point is formed at 15-minute intervals, and a total of 4N sampling points are formed over N consecutive hours.

4. The method for optimizing the allocation of active power in a photovoltaic power plant based on ultra-short-term photovoltaic power prediction according to claim 1, characterized in that, The trained photovoltaic power prediction model predicts the active power output of the photovoltaic power station at the time to be predicted. The implementation method is as follows: The output power of the photovoltaic power station at each sampling point within a consecutive N hours of historical data corresponding to the time to be predicted, along with the average temperature and wind speed within that consecutive N hours, are input into the trained photovoltaic power prediction model to predict the active power output of the photovoltaic power station at the time to be predicted.

5. The method for optimizing the allocation of active power in a photovoltaic power plant based on ultra-short-term photovoltaic power prediction according to claim 1, characterized in that, In step S4, the power deficit ΔP′ is supplied to the photovoltaic power station through the energy storage system and the power plant. PV The implementation method is as follows: Energy storage system supplies ΔP to photovoltaic power plant dis The power plant supplies ΔP′ to the photovoltaic power station PV -ΔP dis .

6. The method for optimizing the allocation of active power in a photovoltaic power plant based on ultra-short-term photovoltaic power prediction according to claim 1, characterized in that, In photovoltaic power plants, photovoltaic arrays with 10°≤T<25° are defined as Class I photovoltaic arrays; Photovoltaic arrays with 25°≤T<40° are defined as Class II photovoltaic arrays; Photovoltaic arrays with 40°≤T are defined as Class III photovoltaic arrays.

7. The method for optimizing the allocation of active power in a photovoltaic power plant based on ultra-short-term photovoltaic power prediction according to claim 1, characterized in that, P Ⅰ =λ1P r 、P Ⅱ =λ2P r 、P Ⅲ =λ3P r , and λ1+λ2+λ3=1; Wherein, λ1, λ2, and λ3 are the absorption coefficients of the first to third types of photovoltaic arrays, respectively.

8. The method for optimizing the allocation of active power in a photovoltaic power plant based on ultra-short-term photovoltaic power prediction according to claim 1, characterized in that, The power constraints for charging and discharging of the energy storage system are as follows: When the energy storage system is charging, it meets the requirements. When the energy storage system discharges, it meets the following requirements: In the formula: and These are the maximum charging and discharging power of the energy storage system, respectively; P ch (t) and P dis (t) represents the charging and discharging power of the energy storage system at time t; σ is the self-discharge rate of the energy storage system; η c η d These represent the charging and discharging efficiencies of the energy storage system, respectively; E max E represents the maximum capacity of the energy storage system. min E represents the minimum capacity of the energy storage system. ess (t-1) represents the remaining power of the energy storage system at time t-1, and Δt is the time difference between time t and time t-1.

9. The method for optimizing the allocation of active power in a photovoltaic power plant based on ultra-short-term photovoltaic power prediction according to claim 1, characterized in that, The state-of-charge constraints of the energy storage system are: SOC min ≤SOC(t)≤SOC max ; Among them, SOC min SOC max These are the lower and upper limits of the state of charge (SOC) of the energy storage system, respectively. SOC(t) is the SOC of the energy storage system at time t, and σ is the self-discharge rate of the energy storage system. c η d These represent the charge and discharge efficiencies of the energy storage system; w c (t), w d (t) represents the on / off and relational numbers of the energy storage system at time t, respectively. bess (t) represents the power of the energy storage system in its charging or discharging state at time t, and P bess (t)>0 indicates the power of the energy storage system in the charging state at time t, P bess (t) < 0 indicates the power of the energy storage system in the discharge state at time t.

10. The method for optimizing the allocation of active power in a photovoltaic power plant based on ultra-short-term photovoltaic power prediction according to claim 1, characterized in that, The expression for the State of Health (SOH) of a battery in an energy storage system is: Among them, Q max C represents the current maximum discharge capacity of the battery. rated This refers to the battery's capacity at the time of manufacture.

Images