Scheduling method of virtual power plant and related device

Abstract

The disclosure provides a scheduling method of a virtual power plant and related equipment. Specifically, the scheduling method comprises obtaining a price constraint condition, load demand data and state data of the power generation unit; determining a first output power of each power generation unit by using a trained deep Q network according to the state data and the load demand data; determining a second output power of each power generation unit by using a trained deep Q network according to the first output power and the price constraint condition; wherein the deep Q network comprises a first deep Q network and a second deep Q network; wherein the first deep Q network is configured to determine the first output power; and the second deep Q network is configured to determine the second output power. The deep Q network is used to fully analyze the load demand data, the operating state of the power generation unit and the price constraint condition, which helps to reduce the load loss and promote the application and development of the virtual power plant in the energy field.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of power grid, and particularly relates to a scheduling method of virtual power plant and related equipment. BACKGROUND

[0002] In the application of virtual power plant (VPP for short), in order to reduce the operation cost of power grid, realize sustainable utilization of energy, and thus improve the application performance of virtual power plant, the scheduling problem thereof has been widely concerned in the field. The scheduling method of virtual power plant provided by the related technology has drawbacks in actual application, which leads to the result of scheduling being difficult to meet the operation demand of virtual power plant. SUMMARY

[0003] Therefore, the present disclosure aims to provide a scheduling method of virtual power plant and related equipment.

[0004] In order to achieve the above purpose, the present disclosure provides a scheduling method of virtual power plant, the virtual power plant comprising at least one power generation unit; the scheduling method specifically comprises:

[0005] determining, according to the state data and the load demand data, first output power of each power generation unit by using a trained deep Q network;

[0006] determining, according to the first output power and the price constraint condition, second output power of each power generation unit by using the trained deep Q network; wherein,

[0007] the deep Q network comprises a first deep Q network and a second deep Q network; wherein, the first deep Q network is configured to determine the first output power; and the second deep Q network is configured to determine the second output power.

[0008] Based on the same inventive concept, the present disclosure further provides a scheduling device of virtual power plant, the virtual power plant comprising at least one power generation unit; the scheduling device comprising:

[0009] an acquisition module configured to acquire a price constraint condition, load demand data and state data of the power generation unit;

[0010] a first output power module configured to determine, according to the state data and the load demand data, first output power of each power generation unit by using a trained deep Q network;

[0011] The second output power module is configured to determine the second output power of each of the power generation units according to the first output power and the price constraint condition by using the trained deep Q network.

[0012] The deep Q network comprises a first deep Q network and a second deep Q network, wherein the first deep Q network is configured to determine the first output power, and the second deep Q network is configured to determine the second output power.

[0013] Based on the same inventive concept, the embodiments of the present disclosure further provide an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the scheduling method according to any one of the above embodiments when executing the program.

[0014] Based on the same inventive concept, the embodiments of the present disclosure further provide a computer program product, comprising computer program instructions, which, when executed on a computer, cause the computer to execute the scheduling method according to any one of the above embodiments.

[0015] As can be seen from the above, the scheduling method of the virtual power plant and the related device provided by the present disclosure, by pre-constructing a deep Q network comprising a first deep Q network and a second deep Q network, determining the first output power of each power generation unit by using the trained first deep Q network according to the state data and the load demand data, and then determining the second output power of each power generation unit by using the second deep Q network according to the first output power and the price constraint condition. By using such a technical solution, the deep Q network is used to fully analyze the load demand data, the operating state of the power generation unit and the price constraint condition, which helps to reduce the load loss and promote the application and development of the virtual power plant in the energy field. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the present disclosure or the related art, the following will briefly introduce the drawings needed to be used in the embodiments or the related art description. Obviously, the drawings in the following description are only embodiments of the present disclosure, and for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0017] Figure 1 Fig. 1 shows a structure schematic diagram of a deep Q network provided by an embodiment of the present disclosure;

[0018] Figure 2 Fig. 2 shows a topological structure schematic diagram of a virtual power plant provided by an embodiment of the present disclosure;

[0019] Figure 3A schematic diagram of a power spot market system architecture is shown according to an embodiment of the present disclosure.

[0020] Figure 4 A schematic diagram of a mechanism of power transaction market operation is shown according to an embodiment of the present disclosure.

[0021] Figure 5 A schematic diagram of a result of upper-layer optimization scheduling is shown according to an embodiment of the present disclosure.

[0022] Figure 6 A schematic diagram of a result of lower-layer optimization scheduling is shown according to an embodiment of the present disclosure.

[0023] Figure 7 A schematic diagram of a comparison of virtual power plant scheduling results is shown according to an embodiment of the present disclosure.

[0024] Figure 8 A schematic diagram of a flow of a virtual power plant scheduling method is shown according to an embodiment of the present disclosure.

[0025] Figure 9 A schematic diagram of a structure of a virtual power plant scheduling apparatus is shown according to an embodiment of the present disclosure.

[0026] Figure 10 A schematic diagram of a structure of an electronic device is shown according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0027] In order to make the objectives, technical solutions and advantages of the present disclosure clearer, the present disclosure is further described in detail below with reference to the accompanying drawings and in conjunction with specific embodiments.

[0028] It should be noted that, unless otherwise defined, technical terms or scientific terms used in the embodiments of the present disclosure should be understood as their common meanings to those skilled in the art to which the present disclosure pertains. The terms "first", "second", and similar terms used in the embodiments of the present disclosure do not denote any order, quantity, or importance, but are merely used to distinguish different components. The terms "include", "contain", and similar terms mean that the elements or objects before the terms encompass the elements or objects listed after the terms and their equivalents, without excluding other elements or objects. The terms "connect" and "connected" and similar terms do not mean only physical or mechanical connections, but can include electrical connections, whether direct or indirect. The terms "upper", "lower", "left", "right", and the like merely represent relative positional relationships, which can change when the absolute positions of the described objects change.

[0029] In order to facilitate understanding of the technical solutions of the present disclosure, some technical terms related to the present disclosure are introduced below.

[0030] Virtual power plant is an Internet of Things technology that aggregates adjustable loads, energy storage, and power sources in different spaces to achieve autonomous coordination and optimization control, and to participate in power system operation and power market transactions. Virtual power plant is mainly composed of three parts: power generation system, energy storage device, and communication system.

[0031] An agent is an entity in a system that can act autonomously and make decisions. In the context of a virtual power plant, each distributed power generation unit can be considered an agent that can adjust its power output based on environmental conditions and scheduling strategies.

[0032] Deep Q-network (DQN) is a reinforcement learning algorithm that combines Q-learning and deep learning. It uses a neural network to approximate the Q function, enabling efficient decision-making in complex environments.

[0033] As described in the background section, the scheduling problem of virtual power plants has received extensive attention in the field. For example, Cheng Xeting et al. proposed a virtual power plant double-layer optimization scheduling method considering the uncertainty risk of new energy output (see Literature 1, Cheng Xeting, Bo Yue-shuang, Jin Yulong, et al. Double-layer scheduling strategy for virtual power plant considering uncertainty risk of new energy output [J]. Modern Electric Power, 2023, 40 (06): 967-975.). This scheduling method analyzes the uncertainty of new energy output under different parameter conditions and quantitatively calculates the corresponding risk value; then, combined with the internal resource state of the virtual power plant, a double-layer optimization scheduling model is constructed; finally, through the solution of the model, the optimal scheduling strategy is obtained. This method is highly dependent on the energy storage equipment in the virtual power plant, however, the capacity and charging and discharging efficiency of the energy storage equipment are limited, which may limit the effectiveness of the double-layer optimization scheduling model. For another example, Pan Ting et al. proposed a virtual power plant double-layer optimization scheduling method based on dynamic time-of-use electricity price (see Literature 2, Pan Ting, Dong Houqi, Wang Yuqing, et al. Double-layer optimization scheduling of virtual power plant based on dynamic time-of-use electricity price [J]. Systems Science and Mathematics, 2024, 44 (02): 304-325.). This scheduling method first constructs a virtual power plant double-layer optimization scheduling model containing a power consumption unit, then analyzes the power plant load side demand in the upper layer of the model and the dynamic time-of-use electricity price in the lower layer, and establishes a target function of minimizing operating cost, thereby obtaining the optimization scheduling result. However, this method does not consider the influence of new energy on the scheduling process, which may lead to a decrease in the accuracy of the model solution. For another example, Zhao Lihang et al. proposed a virtual power plant double-layer optimization scheduling method based on energy economy (see Literature 3, Zhao Lihang, Chang Weiguang, Yang Min, et al. Two-stage energy economic optimization scheduling of virtual power plant in power market environment [J]. China Electric Power, 2022, 55 (10): 14-22.). This scheduling method first constructs a double-layer optimization scheduling model for day-ahead and day-ahead; with the day-ahead forecast data as the reference, the net load parameter is adjusted for the day-ahead scheduling layer, and the energy economy is taken as the target function to obtain the scheduling result. However, in some cases, excessive pursuit of economic benefits may lead to a decrease in the stability of the model solution.

[0034] In view of the drawbacks of the virtual power plant double-layer optimization scheduling method proposed by the related art in actual application, resulting in that the scheduling result is difficult to meet the operation demand of the virtual power plant, the embodiment of the disclosure provides a scheduling method of a virtual power plant and related equipment, by pre-constructing a deep Q network including a first deep Q network and a second deep Q network, according to the state data and the load demand data, the first output power of each power generation unit is determined by using the trained first deep Q network; and then according to the first output power and the price constraint condition, the second output power of each power generation unit is determined by using the second deep Q network. By adopting such a technical solution, the load demand data, the operating state of the power generation unit and the price constraint condition are fully analyzed by using the deep Q network, which helps to reduce the load loss and promote the application and development of the virtual power plant in the energy field.

[0035] In order to make the technical solutions of the disclosure clearer and easier to understand, the following will introduce a scheduling method of a virtual power plant provided by the embodiment of the disclosure in conjunction with the drawings.

[0036] In order to realize the unified scheduling of each distributed power generation unit in the virtual power plant, the embodiment of the disclosure introduces a deep Q network to establish a unified scheduling node state space. It should be noted that the distributed power generation unit can include one or more of a photovoltaic unit that generates electricity by using a photovoltaic panel, a wind power unit that generates electricity by using a fan, and a thermal power unit that generates electricity by using gas, and the disclosure does not limit this. In addition, for a virtual power plant, the number of photovoltaic units, wind power units and thermal power units can each be zero, one or multiple.

[0037] First, define a virtual power plant scheduling node set. The scheduling node set represents all distributed power generation units that need to be scheduled in the virtual power plant, as shown in formula (1):

[0038] (1)

[0039] wherein, V represents the scheduling node set, V 1 represents the i-th distributed power generation unit, n represents the number of schedulable nodes.

[0040] Next, a deep Q-network algorithm is used to assign a corresponding agent to each node in the set of dispatching nodes. It should be noted that the agent's state representation includes, but is not limited to, the dispatching node's current power output, load demand, and uncertainty parameters for wind and photovoltaic power generation. For example, the load demand state represents the amount of power demand that the virtual power plant needs to meet during different time periods and is often correlated with user satisfaction with electricity consumption. For example, the power generation unit output state includes the actual or predicted output of renewable energy generation units such as wind and photovoltaic power generation units, which is affected by natural conditions such as wind speed and sunlight intensity. For example, the power market price state represents the transaction price in the power market during different time periods, which influences the virtual power plant's power purchase and sales decisions. For example, the energy storage device state includes the device's current capacity, charge and discharge efficiency, and capacity limitations. These factors affect the virtual power plant's ability to adjust to imbalances in power supply and demand. For example, the environmental parameter state includes environmental factors such as temperature and humidity, which may affect the efficiency and output of the power generation unit. For example, the equipment operating state includes the health status, maintenance schedule, and failure rate of the power generation equipment. This information is crucial for predicting power generation capacity and planning maintenance activities. The action space corresponds to the scheduling node's operations such as adjusting power generation output, charging and discharging energy storage, etc. By setting a reasonable reward function, the agent can maximize long-term benefits (such as minimizing load loss rate and reducing operating costs) during the optimization scheduling process.

[0041] Then, define the set of agents, namely: A ={ A 1 ,A 2 ,…，A n};in, A represents a collection of agents, A i represents the i-th agent, n Indicates the number of agents.

[0042] Next, with the agent as the core, all states of the scheduling node are merged to generate the state space matrix of the scheduling node, as shown in formula (2):

[0043] (2)

[0044] Among them, S represents the state space matrix of the scheduling node, m Represents the agent dimension.

[0045] Add the executable actions of the scheduling node to the state space matrix. The space matrix of the executable action is designed to be a pair matrix with the state space matrix. Based on the matrix joint relationship, the weight of the scheduling node in the deep Q network is calculated as shown in formula (3):

[0046] (3)

[0047] wherein, denotes the scheduling node weight, denotes the executable action, denotes the reward function of the deep Q network.

[0048] In some embodiments, the corresponding reward function is defined by setting the exploration decay function of the algorithm, as shown in equation (4):

[0049] (4)

[0050] wherein, denotes the exploration decay function, denotes the minimum exploration tolerance, denotes a natural constant, denotes time.

[0051] The deep Q network with double-layer convolution is constructed in the state space where the node weight is defined, and the architecture thereof is shown in Figure 1 .

[0052] As shown in Figure 1 , the deep Q network with double-layer convolution provided by the embodiments of the present disclosure includes a first network and a second network. The first network includes a first convolutional layer and a first fully connected layer, and the second network includes a second convolutional layer and a second fully connected layer. The first convolutional layer and the second convolutional layer are respectively connected with the state and action of the scheduling node, and the execution of the double-layer optimization scheduling strategy in the state space of the scheduling node is realized through the connection of the action parameters. Here, the first network can predict the Q value according to the current state and action, and update its parameters through the gradient descent method.

[0053] In some embodiments, the deep Q network further includes a loss function of the fully connected layer regression, as shown in equation (5):

[0054] (5)

[0055] wherein, L denotes the loss function, d denotes the window size of the fully connected layer, w 0 denotes the weight of the fully connected layer, P (j) denotes the random walk probability of the neighbor node j , and R denotes the regression coefficient.

[0056] During the training process, the network updates its parameters through back propagation according to the loss function (the difference calculated based on the reward function).

[0057] Through the above steps, the initial deep Q network is constructed.

[0058] Based on the state space of the scheduling node and the deep Q network, a double-layer virtual power plant optimization scheduling architecture is designed, and is respectively connected with at least two convolutional layers. It should be noted that the double-layer scheduling of the virtual power plant includes upper-layer optimization scheduling and lower-layer optimization scheduling.

[0059] In the upper-layer optimization scheduling architecture, the disclosure respectively analyzes the load demand of the virtual power plant and the output uncertainty of the wind power and the photovoltaic power, to preliminarily generate the power parameters of the scheduling node.

[0060] Based on the historical data of the virtual power plant, the load demand parameter of the virtual power plant in a unit time is calculated, as shown in formula (6):

[0061] (6)

[0062] wherein, E represents the load demand parameter, P (t) represents t the total power of the virtual power plant at the time, represents a unit time period, represents the user power consumption satisfaction at the corresponding time, and respectively represent the load power consumption before and after the demand response.

[0063] In combination with the output uncertainty of the wind power and the photovoltaic power in the virtual power plant, the corresponding output power characteristics are calculated.

[0064] In some embodiments, the disclosure defines the output power characteristics of the wind power uncertainty according to the real-time wind speed, as shown in formula (7):

[0065] (7)

[0066] wherein, represents the uncertainty power of the wind power, represents the output power of the wind turbine at the rated wind speed, represents a shape parameter, represents a scale parameter, represents the real-time wind speed of the unit environment.

[0067] In some embodiments, the output power characteristics of the photovoltaic power uncertainty are defined according to the real-time illumination, as shown in formula (8):

[0068] (8)

[0069] wherein, represents the uncertainty power of the photovoltaic power, represents the rated output power of the photovoltaic module, denotes the power derating coefficient, denotes the standard illumination, denotes the photovoltaic array capacity, denotes the photovoltaic panel temperature under standard conditions, denotes the real-time temperature of the panel, denotes the temperature coefficient, denotes the real-time illumination intensity.

[0070] The three parameters are combined to generate an upper-layer optimal scheduling strategy, and the corresponding expression can be shown in equation (9):

[0071] (9)

[0072] wherein, denotes the node power parameter of the upper-layer optimal scheduling.

[0073] Finally, on the basis of the upper-layer optimal scheduling strategy, the power trading market environment of the virtual power plant is analyzed at the lower layer to minimize the processing of the net load fluctuation, so as to further correct the power parameter of the scheduling node on the basis of the upper-layer optimal scheduling strategy.

[0074] In some embodiments, a Boolean variable is introduced in the power market trading system to analyze the price constraint of the virtual power plant scheduling, as shown in equation (10):

[0075] (10)

[0076] wherein, H denotes the power trading volume, denotes the maximum trading volume, denotes the power trading price, denotes the Boolean variable.

[0077] Under this constraint, the net load correction value of the power parameter obtained by the upper-layer optimal scheduling is calculated, as shown in equation (11):

[0078] (11)

[0079] wherein, denotes the net load correction value.

[0080] The calculated net load correction value is coupled with the power parameter of the upper-layer optimal scheduling to obtain the power parameter of the lower-layer optimal scheduling, as shown in equation (12):

[0081] (12)

[0082] wherein, denotes the power parameter of the lower-layer optimal scheduling.

[0083] The upper layer and lower layer optimization scheduling results are respectively substituted into the action unit in the scheduling node state space to realize the execution of the scheduling strategy. Here, the power parameters of the upper layer optimization scheduling and the power parameters of the lower layer optimization scheduling can be used as limiting conditions to adjust the scheduling strategy. For example, the power parameters of the upper layer optimization scheduling, the price constraint, etc. can be used as the limiting conditions of the first network; for example, the power parameters of the lower layer optimization scheduling, the price constraint, etc. can be used as the limiting conditions of the second network.

[0084] Next, the scheduling method of the virtual power plant is described in detail in combination with the actual operation architecture of a certain virtual power plant.

[0085] Based on the actual operation architecture of a certain virtual power plant, the corresponding topology structure is constructed, as shown in Figure 2 As shown in Figure 2 , the constructed virtual power plant contains photovoltaic, wind power and gas three kinds of distributed power generation units. The virtual power plant of this architecture is integrated into the actual power spot market system consistent with the actual situation of a certain place. The day-ahead market environment of the virtual power plant is established by using the market system, and the system architecture is as shown in Figure 3

[0086] As shown in Figure 3 , the actual market data collected is used as the actual power trading environment of the virtual power plant. In the experiment of the embodiment, the transaction stage length of the virtual power plant in the power spot market is set to one hour. The power trading market operation mechanism in this experiment is generated as shown in Figure 4

[0087] The operation parameters of each power generation unit of the virtual power plant operation system constructed above are set, as shown in Table 1.

[0088] Table 1 Virtual power plant power generation unit operation parameters

[0089]

[0090] On the basis of the above power generation unit parameter setting, the fixed energy consumption of the electric-gas conversion equipment of the test virtual power plant is defined as 0.179MW.

[0091] Based on the above parameter setting, the scheduling test of the virtual power plant is carried out in the constructed virtual power plant environment.

[0092] According to the virtual power plant scheduling environment set in this experiment, the parameters of the deep Q network in the scheduling method of the embodiment are set, as shown in Table 2.

[0093] Table 2 Deep Q network parameter setting​​​​

[0094]

[0095] The training of the deep Q network provided in the embodiments of the present disclosure is completed according to the parameters in Table 2. On this basis, the scheduling method provided in the embodiments of the present disclosure is used to test the scheduling of the virtual power plant in this environment.

[0096] After analyzing the scheduling method provided in the embodiments of the present disclosure, the optimal scheduling results of the upper layer and the lower layer are obtained respectively, as shown in Figure 5 6

[0097] According to Figure 5 and Figure 6 , in the upper layer scheduling, the power of each unit is preliminarily designed according to the load demand of the virtual power plant and the output uncertainty of the power generation unit. In the lower layer scheduling, the net load fluctuation is minimized under the constructed power trading market operation mechanism, and the energy storage restriction in the virtual power plant is fully considered to obtain the final optimal scheduling result. By analyzing the double-layer optimal scheduling result, it can be seen that the scheduling method of the embodiments of the present disclosure makes full use of the high-efficiency operation period of wind power and photovoltaic power, increases the output in the periods of 6-10 and 10-16 respectively, and increases the output of the gas turbine in the remaining periods. Therefore, the scheduling result obtained by the scheduling method of the embodiments of the present disclosure can fully utilize the output characteristics of the virtual power plant, and has certain feasibility.

[0098] In order to further verify the practical application performance of the scheduling method of the embodiments of the present disclosure, the double-layer scheduling method of the virtual power plant proposed in literature 1, literature 2 and literature 3 is used to carry out comparative experiments respectively, and the time length of each test round is set to 48 h. The scheduling effectiveness is analyzed by calculating the load loss rate of each method in this period, and the scheduling result is shown in Figure 7 .

[0099] As shown in Figure 7 , the load loss rate of the methods proposed in literature 1 and literature 3 is high, indicating that there is a large waste of resources in the scheduling result. The load loss rate of the method in literature 2 and the scheduling method provided in the embodiments of the present disclosure is smaller, but it is found that the method of the present disclosure is more stable. Therefore, the scheduling method of the present disclosure not only has good scheduling effect, but also is more stable, and has higher practical application value in the operation of the virtual power plant.

[0100] Based on the same inventive concept, the embodiments of the present disclosure also provide a scheduling method of a virtual power plant. Here, the virtual power plant includes at least one power generation unit, such as a wind power generation unit, a photovoltaic power generation unit, etc. Figure 8 ​​A flowchart of a scheduling method of a virtual power plant is shown in the embodiment of the present disclosure; as shown in Figure 8 The scheduling method comprises the following steps:

[0101] S801: Obtain price constraint conditions, load demand data, and state data of the power generation units; here, the price constraint conditions can refer to formula (10), which will not be repeated; the state data can include environmental data (such as real-time wind speed, real-time light intensity, real-time temperature, etc.), operation data (such as health status, failure rate, rated power, tip speed ratio, etc.), and the like.

[0102] S803: According to the state data and the load demand data, determine the first output power of each power generation unit by using a trained deep Q network; here, the first output power of each power generation unit can be as shown in Figure 5

[0103] S805: According to the first output power and the price constraint conditions, determine the second output power of each power generation unit by using a trained deep Q network; here, the second output power of each power generation unit can be as shown in Figure 6

[0104] The deep Q network comprises a first deep Q network (such as a first network) and a second deep Q network (such as a second network); wherein the first deep Q network is configured to determine the first output power; and the second deep Q network is configured to determine the second output power.

[0105] With such a technical solution, the load demand data, the state data of the power generation units, and the price constraint conditions are fully analyzed by using the deep Q network, which helps to reduce the load loss and promote the application and development of the virtual power plant in the energy field.

[0106] In some embodiments, the step of obtaining the load demand data specifically comprises:

[0107] Obtain historical data of the virtual power plant; here, the historical data can include virtual power plant total power P (t) , user power consumption satisfaction , , and , and the like.

[0108] According to the historical data, determine the load demand data per unit time according to formula (6). Here, the unit time can be set in advance, such as 1 hour, 2 hours, 4 hours, 12 hours, etc., which is not limited in the present disclosure.

[0109] In some embodiments, the load demand data can be determined according to the historical data of the virtual power plant and the historical data of the power generation units, as shown in Figure 1 ​​, the first deep Q network comprises at least one first convolutional layer and a first fully connected layer; the second deep Q network comprises at least one second convolutional layer and a second fully connected layer; wherein a regression loss function is configured between the first fully connected layer and the second fully connected layer. Here, the regression loss function can be as shown in equation (5).

[0110] In some embodiments, the at least one power generation unit comprises at least one of a wind power unit and a photovoltaic unit;

[0111] Referring to equation (7), the limit condition of the first output power (or the second output power) of the wind power unit includes not higher than the first power (corresponding to the uncertainty power) of the wind power unit; the first power of the wind power unit is determined based on the state data of the wind power unit; the state data at least includes real-time wind speed; and / or

[0112] Referring to equation (8), the limit condition of the first output power (or the second output power) of the photovoltaic unit includes not higher than the second power (corresponding to the uncertainty power) of the photovoltaic unit; the second power of the photovoltaic unit is determined based on the state data of the photovoltaic unit; the state data at least includes real-time light intensity.

[0113] Such technical solutions can ensure that the uncertainty power of the wind power unit and the uncertainty power of the photovoltaic unit participate in the dispatching of the virtual power plant, and the first output power and the second output power are dispatched within the maximum range of the uncertainty power, so that the wind power unit and the photovoltaic unit can fully play their roles.

[0114] In some embodiments, the limit condition of the second deep Q network includes a price constraint condition; wherein, referring to equations (10) and (11), the price constraint condition is used to correct the first output power. In the second deep Q network, considering the price constraint condition, it is beneficial to minimize the net load fluctuation on the basis of sufficient analysis of the market price.

[0115] In some embodiments, forming the deep Q network specifically comprises:

[0116] Based on each power generation unit of the virtual power plant, a node state space matrix (for example, equation (2)) and a weight of the node (for example, equation 3) are constructed;

[0117] Based on the weight and the node state space matrix, an original deep Q network is constructed (refer to Figure 1 );

[0118] Obtain network parameters (for example, table 2) and a training set, train the original deep Q network based on the network parameters and the training set, and obtain the deep Q network.

[0119] In some embodiments, the constructing the node state space matrix and the weight of the node based on each generating unit of the virtual power plant comprises:

[0120] constructing a node set (e.g., formula (1)) by taking each generating unit of the virtual power plant as a node;

[0121] constructing an agent set based on the node set;

[0122] constructing the node state space matrix (e.g., formula (2)) based on the agent set. It should be noted that the above steps can refer to the foregoing and will not be described again.

[0123] It should be noted that the method of the embodiments of the present disclosure can be executed by a single device, such as a computer or a server, etc. The method of the embodiments can also be applied to a distributed scenario, and be completed by multiple devices cooperating with each other. In the case of such a distributed scenario, one of the multiple devices can only execute one or more steps in the method of the embodiments of the present disclosure, and the multiple devices can interact with each other to complete the method.

[0124] It should be noted that some embodiments of the present disclosure have been described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in a different order than the order described above and still achieve desirable results. In addition, the processes depicted in the figures do not necessarily require the particular order shown, or sequential order, to achieve the desired results. In certain implementations, multitasking and parallel processing can be advantageous.

[0125] Based on the same inventive concept, the present disclosure also provides a virtual power plant scheduling device corresponding to any of the above-mentioned embodiment methods.

[0126] Reference Figure 9 , the scheduling device comprises:

[0127] The acquisition module 901 is configured to acquire price constraint conditions, load demand data, and state data of the generating units;

[0128] The first output power module 903 is configured to determine the first output power of each generating unit by using a trained deep Q network according to the state data and the load demand data;

[0129] The second output power module 905 is configured to determine the second output power of each generating unit by using a trained deep Q network according to the first output power and the price constraint conditions; wherein,

[0130] The deep Q network comprises a first deep Q network and a second deep Q network; wherein the first deep Q network is configured to determine the first output power; and the second deep Q network is configured to determine the second output power.

[0131] It should be noted that the deep Q network can be constructed and trained by a network module, and will not be described in detail.

[0132] For the convenience of description, the above device is described by dividing into various modules in terms of functions. Of course, the functions of each module can be implemented in the same or multiple software and / or hardware when implementing the present disclosure.

[0133] The device of the above embodiment is used to implement the corresponding scheduling method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be described here.

[0134] Based on the same inventive concept, corresponding to the method of any of the above embodiments, the present disclosure also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the scheduling method of any of the above embodiments.

[0135] Figure 10 A more specific hardware structure schematic diagram of an electronic device provided by the present embodiment is shown, which can include: a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040 and a bus 1050. The processor 1010, the memory 1020, the input / output interface 1030 and the communication interface 1040 are connected to each other in the device through the bus 1050.

[0136] The processor 1010 can be implemented in the form of a general-purpose CPU (Central Processing Unit, central processor), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), or one or more integrated circuits, etc., for executing related programs to implement the technical solutions provided by the present embodiment.

[0137] The memory 1020 can be implemented in the form of a ROM (Read Only Memory), a RAM (Random Access Memory), a static storage device, a dynamic storage device, etc. The memory 1020 can store an operating system and other application programs, and when the technical solutions provided in the embodiments of the present specification are implemented by software or firmware, the related program codes are stored in the memory 1020 and are called and executed by the processor 1010.

[0138] The input / output interface 1030 is configured to connect an input / output module to realize information input and output. The input / output module can be configured as a component in the device (not shown in the figure) or can be externally connected to the device to provide corresponding functions. The input device can include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output device can include a display, a speaker, a vibrator, an indicator light, etc.

[0139] The communication interface 1040 is configured to connect a communication module (not shown in the figure) to realize communication interaction between the device and other devices. The communication module can realize communication through a wired manner (such as a USB, a network cable, etc.) or through a wireless manner (such as a mobile network, WIFI, Bluetooth, etc.).

[0140] The bus 1050 includes a channel for transmitting information between various components (such as the processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040) of the device.

[0141] It should be noted that although the above device only shows the processor 1010, the memory 1020, the input / output interface 1030, the communication interface 1040, and the bus 1050, in the specific implementation process, the device can also include other components necessary for normal operation. In addition, those skilled in the art can understand that the above device can also only include components necessary for implementing the embodiments of the present specification, and does not necessarily include all the components shown in the figure.

[0142] The electronic device of the above embodiments is used to implement the corresponding scheduling method in any of the preceding embodiments, and has the beneficial effects of the corresponding method embodiments, which are not described here again.

[0143] Based on the same inventive concept, the present disclosure also provides a non-transitory computer-readable storage medium storing computer instructions for causing the computer to execute the scheduling method according to any of the above embodiments.

[0144] The computer readable medium of the embodiments can include permanent and non-permanent, removable and non-removable media, and can be implemented by any method or technology to store information. The information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device.

[0145] The storage medium of the above embodiments stores computer instructions for causing the computer to perform the scheduling method as described in any of the above embodiments, and has the beneficial effects of the corresponding method embodiments, which are not repeated here.

[0146] Based on the same inventive concept, the disclosure also provides a computer program product comprising computer program instructions corresponding to the scheduling method described in any of the above embodiments. In some embodiments, the computer program instructions can be executed by one or more processors of a computer to cause the computer and / or the processor to perform the scheduling method. The processor performing the corresponding step can belong to the corresponding execution subject corresponding to each step in each embodiment of the scheduling method.

[0147] The computer program product of the above embodiments is used to cause the computer and / or the processor to perform the scheduling method as described in any of the above embodiments, and has the beneficial effects of the corresponding method embodiments, which are not repeated here.

[0148] Those skilled in the art should understand that the discussion of any of the above embodiments is only exemplary and is not intended to imply that the scope (including claims) of the disclosure is limited to these examples; under the idea of the disclosure, the above embodiments or technical features between different embodiments can also be combined, the steps can be implemented in any order, and there are many other changes of different aspects of the embodiments of the disclosure as described above. In order to be brief, they are not provided in detail.

[0149] Additionally, to simplify the description and discussion, and so as not to obscure the understanding of the embodiments of the disclosure, the well-known power / ground connections of the integrated circuits (ICs) and other components can or can not be shown in the provided figures. Furthermore, the apparatus can be shown in block diagram form in order to simplify and advance the description of such embodiments and also to highlight the fact that the details regarding how the apparatus is implemented, e.g., in terms of its working details, are highly dependent on the platform within which the embodiments of the disclosure are to be implemented (i.e., these details should be well within the understanding of one of ordinary skill in the art). In situations where detailed circuitry is set forth in order to describe the exemplary embodiments of the disclosure, it should be understood that the disclosure can be practiced with the full understanding and

[0150] Although the present disclosure has been described in connection with certain embodiments, numerous modifications and alterations mentioned above are readily apparent to those of ordinary skill in the art and it is intended that the present disclosure not be limited by the matters described above but ought to be defined in light of the broadest possible scope of the appended claims in combination with the proper interpretations of such claims according to all allowed equivalence concessions. For example, other memory architectures (e.g., dynamic RAM (DRAM)) can use the embodiments discussed.

[0151] It is intended that the embodiments of the present disclosure encompass all such alternatives, modifications, and variations as falling within the broad scope of the appended claims. Accordingly, any and all such modifications, variations or equivalents that fall within the scope of the embodiments of the present disclosure should be intended to be embraced by the claims.

Claims

1. A virtual power plant scheduling method, characterized in that: The virtual power plant includes at least one power generation unit; The scheduling method minimizes net load fluctuations and specifically includes: Obtaining price constraints, load demand data, and status data of the power generation unit; Determining a first output power of each of the power generation units using a trained deep Q network according to the state data and the load demand data; According to the first output power and the price constraint, the second output power of each power generation unit is determined using a trained deep Q network; wherein, The deep Q network includes a first deep Q network and a second deep Q network; wherein the first deep Q network is configured to determine the first output power; the second deep Q network is configured to determine the second output power; the constraint condition of the second deep Q network includes a price constraint condition; wherein the price constraint condition is used to modify the first output power; The at least one power generation unit includes at least one of a wind power unit and a photovoltaic unit; The constraint condition of the first output power of the wind power unit includes not being higher than the first power of the wind power unit; the first power of the wind power unit is determined based on the status data of the wind power unit; the status data at least includes the real-time wind speed; The limiting condition of the first output power of the photovoltaic unit includes not being higher than the second power of the photovoltaic unit; the second power of the photovoltaic unit is determined based on the status data of the photovoltaic unit; the status data at least includes real-time light intensity; Forming the deep Q network specifically includes: Based on each power generation unit of the virtual power plant, construct a node state space matrix and node weights; Constructing an original deep Q network based on the weights and the node state space matrix; Obtain network parameters and a training set, and train the original deep Q network based on the network parameters and the training set to obtain the deep Q network.

2. The scheduling method according to claim 1, characterized in that: The steps for obtaining load demand data include: Obtaining historical data of the virtual power plant; The load demand data per unit time is determined based on the historical data.

3. The scheduling method according to claim 1, characterized in that: The first deep Q network includes at least a first convolutional layer and a first fully connected layer; the second deep Q network includes at least a second convolutional layer and a second fully connected layer.

4. The scheduling method according to claim 1, characterized in that: The step of constructing a node state space matrix and node weights based on each power generation unit of the virtual power plant includes: Taking each power generation unit of the virtual power plant as a node, a node set is constructed; Based on the node set, construct an agent set; Based on the agent set, the node state space matrix is ​​constructed.

5. A scheduling device for a virtual power plant, characterized in that: The virtual power plant includes at least one power generation unit; The dispatching device minimizes the net load fluctuation, specifically including: An acquisition module is configured to: acquire price constraints, load demand data, and status data of the power generation unit; A first output power module is configured to: determine a first output power of each of the power generation units using a trained deep Q network according to the state data and the load demand data; The second output power module is configured to: determine the second output power of each of the power generation units using a trained deep Q network according to the first output power and the price constraint; wherein, The deep Q network includes a first deep Q network and a second deep Q network; wherein the first deep Q network is configured to determine the first output power; the second deep Q network is configured to determine the second output power; the constraint condition of the second deep Q network includes a price constraint condition; wherein the price constraint condition is used to modify the first output power; The at least one power generation unit includes at least one of a wind power unit and a photovoltaic unit; The constraint condition of the first output power of the wind power unit includes not being higher than the first power of the wind power unit; the first power of the wind power unit is determined based on the status data of the wind power unit; the status data at least includes the real-time wind speed; The limiting condition of the first output power of the photovoltaic unit includes not being higher than the second power of the photovoltaic unit; the second power of the photovoltaic unit is determined based on the status data of the photovoltaic unit; the status data at least includes real-time light intensity; The Deep Q Network module is configured as follows: Based on each power generation unit of the virtual power plant, construct a node state space matrix and node weights; Constructing an original deep Q network based on the weights and the node state space matrix; Obtain network parameters and a training set, and train the original deep Q network based on the network parameters and the training set to obtain the deep Q network.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein: The processor implements the scheduling method according to any one of claims 1 to 4 when executing the computer program.

7. A computer program product, characterized in that The method comprises computer program instructions, which, when executed on a computer, enable the computer to execute the scheduling method according to any one of claims 1 to 4.

Images