Output optimization method and device of virtual power plant, equipment and storage medium

Abstract

The invention discloses an output optimization method and device of a virtual power plant, equipment and a storage medium, and relates to the field of new energy, the method starts from two aspects of power generation prediction and power consumption prediction, power generation prediction is realized through prediction of environmental factors, power consumption prediction is realized through prediction of power consumption load, and the power consumption prediction is realized through prediction of power consumption load. And finally, the optimal ratio of each distributed power supply is searched through global optimization, so that the optimal ratio of each distributed power supply is obtained while the scheduling strategy of the power grid is met. And real-time matching can be carried out according to the output condition with strong randomness so as to ensure that the total output (such as peak, valley and equal) of the scheduling strategy does not fluctuate or fluctuates in a relatively small range, and stable operation of the virtual power plant is ensured.

Description

Technical Field

[0001] The present application relates to the field of new energy technologies, and in particular to a method, device, equipment and storage medium for optimizing the output of a virtual power plant. Background Art

[0002] Because the resources used in thermal power generation are non-renewable and thermal power generation causes serious environmental pollution, in recent years, wind and solar power generation have become a growing part of new power systems as a means of supplying new energy. Conventional hydropower units have also shifted from primarily generating electricity to a role that combines power generation with regulation. Among the three mainstream clean energy sources of wind, solar, and hydropower, the real-time power of wind and solar power is significantly affected by climate, weather, and day and night, causing their power generation curves to fluctuate dramatically, with steep rises and falls, and exhibiting significant randomness and volatility. As the clean energy with the best regulation performance, hydropower units have fast start-up and shutdown, flexible regulation, and are the main peak-shaving and frequency-regulating power sources in the power grid. They have strong stability. To ensure the balance and stability of the grid's active power, hydropower units need to adjust their active power output according to the grid's load characteristics and the real-time power of renewable energy sources, smoothing out fluctuations in renewable energy loads to ensure grid stability.

[0003] A virtual power plant is a new type of power coordination and management system that uses information technology and software systems to aggregate and coordinate the optimization of multiple distributed resources, including distributed power sources, energy storage, and adjustable loads. A virtual power plant is not a physical power plant, but rather an innovative energy model that integrates decentralized energy resources through digital technology and smart energy management systems. Using information and communication technologies and software systems, a virtual power plant aggregates geographically dispersed distributed energy resources (such as solar photovoltaics, wind power, and energy storage systems) into a virtual centralized energy system capable of real-time monitoring, analysis, and control of these resources, thereby achieving efficient energy utilization and balancing supply and demand.

[0004] Due to the uncertainty of distributed energy, the power supply reliability and power quality of virtual power plants are affected; when they are integrated into the large power grid on a large scale, random fluctuations are prone to occur, resulting in poor grid connection quality of distributed energy in virtual power plants and difficulty in unified allocation, which leads to certain resistance to the scheduling and optimization of virtual power plants. Summary of the Invention

[0005] The main purpose of this application is to provide a method, device, equipment and storage medium for optimizing the output of a virtual power plant, so as to solve the problem in the prior art that the uncertainty of distributed energy affects the power supply reliability and power quality of the virtual power plant; random fluctuations are easily generated when it is connected to the large power grid on a large scale, resulting in poor grid connection quality of distributed energy in the virtual power plant and difficulty in unified deployment, which leads to certain resistance in the scheduling and optimization of the virtual power plant.

[0006] In order to achieve the above objectives, this application provides the following technical solutions:

[0007] A method for optimizing output of a virtual power plant, the virtual power plant comprising at least two distributed power sources within a predetermined area integrated to a same predetermined monitoring terminal via signal and electrical connections, the method comprising:

[0008] Step S1, obtaining a plurality of environmental parameters of the preset area based on a plurality of preset time periods;

[0009] Step S2: training all environmental parameters through a machine learning machine to obtain a number of future environmental parameters based on a number of preset prediction steps;

[0010] Step S3, obtaining the future output of each distributed power source based on each preset prediction step number through all future environmental parameters;

[0011] Step S4, constructing an output ratio model based on all future outputs;

[0012] Step S5, obtaining a plurality of power loads of the preset area based on a plurality of preset time periods;

[0013] Step S6, training all power loads through a machine learning machine, and obtaining a number of future power loads based on a number of the preset prediction steps;

[0014] Step S7, obtaining the optimal future output ratio of all distributed power sources through the output ratio model according to the future power load of the same preset prediction step number;

[0015] Step S8, obtaining the optimal future output of each distributed power source through global optimization, so that all optimal future outputs meet the optimal future output ratio;

[0016] Step S9: Send each optimal future output to the corresponding distributed power source.

[0017] As a further improvement of the present application, in step S9, each optimal future output is sent to the corresponding distributed power source, and then the following steps are included:

[0018] Step S10, obtaining the future timestamp of each optimal future output respectively;

[0019] Step S20: When each future timestamp arrives, the corresponding distributed power source is adjusted to the corresponding optimal future output.

[0020] As a further improvement of the present application, step S20, when each future timestamp arrives, adjusts the corresponding distributed power source to the corresponding optimal future output, and then includes:

[0021] Step S100: Send all optimal future outputs and all future timestamps to the preset monitoring terminal.

[0022] As a further improvement of the present application, in step S2, all environmental parameters are trained by a machine learning machine, and several future environmental parameters are obtained based on several preset prediction steps, including:

[0023] Step S21, integrating all environmental parameters of the same preset time period into an environmental parameter data set;

[0024] Step S22, performing standard normalization processing on each environmental parameter data set, and obtaining a normalized parameter data set based on one environmental parameter data set;

[0025] Step S23, dividing each normalized parameter data set into a parameter training set and a parameter verification set according to a preset ratio;

[0026] Step S24, defining a neural network model in which the input layer, hidden layer, and output layer are sequentially connected;

[0027] Step S25, inputting all parameter training sets into the input layer in sequence, and performing several trainings through the neural network model;

[0028] Step S26, obtaining the root mean square error of the training results corresponding to the current parameter verification set and the current parameter training set based on each training;

[0029] Step S27, obtaining the minimum error value among all root mean square errors;

[0030] Step S28, obtaining a training result corresponding to the minimum error value as an environmental parameter prediction model;

[0031] Step S29: predicting a number of future environmental parameters based on a number of preset prediction steps using the environmental parameter prediction model.

[0032] As a further improvement of the present application, all distributed power sources include at least one wind power terminal, at least one photovoltaic terminal, and at least one hydropower terminal. Step S4 constructs an output ratio model based on all future outputs, including:

[0033] Step S41, obtaining the future wind power output of the current wind power terminal, the future photovoltaic output of the current photovoltaic terminal, and the future hydropower output of the current hydropower terminal based on the future output of each distributed power source based on each preset prediction step;

[0034] Step S42, performing time series modeling based on the future wind power output, future photovoltaic output, and future hydropower output for each preset prediction step, to obtain wind power output time series samples, photovoltaic output time series samples, and hydropower output time series samples, respectively;

[0035] Step S43: establishing a standard matching model based on power balance constraints, sending-end capacity constraints, receiving-end capacity constraints, transmission security constraints, wind power and photovoltaic power constraints, and curtailment rate constraints;

[0036] Step S44 , substituting the wind power output time series samples, the photovoltaic output time series samples, and the hydropower output time series samples into the standard ratio model to obtain the output ratio model.

[0037] As a further improvement of the present application, in step S6, all power loads are trained by a machine learning machine, and several future power loads are obtained based on several preset prediction steps, including:

[0038] Step S61, integrating all power loads in the same preset time period into a power load data set;

[0039] Step S62: performing standard normalization processing on each power load data set, and obtaining a normalized load data set based on the power load data set;

[0040] Step S63, dividing each normalized load data set into a load training set and a load verification set according to a preset ratio;

[0041] Step S64, defining a neural network model in which the input layer, hidden layer, and output layer are sequentially connected;

[0042] Step S65, inputting all load training sets into the input layer in sequence, and performing several trainings through the neural network model;

[0043] Step S66, obtaining the root mean square error of the training results corresponding to the current load verification set and the current load training set based on each training;

[0044] Step S67, obtaining the minimum error value among all root mean square errors;

[0045] Step S68, obtaining a training result corresponding to the minimum error value as a power load prediction model;

[0046] Step S69: obtaining a plurality of future power loads through prediction of the power load prediction model based on a plurality of preset prediction steps.

[0047] As a further improvement of the present application, step S8, obtaining the optimal future output of each distributed power source through global optimization, so that all optimal future outputs meet the optimal future output ratio, includes:

[0048] Step S81, defining a number of random solutions based on each optimal future output;

[0049] Step S82, defining the optimization results of all random solutions as all optimal future outputs satisfying the optimal future output ratio;

[0050] Step S83, initializing the position of each random solution, and updating the current position and current speed of each random solution respectively;

[0051] Step S84, obtaining the individual optimal solution and the global optimal solution of each random solution based on each update;

[0052] Step S85, respectively determining whether the difference between each individual optimal solution and the previously updated individual optimal solution is less than or equal to a first preset adaptation threshold; if both are less than, executing step S86;

[0053] Step S86, respectively determining whether the difference between each global optimal solution and each global optimal solution updated last time is less than or equal to a second preset adaptation threshold; if both are less than, executing step S87;

[0054] Step S87: Determine whether the optimal solution for all optimal future outputs has been obtained.

[0055] In order to achieve the above objectives, this application also provides the following technical solutions:

[0056] An output optimization device for a virtual power plant, the output optimization device being applied to the output optimization method described above, the output optimization device comprising:

[0057] A preset area environmental parameter acquisition module, used to acquire a plurality of environmental parameters of the preset area based on a plurality of preset time periods;

[0058] A future environmental parameter prediction module is used to train all environmental parameters through a machine learning machine and obtain a number of future environmental parameters based on a number of preset prediction steps;

[0059] A future output acquisition module is used to respectively acquire the future output of each distributed power source based on each preset prediction step number through all future environmental parameters;

[0060] An output ratio model building module is used to build an output ratio model based on all future outputs;

[0061] A preset area power load acquisition module, configured to acquire a plurality of power loads of the preset area based on a plurality of preset time periods;

[0062] A future power load prediction module is used to train all power loads through a machine learning machine and obtain a number of future power loads based on a number of preset prediction steps;

[0063] An optimal future output ratio acquisition module is used to obtain the optimal future output ratio of all distributed power sources through the output ratio model according to the future power load of the same preset prediction step number;

[0064] An optimal future output acquisition module is used to obtain the optimal future output of each distributed power source through global optimization, so that all optimal future outputs meet the optimal future output ratio;

[0065] The optimal future output sending module is used to send each optimal future output to the corresponding distributed power source.

[0066] In order to achieve the above objectives, this application also provides the following technical solutions:

[0067] An electronic device includes a processor and a memory coupled to the processor, wherein the memory stores program instructions that can be executed by the processor; when the processor executes the program instructions stored in the memory, the output optimization method as described above is implemented.

[0068] In order to achieve the above objectives, this application also provides the following technical solutions:

[0069] A storage medium stores program instructions, which, when executed by a processor, can implement the output optimization method described above.

[0070] The present application obtains several environmental parameters of a preset area based on several preset time periods; trains all environmental parameters through a machine learning machine, and obtains several future environmental parameters based on several preset prediction steps; obtains the future output of each distributed power source based on each preset prediction step through all future environmental parameters; constructs an output ratio model based on all future outputs; obtains several power loads of a preset area based on several preset time periods; trains all power loads through a machine learning machine, and obtains several future power loads based on several preset prediction steps; obtains the optimal future output ratio of all distributed power sources through the output ratio model based on the future power loads with the same preset prediction step; obtains the optimal future output of each distributed power source through global optimization, so that all optimal future outputs meet the optimal future output ratio; and sends each optimal future output to the corresponding distributed power source. This application starts from two aspects: power generation prediction and power consumption prediction. It realizes power generation prediction by predicting environmental factors, and then realizes power consumption prediction by predicting power load. It builds a matching model based on the future output of each distributed power source, and finally finds the optimal matching of each distributed power source through global optimization. This allows this application to meet the scheduling strategy of the power grid while being able to match in real time according to the highly random output conditions to ensure that the total output of the scheduling strategy (such as peak, valley, and equal) will not fluctuate or fluctuate within a smaller range, thereby ensuring the stable operation of the virtual power plant. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 This is a flowchart of steps of an embodiment of a method for optimizing output of a virtual power plant according to the present application;

[0072] Figure 2 This is a structural diagram of an embodiment of an output optimization device for a virtual power plant of the present application;

[0073] Figure 3 This is a schematic structural diagram of an embodiment of the electronic device of the present application;

[0074] Figure 4 This is a structural diagram of an embodiment of the storage medium of the present application. DETAILED DESCRIPTION

[0075] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0076] The terms "first", "second" and "third" in this application are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, a feature defined as "first", "second" and "third" may explicitly or implicitly include at least one of such features. In the description of this application, "multiple" means at least two, for example, two, three, etc., unless otherwise clearly and specifically defined. All directional indications in the embodiments of this application (such as up, down, left, right, front, back...) are only used to explain the relative positional relationship, movement, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally also include steps or units that are not listed, or may optionally also include other steps or units that are inherent to these processes, methods, products or devices.

[0077] References to "embodiments" herein mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0078] like Figure 1 As shown, this embodiment provides an embodiment of a method for optimizing the output of a virtual power plant. In this embodiment, the virtual power plant includes at least two distributed power sources within a preset area that are integrated into the same preset monitoring terminal through signal and electrical connections.

[0079] Preferably, the virtual power plant of this embodiment includes a wind power station, a photovoltaic power station, and a hydropower station to facilitate subsequent description, but it is worth noting that the number of the three terminals can be flexibly changed and is not limited to having one of each.

[0080] Specifically, the output optimization method includes the following steps:

[0081] Step S1: Acquire several environmental parameters of a preset area based on several preset time periods.

[0082] Preferably, the environmental parameters can be obtained according to the type of power station in the area, for example, wind speed probability distribution is obtained for wind power stations, solar radiation distribution is obtained for photovoltaic power stations, and water head and flow in hydropower are obtained for hydropower stations.

[0083] Step S2: All environmental parameters are trained by a machine learning machine, and a number of future environmental parameters are obtained based on a number of preset prediction steps.

[0084] Preferably, the machine learning machine of this embodiment can select BP neural network.

[0085] Step S3: obtaining the future output of each distributed power source based on each preset prediction step number through all future environmental parameters.

[0086] Preferably, the output calculated according to the environmental parameters can be directly calculated according to respective formulas or models, for example, wind power output can be calculated by wind speed probability distribution, photovoltaic output can be calculated according to solar radiation distribution, and hydropower output can be calculated according to water head and flow.

[0087] Specifically, wind power output can be determined based on:

[0088] The solution is obtained.

[0089] Wherein, P(V) is the wind turbine output power based on the actual wind speed of the current preset time period; P e The current wind turbine is based on the rated wind speed v e Rated power; v cut-in is the current wind turbine cut-in wind speed; v cut-out The current wind turbine cut-out wind speed.

[0090] Specifically, the photovoltaic output can be determined based on:

[0091] Output (kW) = Solar irradiance (kW / m 2 )×photovoltaic panel area (m 2 )×conversion efficiency (η)×system loss coefficient (β) is calculated.

[0092] Among them, solar irradiance is monitored in real time through meteorological satellites + ground-based photon sensors (spatial resolution ≤ 100 meters, error < 3%), the conversion efficiency is based on the efficiency of mainstream single-crystal PERC modules of 23.5%, and the perovskite-silicon stacked modules reaches 29.8%. The system loss is based on dust obstruction (controlled by intelligent cleaning robots < 2%) and line loss (controlled by silicon carbide inverters < 1.5%).

[0093] Specifically, hydropower output can be calculated based on:

[0094] Output (kW) = 9.81 × head (m) × flow rate (m3 / s) × comprehensive efficiency (η) is calculated.

[0095] Among them, water head (H): Beidou satellite positioning + underwater pressure sensor real-time monitoring, dynamic accuracy reaches ±0.05m, flow (Q) is updated every second according to the AI ​​Acoustic Doppler Current Profiler (ADCP), with an error of <1%, comprehensive efficiency (η): by 2025, the supercritical turbine efficiency will reach 98.3%, the generator efficiency will be 98.8%, and the comprehensive η≈96.5%.

[0096] Step S4: constructing an output ratio model based on all future outputs.

[0097] Preferably, the output ratio model is as follows:

[0098]

[0099] Among them, T is the number of preset prediction steps, is the future output of the i-th distributed generation at the t-th preset prediction step, I out It is the receiving node of all distributed power sources.

[0100] It should be noted that the above formulas and formula symbols are only used to illustrate the principles, and their meanings are not interchangeable with those in other positions.

[0101] Step S5: obtaining a plurality of power loads of a preset area based on a plurality of preset time periods.

[0102] Preferably, the preset time period and the preset number of prediction steps in this embodiment can both be set to one natural hour.

[0103] Step S6: All power loads are trained by a machine learning machine, and several future power loads are obtained based on several preset prediction steps.

[0104] Step S7: Obtain the optimal future output ratio of all distributed power sources through the output ratio model according to the future power load with the same preset prediction step number.

[0105] Step S8: Obtain the optimal future output of each distributed power source through global optimization, so that all optimal future outputs meet the optimal future output ratio.

[0106] Step S9: Send each optimal future output to the corresponding distributed power source.

[0107] Furthermore, in step S9, each optimal future output is sent to the corresponding distributed power source, and then the following steps are also included:

[0108] Step S10: Obtain the future timestamp of each optimal future output respectively.

[0109] Step S20: When each future timestamp arrives, the corresponding distributed power source is adjusted to the corresponding optimal future output.

[0110] Furthermore, in step S20, when each future timestamp arrives, the corresponding distributed power source is adjusted to the corresponding optimal future output, and then the following steps are also included:

[0111] Step S100: Send all optimal future outputs and all future timestamps to a preset monitoring terminal.

[0112] Furthermore, step S2 is to train all environmental parameters through a machine learning machine and obtain a number of future environmental parameters based on a number of preset prediction steps, which specifically includes the following steps:

[0113] Step S21 : integrating all environmental parameters in the same preset time period into an environmental parameter data set.

[0114] In step S22 , each environmental parameter data set is subjected to standard normalization processing, and a normalized parameter data set is obtained based on the environmental parameter data set.

[0115] Preferably, this embodiment prefers the normalization method of zero-mean normalization (Z-score normalization), which gives the mean and standard deviation of the original data to standardize the data, and the processed data conforms to the standard normal distribution, that is, the mean is 0 and the standard deviation is 1. For the normalization method, this embodiment can also use batch normalization. Compared with simple normalization, in previous neural network training, only the input layer data is normalized, but no normalization is performed in the middle layer. Although the data set of the input node is normalized, the data distribution of the input data after matrix multiplication is likely to change greatly, and as the number of hidden layer network layers continues to deepen, the change in data distribution will become greater and greater. Therefore, the normalization processing performed by batch normalization in the middle layer of the neural network makes the training effect better.

[0116] In step S23, each normalized parameter data set is divided into a parameter training set and a parameter verification set according to a preset ratio.

[0117] Preferably, the preset ratio can be set to 8:2, so as to divide the data in the normalized data set into a training set and a sample set in a ratio of 8:2.

[0118] Step S24, defining a neural network model in which the input layer, hidden layer, and output layer are sequentially connected.

[0119] Preferably, the neural network model is represented by the following formula:

[0120]

[0121] Among them, y is the neural network model; x n is the nth input node in the input layer, each input node corresponds to a data in the training set, is the weight from the mth input node of the input layer to the nth input node of the hidden layer; is the bias of the nth input node connected to the hidden layer; is the bias of the output layer; tansig(·) is the activation function; the numbers in the brackets of the symbol subscripts are the number of layers, the subscript (1) is the first layer, that is, the input layer, and the subscripts (1, 2) are the first to second layers, that is, the input layer to the hidden layer.

[0122] It should be noted that the above formulas and formula symbols are only used to illustrate the principles, and their meanings are not interchangeable with those in other positions.

[0123] In step S25, all parameter training sets are sequentially input into the input layer and trained several times through the neural network model.

[0124] Step S26, based on each training, obtain the root mean square error of the training results corresponding to the current parameter verification set and the current parameter training set.

[0125] Step S27: Obtain the minimum error value among all root mean square errors.

[0126] Step S28: Obtain the training result corresponding to the minimum error value as the environmental parameter prediction model.

[0127] Step S29: predicting a number of future environmental parameters based on a number of preset prediction steps using the environmental parameter prediction model.

[0128] Preferably, the training model trains a neural network usually requires a large amount of data, namely a data set; the data set is generally divided into three categories, namely the above-mentioned training set (training set), validation set (validation set) and test set (test set).

[0129] Among them, one epoch is the process of training once with all the samples in the training set. The so-called training once refers to one forward pass and one back pass. When the number of samples in an epoch (i.e., training set) is too large, training once may consume too much time, and it is not necessary to use all the data in the training set for each training. In this case, the entire training set needs to be divided into multiple small blocks, that is, divided into multiple batches for training. An epoch consists of one or more batches. A batch is a part of the training set. Each training process only uses a part of the data, i.e., a batch. The process of training a batch is an iteration.

[0130] Preferably, the neural network training specifically includes a perceptron, which is composed of two layers of neurons. The input layer receives external input signals and transmits them to the output layer. The output layer is MP neurons, and the step function is y j =f(∑ i w i ·x i -θ i ), the step function here is for principle explanation and is not interchangeable with other symbols.

[0131] Preferably, given a training data set, the weight w i (i=1,2,...,n) and training bias θ i It can be obtained through learning, θ i It can be understood as the weight w corresponding to a fixed value of -1,0. i+1 .

[0132] Preferably, in this embodiment, the number of neural network training times can be set to 10,000 times.

[0133] Preferably, the learning rate from the 1st to the 5000th epoch can be set to 0.01, the learning rate from the 5001st to the 7500th epoch can be set to 0.001, and the learning rate from the 7501st to the 10000th epoch can be set to 0.0001.

[0134] It can be understood that the neural network training of this embodiment mainly includes the following ideas:

[0135] ① Initialize the weights and bias items in the network.

[0136] Initializing the parameter values ​​(output unit weights, bias terms and hidden unit weights, bias terms are all model parameters) is to activate forward propagation, obtain the output value of each layer element, and then obtain the value of the loss function.

[0137] ②Activate forward propagation to obtain the output value of each layer and the expected value of the loss function of each layer.

[0138] ③ According to the loss function, calculate the error term of the output unit and the error term of the hidden unit.

[0139] Compute various errors, calculate the gradient of the parameters with respect to the loss function, or calculate partial derivatives using the chain rule of calculus. For partial derivatives of vectors or matrices within a composite function, always multiply the composite function's inner functions with the left derivative. For partial derivatives of scalars within a composite function, either multiply the composite function's inner functions with the right derivative can be used.

[0140] ④Update the weights and bias items in the neural network.

[0141] ⑤ Repeat ② to ④ until the loss function is less than the preset bias or the number of iterations is exhausted, and the parameters output at this time are the current optimal parameters.

[0142] Furthermore, all distributed power sources include at least one wind power terminal, at least one photovoltaic terminal, and at least one hydropower terminal. Step S4 constructs an output ratio model based on all future outputs, specifically including the following steps:

[0143] Step S41 , obtaining the future wind power output of the current wind power terminal, the future photovoltaic output of the current photovoltaic terminal, and the future hydropower output of the current hydropower terminal based on the future output of each distributed power source based on each preset prediction step.

[0144] Step S42 , performing time series modeling on the future wind power output, future photovoltaic output, and future hydropower output based on each preset prediction step, and obtaining wind power output time series samples, photovoltaic output time series samples, and hydropower output time series samples, respectively.

[0145] Preferably, under the above constraints, the three output time series samples are respectively substituted into the output ratio model, and several output capacity ratios are obtained by solving.

[0146] Preferably, the time series can be modeled by ARIMA, wherein the seasonal time series can be modeled by SARIMA. In this embodiment, SAMIRA modeling is preferred, wherein SAMIMA (p, d, q) (P, D, Q, s) has a total of 7 parameters, which can be divided into two categories, one of which is three non-seasonal parameters (p, d, q), and the other is four seasonal parameters (P, D, Q, s). Wherein, p: AR (p), the order of non-seasonal autoregression; d: I (d), the number of one-step difference; q: MA (q), the order of non-seasonal moving average; P: the order of seasonal autoregression; D: the number of seasonal differences; Q: the order of seasonal moving average. Then the SAMIRA model is:

[0147] SARIMA(p,d,q)(P,D,Q):Φ (p) (B)Φ (P) (B S )(1-B) d (1-B S ) D y t =θ (q) (B)Θ (Q) (B S )ε t .

[0148] It should be noted that SAMIRA modeling is a mature existing technology, and its specific meaning will not be repeated in this embodiment. The above formulas and formula symbols are only used to illustrate the principles, and their meanings are not interchangeable with those in other locations.

[0149] Step S43: establishing a standard matching model based on power balance constraints, sending-end capacity constraints, receiving-end capacity constraints, transmission safety constraints, wind power and photovoltaic power constraints, and power curtailment rate constraints.

[0150] Preferably, the constraints are as follows:

[0151] ① Power balance constraint, that is, the balance constraint between power generation and power consumption:

[0152]

[0153] Among them, I in is the sending node of all access terminals, I out The receiving node for all sending ends.

[0154] ②Sending-end capacity constraint, i.e. the generating power at the sending end must meet the switching capacity constraints (substation, switch station):

[0155]

[0156] in, is the maximum (substation, switch station) commutation capacity at the sending end i.

[0157] ③ Capacity constraints at the receiving end, that is, the power consumption at the receiving end must also meet the switching capacity constraints (substation, switch station):

[0158]

[0159] in, is the maximum (substation, switch station) commutation capacity of receiving end j.

[0160] ④ Transmission safety constraints, that is, power constraints on transmission lines are required:

[0161]

[0162] in, is the maximum transmission capacity of the lth transmission line, p l (t) The real-time transmission power of the regional power grid in the above period t.

[0163] ⑤ Wind power and photovoltaic power constraints:

[0164]

[0165] in, is the access capacity of the wind power access terminal, is the access capacity of the optoelectronic access terminal, is the normalized time series value of wind power access output, is the normalized time series value of the photovoltaic access terminal output, and and All are known quantities.

[0166] ⑥ Power curtailment rate constraint, that is, setting an upper limit on the power curtailment rate of each sending end in the regional power grid during the entire period:

[0167]

[0168] Where θ is the power curtailment rate, such as 0.05.

[0169] It should be noted that the above formulas and formula symbols are only used to illustrate the principles, and their meanings are not interchangeable with those in other positions.

[0170] Step S44 , substituting the wind power output time series samples, photovoltaic output time series samples, and hydropower output time series samples into the standard ratio model to obtain an output ratio model.

[0171] Furthermore, step S6 is to train all power loads through a machine learning machine and obtain a number of future power loads based on a number of preset prediction steps, which specifically includes the following steps:

[0172] Step S61 : integrating all power loads in the same preset time period into a power load data set.

[0173] In step S62 , each power load data set is subjected to standard normalization processing, and a normalized load data set is obtained based on the power load data set.

[0174] Step S63: Divide each normalized load data set into a load training set and a load verification set according to a preset ratio.

[0175] Step S64, defining a neural network model in which the input layer, hidden layer, and output layer are sequentially connected.

[0176] In step S65, all load training sets are sequentially input into the input layer and trained several times through the neural network model.

[0177] Step S66: Obtain the root mean square error of the training results corresponding to the current load verification set and the current load training set based on each training.

[0178] Step S67: Obtain the minimum error value among all root mean square errors.

[0179] Step S68: Obtain the training result corresponding to the minimum error value as the power load prediction model.

[0180] Step S69: obtaining a plurality of future power loads through prediction of the power load prediction model based on a plurality of preset prediction steps.

[0181] Preferably, the principles of steps S61 to S69 are the same as those of steps S21 to S29. The specific principles and preferred solutions will not be described in detail in this embodiment. Please refer to the principles of steps S21 to S29 for details.

[0182] Furthermore, step S8 is to obtain the optimal future output of each distributed power source through global optimization, so that all optimal future outputs meet the optimal future output ratio, which specifically includes the following steps:

[0183] Step S81: defining a number of random solutions based on each optimal future output.

[0184] Preferably, all random solutions can be defined according to the following formula:

[0185]

[0186] Among them, P i is the set of all random solutions, p1,p2,...,p i ,…,p N-1 ,p N are each random solution, i is the number of the random solution, N is the number of all random solutions; V i is the set of velocities of all random solutions, v1,v2,...,v i ,…,v N-1 ,v N are the speeds of each random solution respectively.

[0187] Step S82: define the optimization results of all random solutions as all optimal future outputs satisfying the optimal future output ratio.

[0188] Step S83: Initialize the position of each random solution and update the current position and current speed of each random solution respectively.

[0189] Preferably, the current position and current velocity can be updated respectively based on the same random solution according to the following formula:

[0190]

[0191] Among them, v id is the speed of the ith random solution at step d, ω·v id-1 is the velocity inertia of the ith random solution at step d-1, ω is the inertia coefficient, c1·rand·(P best,i -p i ) is the self-perception representation of the i-th random solution, c2·rand·(G best,i -p i ) is the social cognitive representation of the i-th random solution; c1 and c2 are both learning factors, rand is a random number in [0,1], P best,i is the individual optimal solution obtained by the i-th random solution, G best,i is the global optimal solution obtained by the i-th random solution, p id is the i-th random solution at step d, p id-1 is the i-th random solution at step d-1.

[0192] Preferably, the value range of c1 is [0, 0.5], preferably 0.4; the value range of c2 is [0.5, 1], preferably 0.8.

[0193] Step S84: Obtain the individual optimal solution and the global optimal solution of each random solution based on each update.

[0194] Preferably, the inertia coefficient may be linearly decreased once per update step according to the following formula:

[0195]

[0196] Among them, ω id is the inertia coefficient of the i-th random solution after optimization in the d-th step, ω ini is the initial inertia coefficient, Pace current is the current update step number, Pace max is the maximum number of update steps.

[0197] Preferably, the initial inertia coefficient is generally set to 0.5, and the maximum number of update steps is generally set according to actual needs. In this embodiment, it can be set to 10,000 times.

[0198] Step S85 , respectively judging whether the difference between each individual optimal solution and each individual optimal solution updated last time is less than or equal to a first preset adaptation threshold; if both are less than, executing step S86 .

[0199] Step S86: Determine whether the difference between each global optimal solution and each global optimal solution updated last time is less than or equal to a second preset adaptation threshold. If both are less than, execute step S87.

[0200] Preferably, the values ​​of the first preset adaptive threshold and the second preset adaptive threshold need to be adjusted according to the specific problem, and generally need to be adjusted based on the calculation results. If the adaptive threshold is set too small, the algorithm may stop prematurely and fail to obtain the optimal solution; if the adaptive threshold is set too large, the algorithm may be over-updated, wasting computing resources.

[0201] Preferably, the adaptive threshold may also be evaluated by one of the Griewank function, Rastrigin function, Schaffer function, Ackley function, and Rosenbrock function.

[0202] Step S87: Determine whether the optimal solution for all optimal future outputs has been obtained.

[0203] It should be noted that the formulas in the above additional content are for principle explanation only, and the symbolic meanings of the formulas are not interchangeable with other formulas.

[0204] This embodiment obtains several environmental parameters of a preset area based on several preset time periods; trains all environmental parameters through a machine learning machine, and obtains several future environmental parameters based on several preset prediction steps; obtains the future output of each distributed power source based on each preset prediction step through all future environmental parameters; constructs an output ratio model based on all future outputs; obtains several power loads of a preset area based on several preset time periods; trains all power loads through a machine learning machine, and obtains several future power loads based on several preset prediction steps; obtains the optimal future output ratio of all distributed power sources through the output ratio model based on the future power loads with the same preset prediction step; obtains the optimal future output of each distributed power source through global optimization, so that all optimal future outputs meet the optimal future output ratio; and sends each optimal future output to the corresponding distributed power source. This embodiment starts from two aspects: power generation prediction and power consumption prediction. It realizes power generation prediction by predicting environmental factors, and then realizes power consumption prediction by predicting power load. It builds a matching model according to the future output of each distributed power source, and finally finds the optimal matching of each distributed power source through global optimization. This allows this embodiment to meet the scheduling strategy of the power grid while being able to match in real time according to the highly random output conditions to ensure that the total output of the scheduling strategy (such as peak, valley, and equal) will not fluctuate or fluctuate within a smaller range, thereby ensuring the stable operation of the virtual power plant.

[0205] like Figure 2 As shown, this embodiment provides an embodiment of an output optimization device for a virtual power plant. In this embodiment, the output optimization device is applied to the output optimization method as in the above embodiment.

[0206] Specifically, the output optimization device includes a preset area environmental parameter acquisition module 1, a future environmental parameter prediction module 2, a future output acquisition module 3, an output ratio model construction module 4, a preset area power load acquisition module 5, a future power load prediction module 6, an optimal future output ratio acquisition module 7, an optimal future output acquisition module 8, and an optimal future output sending module 9, which are electrically connected in sequence.

[0207] Among them, the preset area environmental parameter acquisition module 1 is used to obtain a number of environmental parameters of the preset area based on a number of preset time periods; the future environmental parameter prediction module 2 is used to train all environmental parameters through a machine learning machine, and obtain a number of future environmental parameters based on a number of preset prediction steps; the future output acquisition module 3 is used to obtain the future output of each distributed power source based on each preset prediction step through all future environmental parameters; the output ratio model construction module 4 is used to construct an output ratio model based on all future outputs; the preset area power load acquisition module 5 is used to obtain a number of power loads of the preset area based on a number of preset time periods. Electric load; the future electricity load prediction module 6 is used to train all electricity loads through a machine learning machine, and obtain several future electricity loads based on several preset prediction steps; the optimal future output ratio acquisition module 7 is used to obtain the optimal future output ratio of all distributed power sources through the output ratio model according to the future electricity load of the same preset prediction step; the optimal future output acquisition module 8 is used to obtain the optimal future output of each distributed power source through global optimization, so that all optimal future outputs meet the optimal future output ratio; the optimal future output sending module 9 is used to send each optimal future output to the corresponding distributed power source.

[0208] Furthermore, the output optimization device also includes an optimal future output future timestamp acquisition module and a distributed power supply adjustment module that are electrically connected in sequence; the optimal future output future timestamp acquisition module is electrically connected to the optimal future output sending module 9.

[0209] Among them, the best future output future timestamp acquisition module is used to obtain the future timestamp of each best future output respectively; the distributed power supply adjustment module is used to adjust the corresponding distributed power supply to the corresponding best future output when each future timestamp arrives.

[0210] Furthermore, the output optimization device also includes a future output and future timestamp sending module electrically connected to the distributed power supply regulation module, and the module is used to send all optimal future outputs and all future timestamps to a preset monitoring terminal.

[0211] Furthermore, the future environmental parameter prediction module 2 specifically includes a first future environmental parameter prediction unit, a second future environmental parameter prediction unit, a third future environmental parameter prediction unit, a fourth future environmental parameter prediction unit, a fifth future environmental parameter prediction unit, a sixth future environmental parameter prediction unit, a seventh future environmental parameter prediction unit, an eighth future environmental parameter prediction unit, and a ninth future environmental parameter prediction unit, which are electrically connected in sequence; the first future environmental parameter prediction unit is electrically connected to the preset area environmental parameter acquisition module 1, and the ninth future environmental parameter prediction unit is electrically connected to the future output acquisition module 3.

[0212] Among them, the first future environmental parameter prediction unit is used to integrate all environmental parameters of the same preset time period into an environmental parameter data set; the second future environmental parameter prediction unit is used to perform standard normalization processing on each environmental parameter data set, and obtain a normalized parameter data set based on an environmental parameter data set; the third future environmental parameter prediction unit is used to divide each normalized parameter data set into a parameter training set and a parameter verification set according to a preset ratio; the fourth future environmental parameter prediction unit is used to define a neural network model in which the input layer, hidden layer, and output layer are signal-connected in sequence; the fifth future environmental parameter prediction unit is used to input all parameter training sets into the input layer in sequence, and perform several trainings through the neural network model; the sixth future environmental parameter prediction unit is used to obtain the root mean square error of the training results corresponding to the current parameter verification set and the current parameter training set based on each training; the seventh future environmental parameter prediction unit is used to obtain the minimum error among all root mean square errors; the eighth future environmental parameter prediction unit is used to obtain the training result corresponding to the minimum error as the environmental parameter prediction model; and the ninth future environmental parameter prediction unit is used to predict several future environmental parameters based on several preset prediction steps through the environmental parameter prediction model.

[0213] Furthermore, the output ratio model construction module 4 specifically includes a first output ratio model construction unit, a second output ratio model construction unit, a third output ratio model construction unit, and a fourth output ratio model construction unit that are electrically connected in sequence; the first output ratio model construction unit is electrically connected to the future output acquisition module 3, and the fourth output ratio model construction unit is electrically connected to the preset area power load acquisition module 5.

[0214] Among them, the first output ratio model construction unit is used to obtain the future wind power output of the current wind power end, the future photovoltaic output of the current photovoltaic end, and the future hydropower output of the current hydropower end based on the future output of each distributed power source based on each preset prediction step; the second output ratio model construction unit is used to perform time series modeling based on the future wind power output, future photovoltaic output, and future hydropower output of each preset prediction step, and obtain wind power output time series samples, photovoltaic output time series samples, and hydropower output time series samples respectively; the third output ratio model construction unit is used to establish a standard ratio model based on power balance constraints, sending end capacity constraints, receiving end capacity constraints, transmission safety constraints, wind power and photovoltaic power constraints, and power abandonment rate constraints; the fourth output ratio model construction unit is used to substitute wind power output time series samples, photovoltaic output time series samples, and hydropower output time series samples into the standard ratio model to obtain an output ratio model.

[0215] Furthermore, the future electricity load forecasting module 6 specifically includes a first future electricity load forecasting unit, a second future electricity load forecasting unit, a third future electricity load forecasting unit, a fourth future electricity load forecasting unit, a fifth future electricity load forecasting unit, a sixth future electricity load forecasting unit, a seventh future electricity load forecasting unit, an eighth future electricity load forecasting unit, and a ninth future electricity load forecasting unit, which are electrically connected in sequence; the first future electricity load forecasting unit is electrically connected to the preset area electricity load acquisition module 5, and the ninth future electricity load forecasting unit is electrically connected to the optimal future output ratio acquisition module 7.

[0216] Among them, the first future electricity load prediction unit is used to integrate all electricity loads in the same preset time period into an electricity load data set; the second future electricity load prediction unit is used to perform standard normalization processing on each electricity load data set, and obtain a normalized load data set based on the electricity load data set; the third future electricity load prediction unit is used to divide each normalized load data set into a load training set and a load verification set according to a preset ratio; the fourth future electricity load prediction unit is used to define a neural network model in which the input layer, hidden layer, and output layer are signal-connected in sequence; the fifth future electricity load prediction unit is used to input all load training sets into the input layer in sequence, and perform several trainings through the neural network model; the sixth future electricity load prediction unit is used to obtain the root mean square error of the training results corresponding to the current load verification set and the current load training set based on each training; the seventh future electricity load prediction unit is used to obtain the minimum error among all root mean square errors; the eighth future electricity load prediction unit is used to obtain the training result corresponding to the minimum error as the electricity load prediction model; and the ninth future electricity load prediction unit is used to predict several future electricity loads based on several preset prediction steps through the electricity load prediction model.

[0217] Furthermore, the best future output acquisition module 8 specifically includes a first best future output acquisition unit, a second best future output acquisition unit, a third best future output acquisition unit, a fourth best future output acquisition unit, a fifth best future output acquisition unit, a sixth best future output acquisition unit, and a seventh best future output acquisition unit, which are electrically connected in sequence; the first best future output acquisition unit is electrically connected to the best future output ratio acquisition module 7, and the seventh best future output acquisition unit is electrically connected to the best future output sending module 9.

[0218] Among them, the first best future output acquisition unit is used to define several random solutions based on each best future output; the second best future output acquisition unit is used to define the optimization result of all random solutions as all best future outputs satisfying the best future output ratio; the third best future output acquisition unit is used to initialize the position of each random solution, and update the current position and current speed of each random solution respectively; the fourth best future output acquisition unit is used to obtain the individual optimal solution and the global optimal solution of each random solution based on each update; the fifth best future output acquisition unit is used to respectively determine whether the difference between each individual optimal solution and each individual optimal solution updated last time is less than or equal to the first preset adaptation threshold; the sixth best future output acquisition unit is used to respectively determine whether the difference between each global optimal solution and each global optimal solution updated last time is less than or equal to the second preset adaptation threshold if both are less than; the seventh best future output acquisition unit is used to determine that the optimal solution of all best future outputs has been obtained if both are less than.

[0219] It should be noted that this embodiment is a functional module embodiment based on the above method embodiment. The optimization, expansion, limitation, example, and principle description of this embodiment can be referred to the above embodiment, and will not be repeated in this embodiment.

[0220] This embodiment obtains several environmental parameters of a preset area based on several preset time periods; trains all environmental parameters through a machine learning machine, and obtains several future environmental parameters based on several preset prediction steps; obtains the future output of each distributed power source based on each preset prediction step through all future environmental parameters; constructs an output ratio model based on all future outputs; obtains several power loads of a preset area based on several preset time periods; trains all power loads through a machine learning machine, and obtains several future power loads based on several preset prediction steps; obtains the optimal future output ratio of all distributed power sources through the output ratio model based on the future power loads with the same preset prediction step; obtains the optimal future output of each distributed power source through global optimization, so that all optimal future outputs meet the optimal future output ratio; and sends each optimal future output to the corresponding distributed power source. This embodiment starts from two aspects: power generation prediction and power consumption prediction. It realizes power generation prediction by predicting environmental factors, and then realizes power consumption prediction by predicting power load. It builds a matching model according to the future output of each distributed power source, and finally finds the optimal matching of each distributed power source through global optimization. This allows this embodiment to meet the scheduling strategy of the power grid while being able to match in real time according to the highly random output conditions to ensure that the total output of the scheduling strategy (such as peak, valley, and equal) will not fluctuate or fluctuate within a smaller range, thereby ensuring the stable operation of the virtual power plant.

[0221] Figure 3Schematic diagram of the structure of an electronic device according to an embodiment of the present application. Figure 3 As shown, the electronic device 10 includes a processor 101 and a memory 102 coupled to the processor 101 .

[0222] The memory 102 stores program instructions for implementing a method for optimizing the output of a virtual power plant according to any of the above embodiments.

[0223] The processor 101 is configured to execute program instructions stored in the memory 102 to optimize the output of the virtual power plant.

[0224] The processor 101 may also be referred to as a CPU (Central Processing Unit). The processor 101 may be an integrated circuit chip having signal processing capabilities. The processor 101 may also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component. The general-purpose processor may be a microprocessor or any conventional processor.

[0225] Furthermore, Figure 4 This is a schematic diagram of the structure of the storage medium of an embodiment of the present application, see Figure 4 The storage medium 11 of the embodiment of the present application stores program instructions 111 that can implement all the above methods, wherein the program instructions 111 can be stored in the above storage medium in the form of a software product, including a number of instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) or a processor to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, or terminal devices such as a computer, a server, a mobile phone, and a tablet.

[0226] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0227] In addition, the functional units in the various embodiments of the present application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units. The above is only an implementation method of the present application and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the description and drawings of this application, or directly or indirectly used in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for optimizing the output of a virtual power plant, wherein the virtual power plant comprises at least two distributed power sources in a predetermined area integrated to a same predetermined monitoring terminal via signal and electrical connections, characterized in that: The output optimization method includes: Step S1, obtaining a plurality of environmental parameters of the preset area based on a plurality of preset time periods; Step S2: training all environmental parameters through a machine learning machine to obtain a number of future environmental parameters based on a number of preset prediction steps; Step S3, obtaining the future output of each distributed power source based on each preset prediction step number through all future environmental parameters; Step S4, constructing an output ratio model based on all future outputs; Step S5, obtaining a plurality of power loads of the preset area based on a plurality of preset time periods; Step S6, training all power loads through a machine learning machine, and obtaining a number of future power loads based on a number of the preset prediction steps; Step S7, obtaining the optimal future output ratio of all distributed power sources through the output ratio model according to the future power load of the same preset prediction step number; Step S8, obtaining the optimal future output of each distributed power source through global optimization, so that all optimal future outputs meet the optimal future output ratio; Step S9: Send each optimal future output to the corresponding distributed power source.

2. The output optimization method according to claim 1, characterized in that: Step S9, sending each optimal future output to the corresponding distributed power source, and then including: Step S10, obtaining the future timestamp of each optimal future output respectively; Step S20: When each future timestamp arrives, the corresponding distributed power source is adjusted to the corresponding optimal future output.

3. The output optimization method according to claim 2, characterized in that: Step S20, when each future time stamp arrives, adjusts the corresponding distributed power source to the corresponding optimal future output, and then includes: Step S100: Send all optimal future outputs and all future timestamps to the preset monitoring terminal.

4. The output optimization method according to claim 1, characterized in that: Step S2: All environmental parameters are trained by a machine learning machine, and several future environmental parameters are obtained based on several preset prediction steps, including: Step S21, integrating all environmental parameters of the same preset time period into an environmental parameter data set; Step S22, performing standard normalization processing on each environmental parameter data set, and obtaining a normalized parameter data set based on one environmental parameter data set; Step S23, dividing each normalized parameter data set into a parameter training set and a parameter verification set according to a preset ratio; Step S24, defining a neural network model in which the input layer, hidden layer, and output layer are sequentially connected; Step S25, inputting all parameter training sets into the input layer in sequence, and performing several trainings through the neural network model; Step S26, obtaining the root mean square error of the training results corresponding to the current parameter verification set and the current parameter training set based on each training; Step S27, obtaining the minimum error value among all root mean square errors; Step S28, obtaining a training result corresponding to the minimum error value as an environmental parameter prediction model; Step S29: predicting a number of future environmental parameters based on a number of preset prediction steps using the environmental parameter prediction model.

5. The output optimization method according to claim 1, wherein all distributed power sources include at least one wind power terminal, at least one photovoltaic terminal, and at least one hydropower terminal, characterized in that: Step S4: constructing an output ratio model based on all future outputs, including: Step S41, obtaining the future wind power output of the current wind power terminal, the future photovoltaic output of the current photovoltaic terminal, and the future hydropower output of the current hydropower terminal based on the future output of each distributed power source based on each preset prediction step; Step S42, performing time series modeling based on the future wind power output, future photovoltaic output, and future hydropower output for each preset prediction step, to obtain wind power output time series samples, photovoltaic output time series samples, and hydropower output time series samples, respectively; Step S43: establishing a standard matching model based on power balance constraints, sending-end capacity constraints, receiving-end capacity constraints, transmission security constraints, wind power and photovoltaic power constraints, and curtailment rate constraints; Step S44 , substituting the wind power output time series samples, the photovoltaic output time series samples, and the hydropower output time series samples into the standard ratio model to obtain the output ratio model.

6. The output optimization method according to claim 1, characterized in that: Step S6: training all power loads through a machine learning machine, and obtaining several future power loads based on several preset prediction steps, including: Step S61, integrating all power loads in the same preset time period into a power load data set; Step S62: performing standard normalization processing on each power load data set, and obtaining a normalized load data set based on the power load data set; Step S63, dividing each normalized load data set into a load training set and a load verification set according to a preset ratio; Step S64, defining a neural network model in which the input layer, hidden layer, and output layer are sequentially connected; Step S65, inputting all load training sets into the input layer in sequence, and performing several trainings through the neural network model; Step S66, obtaining the root mean square error of the training results corresponding to the current load verification set and the current load training set based on each training; Step S67, obtaining the minimum error value among all root mean square errors; Step S68, obtaining a training result corresponding to the minimum error value as a power load prediction model; Step S69: obtaining a plurality of future power loads through prediction of the power load prediction model based on a plurality of preset prediction steps.

7. The output optimization method according to claim 1, characterized in that: Step S8, obtaining the optimal future output of each distributed power source through global optimization, so that all optimal future outputs meet the optimal future output ratio, including: Step S81, defining a number of random solutions based on each optimal future output; Step S82, defining the optimization results of all random solutions as all optimal future outputs satisfying the optimal future output ratio; Step S83, initializing the position of each random solution, and updating the current position and current speed of each random solution respectively; Step S84, obtaining the individual optimal solution and the global optimal solution of each random solution based on each update; Step S85, respectively determining whether the difference between each individual optimal solution and the previously updated individual optimal solution is less than or equal to a first preset adaptation threshold; if both are less than, executing step S86; Step S86, respectively determining whether the difference between each global optimal solution and each global optimal solution updated last time is less than or equal to a second preset adaptation threshold; if both are less than, executing step S87; Step S87: Determine whether the optimal solution for all optimal future outputs has been obtained.

8. An output optimization device for a virtual power plant, the output optimization device being applied to the output optimization method according to any one of claims 1 to 7, characterized in that: The output optimization device comprises: A preset area environmental parameter acquisition module, used to acquire a plurality of environmental parameters of the preset area based on a plurality of preset time periods; A future environmental parameter prediction module is used to train all environmental parameters through a machine learning machine and obtain a number of future environmental parameters based on a number of preset prediction steps; A future output acquisition module is used to respectively acquire the future output of each distributed power source based on each preset prediction step number through all future environmental parameters; An output ratio model building module is used to build an output ratio model based on all future outputs; A preset area power load acquisition module, configured to acquire a plurality of power loads of the preset area based on a plurality of preset time periods; A future power load prediction module is used to train all power loads through a machine learning machine and obtain a number of future power loads based on a number of preset prediction steps; An optimal future output ratio acquisition module is used to obtain the optimal future output ratio of all distributed power sources through the output ratio model according to the future power load of the same preset prediction step number; An optimal future output acquisition module is used to obtain the optimal future output of each distributed power source through global optimization, so that all optimal future outputs meet the optimal future output ratio; The optimal future output sending module is used to send each optimal future output to the corresponding distributed power source.

9. An electronic device, characterized in that: It includes a processor and a memory coupled to the processor, wherein the memory stores program instructions that can be executed by the processor; when the processor executes the program instructions stored in the memory, the output optimization method as described in any one of claims 1 to 7 is implemented.

10. A storage medium, characterized in that: The storage medium stores program instructions, and when the program instructions are executed by the processor, the output optimization method according to any one of claims 1 to 7 can be implemented.

Images