A digital twin construction method for distributed resources of a virtual power plant

Abstract

The application provides a kind of digital twin construction method of virtual power plant distributed resource, comprising the following steps: S1, constructing the digital twin body simulation module model of each power plant in distributed resource;S2, the input variable of simulation module model is constituted into input vector set;S3, the output variable of simulation module model is constituted into output vector set;S4, the model parameter of simulation module model is constituted into model parameter set;S5, data is constituted into sample data set, and optimization objective function is determined;The digital twin simulation module model of photovoltaic power plant, wind power plant adopts multiple input variables, and the historical data of relevant input variables is easy to collect, and accurate prediction of near day (day and several days after it) can be carried out, which is beneficial to accurate prediction of power plant output after model establishment using digital twin simulation module model, and also beneficial to power optimization scheduling and power resource optimization allocation based on accurate prediction of power plant output.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of virtual power plants, and particularly relates to a digital twin construction method for distributed resources of a virtual power plant. BACKGROUND

[0002] A virtual power plant is a power coordination management system that realizes the aggregation and coordinated optimization of DG, energy storage systems, controllable loads, electric vehicles and other DERs through advanced information communication technology and software systems, and participates in the power market and grid operation as a special power plant. The core of the virtual power plant concept can be summarized as "communication" and "aggregation". The key technologies of the virtual power plant mainly include coordinated control technology, intelligent metering technology and information communication technology. The most attractive function of the virtual power plant is that it can aggregate DERs to participate in the operation of the power market and ancillary service market, and provide management and ancillary services for distribution networks and transmission networks.

[0003] Digital twin refers to the use of physical models, sensor updates, operation history and other data to integrate multidisciplinary, multi-physical, multi-scale and multi-probability simulation processes to complete mapping in a virtual space, thereby reflecting the whole life cycle process of the corresponding physical equipment.

[0004] The virtual power plant resources are widely distributed, and in actual use, the prediction of power output of the power plant is inaccurate, which affects the optimal scheduling of power and the optimal allocation of power resources, and at the same time, with the continuous operation of the power plant, the individual optimization is not compatible with the overall optimization of the virtual power plant.

[0005] Therefore, a digital twin construction method for distributed resources of a virtual power plant is proposed. SUMMARY

[0006] Therefore, the embodiments of the present application hope to provide a digital twin construction method for distributed resources of a virtual power plant to solve or alleviate the technical problems existing in the prior art, and at least provide a beneficial choice.

[0007] The technical scheme of the embodiments of the present application is as follows: a digital twin construction method for distributed resources of a virtual power plant, comprising the following steps:

[0008] S1, constructing a digital twin body simulation module model of each power plant in the distributed resources;

[0009] S2, collecting input variables of the simulation module model to form an input vector set;

[0010] S3, collecting output variables of the simulation module model to form an output vector set;

[0011] S4, collecting model parameters of the simulation module model to form a model parameter set;

[0012] S5, the set data constitutes a sample data set, and an optimization objective function is determined;

[0013] S6, a value range of a model parameter is determined to perform parameter optimization to obtain an optimized parameter value;

[0014] S7, simulation calculation and output prediction are performed using a module model, and the optimized sample data is updated.

[0015] Further preferably, in the step S1, when the digital twin simulation module model is constructed, the following steps are included:

[0016] S11, one power plant in the virtual power plant distributed resources is selected, a type of digital twin simulation module model of the power plant is selected, input variables and output variables of the digital twin simulation module model of the power plant are determined, and historical data thereof are collected;

[0017] S12, the collected historical data are divided into training sample data and optimized sample data of the digital twin simulation module model of the power plant;

[0018] S13, the training sample data of the digital twin simulation module model of the power plant are used to train the digital twin simulation module model of the power plant;

[0019] S14, a model parameter benchmark value of the digital twin simulation module model of the power plant is obtained after the training is completed;

[0020] S15, the steps S11-S14 are repeated until the digital twin simulation module models of all power plants in the virtual power plant distributed resources are constructed.

[0021] Further preferably, in the step S2, the input vector set is [D 11 ,D 12 ,…,D 1m1 ,…,D 2m2 ,D 31 ,…,D k1 ,D k2 ,…,D kmk ];

[0022] wherein k is the number of digital twin simulation module models of power plants, [D 11 ,D 12 ,…,D 1m1 ] are m1 input variables of a first digital twin simulation module model of a power plant, [D k1 ,D k2 ,…,D kmk ] are mk input variables of a kth digital twin simulation module model of a power plant; D 2m2 is the number of m2 input variables of a second digital twin simulation module model of a power plant; and D 31m1, m2, …, mk are respectively the number of input variables of the first, second, …, kth power system digital twin simulation module model.

[0023] Further preferably: in the step S3, the output vector set is [C 11 ,C 12 ,…,C 1i ,…,C 2i ,C 31 ,…,C k1 ,C k2 ,…,C ki ];

[0024] wherein [C 11 ,C 12 ,…,C 1i ] are i output variables of the first power system digital twin simulation module model, C 2i is the i output variable of the second power system digital twin simulation module model, C 31 is the first output variable of the third power system digital twin simulation module model, [C k1 ,C k2 ,…,C ki ] are i output variables of the kth power system digital twin simulation module model.

[0025] Further preferably: in the step S4, the model parameter set is [F 11 ,F 12 ,…,F 1n1 ,…,F 2n2 ,F 31 ,…,F k1 ,F k2 ,…,F knk ];

[0026] wherein [F 11 ,F 12 ,…,F 1n1 ] are n1 model parameters of the first power system digital twin simulation module model, F 2n2 is the n2 model parameter of the second power system digital twin simulation module model, F 31 is the first model parameter of the third power system digital twin simulation module model, [F k1 ,F k2 ,…,F knk ] are nk model parameters of the kth power system digital twin simulation module model; n1, n2, …, nk are respectively the number of model parameters of the first, second, …, kth power system digital twin simulation module model.

[0027] Further preferably, in the step S5, the optimization objective is minimization of the sum of absolute values of the model weighted relative errors and Q, and the optimization objective function is:

[0028]

[0029] wherein E is the weighted total output, and

[0030]

[0031] or any one of the equivalent forms, and the equivalent form is:

[0032]

[0033] wherein p is the 1st to the wth set of optimization sample data in the optimization sample data set; r represents the 1st to the kth power plant digital twin simulation module model; δ r is the power plant output weight coefficient corresponding to the rth power plant; q is the 1st to the ith output of the power plant digital twin simulation module model; E prq is the error between the qth output of the rth power plant digital twin simulation module model of the pth set of optimization sample data and the actual output value of the corresponding optimization sample data (i.e., the qth actual output value of the rth power plant of the pth set of optimization sample data); O prq is the qth actual output value of the rth power plant of the pth set of optimization sample data; is the average value of the qth actual output value of the rth power plant in the wth set of optimization sample data; μ q is the output variable weight coefficient.

[0034] For the power plant output weight coefficient δ r and the output variable weight coefficient μ q , respectively,

[0035]

[0036] P r is the historical average output active power of the rth distributed power plant, and for the power plant output weight coefficient δ r , there is:

[0037]

[0038] Further preferably, in the step S6, the value range of each model parameter in the model parameter set is determined; the value range of each model parameter in the model parameter set is between 1 / λ times and λ times of the model parameter reference value; λ is greater than or equal to 2 and less than or equal to 10;

[0039] and after determining the value range of each model parameter in the model parameter set, further comprising:

[0040] The model parameters in the model parameter set are iteratively optimized by using an optimization algorithm; and after the optimization calculation is completed, the optimal solution of the model parameter set is taken as a new model parameter benchmark value.

[0041] Further preferably, in the step S7, when the simulation calculation and the output prediction are performed by using the digital twin simulation module model of each power plant:

[0042] The sample data of the input variables and the output variables of the digital twin simulation module model of each power plant are continuously collected;

[0043] When the amount of the continuously collected sample data reaches or exceeds 20% of the amount of the optimization sample data set, the new collected sample data is used to replace the data in the optimization sample data set by using a rolling method;

[0044] The model parameters of the digital twin simulation module model of each power plant in the distributed resource are continuously integrated and optimized by repeating the steps S6-S7.

[0045] The application also provides a digital twin construction system of a virtual power plant distributed resource, which comprises:

[0046] A construction module, which is configured to construct the digital twin body simulation module model of each power plant in the distributed resource;

[0047] An acquisition module, which is configured to acquire the input variables of the simulation module model to form an input vector set, acquire the output variables of the simulation module model to form an output vector set, and acquire the model parameters of the simulation module model to form a model parameter set;

[0048] A target optimization module, which is configured to collect data to form a sample data set, determine an optimization target function, and determine the value range of the model parameters to perform parameter optimization and obtain an optimized parameter value;

[0049] A prediction module, which is configured to perform simulation calculation and output prediction by using the module model, and update the optimization sample data.

[0050] The application also provides a computer device, which comprises a processor and a memory coupled with the processor, and the memory stores program instructions, which are executed by the processor to make the processor perform the steps of the digital twin construction method of the virtual power plant distributed resource.

[0051] The application has the following advantages due to the above technical solutions:

[0052] The digital twin simulation module model of the photovoltaic power plant and the wind power plant adopts multiple input variables, historical data of relevant input variables is easy to collect, and the near-day (the current day and the next few days) accurate prediction can be performed, which is beneficial to the accurate prediction of power plant output after the model is established by using the digital twin simulation module model, and is also beneficial to the optimal scheduling of power and the optimal allocation of power resources on the basis of accurately predicting the power plant output.

[0053] Two, on the basis of independent modeling of the training data, the digital twin simulation module model of all distributed power plants is uniformly optimized in the whole parameter by using the optimization sample data set, and new historical data is collected and accumulated in the process of simulation, verification and prediction by using the power plant model, part of the old optimization sample data is replaced, and the whole parameter of the digital twin simulation module model of all distributed power plants is continuously intermittently optimized, so that the individual of the distributed power plant is always in a satisfactory operation state, and the individual optimization and the whole optimization of the virtual power plant are compatible and consistent.

[0054] The above summary is only for the purpose of the description and is not intended to limit in any way. In addition to the illustrative aspects, embodiments and features described above, further aspects, embodiments and features of the present application will be apparent from the drawings and the following detailed description. BRIEF DESCRIPTION OF DRAWINGS

[0055] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings needed in the embodiments or the prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0056] Figure 1 The flowchart of the method of the present application. DETAILED DESCRIPTION

[0057] In the following, only some exemplary embodiments are simply described. As those skilled in the art can recognize, the described embodiments can be modified in various different ways without departing from the spirit or scope of the present application. Therefore, the drawings and the description are considered to be exemplary in nature rather than limiting.

[0058] The embodiments of the present application will be described in detail below with reference to the drawings.

[0059] Embodiment one

[0060] As Figure 1As shown, the embodiment of the present application provides a digital twin construction method of distributed resources of a virtual power plant, comprising the following steps:

[0061] S1, constructing a digital twin simulation module model of each power plant in the distributed resources;

[0062] When the digital twin simulation module model is constructed, the following steps are included:

[0063] S11, selecting one power plant in the distributed resources of the virtual power plant, selecting a digital twin simulation module model type of the power plant, determining input variables and output variables of the digital twin simulation module model of the power plant and collecting historical data thereof;

[0064] S12, dividing the collected historical data into training sample data and optimization sample data of the digital twin simulation module model of the power plant;

[0065] S13, training the digital twin simulation module model of the power plant using the training sample data thereof;

[0066] S14, obtaining a model parameter benchmark value of the digital twin simulation module model of the power plant after the training is completed;

[0067] S15, repeating steps S11-S14 until the digital twin simulation module models of all power plants in the distributed resources of the virtual power plant are constructed;

[0068] S2, constructing an input vector set by collecting input variables of the simulation module models;

[0069] The input vector set is [D 11 ,D 12 ,…,D 1m1 ,…,D 2m2 ,D 31 ,…,D k1 ,D k2 ,…,D kmk ];

[0070] Wherein, k is the number of digital twin simulation module models of the power plant, [D 11 ,D 12 ,…,D 1m1 ] is m1 input variables of the first digital twin simulation module model, [D k1 ,D k2 ,…,D kmk ] is mk input variables of the kth digital twin simulation module model; D 2m2 is the number of m2 input variables of the second digital twin simulation module model; and D 31represents the number of the first input variables in the third power plant digital twin simulation module model; m1, m2, ..., mk represent the number of input variables in the first, second, ..., kth power plant digital twin simulation module models, respectively.

[0071] S3. The output vector set is composed of the output variables of the set simulation module model.

[0072] The output vector set is [C 11 C 12 ,…,C 1i ,…,C 2i C 31 ,…,C k1 C k2 ,…,C ki ];

[0073] Among them, [C 11 C 12 ,…,C 1i [i] represents the i-th output variable of the first power plant digital twin simulation module model, C 2i C is the i-th output variable of the second power plant digital twin simulation module model. 31 For the first output variable of the third power plant digital twin simulation module model, [C k1 C k2 ,…,C ki [i] represents the i-th output variable of the k-th power plant digital twin simulation module model;

[0074] S4. The model parameters of the simulation module model constitute the model parameter set;

[0075] The model parameter set is [F 11 ,F 12 ,…,F 1n1 ,…,F 2n2 ,F 31 ,…,F k1 ,F k2 ,…,F knk ];

[0076] Among them, [F 11 ,F 12 ,…,F 1n1 ] represents the n1 model parameters of the first power plant digital twin simulation module model, F 2n2 For the n2th model parameter of the second power plant digital twin simulation module model, F 31 For the first model parameter of the third power plant digital twin simulation module model, [F k1 ,F k2 ,…,F knknk is the number of model parameters of the kth power plant digital twin simulation module model; n1, n2, …, nk are respectively the number of model parameters of the 1st, 2nd, …, kth power plant digital twin simulation module model;

[0077] S5, the set data constitutes a sample data set, and an optimization objective function is determined;

[0078] The optimization objective is the minimization of the weighted relative error absolute value and Q of the model, and the optimization objective function is:

[0079]

[0080] E is the weighted total output, and has

[0081]

[0082] Or any one of the equivalent writings in the above, and the equivalent writing is:

[0083]

[0084] Wherein, p is the 1st to wth group of optimization sample data in the optimization sample data set; r represents the 1st to kth power plant digital twin simulation module model, δ r is the power plant output weight coefficient corresponding to the rth power plant; q is the 1st to ith output of the power plant digital twin simulation module model; E prq is the error between the qth output of the rth power plant digital twin simulation module model of the pth group of optimization sample data and the corresponding actual output value of the optimization sample data (i.e. the qth actual output value of the rth power plant of the pth group of optimization sample data), O prq is the qth actual output value of the rth power plant of the pth group of optimization sample data; is the average value of the qth actual output value of the rth power plant in the wth group of optimization sample data; μ q is the output variable weight coefficient.

[0085] For the power plant output weight coefficient δ r and the output variable weight coefficient μ q , respectively,

[0086]

[0087] P r is the historical average output active power of the rth distributed power plant, and for the power plant output weight coefficient δ r , has:

[0088]

[0089] S6, determine the value range of the model parameters to optimize the parameter values;

[0090] Determine the value range of each model parameter in the model parameter set; the value range of the model parameter in the model parameter set is between 1 / λ times and λ times of the model parameter reference value; λ is greater than or equal to 2 and less than or equal to 10;

[0091] And after determining the value range of each model parameter in the model parameter set, it further includes:

[0092] Using an optimization algorithm to iteratively optimize and calculate the model parameters in the model parameter set; after the optimization calculation is completed, the optimal solution of the model parameter set is taken as the new model parameter reference value

[0093] S7, using the module model to perform simulation calculation and output prediction, and updating the optimization sample data;

[0094] When each power plant digital twin simulation module model is used for simulation calculation and output prediction:

[0095] Continuously collect sample data of input variables and output variables of each power plant digital twin simulation module model;

[0096] When the amount of continuously collected sample data reaches or exceeds 20% of the amount of optimization sample data set data, use the rolling method to replace the data in the optimization sample data set with newly collected sample data;

[0097] Repeat steps S6-S7 to continuously integrate and optimize the parameters of the power plant digital twin simulation module model in each distributed resource;

[0098] The optimization algorithm uses genetic algorithm, or particle swarm algorithm, or wolf swarm algorithm. The end condition of optimization is that the absolute value of the weighted relative error Q of the model is less than the set error threshold ε, or the optimization calculation reaches the set iteration number.

[0099] i is greater than or equal to 1; each power plant digital twin simulation module model output variable includes at least active power output; the active power output included in the output variable is daily average active power output; the output variable weight coefficient corresponding to the daily average active power output is not less than 0.6.

[0100] The input variables of the photovoltaic power plant digital twin simulation module model include the month of the day, the daily maximum climate temperature, the daily average climate temperature, the daily weather state, and the construction age of the photovoltaic power plant.

[0101] The input variables of the wind power plant digital twin simulation module model include daily average wind speed, daily maximum wind speed, and the month (season) of the day.

[0102] The application further provides a digital twin construction system of virtual power plant distributed resources, comprising:

[0103] a construction module, the training module being used for constructing a digital twin simulation module model of each power plant in the distributed resources;

[0104] an acquisition module, the acquisition module being used for acquiring an input variable of the simulation module model to form an input vector set, acquiring an output variable of the simulation module model to form an output vector set, and acquiring a model parameter of the simulation module model to form a model parameter set;

[0105] a target optimization module, the target optimization module being used for collecting data to form a sample data set, determining an optimization target function, and determining a value range of the model parameter to perform parameter optimization and obtain an optimized parameter value;

[0106] a prediction module, the prediction module being used for performing simulation calculation and output prediction by using the module model, and updating the optimization sample data.

[0107] The application further provides a computer device, comprising a processor and a memory coupled with the processor, wherein the memory stores program instructions, and the program instructions are executed by the processor to make the processor execute the steps of the digital twin construction method of virtual power plant distributed resources.

[0108] Embodiment two

[0109] The application further provides an embodiment for practicing the method in embodiment one: constructing a digital twin simulation module model of the power plant, specifically, constructing a digital twin simulation module model of the photovoltaic power plant by using a BP neural network.

[0110] The input variables of the digital twin simulation module model of the photovoltaic power plant are a month in which a day is located M1, a daily maximum climate temperature T1, and a daily weather state A1. The month in which the day is located M1 is equal to a month value, for example, if the data day is in March, M1 is equal to 3; the daily weather state A1 includes five states of sunny, cloudy, overcast, rainy, and snowy, which are represented by numbers 1, 2, 3, 4, and 5 respectively. The output variables of the digital twin simulation module model of the photovoltaic power plant are a daily average active power output P G1 and a daily average reactive power output Q G1 . When the simulation module model in embodiment two is used for prediction, the month in which the day is located M1 is actual data of a predicted day, and the daily maximum climate temperature T1 and the daily weather state A1 are predicted day data estimated according to a weather forecast.

[0111] Embodiment two selects a three-layer BP network with four hidden layer nodes, wherein 3x4, a total of 12 weight coefficients between the input layer and the hidden layer nodes, are V G1 to VG12 , the threshold value of the 4 hidden layer nodes is θ G1 to θ G4 ; the 4x2=8 weight coefficients between the hidden layer nodes and the output nodes are V G13 to V G20 , the threshold value of the 2 output nodes is θ G5 , θ G6 . At this time, the power digital twin simulation module model embodiment two has a total of 26 model parameters, which are weight coefficients V G1 to V G20 , threshold values θ G1 to θ G6 . The activation functions of the output layer and the hidden layer are both unipolar Sigmoid functions.

[0112] The historical data of the power digital twin simulation module model embodiment two are the data of the month M1 in which the day is located, the daily maximum climate temperature T1, and the daily weather state A1 for at least 2 years, as well as the corresponding daily average active power output and daily average reactive power output actual value data. Randomly select a part of them (between 30% and 70%), for example, randomly select half of the data as training sample data, and leave the remaining half of the data as optimization sample data. Using the selected training sample data, the neural network is trained using the gradient descent method or other optimization algorithms such as particle swarm, genetic algorithm, etc., and the training target is to minimize the system average error (minimize the average error sum of squares); suppose the values of the weight coefficients V G1 to V G20 of the neural network after training are V gG1 to V gG20 , and the threshold values θ G1 to θ G6 of the neural network after training are θ gG1 to θ gG6 , then V gG1 to V gG20 and θ gG1 to θ gG6 are the model parameter baseline values of the power digital twin simulation module model embodiment two. In addition, the average value of the actual daily average active power output in the training sample data of embodiment two is calculated, which is taken as the historical average output active power P1 of the power digital twin simulation module model embodiment two; or, the average value of the actual daily average active power output in all historical data (including training sample data and optimization sample data) of embodiment two is calculated, which is taken as the historical average output active power P1 of the power digital twin simulation module model embodiment two.

[0113] Embodiment three

[0114] The application also provides an embodiment for practicing the method according to embodiment one: using the RBF neural network to construct a digital twin simulation module model of a photovoltaic power plant, the input variables being a month in which a day is located M2, diurnal average climate temperature T Q2 , day weather state A2, and service life of the photovoltaic power plant S2. The month in which the day is located M2 has the same meaning as the month in which the day is located M1 in embodiment two, and the day weather state A2 has the same meaning as the day weather state A1 in embodiment two; the diurnal average climate temperature T Q2 is the average value of the temperature measured at the photovoltaic cell of the photovoltaic power plant in the white day interval in one day (uniformly in Celsius), the white day interval is determined according to common sense, the measurement interval is greater than or equal to 30 minutes and less than or equal to 2 hours, for example, the average value of the temperature measured every 1 hour between 5:00 in the morning and 7:00 in the evening (i.e. the white day interval) in one day is taken as the diurnal average climate temperature T Q2 ; the service life of the photovoltaic power plant S2 is 1 in the first year after the construction of the photovoltaic power plant, 2 in the second year, and so on. The output variables of the digital twin simulation module model of the photovoltaic power plant are daily average active power output P G2 and daily average reactive power output Q G2 . When the simulation module model of embodiment three is used for prediction, the month in which the day is located M2 and the service life of the photovoltaic power plant S2 are the actual data of the predicted day, and the diurnal average climate temperature T Q2 and the day weather state A2 are the predicted day data estimated according to weather forecasts and the like.

[0115] The RBF neural network of embodiment three uses five hidden layer nodes with Gaussian functions as the action functions, and the center points corresponding to the four inputs M2, T Q2 , A2 and S2 of the five hidden layer nodes are respectively [Z 11 , Z 12 , Z 13 , Z 14 ], [Z 21 , Z 22 , Z 23 , Z 24 ], [Z 31 , Z 32 , Z 33 , Z 34 ], [Z 41 , Z 42 , Z 43 , Z 44 ] and [Z 51 , Z 52 , Z 53 , Z 54 ]; and the standardization constants (base widths) of the five hidden layer nodes are respectively |φ Z1 , φ z2 , φ z3 , φz4 , φ z5 | 5 hidden layer nodes and 2 output nodes P G2 , Q G2 between the connection weight coefficients are [Y 11 , Y 12 , Y 13 , Y 14 , Y 15 ], [Y 21 , Y 22 , Y 23 , Y 24 , Y 25 ]. At this time, the electrical number twin simulation module model embodiment three has 35 model parameters in total.

[0116] The historical data of embodiment three is the data of the day in the month M2, the diurnal average climate temperature T Q2 , the daily weather state A2, the photovoltaic power plant construction age S2 for at least 2 years, and the corresponding daily average active power output, daily average reactive power output actual value data. Randomly select a part of them (between 30% and 70%), for example, randomly select half of the data as training sample data, and leave the remaining half of the data as optimization sample data. Using the selected training sample data, the center point parameters and the standardization constant (base width) parameters of the Gaussian function are obtained by using the clustering algorithm; the connection weight coefficients between the 5 hidden layer nodes and the 2 output nodes are obtained by using the least square algorithm in the identification theory. The model parameter reference values obtained after training are that the 5 center points are [Z g11 , Z g12 , Z g13 , Z g14 ], [Z g21 , Z g22 , Z g23 , Z g24 ], [Z g31 , Z g32 , Z g33 , Z g34 ], [Z g41 , Z g42 , Z g43 , Z g44 ], [Z g51 , Z g52 , Z g53 , Z g54 ], the standardization constants (base width) of the 5 hidden layer nodes are [φ gZ1 , φ gZ2 , φ gZ3 , φ gZ4 , φ gZ5 ], and the connection weight coefficients between the 5 hidden layer nodes and the 2 output nodes are [Y g11 , Yg12 , Y g13 , Y g14 , Y g15 , Y g21 , Y g22 , Y g23 , Y g24 , Y g25 Similarly, the average value of the actual daily average active power output in the training sample data of Example Three is calculated, which is taken as the historical average output active power P2 of the digital twin simulation module model of the power plant of Example Three; alternatively, the average value of the actual daily average active power output in all historical data (including the training sample data and the optimization sample data) of Example Three is calculated, which is taken as the historical average output active power P2 of the digital twin simulation module model of the power plant of Example Three.

[0117] Example Two adopts a BP neural network to construct the digital twin simulation module model of the power plant, and the input variables can be changed, or added / removed, or changed and added / removed. For example, the daily maximum climate temperature is changed to the daily average climate temperature, or the daily maximum climate temperature and the daily average climate temperature are included in the input; and the service life of the photovoltaic power plant is added to the input. Assuming that the input variables of Example Two include the service life S1 of the photovoltaic power plant and the daily average climate temperature, i.e., the input variables are the month M1 in which the day falls, the daily maximum climate temperature T1, the daily average climate temperature T1, the service life S1 of the photovoltaic power plant, and the daily weather state A1, and the output variables are the daily average active power output P1 and the daily average reactive power output Q1, the input variables are changed and added / removed. Q1 G1 G1 , the three-layer BP network with 6 hidden layer nodes is selected, wherein 30 weight coefficients V1 to V30 between the input layer and the hidden layer nodes are set, 6 threshold values θ1 to θ6 of the hidden layer nodes are set, 12 weight coefficients V1 to V12 between the hidden layer nodes and the output nodes are set, and 2 threshold values θ1 to θ2 of the output nodes are set. H1 H30 H1 H6 H31 H42 H7 H8 At this time, the digital twin simulation module model of the power plant has 50 model parameters.

[0118] ​​​​​​​​​The third embodiment uses an RBF neural network to construct a digital twin simulation module model of a photovoltaic power plant. The input variables can also be changed, or added / removed, or changed and added / removed. For example, the average daily climate temperature is changed to the daily maximum climate temperature; and the input variable of the photovoltaic power plant construction age is deleted simultaneously or separately. It is assumed that the input variables of the third embodiment are the month M2 in which the day is located, the daily maximum climate temperature T2, the daily weather state A2, and the photovoltaic power plant construction age S2; and the output variables are the daily average active power output P G2 and the daily average reactive power output Q G2 . The action functions of the six hidden layer nodes are Gaussian functions, the center points of the six hidden layer nodes corresponding to the inputs M2, T2, A2, and S2 are [B 11 , B 12 , B 13 , B 14 ], [B 21 , B 22 , B 23 , B 24 ], [B 31 , B 32 , B 33 , B 34 ], [B 41 , B 42 , B 43 , B 44 ], [B 51 , B 52 , B 53 , B 54 ], and [B 61 , B 62 , B 63 , B 64 ]; the standardization constants (base widths) of the six hidden layer nodes are |φ B1 , φ B2 , φ B3 , φ B4 , φ B5 , φ B6 |; and the connection weight coefficients between the six hidden layer nodes and the two output nodes P G2 and Q G2 are [YB 11 , Y B12 , Y B13 , Y B14 , Y B15 , Y B16 ], [Y B21 , Y B22 , Y B23 , Y B24 , Y B25 , Y B26At this time, the power digital twin simulation module model embodiment three has 42 model parameters in total.

[0119] Embodiment four

[0120] The present application also provides an embodiment for practicing the method according to embodiment one: constructing a power digital twin simulation module model, specifically, constructing a wind power digital twin simulation module model using a BP neural network. The input variables of the wind power digital twin simulation module model are daily average wind speed S F1 , daily maximum wind speed S F2 , and the season in which the day falls S F3 , wherein the daily average wind speed S F1 is the average value of real-time wind speed in a day, which can be the average value of multiple sampling values or the integral average value of continuous sampling; the season in which the day falls S F3 is taken as 1, 2, 3, and 4 respectively when the data day falls in spring, summer, autumn, and winter. The output variables of the wind power digital twin simulation module model are daily average active power output P G3 and daily average reactive power output Q G3 . When the embodiment four simulation module model is used for prediction, the season in which the day falls S F3 is the actual data of the day to be predicted, and the daily average wind speed S F1 and the daily maximum wind speed S F2 are the data of the day to be predicted estimated according to weather forecasts and the like.

[0121] Embodiment four selects a three-layer BP network with 5 hidden layer nodes, wherein 3x5=15 weight coefficients between the input layer and the hidden layer nodes are denoted as V F1 to V F15 , the thresholds of the 5 hidden layer nodes are denoted as θ F1 to θ F5 ; 5x2=10 weight coefficients between the hidden layer nodes and the output nodes are denoted as V F16 to V F25 , and the thresholds of the 2 output nodes are denoted as θ F6 and θ F7 . At this time, the power digital twin simulation module model embodiment four has 32 model parameters in total, which are the weight coefficients V F1 to V F25 and the thresholds θ F1 to θ F7 . The excitation functions of the output layer and the hidden layer adopt unipolar Sigmoid functions.

[0122] The historical data of the power digital twin simulation module model embodiment four are the daily average wind speed S F1 , the daily maximum wind speed S F2 , and the season in which the day falls S F3Data, and corresponding daily average active power output, daily average reactive power output actual value data. Randomly one part (30% to 70%) of them, for example, randomly half of the data is a training sample data, the remaining half of the data is left as an optimization sample data. Using the selected training sample data, using gradient descent method, or particle swarm, genetic algorithm and other optimization algorithms, the neural network is trained, and the training target is to minimize the system average error (minimum average error square sum); Set the weight coefficient V F1 to V F25 of the neural network after training respectively V gF1 to V gF25 , the threshold value θ F1 to θ F7 of the neural network after training respectively θ gF1 to θ gF7 , V gF1 to V gF25 and θ gF1 to θ gF7 are the model parameter reference values of the digital twin simulation module model embodiment four of the power station. In addition, the average value of the actual daily average active power output in the training sample data of embodiment four is calculated, which is used as the historical average output active power P3 of the digital twin simulation module model embodiment four of the power station; or, the average value of the actual daily average active power output in all historical data (including training sample data and optimization sample data) of embodiment four is calculated, which is used as the historical average output active power P3 of the digital twin simulation module model embodiment four of the power station.

[0123] Embodiment five

[0124] The RBF neural network is used to construct the digital twin simulation module model of the wind power station, the input variables are daily average wind speed S F1 , daily maximum wind speed S F2 , and the month M4 in which the day falls, and the output variables are daily average active power output P G4 and daily average reactive power output Q G4 . When the simulation module model of embodiment five is used for prediction, the month M4 in which the day falls is the actual data of the predicted day, the daily average wind speed S F1 , and the daily maximum wind speed S F2 are estimated according to the weather forecast.

[0125] The RBF neural network of embodiment five uses six hidden layer nodes with Gaussian function as the action function, and the center points corresponding to the inputs S F1 , S F2 , and M4 are respectively [R 11 , R 12 , and R 13 ], [R 21, R 22 , R 23 ], [R 31 , R 32 , R 33 ], [R 41 , R 42 , R 43 ], [R 51 , R 52 , R 53 ], [R 61 , R 62 , R 63 ] ; the standardization constants (base width) of the 6 hidden layer nodes are [φ R1 , φ R2 , φ R3 , φ R4 , φ R5 , φ R6 ] respectively. The connection weight coefficients between the 6 hidden layer nodes and the 2 output nodes P G4 , Q G4 , P M4 are [YR 11 , YR 12 , YR 13 , YR 14 , YR 15 , YR 16 ], [YR 21 , YR 22 , YR 23 , YR 24 , YR 25 , YR 26 ] respectively. At this time, the electric number digital twin simulation module model embodiment five has a total of 36 model parameters.

[0126] The historical data of embodiment five are the daily average wind speed S F1 , the daily maximum wind speed S F2 , the month M4 in which the day falls, and the corresponding daily average active power output and daily average reactive power output in the past 2 years. Randomly select a part of them (between 30% and 70%), for example, randomly select half of the data as training sample data, and leave the remaining half of the data as optimization sample data. Use the selected training sample data to calculate the 6 center point parameters and the standardization constant (base width) parameters of the Gaussian function using the clustering algorithm; use the least square algorithm in identification theory to calculate the connection weight coefficients between the 6 hidden layer nodes and the 2 output nodes. The model parameter reference values obtained after training are that the 6 center points are [R g11 , R g12 , R g13 ], [R g21 , R g22 , R g23 ], [Rg31 , R g32 , R g33 ], [R g41 , R g42 , R g43 ], [R g51 , R g52 , R g53 ], [R g61 , R g62 , R g63 ], the standardization constants (base width) of the 6 hidden layer nodes are [φ gR1 , φ gR2 , φ gR3 , φ gR4 , φ gR5 , φ gR6 ] respectively, and the connection weight coefficients between the 6 hidden layer nodes and the 2 output nodes are [Y gR11 , Y gR12 , Y gR13 , Y gR14 , Y gR15 , Y gR16 ], [Y gR21 , Y gR22 , Y gR23 , Y gR24 , Y gR25 , Y gR26 ] respectively. Similarly, the average value of the actual daily average active power output in the training sample data of Example Five is calculated, which is taken as the historical average output active power P4 of the digital twin simulation module model of the power plant of Example Five; alternatively, the average value of the actual daily average active power output in all historical data of Example Five is calculated, which is taken as the historical average output active power P4 of the digital twin simulation module model of the power plant of Example Five.

[0127] Conclusion:

[0128] The output variables of Examples Two to Five all include daily average active power output and daily average reactive power output. While retaining the daily average active power output, output variables can be added / removed, for example, all daily maximum active power outputs are added, or all daily average reactive power outputs are removed, etc.

[0129] The foregoing method of using BP neural network and RBF neural network to construct the digital twin simulation module model of the photovoltaic power plant and the wind power plant can also be used to construct the digital twin simulation module model of the hydropower plant and the thermal power plant. The digital twin simulation module model of the photovoltaic power plant, the wind power plant, the hydropower plant, and the thermal power plant can also use other neural networks, for example, DRNN diagonal recurrent neural network and CMAC cerebellar neural network model, etc.

[0130] Embodiment six

[0131] The application also provides a model parameter embodiment of integrating and optimizing each power plant digital twin simulation module model in the distributed resources of the virtual power plant.

[0132] The model parameters of the digital twin simulation module models of each power plant in the distributed resources are integrated and optimized, specifically, assuming that the virtual power plant has the three distributed power plants in the foregoing embodiment two, embodiment three and embodiment four, that is, k is equal to 3.

[0133] Step (1), the input variables of all the digital twin simulation module models of the power plants in the distributed resources are collected to form an input vector set. The input variables of embodiment two are the month M1 in which the day is located, the daily maximum climate temperature T1, and the daily weather state A1, a total of three input variables, that is, m1 is equal to 3; the input variables of embodiment three are the month M2 in which the day is located, the daily average climate temperature T Q2 , the daily weather state A2, and the age S2 of the photovoltaic power plant, a total of four input variables (m2 is equal to 4); the input variables of embodiment four are the daily average wind speed S F1 , the daily maximum wind speed S F2 , and the season S F3 in which the day is located, a total of three input variables (m3 is equal to 3). The input vector set formed by the input variables of all the digital twin simulation module models of the power plants in the distributed resources is [M1, T1, A1, M2, T Q2 , A2, S2, S F1 , S F2 , S F3 ].

[0134] Step (2), the output variables in the output variables of all the digital twin simulation module models of the power plants are collected to form an output vector set. The output variables of embodiment two are the daily average active power output P G1 and the daily average reactive power output Q G1 , the output variables of embodiment three are the daily average active power output P G2 and the daily average reactive power output Q G2 , and the output variables of embodiment four are the daily average active power output P G3 and the daily average reactive power output Q G3 , so the output variables of each digital twin simulation module model of the power plant are two, i is equal to 2; the output vector set formed by the output variables of all the digital twin simulation module models of the power plants is [P G1 , Q G1 , P G2 , Q G2 , P G3 , Q G3 ].

[0135] Step (3), the model parameter set is constituted by the model parameters of all the digital twin simulation module models of the power plant. The 26 model parameters (n1 is equal to 26) of the embodiment two are respectively the weight coefficients V G1 to V G20 , the threshold values θ G1 to θ G6 ; the 35 model parameters (n2 is equal to 35) of the embodiment three are respectively the center points [Z 11 , Z 12 , Z 13 , Z 14 ], [Z 21 , Z 22 , Z 23 , Z 24 ], [Z 31 , Z 32 , Z 33 , Z 34 ], [Z 41 , Z 42 , Z 43 , Z 44 ], [Z 51 , Z 52 , Z 53 , Z 54 ], the standardization constants (base width) [φ Z1 , φ Z2 , φ Z3 , φ z4 , φ z5 ], the connection weight coefficients [Y 11 , Y 12 , Y 13 , Y 14 , Y 15 ], [Y 21 , Y 22 , Y 23 , Y 24 , Y 25 ]; the 32 model parameters (n3 is equal to 32) of the embodiment four are respectively the weight coefficients V F1 to V F25 , the threshold values θ F1 to θ F7° The model parameter set constituted by the model parameters of all the digital twin simulation module models of the power plant is

[0136] [V G1 ,…,V G20 ,θ G1 ,…,θ G6 ,Z 11 ,…Z 54 ,φ Z1 ,…,φ z5 ,Y 11 ,…,Y25 V F1 ,…,V F25 ,θ F1 There are a total of 93 model parameters to be optimized.

[0137] Step (4): The optimized sample dataset is constructed by combining the optimized sample data from all digital twin simulation module models of the virtual power plant distributed resources. In Example 6, the optimized sample dataset is constructed by combining the optimized sample data from Examples 2 to 4. The total amount of historical data from the power plant digital twin simulation module models may not be the same, but the number of data sets retained as optimized sample data (i.e., the number of days of daily data) will vary.

[0138] They should be the same. For example, if Examples 2, 3, and 4 collect 800, 1000, and 1100 days of historical data respectively, then 400 days (400 sets) of historical data can be retained as optimization sample data, and the remaining 400, 600, and 700 days of historical data can be used as training sample data for the digital twin simulation module model of Examples 2, 3, and 4 respectively; or 450 days of historical data can be retained as optimization sample data, and the remaining 350, 550, and 650 days of historical data can be used as training sample data for the digital twin simulation module model of Examples 2, 3, and 4 respectively.

[0139] Step 5: Determine the objective function. Assume that Examples 2, 3, and 4 all select 400 days of historical data as the optimization sample data, i.e., there are a total of 400 sets of optimization sample data, and w equals 400; the objective function of Example 6 is...

[0140]

[0141] Where the weighted total output E is

[0142]

[0143] Or equivalently written as

[0144]

[0145] Right now

[0146] E=400μ1(δ1·P W1 +δ2·P W2 +δ3·P W3 )+400μ2(δ1·Q W1 +δ2′Q W2 +δ3·Q W3 (4):

[0147] in, It is P W1is the average value of actual daily average active power output in the 400 sets of optimization sample data of example 1; is Q W1 is the average value of actual daily average reactive power output in the 400 sets of optimization sample data of example 1; is P W2 is the average value of actual daily average active power output in the 400 sets of optimization sample data of example 3; is Q W2 is the average value of actual daily average reactive power output in the 400 sets of optimization sample data of example 3; is P W3 is the average value of actual daily average active power output in the 400 sets of optimization sample data of example 3; is Q W3 is the average value of actual daily average reactive power output in the 400 sets of optimization sample data of example 3.

[0148]

[0149]

[0150]

[0151] The output variables of examples two, three and four are daily average active power output and daily average reactive power output. Considering that daily average active power output plays a greater role in resource allocation and scheduling of a virtual power plant, the output variable weight coefficient corresponding to daily average active power output is not less than 0.6 in the output variable weight coefficient. In example six, the output variable weight coefficient μ1 corresponding to daily average active power output is 0.7, and the output variable weight coefficient μ2 corresponding to daily average reactive power output is 0.3.

[0152] Step 9, determine the value range of each model parameter in the model parameter set. The value rule of each model parameter in the model parameter set is that the value range of a certain model parameter is between 1 / λ of the benchmark value of the model parameter and λ times of the benchmark value. In example six, λ is equal to 5, so the value range of each model parameter in the model parameter set is between 0.2 times (1 / 5) and 5 times of the benchmark value of the model parameter; for example, the value interval of V G1 is [0.2V gG1 5V gG1 ], the value interval of V G2 is [0.2V gG2 5V gG2 ], and the value interval of θ G1 is [0.2θgG1 5θ gG1 ],Z 11 the value interval of θ is [0.2θ g11 5Z g11 ],θ F7 the value interval of θ is [0.2θ gF7 5θ gF7 ]; and so on.

[0153] Step ⑺, the parameters in the model parameter set are iteratively optimized by using an optimization algorithm. In Example 6, the particle swarm optimization algorithm is used for optimization, and the end condition of optimization is that the absolute value of the weighted relative error of the model Q is less than the set error threshold ε, or the optimization iteration reaches the set iteration number. ε is valued between 0.01 and 0.1, and in Example 6, ε is valued at 0.05, and the set iteration number is 1500. The optimization algorithm can also use genetic algorithm, ant colony algorithm, wolf swarm algorithm, etc.

[0154] After the optimization calculation is completed, the optimal solution of the model parameter set is taken as the new model parameter benchmark value.

[0155] Step ⑻, the virtual power plant digital twin simulation module model is simulated and output predicted with the new model parameter benchmark value. The output prediction refers to calculating the output value of the digital twin simulation module model according to the estimated and predicted (i.e. estimated according to the weather forecast) input data of the digital twin simulation module model, and taking the output value as the predicted value to perform power dispatching operation. For example, according to the estimated second day month M1, daily maximum climate temperature T1, daily weather state A1 data, the daily average active power output and the daily average reactive power output of the second day are calculated by using the digital twin simulation module model of Example 1.

[0156] While simulating and predicting the output by using the digital twin simulation module model of each power plant, the sample data of the input variables and output variables of the digital twin simulation module model of each power plant are continuously collected; when the amount of continuously collected sample data reaches or exceeds 20% of the amount of optimization sample data set data, the newly collected sample data is replaced with the optimization sample data set data by using the rolling method, that is, the latest several groups of sample data are replaced with the oldest same number of groups of optimization sample data in the optimization sample data set; for example, the optimization sample data set of Example 6 has a total of 400 groups (400 days) of sample data, when the amount of continuously collected sample data of the input variables and output variables of the digital twin simulation module model of each power plant reaches or exceeds 80 groups (80 days) of sample data, the latest 80 groups (or more than 80 groups) of sample data are replaced with the oldest 80 groups (or more than 80 groups) of optimization sample data in the optimization sample data set. The parameters of the digital twin simulation module model of each distributed resource are continuously integrated and optimized by repeating steps ⑹, ⑺ and ⑻.

[0157] Digital twin is a virtual entity of a physical entity created in a digital way, which simulates, verifies, predicts and controls the whole life cycle of the physical entity by means of historical data, real-time data and algorithmic models.

[0158] The virtual power plant adopts a hardware system architecture of self-discipline decentralization, cloud-edge collaboration and virtual-real integration, takes edge computing services and controlled distributed resources as an atomic node, and makes the gateway carry autonomous operation programs to control the distributed resources on site, so that the atomic node is controllable. The gateway realizes dynamic construction of logical relationship through message subscription / publishing, so that the atomic node is self-disciplined and coordinated. The virtual entity or virtual gateway can participate in communication with the physical entity or gateway as an equal individual, so as to realize virtual-real integration in the coordination mechanism.

[0159] The virtual power plant adopts a three-layer software architecture of connection-aggregation-application. The connection layer is mainly realized by the edge computing services carried by the gateway, which realizes intelligent perception of distributed resources by multiple intelligent modules, and realizes simulation and interaction of distributed resources by digital twin modules. The aggregation layer builds a self-disciplined decentralized architecture through a middleware server, and provides message configuration and forwarding for information interaction of edge computing nodes. The application layer combines various application scenarios of the virtual power plant, customizes business logic and guidance rules of response process, and realizes self-optimization operation of massive distributed resources of the virtual power plant.

[0160] The power plant in the distributed resources of the virtual power plant is used for producing and outputting electric energy. It is the premise of the virtual power plant to aggregate and dispatch, and to intelligently distribute electric power to obtain and accurately predict the ability of each distributed power plant to produce and output electric energy through various ways. On the basis of the existing hardware system architecture and software architecture of the virtual power plant, the quality of the digital twin simulation module model of the distributed power plant used for simulating and predicting power output is the key to accurately verifying, predicting and controlling the virtual power plant.

[0161] The input variables of the digital twin simulation module model of the photovoltaic power plant include the month of the day, the daily maximum climate temperature, the daily average climate temperature, the daily weather state and the service life of the photovoltaic power plant, and the input variables of the digital twin simulation module model of the wind power plant include the daily average wind speed, the daily maximum wind speed and the month (season) of the day, the historical data of the related input variables are easy to collect, and the accurate prediction of the near day (the day and the following days) can be carried out, which is beneficial to the accurate prediction of the power output of the power plant by using the digital twin simulation module model after the model is established, and is also beneficial to the optimal scheduling of the power and the optimal allocation of the power resources on the basis of the accurate prediction of the power output of the power plant. The power plants in the virtual power plant are independently modeled based on the training data, then the overall parameter optimization of the digital twin simulation module models of all the distributed power plants is uniformly carried out by using the optimized sample data set, and new historical data is collected and accumulated in the process of simulation, verification and prediction by using the power plant model, part of the old optimized sample data is replaced, the overall parameter optimization of the digital twin simulation module models of all the distributed power plants is continuously and intermittently carried out, the individual of the distributed power plant is ensured to be in a satisfactory operation state, and the individual optimization and the overall optimization of the virtual power plant are compatible and consistent.

[0162] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of various changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for constructing a digital twin of distributed resources of a virtual power plant, characterized in that, The method comprises the following steps: S1, constructing a digital twin simulation module model of each power plant in the distributed resource; S2, collecting input variables of the simulation module model to form an input vector set; S3, collecting output variables of the simulation module model to form an output vector set; S4, collecting model parameters of the simulation module model to form a model parameter set; S5, collecting optimization sample data of the simulation module model to form an optimization sample data set, and determining an optimization model objective function, wherein the optimization objective is minimization of a weighted relative error absolute value and Q of the model; S6, determining a value range of each model parameter in the model parameter set to perform parameter optimization to obtain a new model parameter benchmark value; S7, performing simulation calculation and output prediction using the module model with the new model parameter benchmark value, and updating the optimization sample data; In the step S1, when the digital twin simulation module model is constructed, the following steps are included: S11, selecting one power plant in the virtual power plant distributed resource, selecting a digital twin simulation module model type of the power plant, determining input variables and output variables of the digital twin simulation module model of the power plant, and collecting historical data of the digital twin simulation module model; S12, dividing the collected historical data into training sample data and optimization sample data of the digital twin simulation module model of the power plant; S13, training the digital twin simulation module model of the power plant using the training sample data of the digital twin simulation module model; S14, obtaining a model parameter benchmark value of the digital twin simulation module model of the power plant after the training is completed; S15, repeating the steps S11-S14 until the digital twin simulation module models of all power plants in the virtual power plant distributed resource are constructed; In the step S2, the input vector set is ; wherein k is the number of power system digital twin simulation module models, m1 is the m1th input variable of the first power system digital twin simulation module model, mk is the mkth input variable of the kth power system digital twin simulation module model; m2 is the m2th input variable of the second power system digital twin simulation module model, m1, m2, …, mk are respectively the number of input variables of the first, second, …, kth power system digital twin simulation module model; In the step S3, the output vector set is ; wherein, is the i-th output variable of the 1st electrical system digital twin simulation module model, is the i-th output variable of the 2nd electrical system digital twin simulation module model, is the 1st output variable of the 3rd electrical system digital twin simulation module model, is the i-th output variable of the k-th electrical system digital twin simulation module model; In the step S4, the model parameter set is ; wherein, n1 model parameters of the 1st power system digital twin simulation module model, n2 model parameters of the 2nd power system digital twin simulation module model, n1 model parameters of the 1st power system digital twin simulation module model, nk model parameters of the kth power system digital twin simulation module model; n1, n2, …, nk are the number of model parameters of the 1st, 2nd, …, kth power system digital twin simulation module model, respectively.

2. The method of claim 1, wherein: In the step S5, the optimization objective is minimization of a weighted relative error absolute value and Q of the model, and the optimization objective function is: ； Wherein, E is a weighted total output, and ； Or equivalently written as: ； Wherein, p is the 1st to the wth group of optimization sample data in the optimization sample data set; r represents the 1st to the kth power plant digital twin simulation module model, is the power plant output weight coefficient corresponding to the rth power plant; q is the 1st to the ith output of the power plant digital twin simulation module model; is the error between the qth output of the rth power plant digital twin simulation module model of the pth group of optimization sample data and the corresponding actual output value of the optimization sample data, is the qth actual output value of the rth power plant of the pth group of optimization sample data; is the average value of the qth actual output value of the rth power plant in the wth group of optimization sample data; is the output variable weight coefficient; For power plant output weight coefficients and output variable weight coefficients , respectively, there are ； ； For the historical average output active power of the rth distributed power plant, for the power plant output weight coefficient , there is: 。 3. The method of claim 1, wherein: In the step S6, the value range of each model parameter in the model parameter set is determined; the value range of each model parameter in the model parameter set is 1 / λ times to λ times of the model parameter benchmark value; λ is greater than or equal to 2 and less than or equal to 10; After the value range of each model parameter in the model parameter set is determined, the following steps are further included: An optimization algorithm is used to iteratively optimize and calculate the model parameters in the model parameter set; After the optimization calculation is completed, the optimal solution of the model parameter set is taken as a new model parameter benchmark value.

4. The method of claim 1, wherein: In the step S7, when each digital twin simulation module model of the power plant is used to perform simulation calculation and output prediction: Sample data of input variables and output variables of each digital twin simulation module model of the power plant are continuously collected; When the amount of the continuously collected sample data reaches or exceeds 20% of the amount of the optimization sample data set, the rolling method is used to replace the data in the optimization sample data set with newly collected sample data; The steps S6-S7 are repeated to continuously integrate and optimize the model parameters of the digital twin simulation module models of the power plants in the distributed resource.

5. The digital twin construction system of distributed resources of a virtual power plant according to any one of claims 1-4, characterized in that, The method comprises the following steps: A construction module is configured to construct a digital twin simulation module model of each power plant in the distributed resource; An acquisition module is configured to acquire an input variable of a simulation module model to form an input vector set, acquire an output variable of the simulation module model to form an output vector set, and acquire a model parameter of the simulation module model to form a model parameter set; A target optimization module is configured to collect data to form a sample data set, determine an optimization target function, and determine a value range of the model parameter to perform parameter optimization to obtain an optimized parameter value; A prediction module is configured to perform simulation calculation and output prediction using the module model, and update the optimized sample data.

6. A computer device, comprising: The computer device comprises a processor and a memory coupled to the processor, and the memory stores program instructions. When the program instructions are executed by the processor, the processor performs the steps of the virtual power plant distributed resource digital twin construction method according to any one of claims 1-4.

Images