A virtual power plant behavior characteristic individualization prediction method and system

Abstract

The application discloses a virtual power plant behavior characteristic individualization prediction method and system, relates to the electrical and automation technical field, and comprises the following steps: constructing each distributed resource model in a virtual power plant; each distributed resource model in the virtual power plant is summarized into a multi-period operation model in the form of a polyhedron constrained by linear inequalities; based on the uncertainty of the distributed resources, the inner approximation method of feasible region flexible aggregation is used to aggregate and model each distributed resource in the virtual power plant, and the flexibility boundary of the overall available resources of the virtual power plant is generated; the coupling relationship between environmental factors and the output of the distributed resources is analyzed by using historical operation data and meteorological data, a distributed resource power prediction model based on an Informer-network network is constructed, and the overall resources of the virtual power plant are predicted. The application effectively solves the complexity problem in virtual power plant resource scheduling, improves the accuracy of power prediction and the flexibility of resource scheduling, and optimizes the power grid operation efficiency.

Description

Technical Field

[0001] This invention relates to the field of electrical and automation technology, specifically to a personalized prediction method and system for the behavioral characteristics of a virtual power plant. Background Technology

[0002] Virtual power plants (VPPs), relying on the internet and modern information and communication technologies, aggregate various types of distributed resources into a unified whole, exhibiting the characteristics of a power plant. They can provide the power grid with diverse services such as energy balancing, backup power, frequency regulation, and peak shaving, which is of great significance for improving the safe and stable operation of the system and promoting the consumption of new energy sources. However, because virtual power plants involve various distributed resources (such as photovoltaic, wind power, energy storage, and adjustable loads), their operating characteristics are complex and interdependent, making accurate modeling and prediction of resource power output a major challenge. Existing prediction methods mostly rely on traditional statistical models or simplified assumptions, which cannot effectively capture complex spatiotemporal coupling relationships and uncertainties, resulting in low prediction accuracy and failing to meet real-time dispatch requirements. Summary of the Invention

[0003] This invention provides a personalized prediction method and system for the behavioral characteristics of virtual power plants, which can effectively solve the problems pointed out in the background art.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0005] A personalized prediction method for the behavioral characteristics of a virtual power plant includes:

[0006] Considering the output characteristics and spatiotemporal properties of different types of distributed renewable energy, we construct models of various distributed resources within a virtual power plant.

[0007] Combining the active power, energy, and regulation services of the distributed resources, a multi-cycle operation model defined by linear inequality constraints is established based on the distributed resource models within the virtual power plant.

[0008] Based on the uncertainty of distributed resources, the internal approximation method of feasible domain flexible aggregation is used to model each distributed resource in the virtual power plant, obtain the flexibility boundary of the overall available resources of the virtual power plant, and determine the overall available resource range of the virtual power plant.

[0009] Based on the flexibility boundary and overall available resource range of the virtual power plant, and using historical operating data and meteorological data, the coupling relationship between environmental factors and distributed resource output is analyzed. A distributed resource power prediction model based on the Informer-network is constructed to predict the overall resources of the virtual power plant.

[0010] Further, each of the distributed resource models can be characterized as:

[0011] energy storage device,

[0012]

[0013] wherein, and are the charging power, discharging power and residual energy of ES i, respectively; is the energy dissipation rate; and are the maximum charging power, maximum discharging power and ramping limit of ES i, respectively; and are the allowed maximum and minimum residual energy constrained by the state of charge boundaries of ES i; are the charging efficiency and discharging efficiency of ES i, respectively, and ΔT is the ambient temperature variation;

[0014] thermally controllable load,

[0015]

[0016] wherein, and are the power consumption and indoor temperature of TCR i, respectively; and are the maximum, minimum power consumption and ramping limit of TCR i, respectively; and are the lower and upper bounds of the ideal indoor temperature, respectively; w TCR (t) is the ambient temperature; is the dissipation rate of TCR i; is the conversion coefficient of the active power of TCR i to temperature; is the influence factor of the ambient temperature;

[0017] delayable load,

[0018]

[0019] wherein, and are the input energy and residual energy of DL i, respectively; and are the dissipation rate and conversion coefficient of DL i, respectively; are the maximum, minimum charging power and rated charging capacity, respectively; denotes the energy required by DL i; and are the start and end times of DL i, respectively.

[0020] Further, in combination with the coupling of active power, energy and regulation service of the distributed resources, based on the model of each distributed resource in the virtual power plant, a multi-period operation model defined by linear inequality constraints is established, further comprising:

[0021] Considering the coupling of active power, energy and regulation service, a general multi-period operation model of a single distributed resource is obtained;

[0022] The constraint conditions of active power, energy and regulation service of each distributed resource are expressed as linear inequalities, and the feasible region of each distributed resource in the multi-period is constructed;

[0023] The linear inequality constraints are integrated into a polyhedral form model to represent the feasible operation space of all resources.

[0024] Further, the general multi-period operation model of a single distributed resource is represented as:

[0025]

[0026] wherein, and e i (t) are the input power, output power and residual energy of a single distributed resource i, respectively; is the regulation upper limit of the regulation service; θ i is the energy dissipation rate, and are the maximum input power, maximum output power and ramping limit of the distributed resource i, e i (t) and are the maximum and minimum residual energy allowed under the state of charge boundary constraint of the distributed resource i, are the energy input efficiency and energy output efficiency of the distributed resource i, w i (t) is the ambient temperature; ω i is the influence factor of ambient temperature, η i is the conversion coefficient of active power to temperature of the distributed resource i.

[0027] Further, based on the uncertainty of the distributed resources, through the inner approximation method of flexible aggregation of feasible regions, each distributed resource in the virtual power plant is modeled to obtain the flexibility boundary of the overall available resources of the virtual power plant, and the range of the overall available resources of the virtual power plant is determined, further comprising:

[0028] A separate feasible region model is established for each distributed resource in the virtual power plant to represent the operation range of the resource in different time periods;

[0029] Minkowski and the feasible region of multiple distributed resources in the virtual power plant are aggregated to obtain the comprehensive feasible region of all resources, and the available resource range of the virtual power plant as a whole is formed;

[0030] For the comprehensive feasible region of each distributed resource, a homogeneous polyhedron-based internal approximation method is adopted to optimize the resource aggregation model by adjusting the scaling factor and the conversion factor, and the flexibility boundary of the available resources of the virtual power plant as a whole is formed.

[0031] Further, the comprehensive feasible region is:

[0032]

[0033] In the formula φ i is the scaling factor of the distributed resource i, is the conversion factor of the distributed resource i, is the basic homogeneous polyhedron of the aggregate k; and is the coefficient matrix for determining the feasible region polyhedron, and X0 is the controllable variable vector in the basic homogeneous polyhedron; X i represents the i-th controllable variable vector, and X is the controllable variable vector of the virtual power plant as a whole after aggregation.

[0034] Further, the flexibility boundary of the available resources of the virtual power plant as a whole is formed, that is, the aggregated feasible region of the virtual power plant aggregation model is:

[0035]

[0036] In the formula, is the aggregated feasible region, is the controllable variable vector, and R k = [r k (t) |t＝1,…,T ] T , respectively, are the input power and output power of the k-th distributed resource; r k (t) is the adjustment service providing range of the k-th distributed resource, and respectively represent the scaling factor and the conversion factor, and are the coefficient matrices for determining the feasible region polyhedron.

[0037] Further, for the historical operation data and the meteorological data, a feature selection method based on the maximum mutual information coefficient is adopted to reduce the input feature dimension of the model, and the calculation formula is:

[0038]

[0039] In the formula, I(a, b) is the mutual information value between variables a and b; A and B are the number of segments divided along the a and b directions, that is, the grid distribution, the values of A and B increase from 1, and meanwhile A*B≤Q is satisfied; the size of Q is 0.6 power of the total amount of data; k(A i ,B i ) is the probability that the coordinate point (a, b) falls into A i 、B i at the same time, k(A i ) and k(B i ) are the probabilities that the coordinate point (a, b) falls into A i 、B i respectively; log2min(A, B) is a normalization factor for controlling the MIC value range in [0, 1], and the greater the MIC value between two variables, the stronger the correlation, and vice versa, the smaller the MIC value, and the smaller the correlation between two variables.

[0040] Further, the distributed resource power prediction model based on the Informer-network network comprises:

[0041] an input variable embedding module, input variables are embedded through data encoding, position encoding and timestamp encoding to generate low-dimensional embedding vectors for subsequent use of an encoder and a decoder;

[0042] a self-attention module for screening more important features from input information;

[0043] an encoder module for extracting long-time dependence in input data;

[0044] a decoder module for generating power prediction output of distributed resources.

[0045] A virtual power plant behavior characteristic individualized prediction system, the system comprises:

[0046] a distributed model construction module, which considers different types of distributed new energy output characteristics and space-time characteristics, and constructs each distributed resource model in the virtual power plant;

[0047] a multi-period model establishment module, which establishes a multi-period operation model defined by linear inequality constraints based on each distributed resource model in the virtual power plant, in combination with the coupling of active power, energy and regulation service of the distributed resources;

[0048] a flexibility boundary generation module, which models each distributed resource in the virtual power plant based on the uncertainty of the distributed resources through an internal approximation method of feasible region flexibility aggregation, obtains the flexibility boundary of the overall available resources of the virtual power plant, and determines the resource range available for the virtual power plant as a whole.

[0049] The distributed model construction module, based on the virtual power plant flexibility boundary and the overall available resource range, analyzes the coupling relationship between environmental factors and distributed resource output by using historical operation data and meteorological data, constructs a distributed resource power prediction model based on an Informer-network network, and predicts the overall resources of the virtual power plant.

[0050] The technical scheme of the present application can achieve the following technical effects:

[0051] The complexity problem in virtual power plant resource scheduling is effectively solved, the accuracy of power prediction and the flexibility of resource scheduling are improved, and the efficiency of power grid operation is optimized. BRIEF DESCRIPTION OF DRAWINGS

[0052] In order to more clearly illustrate the technical scheme in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiment or prior art description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0053] Figure 1 It is a flowchart of a virtual power plant behavior characteristic personalized prediction method;

[0054] Figure 2 It is a virtual power plant distributed resource aggregation schematic diagram;

[0055] Figure 3 It is a flexible aggregation method based on homogeneous polyhedron;

[0056] Figure 4 It is an Informer-network model structure diagram. DETAILED DESCRIPTION

[0057] The technical scheme in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments.

[0058] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs. The terms used in the specification of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0059] Embodiment one

[0060] As Figure 1 shown, the present application provides a virtual power plant behavior characteristic individualized prediction method, the method comprises:

[0061] S1: considering the output characteristics and space-time characteristics of different types of distributed new energy, constructing a model of each distributed resource inside the virtual power plant;

[0062] Specifically, for each type of distributed resource (such as photovoltaic, wind, energy storage device, adjustable load, etc.), it is necessary to analyze their power output characteristics respectively. The power output of photovoltaic and wind power resources is affected by the fluctuation of sunshine intensity and wind speed, which has significant space-time variability; the energy storage device is mainly affected by the charging and discharging characteristics, energy storage limit and device efficiency; and the adjustable load is adjusted dynamically according to the demand change of the power grid, on this basis, for each resource, a corresponding mathematical model is established, which uses linear inequalities and dynamic equations to describe its operation behavior in different time periods. These models not only consider the power output of the resource, but also consider its energy storage, charging and discharging capacity and adjustment service capacity and other factors, so as to accurately reflect the operation characteristics and coupling relationship of each resource inside the virtual power plant, providing important support for subsequent resource aggregation modeling and power prediction.

[0063] S2: based on the model of each distributed resource inside the virtual power plant, a multi-period operation model defined by linear inequality constraints is established combined with the coupling of active power, energy and regulation service of distributed resources;

[0064] Specifically, by considering the coupling relationship of active power, energy and regulation service of each distributed resource inside the virtual power plant, the behavior of these resources is summarized as a multi-period operation model in the form of a polyhedron. This modeling method can accurately describe the output characteristics of various resources in the virtual power plant in different time periods and their interaction, reflecting the complex space-time coupling relationship between resources. By using the polyhedron form of linear inequality constraints, the operation constraints of resources can be converted into a geometric model, so that the operable range of resources can be clearly expressed in the time dimension. This polyhedral model can provide the basis for the scheduling of virtual power plant resources, help optimize resource allocation, and improve the overall efficiency and stability of the power system. At the same time, this method can also effectively deal with external uncertainties (such as weather changes, demand fluctuations, etc.), providing a more robust solution for power system scheduling, and providing a computationally feasible and efficient framework for subsequent scheduling optimization and decision making.

[0065] S3: based on the uncertainty of distributed resources, the distributed resources inside the virtual power plant are modeled by the inner approximation method of flexible aggregation of feasible region, to obtain the flexibility boundary of the overall available resources of the virtual power plant, and determine the range of the overall available resources of the virtual power plant;

[0066] Specifically, firstly, the output characteristics and their uncertainties of each distributed resource in the virtual power plant need to be considered, such as wind power, photovoltaic, energy storage devices and adjustable load, etc. The output of these resources is not only affected by the natural environment (such as weather changes, seasonal fluctuations, etc.), but also by external factors such as grid demand and market price. Therefore, how to reasonably aggregate each distributed resource under the condition of considering uncertainty has become the key to improving the scheduling ability of the virtual power plant. Through the feasible region flexible aggregation internal approximation method, the feasible region of each distributed resource is adjusted and aggregated. Firstly, the feasible region of each resource is modeled as a homogeneous polyhedron, and then the flexibility aggregation of each resource is realized by adjusting the scaling factor and the conversion factor. Through this method, the flexibility boundary of the overall resources in the virtual power plant can be accurately described under uncertain environmental conditions. This flexibility boundary not only considers the dynamic characteristics of the resources themselves, but also reflects the interaction between the resources and the uncertainty of the external environment. Finally, through the aggregated flexibility boundary, more accurate decision support can be provided for the scheduling and resource management of the virtual power plant, ensuring that the virtual power plant can provide stable, reliable and efficient power services when facing uncertainty.

[0067] S4: Based on the flexibility boundary of the virtual power plant and the overall available resource range, the coupling relationship between environmental factors and distributed resource output is analyzed using historical operation data and meteorological data, and a distributed resource power prediction model based on Informer-network network is constructed to predict the overall resources of the virtual power plant.

[0068] Specifically, first, historical operation data of various distributed resources in the virtual power plant (such as power output, load change, etc.) and environment-related meteorological data (such as temperature, humidity, wind speed, light intensity, etc.) need to be collected and preprocessed and cleaned. Then, by analyzing the coupling relationship between environmental factors and distributed resource output, a mathematical model is established to reveal the influence of different environmental factors on the output of each resource. Next, a power prediction model based on Informer-network is constructed. Informer-network is a deep learning-based time series prediction model that can effectively capture long-term dependencies. In this model, historical data and meteorological data are converted into low-dimensional vectors through input embedding layers, and position encoding and timestamp encoding are added to ensure that the model can handle the temporal relationship in time series data. Using self-attention mechanisms and multi-head self-attention layers, the model can focus on important time points and predict future power output based on historical data and external meteorological factors. The output of this model is the power prediction value of the virtual power plant in the future period, which can be used for power dispatching, load balancing, and energy optimization. In addition, Informer-network can provide a certain degree of uncertainty estimation to help power system operators better cope with unforeseen changes. Finally, based on this prediction model, the virtual power plant can achieve more accurate resource scheduling and improve the overall system's operational efficiency and stability.

[0069] The present application proposes a virtual power plant behavior characteristic individualized prediction method considering multi-type power commodity transaction, which significantly improves the accuracy, completeness of the virtual power plant model and the accuracy of the virtual power plant output prediction compared with common methods. It is beneficial to power dispatching, peak load shifting of the power system, and improves the efficiency, economy and stability of the power system.

[0070] As a preferred embodiment of the above embodiment, each distributed resource model can be represented as:

[0071] energy storage device,

[0072]

[0073]

[0074] wherein, and are the charging power, discharging power and remaining energy of ES i, respectively; is the energy dissipation rate; and are the maximum charging power, maximum discharging power and ramping limit of ES i, respectively; and is the allowed maximum and minimum residual energy constrained by the state of charge boundary of ES i; are the charging and discharging efficiency of ES i, respectively;

[0075] thermal controllable load,

[0076]

[0077] wherein, and are the power consumption and indoor temperature of TCR i, respectively; and are the maximum, minimum power consumption and ramping limit of TCR i, respectively; and are the lower and upper bound of the ideal indoor temperature, respectively;w kCR (t) is the ambient temperature; is the dissipation rate of TCR i; is the conversion coefficient of active power to temperature of TCR i; is the influence factor of ambient temperature;

[0078] delayable load,

[0079]

[0080] wherein, and are the input energy and residual energy of DL i, respectively; and are the dissipation rate and conversion coefficient of DL i, respectively; are the maximum, minimum charging power and rated charging capacity, respectively; denotes the energy required by DL i; and are the start and end time of DL i, respectively.

[0081] Specifically, the energy storage device: the energy storage device such as battery can store / release energy, realize the conversion of energy between different periods, consider the daily charging and discharging strategy and the state of charge boundary of the energy storage device, and model the energy storage device; the dynamic characteristics of the thermal controllable load: TCR can be modeled by derivative, and then converted into discrete form; the delayable load: the electric vehicle charging load is a typical DL, and this kind of distributed resource needs to meet the energy demand within the pre-specified time.

[0082] As a preferred embodiment of the above embodiment, based on the coupling of the active power, energy and regulation service of the distributed resource, a multi-period operation model defined by linear inequality constraints is established based on the model of each distributed resource in the virtual power plant, further comprising:

[0083] S21: considering the coupling of active power, energy and regulation service, obtaining a general multi-period operation model of a single distributed resource;

[0084] S22: expressing the constraint conditions of active power, energy and regulation service of each distributed resource as linear inequalities, and constructing the feasible region of each distributed resource in the multi-period;

[0085] S23: integrating each linear inequality constraint into a polyhedral form model to represent the feasible operation space of all resources.

[0086] Specifically, first, in the S21 step, considering the mutual coupling relationship between the active power, energy storage and regulation service of different types of distributed resources (such as energy storage devices, wind power, photovoltaic, adjustable load, etc.) in the virtual power plant, each resource has its specific output characteristics, for example, the charge and discharge power of the energy storage device is affected by the remaining energy and the regulation demand, while the output of the photovoltaic and wind power resources is affected by the fluctuation of environmental factors (such as light, wind speed). Therefore, for each distributed resource, a corresponding multi-period operation model needs to be established, that is, the power output, energy change and the ability to provide regulation service of the resource in different time steps are considered, in this way, the model can reflect the dynamic behavior of the resource and predict its changes and uncertainties in future periods; then, in the S22 step, the constraint conditions of active power, energy and regulation service of each distributed resource need to be converted into the form of linear inequalities, these constraint conditions include: the power output of each resource has maximum and minimum limits, the charge and discharge power of the energy storage device is constrained by the device capacity and the remaining energy, the ability of the regulation service is also constrained by the power change rate (climbing rate) and other physical limitations, all these constraints are described by linear inequalities, which provide clear boundaries for the feasible operation space (feasible region) of each distributed resource in multiple time periods, these linear inequality constraints ensure that the resources will not violate their physical limitations in the actual scheduling process, and provide a basis for optimal scheduling; finally, in the S23 step, all the linear inequality constraints of the distributed resources are integrated to obtain a polyhedral form model, the polyhedral model is a geometric representation method, which represents the constraint conditions of each resource in different time periods as the boundary of the polyhedron, thereby forming the feasible operation space of the resource, this geometric representation method can clearly describe the possible operation states of all resources in the virtual power plant in multiple time periods, and also reflect the mutual influence and constraints between different resources, through the geometric representation of the polyhedron, the overall resource scheduling of the virtual power plant can more accurately calculate the operation range and scheduling limit of each resource, ensuring that efficient resource scheduling and optimization can be carried out under the premise of meeting all resource constraints.

[0087] As a preferred embodiment of the above-mentioned embodiment, the general multi-period operation model of a single distributed resource is represented as:

[0088]

[0089] In the formula, and e i (t) are the input power, output power and residual energy of a single distributed resource i, respectively; is the upper limit of the regulation service; θ i is the energy dissipation rate, and are the maximum input power, maximum output power and ramping limit of the distributed resource i, e i (t) and are the maximum and minimum residual energy allowed under the state of charge boundary constraint of the distributed resource i, are the energy input efficiency and energy output efficiency of the distributed resource i, w i (t) is the ambient temperature; ω i is the influence factor of the ambient temperature, η i is the conversion coefficient of the active power to temperature of the distributed resource i.

[0090] Specifically, the equivalent aggregation model of the virtual power plant considering uncertainty is constructed. In addition to the coupling relationship between power and energy, many distributed resources can further provide regulation services to the power system. Therefore, considering the coupling of active power, energy and regulation services, the general multi-period operation model of a single distributed resource can be obtained.

[0091] The regulation service providing range should be determined comprehensively by the maximum acceptable regulation range set by the virtual power plant, the active variation law and the residual energy state, and the specific formula is as follows:

[0092]

[0093]

[0094] wherein, is the maximum acceptable regulation range, h R is the energy reserve ratio of the specified service, and it is noted that the mentioned regulation service in the actual implementation process can represent different types of auxiliary services, such as frequency regulation capacity or hot standby service. Since the distributed resources are heterogeneous, the core parameters and feasible regions are different. Without loss of generality, considering the active power input, active power output and regulation capacity providing as controllable variables, it can be rearranged into the following compact form, which is essentially a polyhedron constrained by linear inequalities.

[0095]

[0096] wherein represents a controllable variable vector, and R i = [r i (t) |t＝1,…,T ] T ; M i and N i are coefficient matrices.

[0097] As a preferred of the above embodiment, based on the uncertainty of distributed resources, the inner approximation method of flexible aggregation of feasible region is used to model each distributed resource in the virtual power plant, to obtain the flexibility boundary of the overall available resources of the virtual power plant, to determine the range of the overall available resources of the virtual power plant, further comprising:

[0098] S31: establishing an independent feasible region model for each distributed resource in the virtual power plant, representing the operation range of the resource in different time periods;

[0099] S32: using Minkowski sum to aggregate the feasible regions of multiple distributed resources in the virtual power plant, to obtain the comprehensive feasible region of all resources, to form the overall available resource range of the virtual power plant;

[0100] S33: for each distributed resource, using the inner approximation method based on homogeneous polyhedron to optimize the resource aggregation model by adjusting the scaling factor and the conversion factor, to form the flexibility boundary of the overall available resources of the virtual power plant.

[0101] In particular, first, in step S31, an independent feasible region model is established for each distributed resource (such as photovoltaic, wind, energy storage device and adjustable load, etc.) in the virtual power plant, the feasible region of each resource represents its operating range in different time periods, considering factors such as active power, energy state and regulation ability, for example, the feasible region of the energy storage device is limited by the charging and discharging power and the remaining energy, and the power output of wind power and photovoltaic is affected by weather conditions. By modeling these resources, the operating range of each resource in different time periods can be accurately described; in step S32, the Minkowski sum is used to aggregate the feasible regions of multiple distributed resources in the virtual power plant, the Minkowski sum operation combines the feasible regions of each resource into a comprehensive feasible region, forming the available resource range of the virtual power plant as a whole, this comprehensive feasible region reflects the schedulable state of all resources in the virtual power plant in multiple time periods, and ensures the synergistic effect between resources. Through this aggregation method, all resource constraint conditions can be considered uniformly, providing a global operation framework for the overall scheduling of the virtual power plant; finally, in step S33, the comprehensive feasible region of each distributed resource is optimized using the internal approximation method based on homogeneous polyhedron. By adjusting the scaling factor and conversion factor of each resource, the aggregation model of the resource is optimized, this method accurately adjusts the shape of the resource feasible region, enhances the flexibility boundary of the virtual power plant, and ensures that it can adapt to external uncertainties and changes in grid demand. Ultimately, these optimized flexibility boundaries provide a more accurate and efficient basis for resource scheduling and optimization of the virtual power plant.

[0102] As a preferred embodiment of the above embodiment, the comprehensive feasible region is:

[0103]

[0104] wherein φ i is the scaling factor of the distributed resource i, is the conversion factor of the distributed resource i, is the basic homogeneous polyhedron of the aggregate k; and is the coefficient matrix that determines the feasible region polyhedron, X0is the controllable variable vector in the basic homogeneous polyhedron; X i represents the i-th controllable variable vector, X is the controllable variable vector of the virtual power plant as a whole after aggregation.

[0105] As a preferred embodiment of the above embodiment, the flexibility boundary of the available resources of the virtual power plant as a whole, that is, the aggregated feasible region of the aggregation model of the virtual power plant, is:

[0106]

[0107] wherein, is a polyhedral feasible region, is a controllable variable vector, and R k = [r k (t) |t＝1,…,T ] T , are the input power and output power of the kth distributed resource, respectively; r k (t) is the adjustment service providing range of the kth distributed resource, and denote the scaling factor and the conversion factor, respectively, and are the coefficient matrixes for determining the polyhedral feasible region.

[0108] Specifically, based on the homogeneous polyhedral feasible region flexible aggregation method, a virtual power plant aggregation model considering the uncertainty of distributed resources is established, as shown in Figure 2 .

[0109] The determined feasible region (Ω k AGG ) of a group of distributed resources can be characterized by the Minkowski sum of each distributed resource, which is defined as:

[0110]

[0111] wherein, is the Minkowski sum calculation.

[0112] The present application adopts a basic homogeneous polyhedron to approximate the distributed resource polyhedron, in order to enhance the similarity, a basic homogeneous polyhedron with the same structure as the general distributed resource polyhedron is adopted, and the method aims to scale and convert the basic homogeneous polyhedron to calculate an internal approximate polyhedral feasible region For the feasible region of the distributed resources from the same homogeneous polyhedron, the Minkowski sum calculation can be simplified as:

[0113]

[0114] The aggregation method based on the homogeneous polyhedron only needs to perform arithmetic addition when calculating the Minkowski sum, and the calculation efficiency is high, and the aggregation method is as shown in Figure 3 , based on the basic homogeneous polyhedron, considering the uncertainty of the distributed resources, the maximum internal approximate feasible region of each distributed resource can be represented as:

[0115]

[0116] wherein ε∈(0, 1) is the allowed out-of-bound probability.

[0117] by scaling (φ i ) and transformation Basic Homogeneous Polyhedron The chance-constrained programming is used to obtain the inner approximation of the feasible region of each distributed resource under uncertainty. The aim is to enlarge the feasible region of each distributed resource and maximize the aggregated accuracy. To transform the problem into a solvable form, auxiliary variables are introduced The maximum inner approximation problem is transformed into a linear programming problem:

[0118]

[0119] s.t.α i ,U i ≥0;

[0120]

[0121]

[0122] where L dim = 16T-4 is the dimension of the matrix N in (2); U i (s,:) and M i (s,:) represent the s-th row of U i and M i , respectively; is the s-th element of the uncertain matrix N i , where and σ i (s) are the predicted value and the standard deviation caused by the prediction error of N i (s). For ES devices with non-negligible charge-discharge losses, β i (t) | t e {1, …, 2T} should be set to 0 to ensure that the charge-discharge power is simultaneously 0.

[0123] By calculation, the scaling factor and the transformation factor of a single distributed resource are obtained. According to equation (8), the scaling factor and the transformation factor of the VPP aggregation model can be calculated as follows:

[0124]

[0125] For the VPP aggregation model, the aggregated feasible region can be expressed as:

[0126]

[0127] As a preferred embodiment of the above, the historical operation data and meteorological data are reduced in feature dimension by using a feature selection method based on a maximum mutual information coefficient, and the calculation formula is:

[0128]

[0129] In the formula, I(a, b) is the mutual information value between variables a and b; A and B are the number of segmentations along the a and b directions, that is, the grid distribution, the values of A and B increase from 1, and meanwhile satisfy A*B≤Q; the size of Q is the 0.6 power of the total amount of data; k(A i ,B i ) is the probability that the coordinate point (a, b) falls into A i and B i at the same time, k(A i ) and k(B i ) are the probabilities that the coordinate point (a, b) falls into Ai and B i respectively; log2min(A,B) is a normalization factor for controlling the MIC value range to be between [0, 1], the larger the MIC value between two variables, the stronger the correlation, and vice versa, the smaller the MIC value, the smaller the correlation between the two variables.

[0130] Specifically, the maximum mutual information coefficient (MIC) is selected as the feature selection method, mainly because it can capture linear and nonlinear relationships between features. In the power prediction of a virtual power plant, the input data usually includes meteorological data and historical operation data, and there is often a complex nonlinear relationship between these data. Traditional feature selection methods mainly target linear relationships and cannot handle such cases, while MIC can measure both linear and nonlinear correlations and is very suitable for such data. MIC can also effectively handle high-dimensional data by grid partitioning to identify the features with the strongest correlation to the target variable (such as power output), thereby reducing redundant features and improving model calculation efficiency. In addition, MIC can automatically remove irrelevant features, avoiding the interference of redundant information on the model and improving prediction accuracy. Furthermore, MIC can handle different types of data, including continuous and discrete data, making it more flexible in handling various data in a virtual power plant. Therefore, using MIC can help accurately select features related to power prediction, thereby optimizing resource scheduling and power system operation efficiency in a virtual power plant.

[0131] As a preferred embodiment of the above, as shown in Figure 4 , the distributed resource power prediction model based on the Informer-network network includes:

[0132] The input variable embedding module embeds the input variable through data encoding, position encoding and timestamp encoding to generate a low-dimensional embedding vector for subsequent use by the encoder and decoder;

[0133] The self-attention module is used to filter out more important features from the input information.

[0134] The encoder module is used to extract long-time dependencies in the input data.

[0135] The decoder module is used to generate power prediction output of distributed resources.

[0136] Specifically, first, the input variable embedding module converts the original high-dimensional input data into a low-dimensional embedding vector through data encoding, position encoding and timestamp encoding. Data encoding extracts local features of the input data through one-dimensional convolution operation, position encoding adds time series information to each data point through sine and cosine functions, and timestamp encoding converts time information into a fixed-dimensional vector. These encoded embedding vectors are summed and provided as input to the subsequent encoder and decoder modules. Next, the self-attention module uses the self-attention mechanism to filter out the most important features for power prediction. The self-attention mechanism dynamically allocates attention points by calculating the correlation between different parts of the input data, allowing the model to automatically focus on the data area with the strongest relationship to the prediction target. This mechanism can capture complex time-dependent dependencies in the input data and effectively handle long and short-term dependencies in the data. Then, the encoder module extracts long-time dependencies in the input data through multiple self-attention layers to capture global information of the entire sequence. The multi-head self-attention mechanism in the encoder allows the model to process multiple feature subspaces in parallel, thereby more comprehensively understanding the internal structure of the data. In addition, the encoder combines convolution and pooling operations to further improve feature extraction efficiency and reduce computational complexity. Finally, the decoder module generates power prediction values for distributed resources based on the output of the encoder. The decoder uses two multi-head self-attention layers to process the features obtained from the encoder and generates predictions of future power output based on these features. At the same time, to improve processing speed, the decoder directly receives a compressed input vector, thereby speeding up the prediction process.

[0137] The encoder is mainly used to extract long-time dependencies between long sequence inputs, and its main component is the multi-head probabilistic sparse self-attention mechanism. Since there are redundant combinations in the results of the probabilistic sparse self-attention module, a distillation mechanism is needed to extract dominant key features. The formula for the distillation operation from layer j to layer j+1 is:

[0138]

[0139] wherein, is an important operation output in the multi-head probability sparse self-attention block, Conv1d is a one-dimensional convolution operation, the activation function of which is an ELU function, and MaxPool is a pooling operation. Through this distillation operation, the length of the input sequence can be shortened by half, thereby reducing the memory usage and calculation time of the encoder.

[0140] Decoder module (Deco distributed resources)

[0141] The model uses a standard decoder, which is composed of two multi-head self-attention layers. At the same time, in order to alleviate the decline in the processing speed of generating prediction results, an input vector is directly provided to the decoder:

[0142]

[0143] At the same time, the input vector adopts a masked form of multi-head attention mechanism, which can prevent attention to information in future time when processing data at a certain time. Then, the final output is obtained through a fully connected layer. It is worth noting that the decoder structure proposed by this method is a generative structure, that is, it can generate all prediction sequences at a time, greatly reducing the prediction time of decoding. In the encoder and the decoder, the connection between layers adopts a residual and normalization module, so that the network can be deeper, and the problems of gradient disappearance, gradient explosion and deep network degenerating into shallow network can be effectively avoided.

[0144] Embodiment two

[0145] Based on the same inventive concept as the virtual power plant behavior characteristic individualized prediction method in the foregoing embodiment, the application further provides a virtual power plant behavior characteristic individualized prediction system, which comprises:

[0146] A distributed model construction module constructs models of various distributed resources in the virtual power plant by considering the output characteristics and spatio-temporal characteristics of different types of distributed new energy.

[0147] A multi-period model establishment module establishes a multi-period operation model defined by linear inequality constraints based on the models of various distributed resources in the virtual power plant by combining the coupling of active power, energy and regulation services of the distributed resources.

[0148] A flexibility boundary generation module models various distributed resources in the virtual power plant by a feasible region flexible aggregation internal approximation method based on the uncertainty of the distributed resources, obtains the flexibility boundary of the overall available resources of the virtual power plant, and determines the resource range available to the virtual power plant as a whole.

[0149] The distributed model construction module uses historical operation data and meteorological data to analyze the coupling relationship between environmental factors and distributed resource output based on the virtual power plant flexibility boundary and the overall available resource range, and constructs a distributed resource power prediction model based on an Informer-network network to predict the overall resources of the virtual power plant.

[0150] The prediction system in the application can effectively realize the virtual power plant behavior characteristic individualization prediction method, and the technical effects are as described above.

[0151] Although the present application has been described in connection with specific features and embodiments thereof, it will be evident to those of ordinary skill in the art that various modifications and combinations can be made without departing from the spirit and scope of the application. Accordingly, it is intended that the description and drawings be regarded as illustrative rather than restrictive. It is intended that the application cover any and all modifications, changes, combinations, or equivalents that fall within the scope of the application. Obviously, many modifications and variations of the present application are possible in light of the above teachings. It is, therefore, to be understood that within the scope of the application disclosed herein, the application can be practiced otherwise than as specifically described.

Claims

1. A personalized prediction method for the behavioral characteristics of a virtual power plant, characterized in that, The method includes: Considering the output characteristics and spatiotemporal properties of different types of distributed renewable energy, we construct models of various distributed resources within a virtual power plant. Combining the coupling of active power, energy, and regulation services of the distributed resources, and based on the distributed resource models within the virtual power plant, a multi-cycle operation model defined by linear inequality constraints is established. Each of the distributed resource models can be characterized as follows: Energy storage devices Thermally controllable load, Delayed load, Based on the uncertainty of distributed resources, the internal approximation method of feasible domain flexible aggregation is used to model each distributed resource in the virtual power plant, obtain the flexibility boundary of the overall available resources of the virtual power plant, and determine the overall available resource range of the virtual power plant. Based on the flexibility boundary and overall available resource range of the virtual power plant, and using historical operating data and meteorological data, the coupling relationship between environmental factors and distributed resource output is analyzed. A distributed resource power prediction model based on the Informer-network is constructed to predict the overall resources of the virtual power plant.

2. The personalized prediction method for virtual power plant behavioral characteristics according to claim 1, characterized in that, Combining the coupling of active power, energy, and regulation services of the distributed resources, and based on the distributed resource models within the virtual power plant, a multi-cycle operation model defined by linear inequality constraints is established, further including: Considering the coupling of active power, energy, and regulation services, a general multi-cycle operation model for a single distributed resource is obtained; The constraints on the active power, energy, and regulation services of each distributed resource are expressed as linear inequalities, and the feasible region of each distributed resource is constructed over multiple periods. The linear inequality constraints are integrated into a polyhedral model, representing the feasible operation space of all resources.

3. The personalized prediction method for virtual power plant behavior characteristics according to claim 2, characterized in that, The general multi-cycle operation model of a single distributed resource is expressed as follows: Influence factors of degree For distributed resources The conversion coefficient of active power to temperature.

4. The personalized prediction method for virtual power plant behavior characteristics according to claim 1, characterized in that, Based on the uncertainty of distributed resources, an inner approximation method using feasible region flexible aggregation is used to model each distributed resource within the virtual power plant, obtaining the flexibility boundary of the overall available resources of the virtual power plant, and determining the overall available resource range of the virtual power plant, further including: Establish an independent feasible domain model for each distributed resource within the virtual power plant, representing the operational range of the resource in different time periods; Using Minkowski and aggregating the feasible domains of multiple distributed resources within the virtual power plant, a comprehensive feasible domain of all resources is obtained, forming the overall available resource range of the virtual power plant. For each distributed resource, the comprehensive feasible domain is adopted, and the resource aggregation model is optimized by adjusting the scaling factor and the transformation factor to form the flexibility boundary of the overall available resources of the virtual power plant.

5. The personalized prediction method for virtual power plant behavior characteristics according to claim 4, characterized in that, The comprehensive feasible region is: X is the vector of controllable variables for the aggregated virtual power plant.

6. The personalized prediction method for virtual power plant behavior characteristics according to claim 4, characterized in that, The flexibility boundary representation of the overall available resources of the virtual power plant, i.e., the aggregated feasible region of the virtual power plant aggregation model, is as follows: The coefficient matrix of the feasible region polyhedron.

7. The personalized prediction method for virtual power plant behavioral characteristics according to claim 1, characterized in that, For the historical operational data and meteorological data, a feature selection method based on the maximum mutual information coefficient is used to reduce the dimensionality of the model input features. The calculation formula is as follows: The stronger the correlation, the weaker the correlation between the two variables; conversely, the smaller the MIC value, the weaker the correlation between the two variables.

8. The personalized prediction method for virtual power plant behavioral characteristics according to claim 1, characterized in that, The distributed resource power prediction model based on Informer-network includes: The input variable embedding module embeds input variables through data encoding, position encoding, and timestamp encoding to generate low-dimensional embedding vectors for use by the subsequent encoder and decoder. The self-attention module is used to filter out more important features from the input information; The encoder module is used to extract long-term dependencies from the input data; The decoder module is used to generate power prediction outputs for distributed resources.

9. A personalized prediction system for the behavioral characteristics of a virtual power plant, characterized in that, The system includes: The distributed model construction module considers the output characteristics and spatiotemporal properties of different types of distributed new energy sources and constructs models of various distributed resources within the virtual power plant. The multi-period model establishment module, combining the active power, energy, and regulation services of the distributed resources, establishes a multi-period operation model defined by linear inequality constraints based on the distributed resource models within the virtual power plant. Each of the distributed resource models can be characterized as follows: Energy storage devices Thermally controllable load, Delayed load, The flexibility boundary generation module, based on the uncertainty of distributed resources, models each distributed resource in the virtual power plant through the internal approximation method of feasible region flexible aggregation, obtains the flexibility boundary of the overall available resources of the virtual power plant, and determines the overall available resource range of the virtual power plant. The distributed model construction module, based on the flexibility boundary and overall available resource range of the virtual power plant, uses historical operating data and meteorological data to analyze the coupling relationship between environmental factors and distributed resource output, and constructs a distributed resource power prediction model based on the Informer-network to predict the overall resources of the virtual power plant.

Images