A virtual power plant system integrating multiple resources and its flexible scheduling method

Abstract

The present invention discloses a virtual power plant system integrating multiple resources and a flexible scheduling method thereof, relating to the technical field of virtual power plant management. The present invention collects photovoltaic power data, energy storage state of charge data, and load power data in real time, analyzes the photovoltaic fluctuation characteristics, energy storage response capability indicators, and load power data stability indicators corresponding to each type of data, and performs normalization processing; establishes a resource interaction tensor based on the collected data, integrates the grid frequency deviation and line load, and constructs a collaborative optimization model with time lag compensation; updates the grid frequency and line load rate in a fixed time window, dynamically solves the collaborative optimization model, and outputs load regulation and energy storage target scheduling instructions; obtains the grid impedance characteristics and load distribution parameters of the area to be loaded with control instructions, analyzes the scene deviation of the area, and corrects the scheduling value based on the scene deviation. This improves scheduling accuracy and enhances the compatibility of complex power grids.

Description

Technical Field

[0001] The present invention relates to the technical field of virtual power plant management, and in particular to a virtual power plant system integrating multiple resources and a flexible scheduling method thereof. Background Art

[0002] A virtual power plant (VPP) is a control system that uses communication, computing, and scheduling to aggregate large amounts of distributed energy resources for unified management and coordinated optimization. Because it can organically coordinate multiple flexible resources, VPPs have become a research hotspot for new energy management models.

[0003] Existing dispatch schemes mostly focus on single-resource responses or simplified power balance models, failing to comprehensively consider the dynamic characteristics and coupling relationships of various resources. They also lack coordinated optimization of grid frequency and line safety. These approaches often rely on empirical rules or static adjustments to address PV output fluctuations, delayed energy storage responses, and uncertain load variations. These approaches struggle to meet the dual requirements of real-time and precision for coordinated multi-resource dispatch, resulting in suboptimal system dispatch performance.

[0004] Therefore, the present invention discloses a virtual power plant system integrating multiple resources and a flexible scheduling method thereof to solve the above problems. Summary of the Invention

[0005] The purpose of the present invention is to provide a virtual power plant system integrating multiple resources and a flexible scheduling method thereof, so as to solve the problems raised in the prior art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a flexible scheduling method for a virtual power plant integrating multiple resources, the method comprising the following steps:

[0007] S1: Real-time collection of PV power data, energy storage state of charge data, and load power data. Analyze the corresponding PV fluctuation characteristics, energy storage response capability indicators, and load power data stability indicators, and perform normalization processing.

[0008] S2: Based on the collected data, a resource interaction tensor is established, which integrates the grid frequency deviation and line load to construct a collaborative optimization model with time delay compensation.

[0009] S3: Updates the grid frequency and line load factor within a fixed time window, dynamically solves the collaborative optimization model, and outputs load regulation and energy storage target scheduling instructions;

[0010] S4: Obtain the grid impedance characteristics and load distribution parameters of the area to be loaded with the control instruction, analyze the scenario deviation of the area, and correct the scheduling value based on the scenario deviation.

[0011] According to the above scheme, S1 includes the following contents:

[0012] S101: Obtain the photovoltaic power sequence, energy storage charge state data and load power data of the virtual power plant; record the photovoltaic power sequence as P={P(t n )|n∈[1,N]}; where P(t n ) represents the time t n The corresponding photovoltaic power, n represents the time sequence number, N represents the total number of acquisition moments in the photovoltaic power sequence; the time t n The corresponding energy storage charge state is recorded as S (t n ), the energy storage charge state is the ratio of the current remaining power to the maximum capacity; the mth type load power data at time t n The corresponding load power value is recorded as L m (t n The load power data includes industrial load power data, commercial load power data and residential load power data; abnormal data point detection and interpolation filling are performed on the photovoltaic power sequence, energy storage charge state data and load power data respectively;

[0013] S102: Splitting the photovoltaic power sequence, energy storage state of charge data, and load power data based on a preset sliding time window length; analyzing photovoltaic fluctuation characteristics of the split photovoltaic power subsequences, where the photovoltaic fluctuation characteristics of the photovoltaic power subsequences are equal to the maximum photovoltaic power difference of the photovoltaic power subsequences divided by the photovoltaic power mean of the photovoltaic power subsequences;

[0014] Based on the split energy storage state of charge subsequence, the least squares method is used to fit a linear charge and discharge model; in the charge and discharge model, time is the independent variable and the energy storage state of charge is the dependent variable; the slope coefficient in the linear charge and discharge model is recorded as the average charge and discharge characteristic k; based on the average charge and discharge characteristic k combined with the attenuation factor, the energy storage response capability index R is analyzed, R = k × exp (-t d / t c ) / (S max -S min ); where t d represents the energy storage response delay, t c Indicates the system time coefficient; S max Indicates the upper limit of the system's rated energy storage state of charge, S min Indicates the lower limit of the system's rated energy storage charge state; the energy storage response delay and system time coefficient are preset by the system;

[0015] The central difference method is used to analyze the instantaneous load change rate of various load power data at each moment. The load instantaneous change rate subsequence corresponding to the split load power subsequence is obtained, and the absolute value of the ratio of the standard deviation to the mean of the load instantaneous change rate subsequence is recorded as the stability index.

[0016] The photovoltaic fluctuation characteristics, energy storage response capability index and stability index are normalized.

[0017] This application uses anomaly detection and interpolation to ensure data quality and prevent model instability caused by dirty data. It also analyzes photovoltaic fluctuation characteristics, energy storage response indicators, and load stability indicators to provide data support for subsequent analysis.

[0018] According to the above scheme, S2 includes the following contents:

[0019] S201: Based on the fusion of spatial information of photovoltaic power, energy storage state of charge, and various load power data, a coupling tensor is constructed, which is recorded as A{i, j, n}; where i represents the sequence number of the resource type, which includes photovoltaic power, energy storage state of charge, and various load power data; j represents the sequence number of the node;

[0020] S202: Based on the coupling tensor, grid frequency deviation and line load rate, a collaborative optimization model is constructed by weighted minimization of frequency deviation and line overload deviation;

[0021] ;

[0022] Among them, △f (t n ) represents the time t n The corresponding grid frequency deviation is equal to the difference between the real-time grid frequency and the rated grid frequency, f nom Indicates the rated grid frequency; ρ line (t n ) represents the time t n Line load factor, ρ safe represents the line load rate safety threshold, where the line load rate is equal to the ratio of the actual transmission power of the line to the maximum safe transmission power allowed by the line; ω f Represents the frequency deviation coefficient, ω p Indicates the load factor; the sum of the frequency deviation factor and the load factor is 1;

[0023] The constraints include energy storage state of charge range restrictions, load change rate restrictions, and power balance hard constraints; the power balance hard constraint is that the sum of the photovoltaic power of each node plus the sum of the energy storage charging and discharging power in the coupling tensor is equal to the total load power;

[0024] S203: Introduce compensation term into the state equation, and change the time t n+1 The predicted PV power after compensation is recorded as P*(t n+1 ):

[0025] ;

[0026] Where △t represents the data collection time interval, ▽n P(t n ) represents the time t n The corresponding photovoltaic power change rate, β represents the delay attenuation factor, β=1-exp(-t d / t c ).

[0027] The coupled tensor avoids the spatial information loss of traditional models and is used to construct a collaborative optimization model that incorporates grid frequency deviation and line load factor constraints to maintain system stability and security. This model synergistically considers the volatility of renewable energy and the responsiveness of controllable resources, enabling the joint management of distributed generation and loads. It also accounts for the time lag effect of photovoltaic power variations on the system, which helps improve scheduling accuracy.

[0028] According to the above solution, S3 includes the following:

[0029] S301: every preset time length, collect the current grid frequency deviation and line load rate; according to the current grid frequency deviation and line load rate, solve the optimization model through sensitivity feedback iteration to generate the current control instruction. The control instruction includes the output load adjustment amount △L m (t n ) and energy storage target power △P s (t n );

[0030] S302: Update the frequency deviation coefficient before solving the optimization model, ω f (t n ) = ω f0 (1+k f ×|(d△f / dt)(t n )|), where ω f0 is the basic frequency deviation coefficient, k f Represents the frequency sensitivity coefficient; the basic frequency deviation coefficient and the frequency sensitivity coefficient are preset constants, (d△f / dt)(t n ) indicates that at t n The instantaneous rate of change of the grid frequency deviation at time .

[0031] The adaptive weight adjustment strategy enhances control sensitivity; when the system frequency changes drastically, the penalty for frequency deviation is increased and the dispatcher's response priority to frequency is improved.

[0032] According to the above scheme, S4 includes the following contents:

[0033] S401: Obtain the grid equivalent impedance, line resistance-inductance ratio and load concentration of the area to be loaded with the control instruction; the grid equivalent impedance represents the inverse of the grid short-circuit capacity, the line resistance-inductance ratio represents the ratio of line resistance to reactance, and the load concentration is measured using entropy value; based on the grid equivalent impedance Z grid , line resistance-inductance ratio R / X and load concentration C load Construct the feature vector of the control instruction area to be loaded, recorded as Y=[Z grid , R / X, C load ]; extract reference vectors for each scene; the reference vectors for each scene are preset by the system; analyze the similarity between the feature vector of the control instruction area to be loaded and the reference vectors of each scene, and record the reference vector with the highest similarity as the first reference vector; analyze the scene deviation δ based on the feature vector of the control instruction area to be loaded and the first reference vector; the scene deviation is equal to the ratio of the Euclidean norm of the difference between the feature vector of the control instruction area to be loaded and the first reference vector to the maximum value of the element in the first reference vector;

[0034] S402: Modifying the control instruction based on the scene deviation;

[0035] Load correction adjustment △L m *(t n ) = △L m (t n )×min(δ1, δ / δ2), where δ1 represents the amplification instruction coefficient, δ2 represents the scene deviation threshold, and the amplification instruction coefficient and the scene deviation threshold are preset constants;

[0036] Energy storage correction target power △P s *(t n ) = △P s (t n )×(1+δ / (2+α×C load )); wherein α represents the load concentration weight coefficient, which is a system preset constant.

[0037] Scenario deviation correction improves compatibility with complex power grids. The correction strategy appropriately amplifies energy storage power and load regulation when scenario characteristics change significantly, ensuring that control instructions are adaptable to different network conditions.

[0038] Another aspect of the present application provides a virtual power plant system integrating multiple resources, the system being applied to the above-mentioned flexible scheduling method for a virtual power plant integrating multiple resources, the system comprising a data feature analysis module, an optimization model construction module, an instruction update module, and an instruction correction module;

[0039] The data feature analysis module is used to collect photovoltaic power data, energy storage charge state data, and load power data in real time, analyze the photovoltaic fluctuation characteristics, energy storage response capability indicators, and load power data stability indicators corresponding to each type of data, and perform normalization processing;

[0040] The optimization model construction module is used to establish a resource interaction tensor based on the collected data, integrate the grid frequency deviation and line load, and build a collaborative optimization model with time delay compensation;

[0041] The instruction update module is used to update the grid frequency and line load rate in a fixed time window, dynamically solve the collaborative optimization model, and output load adjustment amount and energy storage target scheduling instructions;

[0042] The instruction correction module is used to obtain the grid impedance characteristics and load distribution parameters of the area to be loaded with the control instruction, analyze the scene deviation of the area, and correct the scheduling value based on the scene deviation.

[0043] According to the above solution, the data feature analysis module includes a data acquisition unit and a feature analysis unit;

[0044] The data acquisition unit is used to obtain the photovoltaic power sequence, energy storage charge state data and load power data of the virtual power plant; perform abnormal data point detection and interpolation filling on the photovoltaic power sequence, energy storage charge state data and load power data respectively;

[0045] The feature analysis unit is used to split the photovoltaic power sequence, energy storage state of charge data and load power data based on a preset sliding time window length; analyze the photovoltaic fluctuation characteristics of the split photovoltaic power subsequences; fit the linear charge and discharge model based on the split energy storage state of charge subsequences using the least squares method; analyze the energy storage response capability index based on the average charge and discharge characteristics combined with the attenuation factor; analyze the load instantaneous change rate of the load power data at each moment using the central difference method; obtain the load instantaneous change rate subsequence corresponding to the split load power subsequence, and record the ratio of the standard deviation to the mean of the load instantaneous change rate subsequence as the stability index.

[0046] According to the above solution, the optimization model construction module includes a coupling tensor construction unit and an optimization model construction unit;

[0047] The coupling tensor construction unit is used to fuse spatial information based on photovoltaic power, energy storage charge state and various load power data to construct a coupling tensor;

[0048] The optimization model construction unit constructs a collaborative optimization model based on the coupling tensor, grid frequency deviation and line load rate, using a weighted minimization method of frequency deviation and line overload deviation; the grid frequency deviation is equal to the difference between the real-time grid frequency and the rated grid frequency; the line load rate is equal to the ratio of the actual transmission power of the line to the maximum safe transmission power allowed by the line; the constraints include energy storage charge state range restrictions, load change rate restrictions and power balance hard constraints; the power balance hard constraint is that the sum of the photovoltaic power of each node in the coupling tensor plus the sum of the energy storage charging and discharging power is equal to the total load power.

[0049] According to the above solution, the instruction update module includes a frequency deviation coefficient update unit and an instruction analysis unit;

[0050] The frequency deviation coefficient updating unit is used to update the frequency deviation coefficient based on the power grid frequency deviation before solving the optimization model;

[0051] The instruction analysis unit is used to collect the grid frequency deviation and line load rate at the current moment at a preset time length; solve the optimization model through sensitivity feedback iteration based on the grid frequency deviation and line load rate at the current moment, and generate the control instruction at the current moment.

[0052] According to the above solution, the instruction correction module includes a scene deviation analysis unit and an instruction correction unit;

[0053] The scenario deviation analysis unit is used to obtain the grid equivalent impedance, line resistance-inductance ratio, and load concentration of the control instruction area to be loaded; construct a feature vector of the control instruction area to be loaded based on the grid equivalent impedance, line resistance-inductance ratio, and load concentration; extract a reference vector for each scenario; analyze the similarity between the feature vector of the control instruction area to be loaded and the reference vector of each scenario, and record the reference vector with the highest similarity as the first reference vector; and analyze the scenario deviation based on the feature vector of the control instruction area to be loaded and the first reference vector;

[0054] The instruction correction unit corrects the control instruction based on the scene deviation.

[0055] Compared with the prior art, the beneficial effects of the present invention are as follows: the anomaly detection and interpolation filling of the present application ensure data quality and avoid model instability caused by dirty data; the photovoltaic fluctuation characteristics, energy storage response indicators and load stability indicators are analyzed to provide data support for subsequent analysis; the coupling tensor avoids the spatial information loss of the traditional model, and the tensor is used to build a collaborative optimization model, while incorporating the constraints of grid frequency deviation and line load rate to maintain system stability and security. The model synergistically considers the responsiveness of renewable volatility and controllable resources to achieve joint management of distributed generation and load; considers the time lag effect of photovoltaic power changes on the system, which is conducive to improving scheduling accuracy; the adaptive weight adjustment strategy enhances control sensitivity; when the system frequency changes drastically, the penalty for frequency deviation is increased to increase the scheduling response priority to frequency; the scene deviation correction improves the compatibility of complex power grids, and the correction strategy appropriately amplifies the energy storage power and load adjustment when the scene characteristics change significantly to ensure the adaptability of the control instructions to different network conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0057] Figure 1 This is a flow chart of a flexible scheduling method for a virtual power plant integrating multiple resources according to the present invention;

[0058] Figure 2 This is a structural diagram of a virtual power plant system that integrates multiple resources according to the present invention. DETAILED DESCRIPTION

[0059] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0060] See also Figure 1 The present invention provides a technical solution: a flexible scheduling method for a virtual power plant integrating multiple resources, the method comprising the following steps:

[0061] S1: Real-time collection of PV power data, energy storage state of charge data, and load power data. Analyze the corresponding PV fluctuation characteristics, energy storage response capability indicators, and load power data stability indicators, and perform normalization processing.

[0062] In S1, the following are included:

[0063] S101: Obtain the photovoltaic power sequence, energy storage charge state data and load power data of the virtual power plant; record the photovoltaic power sequence as P={P(t n )|n∈[1,N]}; where P(t n ) represents the time t n The corresponding photovoltaic power, n represents the time sequence number, N represents the total number of acquisition moments in the photovoltaic power sequence; the time t n The corresponding energy storage charge state is recorded as S (t n ), the energy storage charge state is the ratio of the current remaining power to the maximum capacity; the power data of the mth type load at time t n The corresponding load power value is recorded as L m (t n ); Load power data includes industrial load power data, commercial load power data and residential load power data; abnormal data point detection and interpolation filling are performed on photovoltaic power series, energy storage charge state data and load power data respectively;

[0064] S102: Splitting the photovoltaic power sequence, energy storage state of charge data, and load power data based on a preset sliding time window length; analyzing the photovoltaic fluctuation characteristics of the split photovoltaic power subsequences, where the photovoltaic fluctuation characteristics of the photovoltaic power subsequences are equal to the maximum photovoltaic power difference of the photovoltaic power subsequences divided by the photovoltaic power mean of the photovoltaic power subsequences;

[0065] Based on the split energy storage state of charge subsequence, the least squares method is used to fit the linear charge and discharge model; in the charge and discharge model, time is the independent variable and the energy storage state of charge is the dependent variable; the slope coefficient in the linear charge and discharge model is recorded as the average charge and discharge characteristic k; based on the average charge and discharge characteristic k combined with the attenuation factor, the energy storage response capability index R is analyzed, R = k × exp (-t d / t c ) / (S max -S min ); where t d represents the energy storage response delay, t c Indicates the system time coefficient; S max Indicates the upper limit of the system's rated energy storage state of charge, S min Indicates the lower limit of the system's rated energy storage charge state; the energy storage response delay and system time coefficient are preset by the system;

[0066] The central difference method is used to analyze the instantaneous load change rate of various load power data at each moment. The load instantaneous change rate subsequence corresponding to the split load power subsequence is obtained, and the absolute value of the ratio of the standard deviation to the mean of the load instantaneous change rate subsequence is recorded as the stability index.

[0067] The photovoltaic fluctuation characteristics, energy storage response capability index and stability index are normalized.

[0068] S2: Based on the collected data, a resource interaction tensor is established, which integrates the grid frequency deviation and line load to construct a collaborative optimization model with time delay compensation.

[0069] S2 contains the following:

[0070] S201: Based on the fusion of spatial information of PV power, energy storage state of charge, and various load power data, a coupling tensor is constructed, denoted as A{i, j, n}; where i represents the sequence number of the resource type, which includes PV power, energy storage state of charge, and various load power data; j represents the sequence number of the node;

[0071] S202: Based on the coupling tensor, grid frequency deviation and line load rate, a collaborative optimization model is constructed by weighted minimization of frequency deviation and line overload deviation;

[0072] ;

[0073] Among them, △f (t n ) represents the time t n The corresponding grid frequency deviation is equal to the difference between the real-time grid frequency and the rated grid frequency, f nom Indicates the rated grid frequency; ρ line (t n ) represents the time t n Line load factor, ρ safe Indicates the line load rate safety threshold. The line load rate is equal to the ratio of the actual transmission power of the line to the maximum safe transmission power allowed by the line; ω f Represents the frequency deviation coefficient, ω p Indicates the load factor; the sum of the frequency deviation factor and the load factor is 1;

[0074] Example 1: Constraints include energy storage state of charge range restrictions, load change rate restrictions, and power balance hard constraints. The power balance hard constraint is that the sum of the photovoltaic power of each node plus the sum of the energy storage charging and discharging power in the coupling tensor is equal to the total load power. In this example, the specific formula for the power balance hard constraint is:

[0075] ;

[0076] Among them, A{1, j, n} represents the jth node at time t n The photovoltaic power at time t, A{2, j, n} represents the photovoltaic power of the jth node at time t n The energy storage charge state at that time; if i is greater than or equal to 3, then A{i, j, n} represents the power of each type of load; Ploss (t n ) represents the time t n The power loss when charging and discharging; η represents the charge and discharge energy conversion efficiency;

[0077] Example 2; S203: In this example, the state equation is as follows:

[0078] ;

[0079] Where P represents photovoltaic power, S represents the energy storage charge state, f represents the real-time frequency of the power grid, t p represents the photovoltaic response delay, K g represents the frequency power coupling coefficient; D represents the system damping coefficient; θ1 represents the photovoltaic control gain, θ2 represents the energy storage control gain, u comp represents the time delay compensation input;

[0080] Introduce the compensation term into the state equation and change the time t n+1 The predicted PV power after compensation is recorded as P*(t n+1 ):

[0081] ;

[0082] Where △t represents the data collection time interval, ▽ n P(t n ) represents the time t n The corresponding photovoltaic power change rate, β represents the delay attenuation factor, β=1-exp(-t d / t c ).

[0083] S3: Updates the grid frequency and line load factor within a fixed time window, dynamically solves the collaborative optimization model, and outputs load regulation and energy storage target scheduling instructions;

[0084] In S3, include the following:

[0085] S301: Collect the grid frequency deviation and line load rate at the current moment at a preset time interval; solve the optimization model through sensitivity feedback iteration based on the grid frequency deviation and line load rate at the current moment, and generate the control instruction at the current moment. The control instruction includes the output load adjustment amount △L m (t n ) and energy storage target power △P s (t n );

[0086] S302: Update the frequency deviation coefficient before solving the optimization model, ω f (t n ) = ω f0 (1+kf ×|(d△f / dt)(t n )|), where ω f0 is the basic frequency deviation coefficient, k f Indicates the frequency sensitivity coefficient; the basic frequency deviation coefficient and the frequency sensitivity coefficient are preset constants, (d△f / dt)(t n ) indicates that at t n The instantaneous rate of change of the grid frequency deviation at time .

[0087] S4: Obtain the grid impedance characteristics and load distribution parameters of the area to be loaded with the control instruction, analyze the scenario deviation of the area, and correct the scheduling value based on the scenario deviation.

[0088] In S4, include the following:

[0089] S401: Obtain the grid equivalent impedance, line resistance-inductance ratio and load concentration of the area to be loaded with the control instruction; the grid equivalent impedance represents the inverse of the grid short-circuit capacity, the line resistance-inductance ratio represents the ratio of line resistance to reactance, and the load concentration is measured by entropy value; based on the grid equivalent impedance Z grid , line resistance-inductance ratio R / X and load concentration C load Construct the feature vector of the control instruction area to be loaded, recorded as Y=[Z grid , R / X, C load ]; extract the reference vector of each scene; the reference vector of each scene is preset by the system; analyze the similarity between the feature vector of the control instruction area to be loaded and the reference vector of each scene, and record the reference vector with the highest similarity as the first reference vector; analyze the scene deviation δ based on the feature vector of the control instruction area to be loaded and the first reference vector; the scene deviation is equal to the ratio of the Euclidean norm of the difference between the feature vector of the control instruction area to be loaded and the first reference vector to the maximum value of the element in the first reference vector;

[0090] S402: Modifying the control instruction based on the scene deviation;

[0091] Load correction adjustment △L m *(t n ) = △L m (t n )×min(δ1, δ / δ2), where δ1 represents the amplification instruction coefficient, δ2 represents the scene deviation threshold, and the amplification instruction coefficient and the scene deviation threshold are preset constants;

[0092] Energy storage correction target power △P s *(t n ) = △P s (t n )×(1+δ / (2+α×C load)); where α represents the load concentration weight coefficient, which is a system preset constant.

[0093] See also Figure 2 , the present invention provides a technical solution: a virtual power plant system integrating multiple resources, the system including a data feature analysis module, an optimization model construction module, an instruction update module and an instruction correction module;

[0094] The data feature analysis module is used to collect photovoltaic power data, energy storage charge state data, and load power data in real time, analyze the photovoltaic fluctuation characteristics, energy storage response capability indicators, and load power data stability indicators corresponding to each type of data, and perform normalization processing;

[0095] The optimization model construction module is used to establish resource interaction tensors based on collected data, integrate grid frequency deviation and line load, and build a collaborative optimization model with time delay compensation;

[0096] The instruction update module is used to update the grid frequency and line load rate in a fixed time window, dynamically solve the collaborative optimization model, and output load adjustment and energy storage target scheduling instructions;

[0097] The instruction correction module is used to obtain the grid impedance characteristics and load distribution parameters of the area to be loaded with control instructions, analyze the scenario deviation of the area, and correct the scheduling value based on the scenario deviation.

[0098] The data feature analysis module includes a data acquisition unit and a feature analysis unit;

[0099] The data acquisition unit is used to obtain the photovoltaic power sequence, energy storage state of charge data and load power data of the virtual power plant; and perform abnormal data point detection and interpolation filling on the photovoltaic power sequence, energy storage state of charge data and load power data respectively;

[0100] The feature analysis unit is used to split the photovoltaic power sequence, energy storage state of charge data and load power data based on a preset sliding time window length; analyze the photovoltaic fluctuation characteristics of the split photovoltaic power subsequence; fit the linear charge and discharge model based on the split energy storage state of charge subsequence using the least squares method; analyze the energy storage response capability index based on the average charge and discharge characteristics combined with the attenuation factor; analyze the load instantaneous change rate of the load power data at each moment using the central difference method; obtain the load instantaneous change rate subsequence corresponding to the split load power subsequence, and record the ratio of the standard deviation to the mean of the load instantaneous change rate subsequence as the stability index.

[0101] The optimization model building module includes a coupling tensor building unit and an optimization model building unit;

[0102] The coupling tensor construction unit is used to fuse spatial information based on photovoltaic power, energy storage charge state and various load power data to construct a coupling tensor;

[0103] The optimization model construction unit constructs a collaborative optimization model based on the coupling tensor, grid frequency deviation and line load rate, using the weighted minimization method of frequency deviation and line overload deviation; the grid frequency deviation is equal to the difference between the real-time grid frequency and the rated grid frequency; the line load rate is equal to the ratio of the actual transmission power of the line to the maximum safe transmission power allowed by the line; the constraints include the energy storage charge state range limit, the load change rate limit and the power balance hard constraint; the power balance hard constraint is that the sum of the photovoltaic power of each node in the coupling tensor plus the sum of the energy storage charging and discharging power is equal to the total load power.

[0104] The instruction update module includes a frequency deviation coefficient update unit and an instruction analysis unit;

[0105] The frequency deviation coefficient updating unit is used to update the frequency deviation coefficient based on the power grid frequency deviation before solving the optimization model;

[0106] The instruction analysis unit is used to collect the grid frequency deviation and line load rate at the current moment at a preset time interval; solve the optimization model through sensitivity feedback iteration based on the grid frequency deviation and line load rate at the current moment, and generate the control instruction at the current moment.

[0107] The instruction correction module includes a scene deviation analysis unit and an instruction correction unit;

[0108] The scenario deviation analysis unit is used to obtain the grid equivalent impedance, line resistance-inductance ratio, and load concentration of the control instruction area to be loaded; construct a feature vector of the control instruction area to be loaded based on the grid equivalent impedance, line resistance-inductance ratio, and load concentration; extract a reference vector for each scenario; analyze the similarity between the feature vector of the control instruction area to be loaded and the reference vector of each scenario, and record the reference vector with the highest similarity as the first reference vector; and analyze the scenario deviation based on the feature vector of the control instruction area to be loaded and the first reference vector.

[0109] The instruction correction unit corrects the control instruction based on the scene deviation.

[0110] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0111] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. A flexible scheduling method for a virtual power plant integrating multiple resources, characterized in that: The method comprises the following steps: S1: Real-time collection of PV power data, energy storage state of charge data, and load power data. Analyze the corresponding PV fluctuation characteristics, energy storage response capability indicators, and load power data stability indicators, and perform normalization processing. In S1, the following are included: S101: Obtain the photovoltaic power sequence, energy storage charge state data and load power data of the virtual power plant; the photovoltaic power sequence is recorded as P={P(t n )|n∈[1,N]}; where P(t n ) represents the time t n The corresponding photovoltaic power, n represents the time sequence number, N represents the total number of acquisition moments in the photovoltaic power sequence; the time t n The corresponding energy storage charge state is recorded as S (t n ), the energy storage charge state is the ratio of the current remaining power to the maximum capacity; The power data of the mth type load at time t n The corresponding load power value is recorded as L m (t n The load power data includes industrial load power data, commercial load power data and residential load power data; abnormal data point detection and interpolation filling are performed on the photovoltaic power sequence, energy storage charge state data and load power data respectively; S102: Splitting the photovoltaic power sequence, energy storage state of charge data, and load power data based on a preset sliding time window length; analyzing photovoltaic fluctuation characteristics of the split photovoltaic power subsequences, where the photovoltaic fluctuation characteristics of the photovoltaic power subsequences are equal to the maximum photovoltaic power difference of the photovoltaic power subsequences divided by the photovoltaic power mean of the photovoltaic power subsequences; Based on the split energy storage state of charge subsequence, the least squares method is used to fit a linear charge and discharge model; in the charge and discharge model, time is the independent variable and the energy storage state of charge is the dependent variable; the slope coefficient in the linear charge and discharge model is recorded as the average charge and discharge characteristic k; based on the average charge and discharge characteristic k combined with the attenuation factor, the energy storage response capability index R is analyzed, R = k × exp (-t d / t c ) / (S max -S min ); where t d represents the energy storage response delay, t c Indicates the system time coefficient; S max Indicates the upper limit of the system's rated energy storage state of charge, S min Indicates the lower limit of the system's rated energy storage charge state; the energy storage response delay is obtained through the energy storage system step power instruction experiment; the system time coefficient is preset by the system; The central difference method is used to analyze the instantaneous load change rate of various load power data at each moment. The load instantaneous change rate subsequence corresponding to the split load power subsequence is obtained, and the absolute value of the ratio of the standard deviation to the mean of the load instantaneous change rate subsequence is recorded as the stability index. Normalize the photovoltaic fluctuation characteristics, energy storage response capability index and stability index; S2: Based on the collected data, a resource interaction tensor is established, which integrates the grid frequency deviation and line load to construct a collaborative optimization model with time delay compensation. S3: Updates the grid frequency and line load factor within a fixed time window, dynamically solves the collaborative optimization model, and outputs load regulation and energy storage target scheduling instructions; S4: Obtain the grid impedance characteristics and load distribution parameters of the area to be loaded with the control instruction, analyze the scenario deviation of the area, and correct the scheduling value based on the scenario deviation.

2. The method for flexible scheduling of a virtual power plant integrating multiple resources according to claim 1, characterized in that: S2 contains the following: S201: Based on the fusion of spatial information of photovoltaic power, energy storage state of charge, and various load power data, a coupling tensor is constructed, which is recorded as A{i, j, n}; where i represents the sequence number of the resource type, which includes photovoltaic power, energy storage state of charge, and various load power data; j represents the sequence number of the node; S202: Based on the coupling tensor, grid frequency deviation and line load rate, a collaborative optimization model is constructed by weighted minimization of frequency deviation and line overload deviation; The constraints include energy storage state of charge range restrictions, load change rate restrictions, and power balance hard constraints; the power balance hard constraint is that the sum of the photovoltaic power of each node plus the sum of the energy storage charging and discharging power in the coupling tensor is equal to the total load power; S203: Introduce compensation term into the state equation, and change the time t n+1 The predicted PV power after compensation is recorded as P*(t n+1 ).

3. The method for flexible scheduling of a virtual power plant integrating multiple resources according to claim 2, characterized in that: In S3, include the following: S301: every preset time length, collect the current grid frequency deviation and line load rate; according to the current grid frequency deviation and line load rate, solve the optimization model through sensitivity feedback iteration to generate the current control instruction; the control instruction includes the output load adjustment amount ΔL m (t n ) and energy storage target power △P s (t n ); S302: Update the frequency deviation coefficient before solving the optimization model, ω f (t n ) = ω f0 (1+k f ×|(d△f / dt)(t n )|), where ω f0 is the basic frequency deviation coefficient, k f Represents the frequency sensitivity coefficient; the basic frequency deviation coefficient and the frequency sensitivity coefficient are preset constants, (d△f / dt)(t n ) indicates that at t n The instantaneous rate of change of the grid frequency deviation at time .

4. The method for flexible scheduling of a virtual power plant integrating multiple resources according to claim 3, characterized in that: In S4, include the following: S401: Obtain the grid equivalent impedance, line resistance-inductance ratio and load concentration of the area to be loaded with the control instruction; the grid equivalent impedance represents the inverse of the grid short-circuit capacity, the line resistance-inductance ratio represents the ratio of line resistance to reactance, and the load concentration is measured using entropy value; based on the grid equivalent impedance Z grid , line resistance-inductance ratio R / X and load concentration C load Construct the feature vector of the control instruction area to be loaded, recorded as Y=[Z grid , R / X, C load ]; extract reference vectors for each scene; the reference vectors for each scene are preset by the system; analyze the similarity between the feature vector of the control instruction area to be loaded and the reference vectors of each scene, and record the reference vector with the highest similarity as the first reference vector; analyze the scene deviation δ based on the feature vector of the control instruction area to be loaded and the first reference vector; the scene deviation is equal to the ratio of the Euclidean norm of the difference between the feature vector of the control instruction area to be loaded and the first reference vector to the maximum value of the element in the first reference vector; S402: Modifying the control instruction based on the scene deviation; Load correction adjustment △L m * (t n ) = △L m (t n )×min(δ1, δ / δ2), where δ1 represents the amplification instruction coefficient, δ2 represents the scene deviation threshold, and the amplification instruction coefficient and the scene deviation threshold are preset constants; Energy storage correction target power △P s * (t n ) = △P s (t n )×(1+δ / (2+α×C load )); wherein α represents the load concentration weight coefficient, which is a system preset constant.

5. A virtual power plant system integrating multiple resources, wherein the system is applied to implement a flexible scheduling method for a virtual power plant integrating multiple resources as claimed in any one of claims 1 to 4, characterized in that: The system includes a data feature analysis module, an optimization model construction module, an instruction update module and an instruction correction module; The data feature analysis module is used to collect photovoltaic power data, energy storage charge state data, and load power data in real time, analyze the photovoltaic fluctuation characteristics, energy storage response capability indicators, and load power data stability indicators corresponding to each type of data, and perform normalization processing; The optimization model construction module is used to establish a resource interaction tensor based on the collected data, integrate the grid frequency deviation and line load, and build a collaborative optimization model with time delay compensation; The instruction update module is used to update the grid frequency and line load rate in a fixed time window, dynamically solve the collaborative optimization model, and output load adjustment amount and energy storage target scheduling instructions; The instruction correction module is used to obtain the grid impedance characteristics and load distribution parameters of the area to be loaded with the control instruction, analyze the scene deviation of the area, and correct the scheduling value based on the scene deviation.

6. The virtual power plant system integrating multiple resources according to claim 5, characterized in that: The data feature analysis module includes a data acquisition unit and a feature analysis unit; The data acquisition unit is used to obtain the photovoltaic power sequence, energy storage charge state data and load power data of the virtual power plant; perform abnormal data point detection and interpolation filling on the photovoltaic power sequence, energy storage charge state data and load power data respectively; The feature analysis unit is used to split the photovoltaic power sequence, energy storage state of charge data and load power data based on a preset sliding time window length; analyze the photovoltaic fluctuation characteristics of the split photovoltaic power subsequences; fit the linear charge and discharge model based on the split energy storage state of charge subsequences using the least squares method; analyze the energy storage response capability index based on the average charge and discharge characteristics combined with the attenuation factor; analyze the load instantaneous change rate of the load power data at each moment using the central difference method; obtain the load instantaneous change rate subsequence corresponding to the split load power subsequence, and record the ratio of the standard deviation to the mean of the load instantaneous change rate subsequence as the stability index.

7. The virtual power plant system integrating multiple resources according to claim 5, characterized in that: The optimization model construction module includes a coupling tensor construction unit and an optimization model construction unit; The coupling tensor construction unit is used to fuse spatial information based on photovoltaic power, energy storage charge state and various load power data to construct a coupling tensor; The optimization model construction unit constructs a collaborative optimization model based on the coupling tensor, grid frequency deviation and line load rate, using a weighted minimization method of frequency deviation and line overload deviation; the grid frequency deviation is equal to the difference between the real-time grid frequency and the rated grid frequency; the line load rate is equal to the ratio of the actual transmission power of the line to the maximum safe transmission power allowed by the line; the constraints include energy storage charge state range restrictions, load change rate restrictions and power balance hard constraints; the power balance hard constraint is that the sum of the photovoltaic power of each node in the coupling tensor plus the sum of the energy storage charging and discharging power is equal to the total load power.

8. The virtual power plant system integrating multiple resources according to claim 5, characterized in that: The instruction update module includes a frequency deviation coefficient update unit and an instruction analysis unit; The frequency deviation coefficient updating unit is used to update the frequency deviation coefficient based on the power grid frequency deviation before solving the optimization model; The instruction analysis unit is used to collect the grid frequency deviation and line load rate at the current moment at a preset time length; solve the optimization model through sensitivity feedback iteration based on the grid frequency deviation and line load rate at the current moment, and generate the control instruction at the current moment.

9. The virtual power plant system integrating multiple resources according to claim 5, characterized in that: The instruction correction module includes a scene deviation analysis unit and an instruction correction unit; The scenario deviation analysis unit is used to obtain the grid equivalent impedance, line resistance-inductance ratio, and load concentration of the control instruction area to be loaded; construct a feature vector of the control instruction area to be loaded based on the grid equivalent impedance, line resistance-inductance ratio, and load concentration; extract a reference vector for each scenario; analyze the similarity between the feature vector of the control instruction area to be loaded and the reference vector of each scenario, and record the reference vector with the highest similarity as the first reference vector; and analyze the scenario deviation based on the feature vector of the control instruction area to be loaded and the first reference vector; The instruction correction unit corrects the control instruction based on the scene deviation.

Images