A virtual power plant power regulation method and system based on response deviation closed-loop feedback

Abstract

The present application relates to the technical field of power regulation, and particularly relates to a virtual power plant power regulation method and system based on response deviation closed-loop feedback, which comprises the following steps: obtaining real-time active power response values and instruction power of each component, and calculating response deviation of adjacent sampling points; constructing a response deviation inertia mapping index; calculating a deviation accumulation dynamic correlation index based on the response deviation inertia mapping index; constructing a sensitivity adaptive feedback gain index by using normalized values of the response deviation inertia mapping index and the deviation accumulation dynamic correlation index; mapping the sensitivity adaptive feedback gain index to a basic penalty parameter weight in a distributed optimization objective function, obtaining power regulation instructions for each component by solving the improved objective function, and realizing closed-loop correction of power deviation. The present application ensures the accuracy of virtual power plant output and the stability of system frequency.

Description

Technical Field

[0001] This invention relates to the field of power regulation technology, and in particular to a virtual power plant power regulation method and system based on response deviation closed-loop feedback. Background Technology

[0002] With the green and low-carbon transformation of the energy structure, distributed power sources, energy storage systems, and controlled loads are being connected to the grid on a large scale. Virtual power plants, as an effective means of collaboratively optimizing various types of distributed components, have become a key support for improving the flexibility and responsiveness of the power system. In actual dispatching, due to the characteristics of distributed components such as high randomness, wide spatial distribution, and uncertain communication delays, virtual power plants often exhibit deviations between predicted output and actual response when executing power commands issued by the upper-level grid. Therefore, how to achieve closed-loop correction of power deviations through efficient power regulation methods in the dynamic game among massive heterogeneous components, ensuring the accuracy of virtual power plant output and the stability of system frequency, is a core requirement that urgently needs to be addressed in the construction of new power systems.

[0003] In existing research on power allocation and regulation in virtual power plants, the Alternating Direction Method of Multipliers (ADMM) effectively addresses the privacy protection and collaborative control of components by decomposing the global optimization problem into multiple sub-problems for parallel solution. While the ADMM algorithm exhibits good convergence and scalability, in real-time power regulation scenarios within virtual power plants, its step size parameter is typically set to a fixed value, lacking the ability to adapt to dynamic changes in grid frequency fluctuations and component response deviations. This results in slow convergence speeds under conditions of drastic fluctuations in component output, making it difficult to meet the real-time power compensation requirements at the second or even millisecond level, and easily leading to secondary deviation fluctuations in the power system due to response lag. Summary of the Invention

[0004] To ensure the accuracy of virtual power plant output and the stability of system frequency, this invention provides a virtual power plant power regulation method and system based on response deviation closed-loop feedback.

[0005] In a first aspect, the present invention provides a virtual power plant power regulation method based on response deviation closed-loop feedback, employing the following technical solution: A virtual power plant power regulation method based on closed-loop feedback of response deviation includes the following steps: acquiring the real-time active power response value and command power of each component, and calculating the response deviation of adjacent sampling points; constructing a response deviation inertial mapping index based on the response deviation sequence within a preset sliding window and the deviation change of adjacent sampling points, wherein the response deviation inertial mapping index is positively correlated with both the response deviation and the deviation change; calculating a deviation accumulation dynamic correlation index based on the response deviation inertial mapping index, wherein the deviation accumulation dynamic correlation index is positively correlated with the absolute value of the accumulated deviation; constructing a sensitivity adaptive feedback gain index by using the normalized values ​​of the response deviation inertial mapping index and the deviation accumulation dynamic correlation index, combined with the response deviation and the response deviation change at the current moment; mapping the sensitivity adaptive feedback gain index to the weight of the basic penalty parameter in the distributed optimization objective function, and obtaining the power regulation command issued to each component by solving the improved objective function, thereby realizing closed-loop correction of the power deviation.

[0006] To address the issues of fixed parameters and difficulty in adapting to component physical response lags and communication delays in existing virtual power plant power regulation algorithms, this invention constructs response deviation inertia mapping indices and deviation accumulation dynamic correlation indices. These indices can capture the instantaneous inertial trend and historical accumulation effect of deviations in real time. By transforming these indices into sensitivity adaptive feedback gains and mapping them to the penalty parameters of the distributed optimization objective function, a shift from static optimization to dynamic closed-loop feedback control is achieved. This effectively solves the power regulation overshoot or oscillation problems caused by response lags, improving the accuracy of the virtual power plant's response to higher-level commands and the stability of the system frequency.

[0007] Preferably, the method for calculating the response deviation of adjacent sampling points is as follows: the collected data is denoised using a median filtering algorithm, and the historical sequence is truncated and normalized using a sliding window technique before the response deviation of adjacent sampling points is calculated.

[0008] By introducing median filtering and sliding window normalization mechanisms, abnormal spike data caused by communication disturbances can be effectively eliminated before calculating the deviation, and the dimensional differences between different capacity components can be eliminated, ensuring the purity and comparability of subsequent deviation calculations, thereby improving the robustness of the entire feedback control system.

[0009] Preferably, the calculation steps for the response deviation inertia mapping index are as follows: multiply the response deviation of each sampling point within the sliding window by the change in deviation of adjacent sampling points, and divide by the time length of the sliding window to obtain the inertia component of each sampling point; calculate the mean of the sum of squares of the inertia components of all sampling points; and take the square root of the mean to obtain the response deviation inertia mapping index.

[0010] By quantifying the response deviation inertia mapping index, the product of the deviation value and its change is used to characterize the diffusion trend of the deviation. This index can more sensitively identify the nonlinear deviation surge caused by physical inertia, providing the algorithm with predictive ability, enabling it to perceive the ease or difficulty of adjustment in the early stage of deviation expansion.

[0011] Preferably, the calculation steps for the deviation accumulation dynamic correlation index are as follows: calculate the cumulative sum of the product of the response deviation and the sampling time of each sampling point within the sliding window, divide the cumulative sum by the product of the cumulative attenuation coefficient and the total rated power of the virtual power plant, and use the exponent of the natural constant. Multiply the exponent by the response deviation inertia mapping index to obtain the deviation accumulation dynamic correlation index.

[0012] This invention constructs a dynamic correlation index for deviation accumulation, which uses integral effects and exponential functions to measure long-term small deviations. It can detect static residuals that the system has failed to eliminate over a long period of time, and generate stronger adjustment signals through the accumulation effect, thereby effectively eliminating the adjustment dead zone and preventing the system from falling into steady-state error.

[0013] Preferably, the calculation steps for the sensitivity adaptive feedback gain index are as follows: add the normalized value of the response deviation inertial mapping index to the normalized value of the deviation accumulation dynamic correlation index to obtain the first intermediate value; divide the sum of the response deviation and the change in response deviation at the current moment by the sum of the response deviation and the preset parameter, and then add it to the preset constant to obtain the second intermediate value; multiply the first intermediate value and the second intermediate value to obtain the sensitivity adaptive feedback gain index.

[0014] This invention generates a sensitivity adaptive feedback gain by combining the normalized inertial index with the accumulated index and introducing a correction term based on the deviation and its rate of change at the current moment. This gain can rapidly increase when the deviation and the deviation derivative are in the same direction, enabling the control system to automatically apply a greater adjustment force when the risk of system instability increases, thus realizing intelligent dynamic allocation of adjustment intensity.

[0015] Preferably, the expression for the distributed optimization objective function of the virtual power plant is:

[0016] in: This represents the distributed optimization objective function for a virtual power plant. This represents the total number of units in the virtual power plant. For the first The operating cost function of each component; For the first The adjustment power value to be allocated to each component; The baseline parameter of the operating cost function has a value of 1 and the same unit as the operating cost. Let t be the command power issued by the upper-level power grid to the j-th component; These are the basic penalty parameters for the ADMM algorithm; The sensitivity adaptive feedback gain index at time t; For time t, based on the first A predictive compensation operator is constructed from the derivative and integral term of the response deviation of individual components; Reference parameters indicating power.

[0017] This invention improves the traditional distributed optimization objective function of virtual power plants by dynamically mapping the adaptive feedback gain to the weights of the basic penalty parameters and introducing a predictive compensation operator. This allows the ADMM algorithm to be no longer limited to a fixed step size during the iterative solution process, but to automatically adjust the convergence speed and direction according to the real-time response deviation state. This ensures that the algorithm can converge quickly and offset power deviations in scenarios with multiple heterogeneous components.

[0018] Preferably, the expression for the prediction compensation operator is:

[0019] in, For time t, based on the first A prediction compensation operator is constructed from the derivatives and integral terms of the response deviations of individual components. These are the preset first and second proportional coefficients, respectively. For time t, the first Response deviation of individual components For time t, the first The change in response deviation of individual components.

[0020] This invention clarifies the composition of the prediction compensation operator in the objective function. By weighting the deviation and its derivative terms, the system can compensate in advance in the optimization objective before the actual response deviation causes a serious lag, effectively offsetting the response delay caused by the physical characteristics of the components and improving the dynamic performance of real-time power tracking.

[0021] Preferably, the method for obtaining the operating cost function is as follows: the historical power of the component and the operating cost are fitted as a quadratic function using the least squares method; wherein, the operating cost is calculated based on maintenance losses and equivalent depreciation, and the equivalent depreciation is the ratio of the component value to its expected service life.

[0022] By fitting historical data using the least squares method to obtain the operating cost function and comprehensively considering maintenance losses and equivalent depreciation, the power allocation of the virtual power plant is not only based on the theoretical model, but also on the actual full life cycle cost of the components. This ensures that the power regulation strategy maximizes economic benefits while meeting technical indicators and extending the service life of the equipment.

[0023] Preferably, the closed-loop correction method is as follows: when the system detects that the response lag has led to an increase in the response deviation, the power allocation share of the energy storage components is automatically optimized in the next iteration cycle to offset the deviation by increasing the gain of the penalty term in the objective function.

[0024] When a deviation is detected in a slow-response unit, the fast-response unit is automatically forced to take on a larger share of regulation by increasing the penalty term. This mechanism utilizes the complementary nature of heterogeneous resources to automatically fill the power gap during the dynamic regulation process, ensuring the smoothness and accuracy of the overall power output of the virtual power plant.

[0025] Secondly, the present invention provides a virtual power plant power regulation system based on response deviation closed-loop feedback, which adopts the following technical solution: A virtual power plant power regulation system based on response deviation closed-loop feedback includes a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement the virtual power plant power regulation method based on response deviation closed-loop feedback as described above.

[0026] The aforementioned virtual power plant power regulation method based on closed-loop feedback of response deviation is generated into a computer program and stored in memory for loading and execution by the processor. Thus, a system is built based on the memory and processor for convenient use.

[0027] The present invention has the following technical effects: To address the problem that traditional ADMM algorithms have fixed parameters and are unable to cope with the lag in the physical response of components, this invention constructs a response deviation inertia mapping and a dynamic correlation index of deviation accumulation to generate a sensitivity adaptive feedback gain in real time. This gain dynamically corrects the penalty weights in the distributed optimization objective function, transforming static optimization into dynamic control with closed-loop feedback characteristics. This achieves rapid convergence and precise cancellation of power deviation, ensuring the accuracy of the virtual power plant output and the stability of the system frequency. Attached Figure Description

[0028] Figure 1 This is a flowchart of a virtual power plant power regulation method based on response deviation closed-loop feedback according to the present invention.

[0029] Figure 2 This is a power tracking comparison chart of the present invention.

[0030] Figure 3 This is a comparison chart of the convergence speed of the present invention.

[0031] Figure 4 This is the indicator correlation evolution diagram of the present invention. Detailed Implementation

[0032] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0033] This invention discloses a virtual power plant power regulation method based on response deviation closed-loop feedback, referring to... Figure 1 This includes the following steps: S1: Multidimensional data acquisition and preprocessing.

[0034] Multi-dimensional data acquisition is performed by edge computing nodes deployed in the virtual power plant control center. The real-time active power response values ​​of each component are obtained in real time by utilizing inverters, battery management systems (BMS), and intelligent power meters and high-precision phasor measurement units (PMUs) installed on the distributed power source. and the commanded power issued by the superior power grid The data acquisition method employs Modbus-TCP or IEC61850 communication protocols, and synchronous data upload is achieved via industrial Ethernet to ensure data timeliness.

[0035] Median filtering is used to remove abnormal spikes caused by communication disturbances; then, based on the current control cycle, a sliding window technique is used to truncate data of length [missing information]. The historical sequence was analyzed, and each sequence was linearly normalized to eliminate the dimensional influence between different capacity components. Finally, the response deviation at sampling point t was calculated. 。

[0036] S2: Construct a response bias inertia mapping index.

[0037] During the regulation of a virtual power plant, the physical response lag of distributed components causes the response deviation to exhibit a nonlinear jump in the initial stage of command change, and a deviation tail pattern to emerge during regulation, meaning the disappearance of the deviation lags behind the issuance of the command. This unique inertial characteristic prevents the ADMM algorithm from accurately capturing the instantaneous evolution trend of the deviation when updating the Lagrange multipliers, resulting in overshoot or oscillation in the regulation command. Therefore, this embodiment constructs a response deviation inertia mapping index to quantify the degree of deviation between the deviation change rate and the target deviation, reflecting the ease with which the system overcomes response inertia.

[0038] With the sampling point at time t as the endpoint, a construction length of... If a sliding window is used, then the response bias sequence corresponding to the sliding window is: The expression for the response deviation inertia mapping index is:

[0039] in, The response deviation inertial mapping index at time t; The first digit within the sliding window corresponding to time t Response deviation at each sampling point; Let be the change in deviation between adjacent sampling points within the sliding window at time t, i.e. ; The time length corresponding to the sliding window, with a preset value range of 0.02s to 0.1s; This represents the number of sampling points within the sliding window, with a value of 10.

[0040] The greater the physical inertia of the components within the virtual power plant and the slower the response, the greater the deviation. and its rate of change It will be at a high level at the same time, leading to The numerical value increases significantly. Compared to directly using absolute deviation modeling, this indicator, by introducing the product of the rate of change, can predict the diffusion trend of the deviation and provide adjustment gain support for the algorithm.

[0041] S3: Construct dynamic correlation indicators for deviation accumulation.

[0042] Since virtual power plant regulation is a continuous-time process, response deviations not only possess instantaneous inertia but also have a cumulative effect. According to the principles of energy conservation and feedback control, persistent small deviations can be transformed into significant electrical energy differences through integral effects, resulting in residual deviation accumulation. This characteristic manifests as a continuous increase in the time-domain offset of the data sequence. If the ADMM algorithm fails to detect this cumulative trend in a timely manner, it can lead to a static deviation at the global convergence point. Therefore, by introducing an integral gain factor, a dynamic correlation index for deviation accumulation is constructed. This is used to measure the urgency of eliminating static residuals in a system.

[0043] The formula for calculating the dynamic correlation index of deviation accumulation is:

[0044] in, Accumulate dynamic correlation indicators for the deviation at time t; The response deviation inertial mapping index at time t; The first digit within the sliding window corresponding to time t Response deviation at each sampling point; This represents the time length corresponding to the sliding window. This represents the total rated power of the virtual power plant. To accumulate the attenuation coefficient, in this embodiment, its value is 1.5, in seconds, to eliminate dimensional differences; exp represents an exponential function with base e.

[0045] When a static deviation persists in the system and cannot be eliminated, the absolute value of the cumulative deviation in the exponential term increases, leading to... It increases exponentially. The dynamic correlation index of deviation accumulation not only inherits the response deviation inertia mapping index, but also... The sensitivity to instantaneous changes further strengthens the penalty for historical biases. Compared to solely considering real-time values, this indicator can effectively guide the algorithm to eliminate static adjustment dead zones and improve adjustment accuracy.

[0046] S4: Construct a sensitivity-adaptive feedback gain index.

[0047] Virtual power plant power regulation also exhibits an asymmetric response sensitivity characteristic, meaning that the required regulation force differs between the deviation reduction phase and the deviation expansion phase. According to the power system frequency stability criterion, when the deviation and its derivative are in the same direction, the risk of system instability surges dramatically. This pattern manifests as follows: and Nonlinear coupling in the time domain distribution. To transform this complex characteristic into control parameters usable by the algorithm, this embodiment constructs a sensitivity adaptive feedback gain index. .

[0048] Using the linear function normalization algorithm to and After normalization, the expression for the sensitivity adaptive feedback gain index is:

[0049] in, The sensitivity adaptive feedback gain index at time t; The normalized value of the dynamic correlation index is accumulated for the deviation at time t; The normalized value of the response deviation inertial mapping index at time t; Let be the response deviation at time t; Let be the change in response deviation at time t. ; This is a preset parameter, set to 1 to prevent the denominator from being zero.

[0050] When the deviation is widening and both inertia and accumulation are significant, the correction term... This will increase, causing the calculation results to increase rapidly. This indicator integrates the current status, trend, and history of the deviation, and can accurately reflect the intensity of the current power regulation's demand for feedback gain.

[0051] S5: Improvement of ADMM algorithm and implementation of closed-loop regulation.

[0052] In the virtual power plant power regulation architecture based on the ADMM algorithm, this step will use the sensitivity adaptive feedback gain index obtained in step S4. This is directly mapped to the weights of the basic penalty parameters in the objective function. Within each scheduling cycle, the improved distributed optimization objective function for the virtual power plant is constructed as follows:

[0053] in: This represents the distributed optimization objective function for a virtual power plant. This represents the total number of units in the virtual power plant. For the first The operating cost function of each component; For the first The adjustment power value to be allocated to each component; Let t be the command power issued by the upper-level power grid to the j-th component; The initial value of this parameter is a constant between 0.1 and 1.0, based on convergence theory. In this embodiment, the value is 0.5. The sensitivity adaptive feedback gain index at time t; For time t, based on the first A prediction compensation operator is constructed from the derivative and integral term of the response deviation of individual components, and then... Calculated, where These are the preset first and second scaling factors, with values ​​of 0.05 and 0.01 respectively. For time t, the first Response deviation of individual components For time t, the first The change in response deviation of individual components; The baseline parameter of the operating cost function has a value of 1 and the same unit as the operating cost. A reference parameter representing power, with a value of 1 and the same unit as power.

[0054] Among them, the first The method for obtaining the operating cost function of an individual component is as follows: [The text abruptly ends here, likely due to an incomplete sentence or a formatting error.] The historical power and operating cost of each component were fitted using the least squares method to obtain the fourth function. The operating cost function of each component. Operating costs can be represented by maintenance depreciation and equivalent depreciation. Equivalent depreciation is calculated as the ratio of the component's value to its expected useful life.

[0055] The above formula is passed through The penalty term gain of the ADMM algorithm was dynamically adjusted, enabling the algorithm to automatically increase the penalty on power when the deviation evolves drastically, thus forcing the power of each component (distributed power source, energy storage, etc.) to be reduced. Fast command power To move closer. At the same time, the formula introduces a compensation operator based on deviation derivative and integral information. This allows the algorithm to predict and offset potential response residuals during the optimization process of solving subproblems, transforming the original static optimization process into a dynamic control process with closed-loop feedback characteristics. This solves the problem of mismatch between convergence step size and physical response speed caused by fixed parameters in traditional ADMM, ensuring high convergence and real-time performance of the algorithm in complex fluctuating environments.

[0056] The improved algorithm is deployed in the virtual power plant coordination controller, which performs real-time calculations. Furthermore, the objective function is modified, and the parallel computing characteristics of the improved ADMM algorithm are utilized to dynamically issue power adjustment commands, that is, to optimize the calculation of the objective function when it is minimized. The value of . When the system detects a response lag caused by the inertia of components, it automatically optimizes the share of fast-response components such as energy storage in the next iteration cycle by increasing the compensation term loss in the objective function, thereby offsetting the power deviation and ultimately achieving precise closed-loop regulation of the virtual power plant's power.

[0057] The technical effects of this invention can also be illustrated in conjunction with the accompanying drawings, such as... Figure 2 As shown, when the instruction undergoes a step change, such as a sudden increase in adjustment requirements, the curves of existing technologies will exhibit significant delay and overshoot; however, under the action of the prediction compensation operator, the present invention can more quickly fit the instruction curve and significantly reduce the fluctuation amplitude.

[0058] like Figure 3 As shown, existing technologies have a slow convergence speed and a gradual decrease in residuals, making it difficult to achieve a low convergence level within 50 iterations. The improved algorithm of this invention exhibits extremely high convergence efficiency, with the residuals decreasing rapidly within a few iterations and stabilizing below the convergence threshold, reflecting that the algorithm can meet the real-time compensation requirements of virtual power plants at the second or even millisecond level.

[0059] like Figure 4 As shown, whenever the deviation curve exhibits a significant jump or peak, the gain index instantly generates an extremely high pulse, indicating that the system can automatically increase the adjustment intensity based on the magnitude and rate of change of the deviation. It maintains stability when the deviation is small and forces each unit to respond quickly when the deviation increases, thus achieving the effect of intelligent dynamic allocation of adjustment intensity.

[0060] This invention also discloses a virtual power plant power regulation system based on response deviation closed-loop feedback, including a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement a virtual power plant power regulation method based on response deviation closed-loop feedback according to the present invention.

[0061] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0062] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A virtual power plant power regulation method based on response deviation closed-loop feedback, characterized in that, The steps include: acquiring the real-time active power response value and command power of each component, and calculating the response deviation between adjacent sampling points; constructing a response deviation inertial mapping index based on the response deviation sequence within a preset sliding window and the deviation change between adjacent sampling points, wherein the response deviation inertial mapping index is positively correlated with both the response deviation and the deviation change; and calculating a deviation accumulation dynamic correlation index based on the response deviation inertial mapping index, wherein the deviation accumulation dynamic correlation index is positively correlated with the absolute value of the accumulated deviation. By utilizing the normalized values ​​of the response deviation inertia mapping index and the deviation accumulation dynamic correlation index, and combining the response deviation and the change in response deviation at the current moment, a sensitivity adaptive feedback gain index is constructed. The sensitivity adaptive feedback gain index is mapped to the weight of the basic penalty parameter in the distributed optimization objective function. By solving the improved objective function, the power adjustment command for each component is obtained, and the closed-loop correction of the power deviation is achieved.

2. The virtual power plant power regulation method based on response deviation closed-loop feedback according to claim 1, characterized in that, The method for calculating the response deviation between adjacent sampling points is as follows: the collected data is denoised using a median filtering algorithm, and the historical sequence is truncated and normalized using a sliding window technique before the response deviation between adjacent sampling points is calculated.

3. The virtual power plant power regulation method based on response deviation closed-loop feedback according to claim 1, characterized in that, The calculation steps for the response deviation inertia mapping index are as follows: multiply the response deviation of each sampling point within the sliding window by the change in deviation of the adjacent sampling points, and then divide by the time length of the sliding window to obtain the inertia component of each sampling point. Calculate the mean of the sum of squares of the inertia components of all sampling points, and take the square root of the mean to obtain the response deviation inertia mapping index.

4. The virtual power plant power regulation method based on response deviation closed-loop feedback according to claim 1, characterized in that, The calculation steps for the deviation accumulation dynamic correlation index are as follows: calculate the cumulative sum of the product of the response deviation and the sampling time of each sampling point within the sliding window, divide the cumulative sum by the product of the cumulative attenuation coefficient and the total rated power of the virtual power plant, and use the exponent of the natural constant. Multiply the exponent by the response deviation inertia mapping index to obtain the deviation accumulation dynamic correlation index.

5. The virtual power plant power regulation method based on response deviation closed-loop feedback according to claim 1, characterized in that, The calculation steps for the sensitivity adaptive feedback gain index are as follows: add the normalized value of the response deviation inertial mapping index to the normalized value of the deviation accumulation dynamic correlation index to obtain the first intermediate value; divide the sum of the response deviation and the change in response deviation at the current moment by the sum of the response deviation and the preset parameter, and then add it to the preset constant to obtain the second intermediate value; multiply the first intermediate value and the second intermediate value to obtain the sensitivity adaptive feedback gain index.

6. The virtual power plant power regulation method based on response deviation closed-loop feedback according to claim 1, characterized in that, The expression for the distributed optimization objective function of the virtual power plant is: in: This represents the distributed optimization objective function for a virtual power plant. This represents the total number of units in the virtual power plant. For the first The operating cost function of each component; For the first The adjustment power value to be allocated to each component; The baseline parameter of the operating cost function has a value of 1 and the same unit as the operating cost. Let t be the command power issued by the upper-level power grid to the j-th component; These are the basic penalty parameters for the ADMM algorithm; The sensitivity adaptive feedback gain index at time t; For time t, based on the first A predictive compensation operator is constructed from the derivative and integral term of the response deviation of individual components; Reference parameters indicating power.

7. The virtual power plant power regulation method based on response deviation closed-loop feedback according to claim 6, characterized in that, The expression for the prediction compensation operator is: in, For time t, based on the first A prediction compensation operator is constructed from the derivatives and integral terms of the response deviations of individual components. These are the preset first and second proportional coefficients, respectively. For time t, the first Response deviation of individual components For time t, the first The change in response deviation of individual components.

8. The virtual power plant power regulation method based on response deviation closed-loop feedback according to claim 6, characterized in that, The method for obtaining the operating cost function is as follows: the historical power of the component and the operating cost are fitted as a quadratic function using the least squares method; whereby the operating cost is calculated based on maintenance losses and equivalent depreciation, and the equivalent depreciation is the ratio of the component's value to its expected service life.

9. The virtual power plant power regulation method based on response deviation closed-loop feedback according to claim 1, characterized in that, The closed-loop correction method is as follows: when the system detects that the response lag has led to an increase in the response deviation, the power allocation share of the energy storage components is automatically optimized in the next iteration cycle to offset the deviation by increasing the gain of the penalty term in the objective function.

10. A virtual power plant power regulation system based on response deviation closed-loop feedback, characterized in that, include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement a virtual power plant power regulation method based on response deviation closed-loop feedback according to any one of claims 1-9.

Images